WO2022217597A1 - 一种用于电机驱动器的故障预警方法和装置 - Google Patents

一种用于电机驱动器的故障预警方法和装置 Download PDF

Info

Publication number
WO2022217597A1
WO2022217597A1 PCT/CN2021/087861 CN2021087861W WO2022217597A1 WO 2022217597 A1 WO2022217597 A1 WO 2022217597A1 CN 2021087861 W CN2021087861 W CN 2021087861W WO 2022217597 A1 WO2022217597 A1 WO 2022217597A1
Authority
WO
WIPO (PCT)
Prior art keywords
motor driver
data
vehicle
model
motor
Prior art date
Application number
PCT/CN2021/087861
Other languages
English (en)
French (fr)
Inventor
杜若阳
于卫
汪浩
唐诗
龙纲
Original Assignee
华为数字能源技术有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华为数字能源技术有限公司 filed Critical 华为数字能源技术有限公司
Priority to CN202180002804.3A priority Critical patent/CN115398439A/zh
Priority to PCT/CN2021/087861 priority patent/WO2022217597A1/zh
Publication of WO2022217597A1 publication Critical patent/WO2022217597A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
    • G01D21/00—Measuring or testing not otherwise provided for
    • G01D21/02—Measuring two or more variables by means not covered by a single other subclass
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02P—CONTROL OR REGULATION OF ELECTRIC MOTORS, ELECTRIC GENERATORS OR DYNAMO-ELECTRIC CONVERTERS; CONTROLLING TRANSFORMERS, REACTORS OR CHOKE COILS
    • H02P23/00—Arrangements or methods for the control of AC motors characterised by a control method other than vector control
    • H02P23/14—Estimation or adaptation of motor parameters, e.g. rotor time constant, flux, speed, current or voltage

Definitions

  • the embodiments of the present application relate to the field of power electronic power conversion, and in particular, to a fault early warning method and device for a motor driver.
  • the motor driver is used for drive, idle and brake control of the motor.
  • the motor driver may include a power semiconductor module, which is the core component of the electric vehicle powertrain and is also a component that is prone to failure.
  • the life of the motor driver is evaluated by monitoring the temperature, temperature amplitude change, cycle number and other parameters of the motor driver, and the life model can be used to evaluate the motor driver.
  • the life model can include: Coffin-Manson model , Norris-Landzberg model and Bayerer model.
  • the input parameters of the life model are obtained by performing life tests on a small number of motor drives.
  • the life model can only represent the overall situation in a statistical sense, and has no practical value for each individual motor drive. Therefore, the above-mentioned lifetime model is less general and cannot be used for fault diagnosis of a single motor drive.
  • Embodiments of the present application provide a fault early warning method and device for a motor driver, which are used to realize fault early warning of the motor driver.
  • an embodiment of the present application provides a fault early warning method for a motor driver, including:
  • a fault warning is performed on the motor driver according to the first abnormality degree parameter.
  • a fault warning can be given to the motor driver according to the running state data when the motor driver is running.
  • the running state data of the motor driver is collected when the motor driver is running.
  • the running state data can reflect the real running state of the motor driver, and the fault warning based on the running state data can realize the advance prediction of the failure of the motor driver and realize the fault early warning of the motor driver.
  • the method further includes:
  • a fault warning is performed on the motor driver according to the second abnormality degree parameter.
  • the static data of the motor driver is obtained when the vehicle is in the stop stage, and the motor driver can be warned of faults according to the static data.
  • the static data of the motor driver is the data collected from the motor driver when the vehicle is stopped.
  • the static data can reflect the real static state of the motor driver. Based on the static data, the fault early warning can realize the advance prediction of the failure of the motor driver and realize the fault early warning of the motor driver.
  • the vehicle is in a stop phase, including at least one of the following:
  • the vehicle is in the charging phase, the vehicle is in the ignition phase, and the vehicle is in the ignition phase.
  • static data is sampled during the vehicle charging stage, the vehicle ignition stage, and the vehicle flameout stage, so that the fault diagnosis device can provide an emergency response function in the event of a network disconnection according to the static data.
  • the performing a fault warning to the motor driver according to the first abnormality degree parameter includes:
  • abnormal conditions can be preset, and the abnormal conditions can be dynamically updated through the AI diagnosis model, so as to ensure that the abnormal conditions can be used for fault judgment and the accuracy of fault judgment.
  • the abnormal condition After the abnormal condition is dynamically updated, it can be determined whether the first abnormality degree satisfies the dynamically updated abnormality condition.
  • the first abnormality degree parameter satisfies the updated abnormality condition, it indicates that the motor driver may be faulty.
  • the drive gives an early warning of failure.
  • the accuracy of the fault warning can be further improved.
  • the method further includes:
  • the inputting of the running state data into the preset artificial intelligence AI diagnostic model includes:
  • the running state data may be cleaned first, for example, the data that meets the cleaning conditions may be eliminated from the running state data.
  • the cleaning condition may be that the data in the running state is incomplete, or an indicator of the data in the running state exceeds the standard.
  • the method further includes:
  • the cloud server obtains training sample data of multiple vehicles of the same type and training sample data of multiple vehicles of different types, and the motor driver is installed on the vehicles;
  • the cloud server uses the training sample data of the multiple vehicles of the same type and the training sample data of the multiple vehicles of different types to perform model training based on the fault warning task, so as to obtain the preset AI diagnosis model .
  • the cloud server uses multiple training sample data, and the multiple training sample data may include training sample data of multiple vehicles of the same type and training sample data of multiple vehicles of different types, so that the obtained AI diagnosis
  • the model is an AI global model, which is oriented to specific fault warning tasks.
  • the model is trained based on all the collected vehicle terminal equipment side data, and based on this model, the end side reasoning results of each vehicle terminal equipment are given.
  • the method further includes:
  • the cloud server obtains training sample data of one vehicle or training sample data of multiple vehicles of the same type, and the motor driver is installed on the vehicle;
  • the cloud server uses the training sample data of the one vehicle or the training sample data of multiple vehicles of the same type to perform model training based on a fault warning task to obtain the preset AI diagnosis model.
  • the cloud server uses multiple training sample data, and the multiple training sample data may include the training sample data of one vehicle or the training sample data of multiple vehicles of the same type, so that the obtained AI diagnosis model is an AI partial model.
  • the model is for specific fault warning tasks.
  • the AI local model may be a certain type of vehicle terminal equipment or even a certain vehicle terminal equipment. Based on this model, each end-side reasoning result is given.
  • the cloud server when the cloud server outputs the first abnormality degree parameter of the motor driver through the AI diagnosis model, the motor driver is warned of a fault according to the first abnormality degree parameter, include:
  • the cloud server sends the fault warning information to the vehicle terminal device.
  • the cloud server has model training capabilities and decision-making inference capabilities, and the vehicle terminal device can perform fault warning according to the decision result of the cloud server.
  • the motor driver can be fault warning.
  • the method further includes: the vehicle terminal device receives the AI diagnosis model sent by the cloud server;
  • the inputting the running state data into a preset artificial intelligence AI diagnosis model, and outputting the first abnormality parameter of the motor driver through the AI diagnosis model including:
  • the vehicle terminal device inputs the running state data into the AI diagnosis model, and outputs the first abnormality degree parameter of the motor driver through the AI diagnosis model.
  • the cloud server has the model training capability
  • the vehicle terminal device has the decision-making inference capability
  • the vehicle terminal device can receive the AI diagnosis model of the cloud server, so that the vehicle terminal device can output the decision result according to the AI diagnosis model.
  • the fault early warning of the motor driver can be realized through the interaction between the cloud server and the vehicle terminal equipment.
  • the method further includes:
  • the cloud server obtains training sample data of multiple different types of vehicles, or training sample data of one vehicle, or training sample data of multiple vehicles of the same type;
  • the cloud server performs data association extraction on the training sample data of the multiple vehicles of different types, or the training sample data of one vehicle, or the training sample data of multiple vehicles of the same type, so as to obtain a pre-training model,
  • the pre-training model is used to represent the extracted data association;
  • the cloud server sends the pre-training model to the vehicle terminal device.
  • the cloud server does not need to train the AI diagnostic model.
  • the cloud server only needs to extract the data association relationship, and after obtaining the pre-training model, send the pre-training model to the vehicle terminal device.
  • the method further includes:
  • the vehicle terminal device receives the pre-trained model from the cloud server
  • the vehicle terminal device performs model parameter optimization processing on the pre-training model according to the preset label data, so as to obtain the preset AI diagnosis model.
  • the vehicle terminal device performs model parameter optimization processing on the pre-trained model according to the preset label data.
  • the model parameter optimization process can also be called fine-tuning, that is, the vehicle terminal device can perform model training to obtain the preset AI diagnostic model.
  • the running state data includes: dynamic data collected by default when the motor driver is running; or,
  • the running state data includes: dynamic data collected through the CAN bus of the controller area network.
  • the running state data includes: dynamic data obtained by high-frequency sampling when the vehicle is in a running phase, and the motor driver is installed on the vehicle.
  • the vehicle has multiple states, for example, the vehicle is in a running phase, and the running phase refers to a phase after the vehicle is started to spark, for example, the running phase means that the vehicle is in a running phase.
  • the dynamic data obtained by performing high-frequency sampling when the vehicle is in the running phase may be the aforementioned running state data, and the data items and data amounts of the high-frequency sampling will be exemplified in subsequent embodiments.
  • the high-frequency sampling refers to that the frequency of data collection is greater than a preset frequency threshold.
  • the running state data includes at least one of the following: parameters of the motor driver, parameters of the cooling system, parameters of the motor, and parameters of the battery.
  • the running state data of the motor driver when the motor driver is running in the embodiments of the present application can be implemented in various manners. For example, when the motor driver is running, data can be collected for the motor driver to obtain the parameters of the motor driver, or when the motor driver is running Data collection may be performed for the cooling system to obtain parameters of the cooling system, or data collection may be performed for the motor to obtain the parameters of the motor when the motor driver is running, or data collection may be performed for the battery to obtain the parameters of the battery when the motor driver is running.
  • the parameters of the motor driver include at least one of the following: DC bus voltage, DC bus current, low-voltage power supply voltage, vehicle driving status command, vehicle status, motor driver temperature, rotor position electrical angle, The rms value of the three-phase current of the motor driver, the three-phase current of the motor driver, the sampled value of the three-phase current of the motor driver, the derating status of the motor driver; and/or,
  • the parameters of the cooling system include at least one of the following: coolant flow rate, coolant temperature, oil pump target rotational speed, oil pump state, oil pump actual rotational speed, oil pump supply voltage, oil pump power semiconductor device temperature, oil pump current; and/or,
  • the parameters of the motor include at least one of the following: electromagnetic frequency, motor working mode command, motor controller working state, motor target torque command, motor current torque, motor target speed command, motor current speed, motor temperature, motor direct Shaft voltage, motor direct axis current reference value, motor direct axis current feedback value, motor quadrature axis voltage, motor quadrature axis current reference value, motor quadrature axis current feedback value, motor current calibration torque; and/or,
  • the parameters of the battery include at least one of the following: rated output voltage, battery capacity, and maximum output current.
  • an embodiment of the present application further provides a fault early warning device for a motor driver, which is characterized in that it includes:
  • the acquisition module is used to acquire the running state data of the motor driver when it is running;
  • the reasoning module is used to input the running state data into a preset artificial intelligence AI diagnosis model, and output the first abnormality degree parameter of the motor driver through the AI diagnosis model, wherein the AI diagnosis model is used for according to Extracting data features from the running state data, and performing decision inference based on the data features;
  • An early warning module configured to give a fault early warning to the motor driver according to the first abnormality degree parameter.
  • the acquiring module is further configured to acquire static data obtained by performing low-frequency sampling on the motor driver when the vehicle is in a stop phase, and the motor driver is installed on the vehicle;
  • the reasoning module is further configured to input the static data into the AI diagnosis model, and output the second abnormality parameter of the motor driver through the AI diagnosis model;
  • the early warning module is further configured to perform a fault early warning on the motor driver according to the second abnormality degree parameter.
  • the vehicle is in a stop phase, including at least one of the following:
  • the vehicle is in the charging phase, the vehicle is in the ignition phase, and the vehicle is in the ignition phase.
  • the early warning module is configured to dynamically update the preset abnormal conditions through the AI diagnostic model to obtain the updated abnormal conditions; when the first abnormality degree parameter satisfies all When the updated abnormal condition is detected, a fault warning is performed on the motor driver.
  • the apparatus further includes: a data processing module, configured to perform cleaning processing on the running status data after the acquisition module acquires the running state data of the motor driver during operation, and obtain the cleaned and processed data. running state data;
  • the reasoning module is used for inputting the running state data after the cleaning process into the AI diagnosis model.
  • the apparatus is specifically a cloud server; the apparatus further includes: a training module,
  • the acquisition module is further configured to acquire training sample data of multiple vehicles of the same type and training sample data of multiple vehicles of different types, and the motor driver is installed on the vehicles;
  • the training module is configured to use the training sample data of the multiple vehicles of the same type and the training sample data of the multiple vehicles of different types to perform model training based on the fault warning task, so as to obtain the preset AI diagnostic model.
  • the apparatus is specifically a cloud server; the apparatus further includes: a training module,
  • the acquisition module is further configured to acquire training sample data of one vehicle or training sample data of multiple vehicles of the same type, and the motor driver is installed on the vehicle;
  • the training module is configured to use the training sample data of the one vehicle or the training sample data of multiple vehicles of the same type to perform model training based on a fault warning task to obtain the preset AI diagnosis model.
  • the apparatus is specifically a cloud server; the early warning module is configured to generate fault early warning information according to the first abnormality degree parameter; and send the fault early warning information to a vehicle terminal device.
  • the apparatus is specifically a vehicle terminal device
  • the obtaining module is further configured to receive the AI diagnosis model sent by the cloud server.
  • the apparatus is specifically a cloud server; the apparatus further includes: a training module,
  • the obtaining module is further configured to obtain training sample data of a plurality of vehicles of different types, or training sample data of one vehicle, or training sample data of multiple vehicles of the same type;
  • the training module is used to extract the data association relationship on the training sample data of the plurality of vehicles of different types, or the training sample data of one vehicle, or the training sample data of multiple vehicles of the same type, so as to obtain a pre-determined model. Training a model, where the pre-training model is used to represent the extracted data association; sending the pre-training model to the vehicle terminal device.
  • the apparatus is specifically a vehicle terminal device; the apparatus further includes: a training module,
  • the training module is used to receive the pre-training model from the cloud server, wherein the pre-training module is used to represent the training sample data of the cloud server for a plurality of vehicles of different types, or the training sample data of one vehicle, Or a data association relationship obtained by extracting training sample data of multiple vehicles of the same type; performing model parameter optimization processing on the pre-trained model according to preset label data to obtain the preset AI diagnostic model.
  • the running state data includes: dynamic data collected by default when the motor driver is running; or,
  • the running state data includes: dynamic data collected through the CAN bus of the controller area network.
  • the running state data includes: dynamic data obtained by high-frequency sampling when the vehicle is in a running phase, and the motor driver is installed on the vehicle.
  • the running state data includes at least one of the following: parameters of the motor driver, parameters of the cooling system, parameters of the motor, and parameters of the battery.
  • the parameters of the motor driver include at least one of the following: DC bus voltage, DC bus current, low-voltage power supply voltage, vehicle driving status command, vehicle status, motor driver temperature, rotor position electrical angle, The rms value of the three-phase current of the motor driver, the three-phase current of the motor driver, the sampled value of the three-phase current of the motor driver, the derating status of the motor driver; and/or,
  • the parameters of the cooling system include at least one of the following: coolant flow rate, coolant temperature, oil pump target rotational speed, oil pump state, oil pump actual rotational speed, oil pump supply voltage, oil pump power semiconductor device temperature, oil pump current; and/or,
  • the parameters of the motor include at least one of the following: electromagnetic frequency, motor working mode command, motor controller working state, motor target torque command, motor current torque, motor target speed command, motor current speed, motor temperature, motor direct Shaft voltage, motor direct axis current reference value, motor direct axis current feedback value, motor quadrature axis voltage, motor quadrature axis current reference value, motor quadrature axis current feedback value, motor current calibration torque; and/or,
  • the parameters of the battery include at least one of the following: rated output voltage, battery capacity, and maximum output current.
  • the component modules of the fault warning device for the motor drive may also perform the steps described in the first aspect and various possible implementation manners. Description of possible implementations.
  • the embodiments of the present application further provide a fault early warning method for a motor driver, the method comprising:
  • a fault warning is performed on the motor driver according to the second abnormality degree parameter.
  • the vehicle is in a stop phase, including at least one of the following:
  • the vehicle is in the charging phase, the vehicle is in the ignition phase, and the vehicle is in the ignition phase.
  • the performing a fault warning to the motor driver according to the second abnormality degree parameter includes:
  • the method further includes:
  • the inputting the static data into the preset artificial intelligence AI diagnosis model includes:
  • the static data after cleaning is input into the AI diagnosis model.
  • the method further includes:
  • the cloud server obtains training sample data of multiple vehicles of the same type and training sample data of multiple vehicles of different types, and the motor driver is installed on the vehicles;
  • the cloud server uses the training sample data of the multiple vehicles of the same type and the training sample data of the multiple vehicles of different types to perform model training based on the fault warning task, so as to obtain the preset AI diagnosis model .
  • the method further includes:
  • the cloud server obtains training sample data of one vehicle or training sample data of multiple vehicles of the same type, and the motor driver is installed on the vehicle;
  • the cloud server uses the training sample data of the one vehicle or the training sample data of multiple vehicles of the same type to perform model training based on a fault warning task to obtain the preset AI diagnosis model.
  • the cloud server when the cloud server outputs the first abnormality degree parameter of the motor driver through the AI diagnostic model, the motor driver is warned of a fault according to the second abnormality degree parameter, include:
  • the cloud server generates fault warning information according to the second abnormality parameter
  • the cloud server sends the fault warning information to the vehicle terminal device.
  • the method further includes: the vehicle terminal device receives the AI diagnosis model sent by the cloud server;
  • the vehicle terminal device inputs the static data into the AI diagnosis model, and outputs the second abnormality degree parameter of the motor driver through the AI diagnosis model.
  • the method further includes:
  • the cloud server obtains training sample data of multiple different types of vehicles, or training sample data of one vehicle, or training sample data of multiple vehicles of the same type;
  • the cloud server performs data association extraction on the training sample data of the multiple vehicles of different types, or the training sample data of one vehicle, or the training sample data of multiple vehicles of the same type, so as to obtain a pre-training model,
  • the pre-training model is used to represent the extracted data association;
  • the cloud server sends the pre-training model to the vehicle terminal device.
  • the method further includes:
  • the vehicle terminal device receives the pre-trained model from the cloud server
  • the vehicle terminal device performs model parameter optimization processing on the pre-training model according to the preset label data, so as to obtain the preset AI diagnosis model.
  • the embodiments of the present application further provide a fault warning device for a motor driver, the device comprising:
  • an acquisition module configured to acquire static data obtained by performing low-frequency sampling on the motor driver when the vehicle is in a stop phase, and the motor driver is installed on the vehicle;
  • an inference module configured to input the static data into the AI diagnosis model, and output the second abnormality parameter of the motor driver through the AI diagnosis model;
  • an early warning module configured to perform a fault early warning on the motor driver according to the second abnormality degree parameter.
  • the vehicle is in a stop phase, including at least one of the following:
  • the vehicle is in the charging phase, the vehicle is in the ignition phase, and the vehicle is in the ignition phase.
  • the early warning module is configured to dynamically update the preset abnormal condition through the AI diagnostic model, so as to obtain the updated abnormal condition; when the second abnormality degree parameter satisfies all When the updated abnormal condition is detected, a fault warning is performed on the motor driver.
  • the apparatus further includes: a data processing module, configured to: after the acquisition module acquires static data obtained by performing low-frequency sampling on the motor driver when the vehicle is in a stop phase, Perform cleaning to obtain static data after cleaning;
  • the reasoning module is used for inputting the cleaned static data into the AI diagnosis model.
  • the apparatus is specifically a cloud server; the apparatus further includes: a training module,
  • the acquisition module is further configured to acquire training sample data of multiple vehicles of the same type and training sample data of multiple vehicles of different types, and the motor driver is installed on the vehicles;
  • the training module is configured to use the training sample data of the multiple vehicles of the same type and the training sample data of the multiple vehicles of different types to perform model training based on the fault warning task, so as to obtain the preset AI diagnostic model.
  • the apparatus is specifically a cloud server; the apparatus further includes: a training module,
  • the acquisition module is further configured to acquire training sample data of one vehicle or training sample data of multiple vehicles of the same type, and the motor driver is installed on the vehicle;
  • the training module is configured to use the training sample data of the one vehicle or the training sample data of multiple vehicles of the same type to perform model training based on a fault warning task to obtain the preset AI diagnosis model.
  • the apparatus is specifically a cloud server; the early warning module is configured to generate fault early warning information according to the second abnormality degree parameter; and send the fault early warning information to the vehicle terminal device.
  • the apparatus is specifically a vehicle terminal device
  • the obtaining module is further configured to receive the AI diagnosis model sent by the cloud server.
  • the apparatus is specifically a cloud server; the apparatus further includes: a training module,
  • the obtaining module is further configured to obtain training sample data of a plurality of vehicles of different types, or training sample data of one vehicle, or training sample data of multiple vehicles of the same type;
  • the training module is used to extract the data association relationship on the training sample data of the plurality of vehicles of different types, or the training sample data of one vehicle, or the training sample data of multiple vehicles of the same type, so as to obtain a pre-determined model. Training a model, where the pre-training model is used to represent the extracted data association; sending the pre-training model to the vehicle terminal device.
  • the apparatus is specifically a vehicle terminal device; the apparatus further includes: a training module,
  • the training module is used to receive the pre-training model from the cloud server, wherein the pre-training module is used to represent the training sample data of the cloud server for a plurality of vehicles of different types, or the training sample data of one vehicle, Or a data association relationship obtained by extracting training sample data of multiple vehicles of the same type; performing model parameter optimization processing on the pre-trained model according to preset label data to obtain the preset AI diagnostic model.
  • an embodiment of the present application provides a computer-readable storage medium, where instructions are stored in the computer-readable storage medium, and when the computer-readable storage medium runs on a computer, the computer executes the first aspect or the third aspect. method described.
  • an embodiment of the present application provides a computer program product containing instructions, which, when executed on a computer, cause the computer to execute the method described in the first aspect or the third aspect.
  • an embodiment of the present application provides a communication device.
  • the communication device may include a cloud server, a vehicle terminal device, or an entity such as a chip.
  • the communication device includes: a processor and a memory; the memory is used to store instructions; The processor is configured to execute the instructions in the memory, causing the communication device to perform the method as described in any of the foregoing first or third aspects.
  • the present application provides a chip system
  • the chip system includes a processor for supporting the fault early warning device to implement the functions involved in the first aspect or the third aspect, for example, sending or processing the above method. the data and/or information involved.
  • the chip system further includes a memory for storing necessary program instructions and data of the fault warning device.
  • the chip system may be composed of chips, or may include chips and other discrete devices.
  • the present application provides a cloud server
  • the cloud server includes a processor for supporting the fault early warning device to implement the functions involved in the first aspect or the third aspect, for example, sending or processing the above method. the data and/or information involved.
  • the cloud server further includes a memory for storing necessary program instructions and data of the fault warning device.
  • the present application provides a vehicle terminal device, and the chip system includes a processor, which is used to support the fault early warning device to implement the functions involved in the first aspect or the third aspect, for example, send or process the above method. the data and/or information involved.
  • the vehicle terminal device further includes a memory for storing necessary program instructions and data of the fault warning device.
  • an embodiment of the present application further provides a fault early warning system.
  • the fault early warning system may include the cloud server described in the ninth aspect and the vehicle terminal device described in the tenth aspect.
  • an embodiment of the present application provides a vehicle networking device
  • the vehicle networking device may include a vehicle networking server, a roadside unit, a vehicle networking communication device, or a chip and other entities
  • the vehicle networking device includes: a processor.
  • the IoV device further includes: a memory; the memory is used to store instructions; the processor is used to execute the instructions in the memory, so that the IoV device executes the first aspect or The method of any one of the third aspects.
  • the embodiments of the present application have the following advantages:
  • first obtain the running state data of the motor driver during operation then input the running state data into a preset AI diagnosis model, and output the first abnormality degree parameter of the motor driver through the AI diagnosis model, wherein the AI diagnosis
  • the model is used for extracting data features according to the running state data, and making decision inference based on the data features; according to the first abnormality degree parameter, a fault warning is performed for the motor driver. Since the vehicle is in a running state when the motor driver is running in the embodiment of the present application, a fault warning can be given to the motor driver according to the running state data when the motor driver is running.
  • the running state data of the motor driver is the data collected when the motor driver is running. The data can reflect the real operating state of the motor driver. Based on the operating state data, fault warning can be performed, which can realize the advance prediction of the failure of the motor driver and realize the fault early warning of the motor driver.
  • FIG. 1 is a schematic block diagram of a flowchart of a fault early warning method for a motor driver provided by an embodiment of the present application;
  • FIG. 2 is a schematic block diagram of a flowchart of a fault early warning method for a motor driver provided by an embodiment of the present application;
  • FIG. 3 is a schematic block diagram of a flowchart of a fault early warning method for a motor driver provided by an embodiment of the present application;
  • FIG. 4 is a schematic block diagram of a flowchart of a fault early warning method for a motor driver provided by an embodiment of the present application;
  • FIG. 5 is a schematic block diagram of a flowchart of a fault early warning method for a motor driver according to an embodiment of the present application
  • FIG. 6 is a schematic block diagram of a flowchart of a fault early warning method for a motor driver provided by an embodiment of the present application
  • FIG. 7 is a schematic structural diagram of the composition of a cloud server and a vehicle terminal device according to an embodiment of the present application.
  • FIG. 8 is a schematic flowchart of an execution flow of an AI diagnosis model provided by an embodiment of the present application.
  • FIG. 9 is a schematic flowchart of an execution flow of an AI diagnosis model provided by an embodiment of the present application.
  • FIG. 10 is a schematic flowchart of a collection process of running state data and static data provided by an embodiment of the present application
  • FIG. 11a is a schematic structural diagram of a fault warning device for a motor driver according to an embodiment of the present application.
  • FIG. 11b is a schematic structural diagram of a fault warning device for a motor driver provided by an embodiment of the application.
  • FIG. 11c is a schematic structural diagram of a fault warning device for a motor driver provided by an embodiment of the present application.
  • FIG. 12 is a schematic structural diagram of a fault warning device for a motor driver provided by an embodiment of the present application.
  • Embodiments of the present application provide a fault early warning method and device for a motor driver, which are used to implement fault early warning for the motor driver.
  • the embodiment of the present application provides a fault early warning method for a motor driver, which can perform fault early warning on the motor driver and improve the fault early warning accuracy of the motor driver.
  • the motor driver is installed on the vehicle, and the motor driver is an electronic control device on the vehicle.
  • the vehicle can be a new energy vehicle, a smart car, a vehicle terminal device, etc.
  • the motor driver can also be called a motor control unit (motor control unit, MCU) .
  • a motor driver may include one or more power semiconductor devices.
  • a power semiconductor device refers to a semiconductor device that implements a circuit switching function.
  • a power semiconductor device may be a switch tube.
  • each switch tube is a metal oxide semiconductor field.
  • the effect transistor metal-oxide-semiconductor field-effect transistor, MOSFET
  • MOSFET metal-oxide-semiconductor field-effect transistor
  • each switch can also be other semiconductor devices such as an insulated gate bipolar transistor (insulated gate bipolar transistor, IGBT), etc.
  • the switch tube may also be a diode (Diode), etc.
  • the switch tube is mainly applied to an IGBT for illustration.
  • the motor driver realizes the control of the motor by controlling the rotation angle and the running speed of the motor.
  • the motor driver can be driven by relays or power transistors, and can also be driven by thyristors or power MOS field effect transistors. speed, forward and reverse control of DC motors, etc.
  • the vehicle is in a running state when the motor driver is running, and a fault warning can be given to the motor driver according to the running state data when the motor driver is running.
  • the running state data of the motor driver is the data collected when the motor driver is running, the running state The data can reflect the real operating state of the motor driver. Based on the operating state data, fault warning can be performed, which can realize the advance prediction of the failure of the motor driver and realize the fault early warning of the motor driver.
  • the fault early warning method for a motor driver is implemented by a fault early warning device for a motor driver (hereinafter referred to as a fault early warning device), and the fault early warning device can store pre-trained artificial intelligence (artificial intelligence, AI) diagnostic model, the AI diagnostic model can be a diagnostic model obtained after model training using an AI diagnostic algorithm.
  • AI diagnostic model has the ability to extract data features, and based on the data features, it can perform decision-making inference and output the abnormality of the motor driver. Through the analysis of the abnormality parameter, it can be determined whether there is a fault in the motor driver, so as to give a fault early warning to the motor driver.
  • the fault early warning device has various implementation manners, and the fault early warning device may be a server independent of the vehicle, or a terminal device integrated with the vehicle, which is not limited here.
  • the fault warning device may be a cloud server, and the cloud server may also be called a cloud platform or a cloud platform.
  • the pre-trained AI diagnosis model can be stored in the cloud server, so that the cloud server can notify the vehicle terminal device of the inference result after the AI diagnosis model completes the decision inference.
  • the fault warning device may also be a vehicle terminal device, which may be a controller of the vehicle, or the vehicle terminal device may be a terminal device integrated in the vehicle, and the vehicle terminal device may also be referred to as a terminal-side terminal device.
  • the vehicle terminal device can obtain the inference results from the cloud server, so that the vehicle terminal device can give a fault warning to the motor driver.
  • the fault warning device may be a vehicle terminal device, the vehicle terminal device may obtain a pre-training model from a cloud server, and then the vehicle terminal device generates an AI diagnostic model according to the pre-training model, so that the vehicle terminal device can use the AI diagnostic model to detect the motor.
  • the drive gives an early warning of failure.
  • the embodiments of the present application do not limit the implementation manner of the fault early warning device.
  • an embodiment of the present application provides a fault early warning method for a motor driver, which mainly includes the following steps:
  • the motor driver is installed in the vehicle, and the motor driver needs to run during the operation of the vehicle.
  • the fault early warning device can obtain the running state data of the motor driver when it is running, and the running state data can reflect the real running state of the motor driver.
  • the embodiment of the present application does not limit the specific collection process of the running state data of the motor driver and the collection range of the data items and data parameters included in the running state data, for example, it can be comprehensively determined according to the type and function of the motor driver, the running state of the vehicle, etc.
  • Running state data, the running state data will be described in detail in subsequent embodiments.
  • the fault early warning device can be a cloud server, and the vehicle terminal device can collect the running state data of the motor driver, and then the vehicle terminal device can send the running state data of the motor driver to the cloud server through a communication interface or a wireless network.
  • the fault early warning device may be a vehicle terminal device, and the vehicle terminal device is provided with a data measurement module, and the data measurement module can collect the running state data of the motor driver.
  • the running state data includes: dynamic data collected by default when the motor driver is running; or,
  • the running state data includes: dynamic data collected through the controller area network (CAN) bus.
  • CAN controller area network
  • the running state data includes the dynamic data collected by default when the motor driver is running, and the dynamic data will be collected by default when the motor driver is running. Except for the original control function of the vehicle, the default collected dynamic data can be used in the embodiment of the present application.
  • the collected dynamic data is used for fault early warning of the motor driver.
  • the running state data includes dynamic data collected through the CAN bus, and the motor controller is connected to the CAN bus, so dynamic data can be collected through the CAN bus to obtain the above-mentioned running state data.
  • the embodiments of the present application are not limited to the above two data collection methods. Through the above data collection methods, the running state data of the motor driver during operation can be obtained without the need to upgrade the single-board hardware of the vehicle and the motor driver, which improves the operation performance. data collection efficiency.
  • the embodiments of the present application do not limit the specific collection process of the running state data of the motor driver and the collection range of the data items and data parameters included in the running state data.
  • the running state data includes: dynamic data obtained by high-frequency sampling when the vehicle is in the running phase, and the motor driver is installed on the vehicle.
  • the vehicle has multiple states, for example, the vehicle is in a running phase
  • the running phase refers to a phase after the vehicle is started after ignition, for example, the running phase means that the vehicle is in a running phase.
  • the dynamic data obtained by performing high-frequency sampling when the vehicle is in the running phase may be the aforementioned running state data, and the data items and data amounts of the high-frequency sampling will be exemplified in subsequent embodiments.
  • the high-frequency sampling refers to that the frequency of data collection is greater than a preset frequency threshold, and the specific value of the frequency threshold is not limited in this embodiment of the present application.
  • the running state data includes at least one of the following: parameters of the motor driver, parameters of the cooling system, parameters of the motor, and parameters of the battery.
  • the running state data of the motor driver when the motor driver is running in the embodiment of the present application can be realized in various ways.
  • data can be collected for the motor driver to obtain the parameters of the motor driver, or when the motor driver is running, the cooling The system collects data to obtain the parameters of the cooling system, or can collect data from the motor to obtain the parameters of the motor when the motor driver is running, or collect data from the battery to obtain the parameters of the battery when the motor driver is running.
  • specific parameter names and parameter contents in the parameters of the motor driver, the parameters of the cooling system, the parameters of the motor, and the parameters of the battery are not limited.
  • parameters of the motor driver, parameters of the cooling system, parameters of the motor, and parameters of the battery can be obtained through data collection. Since the parameters of the motor driver, the parameters of the cooling system, the parameters of the motor, and the parameters of the battery are all data collected during the operation of the motor driver, any of the parameters of the motor driver, the parameters of the cooling system, the parameters of the motor, and the parameters of the battery Each item can reflect the real operating state of the motor driver, and the fault warning based on the above parameters can realize the advance prediction of the failure of the motor driver and realize the fault early warning of the motor driver.
  • the parameters of the motor driver include at least one of the following: DC bus voltage, DC bus current, low voltage power supply voltage, vehicle driving status command, vehicle status, motor driver temperature, rotor position electrical angle , the rms value of the three-phase current of the motor driver, the three-phase current of the motor driver, the sampling value of the three-phase current of the motor driver, the derating status of the motor driver; and/or,
  • the parameters of the cooling system include at least one of the following: coolant flow rate, coolant temperature, oil pump target speed, oil pump status, oil pump actual speed, oil pump supply voltage, power semiconductor device temperature of the oil pump, and oil pump current; and/or,
  • the parameters of the motor include at least one of the following: electromagnetic frequency, motor working mode command, motor controller working state, motor target torque command, motor current torque, motor target speed command, motor current speed, motor temperature, motor direct axis voltage , motor direct axis current given value, motor direct axis current feedback value, motor quadrature axis voltage, motor quadrature axis current given value, motor quadrature axis current feedback value, motor current calibration torque; and/or,
  • the parameters of the battery include at least one of the following: rated output voltage, battery capacity, and maximum output current.
  • the three-phase currents of the motor driver may be the currents of the U, V and W phases.
  • the fault diagnosis apparatus may pre-train an AI diagnosis model, and after the fault diagnosis apparatus acquires the running state data, the AI diagnosis model is used for decision inference.
  • the running state data is input into the AI diagnostic model
  • the AI diagnostic model extracts data features according to the running state data
  • the data features are obtained in advance by the AI diagnostic model according to the input running state data
  • the data features are the AI diagnostic model The basis for making decision-making inferences.
  • the AI diagnosis model may be a machine learning model, for example, the AI diagnosis model may be a neural network model obtained by pre-training, or the AI diagnosis model may also be other machine learning models, such as a linear regression model, a decision tree model, etc.,
  • the AI diagnostic model performs decision inference based on the extracted data features.
  • the data content included in the data feature is not limited in this embodiment of the present application.
  • the AI diagnostic model After the AI diagnostic model performs decision-making inference, it can output the first abnormality degree parameter of the motor driver, where the first abnormality degree parameter is an abnormality defined for distinguishing other abnormality degree parameters (such as the second abnormality degree parameter) appearing in subsequent embodiments degree parameter.
  • the first abnormality degree parameter is a parameter used to measure the abnormality of the motor driver, and the first abnormality degree parameter may also be referred to as a first abnormality factor score, a first abnormality degree score, or the like.
  • the fault diagnosis apparatus executes the foregoing step 102.
  • the fault diagnosis apparatus may be a cloud server, or the fault diagnosis apparatus may be a vehicle terminal device.
  • step 101 obtains the running state data of the motor driver during operation
  • the method provided by the embodiments of the present application may further include the following steps:
  • the running state data is cleaned to obtain the cleaned running state data.
  • the fault diagnosis device may first perform cleaning processing on the running state data, for example, may first eliminate data that meets the cleaning conditions in the running state data.
  • the cleaning condition may be that the data in the running state is incomplete, or an indicator of the data in the running state exceeds the standard.
  • inputting the running state data into the preset AI diagnosis model in the aforementioned step 102 includes:
  • the fault diagnosis device obtains the cleaned and processed running state data, it inputs the cleaned and processed running state data into the AI diagnostic model, and uses the AI diagnostic model to extract the data features of the cleaned and processed running state data, The running state data is cleaned and then input into the AI diagnosis model, which can improve the reasoning efficiency of the AI diagnosis model.
  • the working condition of the motor driver can be measured according to the first abnormality degree parameter, so as to identify whether the motor driver has an abnormality, and when the motor driver has an abnormality
  • a fault warning is performed on the motor driver, so as to realize the advance prediction of the failure of the motor driver.
  • fault early warning means that a certain time before the fault occurs, it is predicted that the fault will occur in the future, and an alarm signal is issued.
  • the method provided by the embodiment of the present application can be applied to an electric drive system of an automobile, to diagnose the failure of the electric drive system of the automobile caused by the motor driver in advance and give an early warning, so that the user can perform maintenance before the failure occurs, and avoid situations such as the vehicle breaking down during driving.
  • step 103 performs a fault warning for the motor driver according to the first abnormality degree parameter, including:
  • the fault diagnosis device can preset abnormal conditions, and the abnormal conditions can be dynamically updated through the AI diagnosis model, so as to ensure that the abnormal conditions can be used for fault judgment and ensure the accuracy of fault judgment.
  • the abnormal condition After the abnormal condition is dynamically updated, it can be determined whether the first abnormality degree satisfies the dynamically updated abnormality condition. When the first abnormality degree parameter satisfies the updated abnormality condition, it indicates that the motor driver may be faulty.
  • the drive gives an early warning of failure.
  • the accuracy of the fault warning can be further improved.
  • the running state data of the motor driver during operation is obtained first, and then the running state data is input into the preset AI diagnosis model, and the first abnormality parameter of the motor driver is output through the AI diagnosis model, wherein,
  • the AI diagnostic model is used to extract data features according to the running state data, and perform decision-making inference based on the data features; according to the first abnormality degree parameter, the motor driver is warned of faults. Since the vehicle is in a running state when the motor driver is running in the embodiment of the present application, a fault warning can be given to the motor driver according to the running state data when the motor driver is running.
  • the running state data of the motor driver is the data collected when the motor driver is running. The data can reflect the real operating state of the motor driver. Based on the operating state data, fault warning can be performed, which can realize the advance prediction of the failure of the motor driver and realize the fault early warning of the motor driver.
  • the embodiment provides a fault early warning method for a motor driver, which mainly includes the following steps:
  • the vehicle has multiple states, for example, the vehicle is in a stop phase, which may also be called a stationary phase.
  • the low-frequency sampling means that the frequency of data collection is lower than a preset frequency threshold, and the value of the frequency threshold is not limited in this embodiment of the present application.
  • the fault diagnosis device can be a vehicle terminal device.
  • the vehicle terminal device can perform low-frequency sampling on the motor driver to obtain static data, which can be used for decision-making inference of the AI diagnostic model.
  • the fault diagnosis device may be a cloud server.
  • the vehicle terminal device may perform low-frequency sampling on the motor driver to obtain static data, and the vehicle terminal device may send the static data to the cloud server, thereby The cloud server can receive the static data from the vehicle terminal device.
  • the vehicle is in the stop phase, including at least one of the following: the vehicle is in the charging phase, the vehicle is in the ignition phase, and the vehicle is in the ignition off phase.
  • static data is sampled in the vehicle charging stage, the vehicle ignition stage, and the vehicle flameout stage, so that the fault diagnosis device can provide an emergency response function in the case of a network disconnection according to the static data.
  • the fault diagnosis apparatus may pre-train an AI diagnosis model, and after acquiring the static data, the fault diagnosis apparatus uses the AI diagnosis model to perform decision inference.
  • the static data is input into the AI diagnostic model
  • the AI diagnostic model extracts data features according to the static data
  • the data features are obtained in advance by the AI diagnostic model according to the input static data
  • the data features are the decision-making reasoning performed by the AI diagnostic model basis.
  • the AI diagnosis model may be a machine learning model, for example, the AI diagnosis model may be a neural network model obtained by pre-training, or the AI diagnosis model may also be other machine learning models, such as a linear regression model, a decision tree model, etc.
  • the AI diagnostic model performs decision inference based on the extracted data features.
  • the data content included in the data feature is not limited in this embodiment of the present application.
  • the AI diagnosis model After the AI diagnosis model performs decision inference, it can output the second abnormality parameter of the motor driver.
  • the second abnormality degree parameter is a parameter used to measure the abnormality of the motor driver, and the second abnormality degree parameter may also be referred to as a second abnormality factor score, a second abnormality degree score, or the like.
  • the working condition of the motor driver can be measured according to the second abnormality degree parameter, so as to identify whether the motor driver has an abnormal condition.
  • the fault occurs, it is determined that the motor driver has a fault, and a fault warning is performed on the motor driver, so as to realize the advance prediction of the failure of the motor driver.
  • the method provided by the embodiment of the present application can be applied to an electric drive system of an automobile, to diagnose the failure of the electric drive system of the automobile caused by the motor driver in advance and give an early warning, so that the user can perform maintenance before the failure occurs, and avoid situations such as the vehicle breaking down during driving.
  • the fault diagnosis device can obtain the aforementioned static data and running state data, and input both the static data and running state data into the AI diagnosis model. Decision-making reasoning, output the third abnormality parameter, and finally give a fault warning to the motor driver according to the third abnormality parameter.
  • the static data of the motor driver is obtained when the vehicle is in the stop stage, and a fault warning can be performed on the motor driver according to the static data.
  • the collected data and static data can reflect the real static state of the motor driver. Based on the static data, fault warning can be performed to predict the failure of the motor driver in advance and realize the fault early warning of the motor driver.
  • FIGS. 1 and 2 a fault early warning method for a motor driver executed by a fault diagnosis device is used for illustration.
  • FIG. 3 an interaction process between the cloud server and the vehicle terminal device in the embodiment of the present application is introduced, as shown in FIG. 3 .
  • the embodiment of the present application provides a fault early warning method for a motor driver, which mainly includes the following steps:
  • the cloud server obtains training sample data of multiple vehicles of the same type and training sample data of multiple vehicles of different types, and the motor driver is installed on the vehicle.
  • the cloud server can interact with multiple vehicle terminal devices, and one or more motor drivers can be installed in each vehicle.
  • the cloud server can obtain multiple training sample data, and the training sample data can be used for model training.
  • the plurality of training sample data may be training sample data of a plurality of vehicles of the same type and training sample data of a plurality of vehicles of different types, and the type refers to the type of the vehicle.
  • the data type and data content of the training sample data are not limited.
  • the cloud server uses the training sample data of multiple vehicles of the same type and the training sample data of multiple different types of vehicles to perform model training based on the fault warning task, so as to obtain a preset AI diagnosis model.
  • the cloud server obtains training sample data of multiple vehicles of the same type and training sample data of multiple different types of vehicles.
  • the cloud server can pre-define the training task as a fault warning task, and the cloud server uses the data of multiple vehicles of the same type.
  • the training sample data and the training sample data of a plurality of different types of vehicles are used for model training.
  • the specific training process of the model is not described in detail in the embodiment of the present application.
  • a preset AI diagnosis model can be obtained.
  • the cloud server first uses the training sample data of multiple vehicles of the same type for model training, and then generalizes the training sample data of different types of vehicles.
  • the model training can be completed and the preset AI diagnosis model can be output.
  • the cloud server acquires the running state data of the motor driver when the motor driver is running and sent by the vehicle terminal device.
  • the cloud server inputs the running state data into a preset artificial intelligence AI diagnosis model, and outputs the first abnormality degree parameter of the motor driver through the AI diagnostic model, wherein the AI diagnostic model is used to extract data features according to the running state data, and pass the data through the AI diagnostic model. Data features for decision inference.
  • the cloud server may use the AI diagnosis model to perform decision inference, and output the first abnormality degree parameter of the motor driver through the AI diagnosis model.
  • the implementation manner of step 303 to step 304 is similar to the implementation manner of step 101 to step 102 in the foregoing embodiment, and details are described in the foregoing description of the embodiment, which is not limited herein.
  • the cloud server generates fault warning information according to the first abnormality degree parameter.
  • the cloud server sends fault warning information to the vehicle terminal device.
  • the cloud server After the cloud server outputs the first anomaly degree parameter through the AI diagnosis model, the cloud server generates fault warning information according to the first anomaly degree parameter, and the fault warning information may be the result of the cloud server's reasoning and decision based on the first anomaly degree parameter.
  • the server sends the fault warning information to the vehicle terminal device, so that the vehicle terminal device can perform fault warning on the motor driver according to the fault warning information.
  • the vehicle terminal device receives the fault warning information from the cloud server.
  • the vehicle terminal equipment performs fault warning on the motor driver according to the fault warning information.
  • the cloud server uses multiple training sample data
  • the multiple training sample data may include training sample data of multiple vehicles of the same type and training sample data of multiple vehicles of different types
  • the resulting AI diagnostic model is an AI global model, which is oriented to specific fault warning tasks.
  • the model is trained based on all collected vehicle terminal equipment side data, and based on this model, the end point of each vehicle terminal equipment is given.
  • the side inference results are sent to the vehicle terminal equipment as a decision-making strategy.
  • the vehicle terminal equipment does not do training and reasoning, but directly executes the inference results of the cloud server. Therefore, the fault warning of the motor driver can be completed through the interaction between the cloud server and the vehicle terminal equipment.
  • the cloud server in the embodiment of the present application has model training capabilities and decision-making inference capabilities, and the vehicle terminal device can perform fault warning according to the decision result of the cloud server. Through the interaction between the cloud server and the vehicle terminal device, the Realize the fault early warning of the motor driver.
  • an embodiment of the present application provides a fault early warning method for a motor driver, which mainly includes the following steps:
  • the cloud server obtains training sample data of multiple vehicles of the same type and training sample data of multiple vehicles of different types, and the motor driver is installed on the vehicle.
  • the cloud server uses the training sample data of multiple vehicles of the same type and the training sample data of multiple vehicles of different types to perform model training based on the fault warning task, so as to obtain a preset AI diagnosis model.
  • the cloud server sends the AI diagnosis model to the vehicle terminal device.
  • the cloud server may send the trained AI diagnosis model to the vehicle terminal device.
  • the vehicle terminal device receives the AI diagnosis model sent by the cloud server.
  • the vehicle terminal device does not have the model training capability, and the vehicle terminal device can receive the AI diagnosis model from the cloud server, so that the vehicle terminal device can use the AI diagnosis model to perform decision inference. Specifically, the vehicle terminal device can execute Subsequent steps 405 to 407 follow.
  • the vehicle terminal device acquires the running state data of the motor driver during running.
  • the vehicle terminal device inputs the running state data into the preset AI diagnosis model, and outputs the first abnormality degree parameter of the motor driver through the AI diagnosis model, wherein the AI diagnosis model is used to extract data features according to the running state data, and pass the data through the data. Features for decision inference.
  • the vehicle terminal device performs a fault warning for the motor driver according to the first abnormality degree parameter.
  • steps 405 to 407 are similar to the implementation manners of steps 101 to 103 in the foregoing embodiments.
  • steps 405 to 407 are similar to the implementation manners of steps 101 to 103 in the foregoing embodiments.
  • the cloud server uses multiple training sample data
  • the multiple training sample data may include training sample data of multiple vehicles of the same type and training sample data of multiple vehicles of different types
  • the resulting AI diagnostic model is an AI global model, which is designed for specific fault warning tasks.
  • the model is trained based on all the collected vehicle terminal equipment side data, and the cloud server sends the trained AI diagnostic model to the vehicle terminal.
  • Equipment, vehicle terminal equipment is based on this model and gives the end-side inference results of each vehicle terminal equipment.
  • the vehicle terminal equipment does not do training, but directly uses the AI diagnostic model to infer and execute the inference results. Therefore, the cloud server and vehicle terminal equipment can be used for inference. interaction to complete the fault early warning of the motor driver.
  • the cloud server has the model training capability
  • the vehicle terminal device has the decision-making inference capability
  • the vehicle terminal device can receive the AI diagnosis model of the cloud server, so that the vehicle terminal device can diagnose the model according to the AI
  • the decision result is output, and the motor driver can be warned of faults.
  • the fault early warning of the motor driver can be realized.
  • an embodiment of the present application provides a fault early warning method for a motor driver, which mainly includes the following steps:
  • the cloud server acquires training sample data of one vehicle or training sample data of multiple vehicles of the same type, and the motor driver is installed on the vehicle.
  • the cloud server can interact with one vehicle terminal device or multiple vehicle terminal devices of the same type, and one or more motor drivers can be installed in each vehicle.
  • the cloud server can obtain multiple training sample data, and the training sample data can be used for model training.
  • the plurality of training sample data may be the training sample data of one vehicle or the training sample data of multiple vehicles of the same type, and the type refers to the type of the vehicle.
  • the data type and data content of the training sample data are not limited.
  • the cloud server uses the training sample data of one vehicle or the training sample data of multiple vehicles of the same type to perform model training based on the fault warning task, so as to obtain a preset AI diagnosis model.
  • the cloud server obtains the training sample data of one vehicle or the training sample data of multiple vehicles of the same type.
  • the cloud server can pre-define the training task as a fault warning task, and the cloud server uses the training sample data of one vehicle or the same type of multiple vehicles.
  • Model training is performed on the training sample data of each vehicle, and the specific training process of the model is not described in detail in the embodiment of the present application, and a preset AI diagnosis model can be obtained after the model training is completed.
  • the cloud server first uses the training sample data of one vehicle for model training, and then generalizes the training sample data of multiple vehicles of the same type. Finally, the training of the model can be completed and the preset AI diagnosis model can be output.
  • the cloud server acquires the running state data of the motor driver when the motor driver is running and sent by the vehicle terminal device.
  • the cloud server inputs the running state data into a preset artificial intelligence AI diagnosis model, and outputs the first abnormality degree parameter of the motor driver through the AI diagnosis model, wherein the AI diagnosis model is used to extract data features according to the running state data, and pass the data through the AI diagnostic model. Data features for decision inference.
  • the cloud server generates fault warning information according to the first abnormality degree parameter.
  • the cloud server sends fault warning information to the vehicle terminal device.
  • the vehicle terminal device receives the fault warning information from the cloud server.
  • the vehicle terminal equipment performs a fault warning on the motor driver according to the fault warning information.
  • steps 503 to 508 are similar to the implementation manners of steps 303 to 308 in the foregoing embodiments.
  • steps 503 to 508 are similar to the implementation manners of steps 303 to 308 in the foregoing embodiments.
  • the cloud server uses a plurality of training sample data, and the plurality of training sample data may include the training sample data of one vehicle or the training sample data of multiple vehicles of the same type, thereby obtaining
  • the AI diagnostic model is an AI partial model, which is oriented to specific fault warning tasks.
  • the AI partial model may be a certain type of vehicle terminal equipment or even a certain vehicle terminal equipment. Based on this model, each terminal is given.
  • the side inference results are sent to the vehicle terminal equipment as a decision-making strategy.
  • the vehicle terminal equipment does not do training and reasoning, but directly executes the inference results of the cloud server. Therefore, the fault warning of the motor driver can be completed through the interaction between the cloud server and the vehicle terminal equipment.
  • the cloud server in the embodiment of the present application has model training capabilities and decision-making inference capabilities, and the vehicle terminal device can perform fault warning according to the decision result of the cloud server. Through the interaction between the cloud server and the vehicle terminal device, the Realize the fault early warning of the motor driver.
  • an embodiment of the present application provides a fault early warning method for a motor driver, which mainly includes the following steps:
  • the cloud server acquires training sample data of multiple vehicles of different types, or training sample data of one vehicle, or training sample data of multiple vehicles of the same type.
  • the cloud server can interact with different types of vehicle terminal devices, or one vehicle terminal device or multiple vehicle terminal devices of the same type, and one or more motor drivers can be installed in each vehicle.
  • the cloud server can obtain multiple training sample data, and the training sample data can be used to extract data associations.
  • the multiple training sample data may be training sample data of multiple different types of vehicles, or training sample data of one vehicle or training sample data of multiple vehicles of the same type, and the type refers to the type of the vehicle.
  • the data type and data content of the training sample data are not limited.
  • the cloud server performs data association extraction on the training sample data of a plurality of vehicles of different types, or the training sample data of one vehicle, or the training sample data of multiple vehicles of the same type, so as to obtain a pre-training model and pre-training Models are used to represent the extracted data associations.
  • the cloud server is not oriented to specific tasks or functions, and extracts data associations from training sample data of multiple different types of vehicles, or extracts data associations from training sample data of one vehicle, or extracts data associations from multiple training sample data of the same type.
  • the data association relationship is extracted from the training sample data of the vehicle to obtain a pre-trained model.
  • the data association relationship may be the relationship between each data dimension and its characteristics in a certain time sequence and working condition in the multiple training sample data.
  • the cloud server sends the pre-training model to the vehicle terminal device.
  • the cloud server may send the pre-training model to the vehicle terminal device.
  • the vehicle terminal device receives the pre-trained model from the cloud server.
  • the vehicle terminal device performs model parameter optimization processing on the pre-trained model according to the preset label data, so as to obtain a preset AI diagnosis model.
  • the cloud server does not need to train the AI diagnostic model, and the cloud server only needs to extract the data association relationship, and after obtaining the pre-training model, send the pre-training model to the vehicle terminal equipment, and the vehicle terminal equipment
  • the label data performs model parameter optimization processing on the pre-trained model.
  • the model parameter optimization process can also be called fine-tuning, that is, the vehicle terminal equipment can perform model training to obtain a preset AI diagnosis model. For example, after the vehicle terminal device receives the pre-training model, the vehicle terminal device fine-tunes the pre-training model according to the fault warning task and combined with the preset label data, so as to realize the fault warning task with low resource consumption.
  • the vehicle terminal device acquires the running state data of the motor driver during running.
  • the vehicle terminal device inputs the running state data into the preset AI diagnosis model, and outputs the first abnormality degree parameter of the motor driver through the AI diagnosis model, wherein the AI diagnosis model is used to extract data features according to the running state data, and pass the data through the data Features for decision inference.
  • the vehicle terminal device performs a fault warning for the motor driver according to the first abnormality degree parameter.
  • steps 606 to 608 are similar to the implementation manners of steps 101 to 103 in the foregoing embodiments. For details, refer to the foregoing descriptions of the embodiments, which are not limited herein.
  • the cloud server uses a plurality of training sample data, and the cloud server can obtain a pre-training model, and deliver the pre-training model to the vehicle terminal device as a training initialization parameter of the vehicle terminal device.
  • the vehicle terminal device performs model parameter optimization based on a small amount of data collected, that is, the process of fine-tuning or transfer learning, so that the AI diagnosis model can be obtained.
  • the device directly uses the AI diagnostic model for inference and executes the inference results, so it can complete the fault warning of the motor driver through the interaction between the cloud server and the vehicle terminal device.
  • the cloud server sends a pre-training model to the vehicle terminal device, the vehicle terminal device generates an AI diagnosis model according to the pre-training model, and the vehicle terminal device has decision-making and reasoning capabilities, so the vehicle terminal device The decision result can be output according to the AI diagnosis model, and the fault warning of the motor driver can be carried out.
  • the fault early warning of the motor driver can be realized.
  • the embodiment of the present application proposes a fault early warning method for a motor driver.
  • the embodiment of the present application is not only applicable to the motor driver, but also to all power electronic components that use power semiconductor modules, such as chargers, DC/DC converters Converter (direct-current/direct-current converter, DC/DC), etc.
  • the technical solutions provided by the embodiments of the present application can realize the intelligent diagnosis of the motor driver at the whole machine level.
  • the motor driver is a power semiconductor device as an example to illustrate.
  • the running state data of the power semiconductor device is collected in the vehicle running state, the data characteristics are judged by the AI diagnosis model, and the abnormality degree of the data characteristics is scored, and the abnormality degree exceeds the standard. Power semiconductor devices are identified to enable advance prediction of individual power semiconductor device failures.
  • FIG. 7 is a schematic structural diagram of a cloud server and a vehicle terminal device according to an embodiment of the present application.
  • the cloud server may also be called a cloud, and the cloud server includes at least one of the following modules: a data processing module, a data storage module, a model training module, and a cloud inference module.
  • the vehicle terminal device may also be called end-side or board-side, and the vehicle terminal device includes at least one of the following modules: a data measurement module, a fine-tuning module, a policy execution module, and an end-side reasoning module.
  • the data measurement module is used to collect training sample data of the power semiconductor device.
  • the data storage module is used to store the massive data reported by the terminal side, which is convenient for subsequent processing.
  • the data processing module is used to clean and process the massive data, which is convenient for the training of the subsequent AI diagnosis model.
  • the model training module is used to perform fault warning or state calculation, and the following examples will be explained with fault warning scenarios.
  • the model training module on the cloud server has the following three implementations:
  • the AI diagnostic model is a general model. All vehicles share the same model, which is oriented to specific tasks or functions, such as fault warning tasks or capacity estimation tasks.
  • the model can be directly used for the result inference of these scenarios. Extract the feature information (such as fault or capacity-related features) in the training sample car, train a model, and the cloud inference module can infer and predict the data of each car on the cloud server based on the model to obtain the inference result, and the inference result ( For example, fault warning, etc.) is used as a strategy, and is sent to the vehicle terminal device, and the policy execution module in the vehicle terminal device performs control and processing according to the strategy of the cloud server.
  • the feature information such as fault or capacity-related features
  • the cloud inference module can infer and predict the data of each car on the cloud server based on the model to obtain the inference result, and the inference result ( For example, fault warning, etc.) is used as a strategy, and is sent to the vehicle terminal device, and the policy execution module in the vehicle terminal device perform
  • the AI diagnostic model is a local model.
  • a class of vehicles shares a model, or a vehicle uses a single model.
  • the local models are similar to the aforementioned general models, and these models are also task-oriented.
  • the cloud server extracts features based on the data of a certain type of vehicle or each vehicle, and trains an AI diagnosis model. Based on these models, the cloud server performs inference and prediction for the corresponding vehicle to obtain inference results, and the inference results (such as fault early warning) are used. etc.) as a policy, and is sent to the vehicle terminal device, and the policy execution module in the vehicle terminal device performs control processing according to the policy of the cloud server.
  • the pre-training model is not oriented to specific tasks or functions. It learns the relationship between various data dimensions and their characteristics on the device side in certain time series and working conditions from massive data to form a pre-training model.
  • the fine-tuning module of the vehicle terminal equipment can fine-tune the pre-trained model in combination with certain labeled data to obtain an AI diagnostic model, which can realize tasks such as fault warning and capacity estimation under low resource consumption.
  • the cloud server stores a large amount of data such as timing, operating conditions, driving behavior, vehicle position, total current, total voltage, temperature, motor status, alarm type and level, and fault status.
  • the AI diagnostic model of this model has both general information learned through big data and its own unique information.
  • the general information refers to the data distribution among various measurement values learned by the cloud server from multiple running vehicles, and the general information reflects the relationship between these measurement data under the condition of big data.
  • the unique information refers to the measurement value of the vehicle terminal equipment. After fine-tuning or transfer learning combined with the general information of the pre-training model, the unique AI diagnosis model on the terminal side can be obtained.
  • the cloud inference module performs the inference function of the cloud server, and provides inference results for tasks such as fault warning or capacity estimation based on the general model or local model trained in the cloud.
  • the fine-tuning module performs end-side fine-tuning (i.e. transfer learning), based on the cloud-based pre-training model and fine-tuning with the data of the vehicle terminal equipment to obtain an AI diagnostic model for specific tasks (such as fault warning tasks).
  • end-side fine-tuning i.e. transfer learning
  • the device-side inference module performs the inference function of the device-side and can support inference in two cases: one is that the general model or local model of the cloud server is lightweight and then sinks to the vehicle terminal device, and the vehicle terminal device is based on the AI sent by the cloud server. Diagnose models for inference. The other is that the vehicle terminal equipment performs inference after obtaining the AI diagnosis model based on the cloud-based pre-training model and fine-tuning (transfer learning).
  • the policy execution module is used to take measures to process the inference results of safety warning (eg, fault warning task) or state estimation (eg, capacity estimation task).
  • safety warning eg, fault warning task
  • state estimation eg, capacity estimation task
  • training reasoning on the cloud and training reasoning on the terminal side may be adopted, or reasoning may be performed only by the cloud, or reasoning may be performed only by the terminal side, and the specific implementation manner is not limited.
  • the default running state data of the motor driver is used as the input of the AI diagnosis model, the abnormal factor score of the motor driver is calculated, and the risk level of the motor driver failure in the future is judged by the abnormal factor score.
  • An early warning signal is issued before a real fault occurs, so that users can detect and repair in advance and avoid damage during driving.
  • the input data of the AI diagnostic model include: bus voltage V_dc, three-phase current I_phase, IGBT temperature T_jav, coolant flow L, coolant temperature T_fluid, and electromagnetic frequency f_em.
  • the input running state data is shown in Table 1 below. These data are the data sampled by default when the MCU is running, and can be directly obtained from the CAN bus. For different vehicles, the specifications of the collected running state data may be different. This is only an example, and there is no limitation. As long as the data content that can be collected on the CAN bus can be used as the input data of the AI diagnostic model.
  • Table 1 is an example of some items of the collected running state data.
  • the test items can be added to correspond to the failure modes to be detected:
  • the AI diagnostic model performs decision inference according to the above input data, and outputs the first abnormality degree parameter.
  • decision inference process of the AI diagnostic model please refer to the description of the model inference process shown in Figure 7 above, which will not be repeated here.
  • additional sampling sensors or circuits are used in the motor driver to extract specific key data as the input data of the AI diagnostic model, and to calculate the abnormal factor score of the power semiconductor device. Determine the risk level of power semiconductor devices in the future, and issue early warning signals for products with high risk levels before the actual failure occurs, so that users can detect and repair in advance to avoid damage during driving.
  • the input data of the AI diagnosis model may include: breakdown voltage Brv_ce, CE leakage current I_ces, GE leakage current I_ges, threshold voltage Vth, IGBT saturation voltage drop V_cesat , diode conduction voltage drop V_f, IGBT parasitic capacitance Cies, Coes, Cres, module switching loss Eon, Eoff, Err.
  • breakdown voltage Brv_ce CE leakage current I_ces
  • GE leakage current I_ges threshold voltage Vth
  • IGBT saturation voltage drop V_cesat diode conduction voltage drop V_f
  • IGBT parasitic capacitance Cies Coes, Cres, module switching loss Eon, Eoff, Err.
  • test circuit used in the embodiments of the present application has various implementation manners.
  • the test circuit may be an Iges test circuit, a Vth test circuit, a Bvces and Ices test circuit, an Ices low-end test circuit, and a Vcesat and Vf test circuit.
  • the types of test circuits are not limited in the embodiments of the present application.
  • the input data of the AI diagnostic model are shown in Tables 2 and 3 below for IGBT and MOS power semiconductor devices respectively. These data are not the data sampled by default when the MCU is running. Additional sensors or sampling circuits can be added to the MCU to achieve data collection.
  • Table 2 shows the test items of the IGBT module, the real test items can be added, and the test items correspond to the failure modes to be detected:
  • Table 3 shows some test items of MOSFET, the real items can be added, and the test items correspond to the failure modes to be detected:
  • FIG. 10 a schematic flowchart of a collection process of running state data and static data provided by an embodiment of the present application.
  • the driving data of the power semiconductor device is dynamically sampled in the running state of the vehicle, and the decision inference based on the AI diagnostic model is performed on the cloud platform.
  • the data acquisition process in this embodiment requires the power semiconductor device to be completed under operating conditions, that is, the power semiconductor device is performed under the condition that the entire power semiconductor device is in a high-frequency switching operating mode. For example, key data sampling is performed during the operation phase of the vehicle.
  • the key data sampling in the stages of car charging, car ignition, and car shutdown mainly includes the following two stages: In the first stage, the collected data is transferred to the cloud platform, and the big data-based AI diagnosis model is carried out on the cloud platform. In the second stage, the cloud platform sends the inference results to the vehicle terminal equipment, and the vehicle terminal equipment performs data preprocessing and compression, and provides emergency response functions in the event of a network disconnection.
  • the data acquisition work in this embodiment requires the power semiconductor to be performed under static conditions, that is, the power semiconductor device is not in the low-frequency switching operating mode, such as in the stage of car charging, car ignition, and car ignition, to carry out the above key data. sampling.
  • the cloud and/or the terminal side have the training and reasoning capabilities of the AI diagnostic model.
  • the AI diagnostic model is used to determine the data features, and the abnormality score is performed on the features, so as to identify the motor driver whose abnormality exceeds the standard, so as to realize the advance prediction of the failure of the motor driver. Diagnose the failure of the electric drive system of the car due to the motor driver in advance and give an early warning, so that the user can perform maintenance before the failure occurs and avoid breaking down during driving.
  • the terminal has local training and decision-making inference capabilities under the constraints of limited computing power, and the accuracy of models such as fault prediction, as well as the real-time and reliability of inference have been improved.
  • a fault early warning device 1100 for a motor driver may include: an acquisition module 1101, a reasoning module 1102, and an early warning module 1103, wherein,
  • the acquisition module is used to acquire the running state data of the motor driver when it is running;
  • the reasoning module is used to input the running state data into a preset artificial intelligence AI diagnosis model, and output the first abnormality degree parameter of the motor driver through the AI diagnosis model, wherein the AI diagnosis model is used for according to Extracting data features from the running state data, and performing decision inference based on the data features;
  • An early warning module configured to give a fault early warning to the motor driver according to the first abnormality degree parameter.
  • the acquisition module is further configured to acquire static data obtained by performing low-frequency sampling on the motor driver when the vehicle is in a stop phase, and the motor driver is installed on the vehicle;
  • the reasoning module is further configured to input the static data into the AI diagnosis model, and output the second abnormality parameter of the motor driver through the AI diagnosis model;
  • the early warning module is further configured to perform a fault early warning on the motor driver according to the second abnormality degree parameter.
  • the vehicle is in a stop phase, including at least one of the following:
  • the vehicle is in the charging phase, the vehicle is in the ignition phase, and the vehicle is in the ignition phase.
  • the early warning module is configured to dynamically update the preset abnormal conditions through the AI diagnostic model, so as to obtain the updated abnormal conditions; when the first abnormality degree parameter satisfies all When the updated abnormal condition is detected, a fault warning is performed on the motor driver.
  • the apparatus further includes: a data processing module 1104, configured to process the running state data on the running state data after the obtaining module obtains the running state data of the motor driver during operation. Cleaning process to obtain running state data after cleaning process;
  • the reasoning module is used for inputting the running state data after the cleaning process into the AI diagnosis model.
  • the apparatus is specifically a cloud server; as shown in FIG. 11c, the apparatus further includes: a training module 1105,
  • the acquisition module is further configured to acquire training sample data of multiple vehicles of the same type and training sample data of multiple vehicles of different types, and the motor driver is installed on the vehicles;
  • the training module is configured to use the training sample data of the multiple vehicles of the same type and the training sample data of the multiple vehicles of different types to perform model training based on the fault warning task, so as to obtain the preset AI diagnostic model.
  • the apparatus is specifically a cloud server; the apparatus further includes: a training module,
  • the acquisition module is further configured to acquire training sample data of one vehicle or training sample data of multiple vehicles of the same type, and the motor driver is installed on the vehicle;
  • the training module is configured to use the training sample data of the one vehicle or the training sample data of multiple vehicles of the same type to perform model training based on a fault warning task to obtain the preset AI diagnosis model.
  • the device is specifically a cloud server; the early warning module is configured to generate fault early warning information according to the first abnormality degree parameter; and send the fault early warning information to a vehicle terminal device.
  • the apparatus is specifically a vehicle terminal device
  • the obtaining module is further configured to receive the AI diagnosis model sent by the cloud server.
  • the apparatus is specifically a cloud server; the apparatus further includes: a training module,
  • the obtaining module is further configured to obtain training sample data of a plurality of vehicles of different types, or training sample data of one vehicle, or training sample data of multiple vehicles of the same type;
  • the training module is used to extract the data association relationship on the training sample data of the plurality of vehicles of different types, or the training sample data of one vehicle, or the training sample data of multiple vehicles of the same type, so as to obtain a pre-determined model. Training a model, where the pre-training model is used to represent the extracted data association; sending the pre-training model to the vehicle terminal device.
  • the apparatus is specifically a vehicle terminal device; the apparatus further includes: a training module,
  • the training module is used to receive the pre-training model from the cloud server, wherein the pre-training module is used to represent the training sample data of the cloud server for a plurality of vehicles of different types, or the training sample data of one vehicle, Or a data association relationship obtained by extracting training sample data of multiple vehicles of the same type; performing model parameter optimization processing on the pre-trained model according to preset label data to obtain the preset AI diagnostic model.
  • the running state data includes: dynamic data collected by default when the motor driver is running; or,
  • the running state data includes: dynamic data collected through the CAN bus of the controller area network.
  • the running state data includes: dynamic data obtained by high-frequency sampling when the vehicle is in a running phase, and the motor driver is installed on the vehicle.
  • the operating state data includes at least one of the following: parameters of the motor driver, parameters of the cooling system, parameters of the motor, and parameters of the battery.
  • the parameters of the motor driver include at least one of the following: DC bus voltage, DC bus current, low voltage power supply voltage, vehicle driving status command, vehicle status, motor driver temperature, rotor position electrical angle, The rms value of the three-phase current of the motor driver, the three-phase current of the motor driver, the sampled value of the three-phase current of the motor driver, the derating status of the motor driver; and/or,
  • the parameters of the cooling system include at least one of the following: coolant flow rate, coolant temperature, oil pump target rotational speed, oil pump state, oil pump actual rotational speed, oil pump supply voltage, oil pump power semiconductor device temperature, oil pump current; and/or,
  • the parameters of the motor include at least one of the following: electromagnetic frequency, motor working mode command, motor controller working state, motor target torque command, motor current torque, motor target speed command, motor current speed, motor temperature, motor direct Shaft voltage, motor direct axis current reference value, motor direct axis current feedback value, motor quadrature axis voltage, motor quadrature axis current reference value, motor quadrature axis current feedback value, motor current calibration torque; and/or,
  • the parameters of the battery include at least one of the following: rated output voltage, battery capacity, and maximum output current.
  • the running state data of the motor driver during operation is obtained first, and then the running state data is input into the preset AI diagnosis model, and the first abnormality parameter of the motor driver is output through the AI diagnosis model, wherein,
  • the AI diagnostic model is used to extract data features according to the running state data, and perform decision-making inference based on the data features; according to the first abnormality degree parameter, the motor driver is warned of faults. Since the vehicle is in a running state when the motor driver is running in the embodiment of the present application, a fault warning can be given to the motor driver according to the running state data when the motor driver is running.
  • the running state data of the motor driver is the data collected when the motor driver is running. The data can reflect the real operating state of the motor driver. Based on the operating state data, fault warning can be performed, which can realize the advance prediction of the failure of the motor driver and realize the fault early warning of the motor driver.
  • the embodiment of the present application also provides a fault early warning device for a motor driver, the device comprising:
  • an acquisition module configured to acquire static data obtained by performing low-frequency sampling on the motor driver when the vehicle is in a stop phase, and the motor driver is installed on the vehicle;
  • an inference module configured to input the static data into the AI diagnosis model, and output the second abnormality parameter of the motor driver through the AI diagnosis model;
  • an early warning module configured to perform a fault early warning on the motor driver according to the second abnormality degree parameter.
  • the vehicle is in a stop phase, including at least one of the following:
  • the vehicle is in the charging phase, the vehicle is in the ignition phase, and the vehicle is in the ignition phase.
  • the early warning module is configured to dynamically update the preset abnormal condition through the AI diagnostic model, so as to obtain the updated abnormal condition; when the second abnormality degree parameter satisfies all When the updated abnormal condition is detected, a fault warning is performed on the motor driver.
  • the apparatus further includes: a data processing module, configured to: after the acquisition module acquires static data obtained by performing low-frequency sampling on the motor driver when the vehicle is in a stop phase, Perform cleaning to obtain static data after cleaning;
  • the reasoning module is used for inputting the cleaned static data into the AI diagnosis model.
  • the apparatus is specifically a cloud server; the apparatus further includes: a training module,
  • the acquisition module is further configured to acquire training sample data of multiple vehicles of the same type and training sample data of multiple vehicles of different types, and the motor driver is installed on the vehicles;
  • the training module is configured to use the training sample data of the multiple vehicles of the same type and the training sample data of the multiple vehicles of different types to perform model training based on the fault warning task, so as to obtain the preset AI diagnostic model.
  • the apparatus is specifically a cloud server; the apparatus further includes: a training module,
  • the acquisition module is further configured to acquire training sample data of one vehicle or training sample data of multiple vehicles of the same type, and the motor driver is installed on the vehicle;
  • the training module is configured to use the training sample data of the one vehicle or the training sample data of multiple vehicles of the same type to perform model training based on a fault warning task to obtain the preset AI diagnosis model.
  • the apparatus is specifically a cloud server; the early warning module is configured to generate fault early warning information according to the second abnormality degree parameter; and send the fault early warning information to the vehicle terminal device.
  • the apparatus is specifically a vehicle terminal device
  • the obtaining module is further configured to receive the AI diagnosis model sent by the cloud server.
  • the apparatus is specifically a cloud server; the apparatus further includes: a training module,
  • the obtaining module is further configured to obtain training sample data of a plurality of vehicles of different types, or training sample data of one vehicle, or training sample data of multiple vehicles of the same type;
  • the training module is used to extract the data association relationship on the training sample data of the plurality of vehicles of different types, or the training sample data of one vehicle, or the training sample data of multiple vehicles of the same type, so as to obtain a pre-determined model. Training a model, where the pre-training model is used to represent the extracted data association; sending the pre-training model to the vehicle terminal device.
  • the apparatus is specifically a vehicle terminal device; the apparatus further includes: a training module,
  • the training module is used to receive the pre-training model from the cloud server, wherein the pre-training module is used to represent the training sample data of the cloud server for a plurality of vehicles of different types, or the training sample data of one vehicle, Or a data association relationship obtained by extracting training sample data of multiple vehicles of the same type; performing model parameter optimization processing on the pre-trained model according to preset label data to obtain the preset AI diagnostic model.
  • Embodiments of the present application further provide a computer storage medium, wherein the computer storage medium stores a program, and the program executes some or all of the steps described in the above method embodiments.
  • the fault early warning device 1200 for a motor driver includes:
  • the receiver 1201, the transmitter 1202, the processor 1203 and the memory 1204 (wherein the number of the processors 1203 in the fault early warning device 1200 for the motor driver can be one or more, and one processor is taken as an example in FIG. 12).
  • the receiver 1201 , the transmitter 1202 , the processor 1203 , and the memory 1204 may be connected by a bus or in other ways, wherein the connection by a bus is taken as an example in FIG. 12 .
  • Memory 1204 may include read-only memory and random access memory, and provides instructions and data to processor 1203 .
  • a portion of memory 1204 may also include non-volatile random access memory (NVRAM).
  • NVRAM non-volatile random access memory
  • the memory 1204 stores operating system and operation instructions, executable modules or data structures, or a subset thereof, or an extended set thereof, wherein the operation instructions may include various operation instructions for implementing various operations.
  • the operating system may include various system programs for implementing various basic services and handling hardware-based tasks.
  • the processor 1203 controls the operation of the fault warning device for the motor driver, and the processor 1203 may also be referred to as a central processing unit (CPU).
  • the components of the fault early warning device for a motor driver are coupled together through a bus system, where the bus system may include a power bus, a control bus, and a status signal bus in addition to a data bus.
  • the various buses are referred to as bus systems in the figures.
  • the methods disclosed in the above embodiments of the present application may be applied to the processor 1203 or implemented by the processor 1203 .
  • the processor 1203 may be an integrated circuit chip, which has signal processing capability. In the implementation process, each step of the above-mentioned method can be completed by an integrated logic circuit of hardware in the processor 1203 or an instruction in the form of software.
  • the above-mentioned processor 1203 may be a general-purpose processor, a digital signal processor (digital signal processing, DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (field-programmable gate array, FPGA) or Other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
  • DSP digital signal processing
  • ASIC application specific integrated circuit
  • FPGA field-programmable gate array
  • Other programmable logic devices discrete gate or transistor logic devices, discrete hardware components.
  • a general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
  • the steps of the method disclosed in conjunction with the embodiments of the present application may be directly embodied as executed by a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor.
  • the software modules may be located in random access memory, flash memory, read-only memory, programmable read-only memory or electrically erasable programmable memory, registers and other storage media mature in the art.
  • the storage medium is located in the memory 1204, and the processor 1203 reads the information in the memory 1204, and completes the steps of the above method in combination with its hardware.
  • the receiver 1201 can be used to receive input digital or character information, and to generate signal input related to the relevant settings and function control of the fault warning device for the motor driver.
  • the transmitter 1202 can include a display device such as a display screen, and the transmitter 1202 can It is used to output digital or character information through an external interface.
  • the processor 1203 is configured to execute the method steps shown in FIG. 1 to FIG. 6 as follows.
  • the chip when the fault warning device for the motor driver is a chip in the terminal, the chip includes: a processing unit and a communication unit, the processing unit may be, for example, a processor, and the communication unit may be, for example, a It is an input/output interface, pin or circuit, etc.
  • the processing unit can execute the computer-executed instructions stored in the storage unit, so that the chip in the terminal executes the method of any one of the above-mentioned first aspect.
  • the storage unit is a storage unit in the chip, such as a register, a cache, etc., and the storage unit may also be a storage unit in the terminal located outside the chip, such as a read-only memory (read only memory). -only memory, ROM) or other types of static storage devices that can store static information and instructions, random access memory (RAM), etc.
  • the processor mentioned in any one of the above may be a general-purpose central processing unit, a microprocessor, an ASIC, or one or more integrated circuits for controlling the execution of the program of the method in the first aspect.
  • the device embodiments described above are only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be A physical unit, which can be located in one place or distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution in this embodiment.
  • the connection relationship between the modules indicates that there is a communication connection between them, which can be specifically implemented as one or more communication buses or signal lines.
  • U disk mobile hard disk
  • ROM read-only memory
  • RAM magnetic disk or optical disk
  • a computer device which may be a personal computer, server, or network device, etc.
  • the computer program product includes one or more computer instructions.
  • the computer may be a general purpose computer, special purpose computer, computer network, or other programmable device.
  • the computer instructions may be stored in or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be downloaded from a website site, computer, server, or data center Transmission to another website site, computer, server, or data center is by wire (eg, coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (eg, infrared, wireless, microwave, etc.).
  • wire eg, coaxial cable, fiber optic, digital subscriber line (DSL)
  • wireless eg, infrared, wireless, microwave, etc.
  • the computer-readable storage medium may be any available medium that can be stored by a computer, or a data storage device such as a server, data center, etc., which includes one or more available media integrated.
  • the usable media may be magnetic media (eg, floppy disks, hard disks, magnetic tapes), optical media (eg, DVD), or semiconductor media (eg, Solid State Disk (SSD)), and the like.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Software Systems (AREA)
  • Medical Informatics (AREA)
  • Evolutionary Computation (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Power Engineering (AREA)
  • Testing And Monitoring For Control Systems (AREA)

Abstract

一种电机驱动器的故障预警方法和装置，用于实现电机驱动器的故障预警。本申请实施例提供一种用于电机驱动器的故障预警方法，包括：获取电机驱动器运行时的运行态数据；将所述运行态数据输入到预设的人工智能AI诊断模型，通过所述AI诊断模型输出所述电机驱动器的第一异常度参数，其中，所述AI诊断模型用于根据所述运行态数据提取数据特征，并通过所述数据特征进行决策推理；根据所述第一异常度参数对所述电机驱动器进行故障预警。

Description

一种用于电机驱动器的故障预警方法和装置 技术领域
本申请实施例涉及电力电子功率变换领域，尤其涉及一种用于电机驱动器的故障预警方法和装置。
背景技术
电机驱动器用于对电机的驱动、怠速和制动控制。电机驱动器可以包括功率半导体模组，功率半导体模组是电动汽车动力总成的核心器件，同时也是容易失效的部件。
在目前的对电机驱动器的故障诊断方案中，主要通过监测电机驱动器的温度、温度幅值变化、循环数等参数，利用寿命模型对电机驱动器进行寿命评估，该寿命模型可以包括：Coffin-Manson模型，Norris-Landzberg模型及Bayerer模型。
上述故障诊断方案中，寿命模型的输入参数通过对少量的电机驱动器进行寿命测试之后得到，该寿命模型只能表征统计意义上的整体情况，对于每一个电机驱动器的个体没有实用价值。因此，上述的寿命模型通用性较差且无法针对单个电机驱动器进行故障诊断。
发明内容
本申请实施例提供了一种用于电机驱动器的故障预警方法和装置，用于实现电机驱动器的故障预警。
为解决上述技术问题，本申请实施例提供以下技术方案：
第一方面，本申请实施例提供一种用于电机驱动器的故障预警方法，包括：
获取电机驱动器运行时的运行态数据；
将所述运行态数据输入到预设的人工智能AI诊断模型，通过所述AI诊断模型输出所述电机驱动器的第一异常度参数，其中，所述AI诊断模型用于根据所述运行态数据提取数据特征，并通过所述数据特征进行决策推理；
根据所述第一异常度参数对所述电机驱动器进行故障预警。
在该方案中，由于本申请实施例中电机驱动器运行时车辆处于运行状态，根据电机驱动器运行时的运行态数据可以对电机驱动器进行故障预警，电机驱动器的运行态数据是电机驱动器运行时采集到的数据，运行态数据能够反映电机驱动器的真实运行状态，基于该运行态数据进行故障预警，可以实现对电机驱动器失效的提前预测，实现对电机驱动器的故障预警。
在一种可能的实现方式中，所述方法还包括：
获取车辆处于停止阶段时对所述电机驱动器进行低频采样得到的静态数据，所述电机驱动器安装在所述车辆上；
将所述静态数据输入到所述AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第二异常度参数；
根据所述第二异常度参数对所述电机驱动器进行故障预警。
在该方案中，本申请实施例中车辆处于停止阶段时获取电机驱动器的静态数据，根据 该静态数据可以对电机驱动器进行故障预警，电机驱动器的静态数据是车辆停止时对电机驱动器采集到的数据，静态数据能够反映电机驱动器的真实静止状态，基于该静态数据进行故障预警，可以实现对电机驱动器失效的提前预测，实现对电机驱动器的故障预警。
在一种可能的实现方式中，所述车辆处于停止阶段，至少包括如下一种：
所述车辆处于充电阶段、所述车辆处于打火阶段、所述车辆处于熄火阶段。
在该方案中，在车辆充电阶段、车辆打火阶段、车辆熄火阶段时进行静态数据的采样，从而故障诊断装置可以根据该静态数据提供断网情况下的紧急响应功能。
在一种可能的实现方式中，所述根据所述第一异常度参数对所述电机驱动器进行故障预警，包括：
通过所述AI诊断模型对预设的异常条件进行动态更新，以得到更新后的异常条件；
当所述第一异常度参数满足所述更新后的异常条件时，对所述电机驱动器进行故障预警。
在该方案中，可以预设异常条件，并且该异常条件可以通过AI诊断模型进行动态更新，以保证该异常条件能够用于故障判断，保证故障判断的准确率。在对异常条件进行动态更新之后，可以判断第一异常度是否满足动态更新后的异常条件，当第一异常度参数满足更新后的异常条件时，说明该电机驱动器可能存在故障，此时对电机驱动器进行故障预警。本申请实施例中通过对异常条件进行动态更新，可以进一步提高故障预警的准确度。
在一种可能的实现方式中，所述获取电机驱动器运行时的运行态数据之后，所述方法还包括：
对所述运行态数据进行清洗处理，得到清洗处理后的运行态数据；
所述将所述运行态数据输入到预设的人工智能AI诊断模型包括：
将所述清洗处理后的运行态数据输入到所述AI诊断模型。
在该方案中，在获取到多个运行态数据之后，可以先对该运行态数据进行清洗处理，例如可以先对运行态数据中符合清洗条件的数据进行剔除。该清洗条件可以是运行态数据不完整，或者运行态数据的某项指标超标等。获取到清洗处理后的运行态数据之后，将清洗处理后的运行态数据输入到AI诊断模型中，通过AI诊断模型对清洗处理后的运行态数据进行数据特征的提取，通过对运行态数据进行清洗处理之后再输入到AI诊断模型，可以提高AI诊断模型的推理效率。
在一种可能的实现方式中，所述方法还包括：
云端服务器获取同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
所述云端服务器使用所述同类型的多个车辆的训练样本数据和所述多个不同类型的车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
在该方案中，云端服务器使用了多个训练样本数据，该多个训练样本数据可以包括同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，从而得到的AI诊断模型是AI全局模型，该模型是面向具体的故障预警任务进行的，该模型基于所有采集的车辆终端设备侧数据训练，基于该模型并给出每个车辆终端设备的端侧推理结果。
在一种可能的实现方式中，所述方法还包括：
云端服务器获取一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
所述云端服务器使用所述一个车辆的训练样本数据或者所述同类型的多个车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
在该方案中，云端服务器使用了多个训练样本数据，该多个训练样本数据可以包括一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，从而得到的AI诊断模型是AI局部模型，该模型是面向具体的故障预警任务进行的，该AI局部模型可能是某个型号的车辆终端设备甚至是某个车辆终端设备，基于该模型并给出每个端侧推理结果。
在一种可能的实现方式中，当云端服务器通过所述AI诊断模型输出所述电机驱动器的第一异常度参数时，所述根据所述第一异常度参数对所述电机驱动器进行故障预警，包括：
所述云端服务器根据所述第一异常度参数生成故障预警信息；
所述云端服务器向车辆终端设备发送所述故障预警信息。
在该方案中，本申请实施例中云端服务器具有模型训练能力和决策推理能力，车辆终端设备可以按照云端服务器的决策结果进行故障预警，通过云端服务器和车辆终端设备的交互，可以实现对电机驱动器的故障预警。
在一种可能的实现方式中，当云端服务器向车辆终端设备发送所述AI诊断模型时，所述方法还包括：所述车辆终端设备接收所述云端服务器发送的AI诊断模型；
所述将所述运行态数据输入到预设的人工智能AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第一异常度参数，包括：
所述车辆终端设备将所述运行态数据输入到所述AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第一异常度参数。
在该方案中，本申请实施例中云端服务器具有模型训练能力，车辆终端设备具有决策推理能力，车辆终端设备可以接收云端服务器的AI诊断模型，从而车辆终端设备可以根据AI诊断模型输出决策结果，对电机驱动器进行故障预警，通过云端服务器和车辆终端设备的交互，可以实现对电机驱动器的故障预警。
在一种可能的实现方式中，所述方法还包括：
云端服务器获取多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据；
所述云端服务器对所述多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据，进行数据关联关系提取，以得到预训练模型，所述预训练模型用于表示提取到的数据关联关系；
所述云端服务器向车辆终端设备发送所述预训练模型。
在该方案中，云端服务器不需要进行AI诊断模型的训练，云端服务器只需要提取数据关联关系，得到预训练模型之后，向车辆终端设备发送该预训练模型。
在一种可能的实现方式中，所述方法，还包括：
所述车辆终端设备接收来自所述云端服务器的预训练模型；
所述车辆终端设备根据预设的标签数据对所述预训练模型进行模型参数寻优处理，以得到所述预设的AI诊断模型。
在该方案中，车辆终端设备根据预设的标签数据对预训练模型进行模型参数寻优处理，模型参数寻优处理也可以称为微调，即车辆终端设备可以进行模型训练，以得到预设的AI诊断模型。
在一种可能的实现方式中，所述运行态数据包括：所述电机驱动器运行时默认采集的动态数据；或者，
所述运行态数据包括：通过控制器局域网络CAN总线采集的动态数据。
在一种可能的实现方式中，所述运行态数据包括：车辆处于运行阶段进行高频采样得到的动态数据，所述电机驱动器安装在所述车辆上。
在该方案中，车辆具有多种状态，例如车辆处于运行阶段，该运行阶段是指车辆在打火启动之后的阶段，例如运行阶段是指车辆处于行驶阶段。在车辆处于运行阶段时进行高频采样得到的动态数据可以是前述的运行态数据，后续实施例中对于高频采样的数据项以及数据量进行举例说明。其中，高频采样是指数据采集的频率大于预设的频率阈值。
在一种可能的实现方式中，所述运行态数据至少包括如下一种：电机驱动器的参数、冷却系统的参数、电机的参数、电池的参数。
在该方案中，本申请实施例中电机驱动器运行时的运行态数据具有多种实现方式，例如在电机驱动器运行时可以针对电机驱动器进行数据采集以得到电机驱动器的参数，或者在电机驱动器运行时可以针对冷却系统进行数据采集以得到冷却系统的参数，或者在电机驱动器运行时可以针对电机进行数据采集以得到电机的参数，或者在电机驱动器运行时可以针对电池进行数据采集以电池的参数。
在一种可能的实现方式中，所述电机驱动器的参数包括如下至少一种：直流母线电压、直流母线电流、低压电源电压、车辆行驶状态命令、车辆状态、电机驱动器温度、转子位置电角度、电机驱动器三相电流有效值、电机驱动器三相电流、电机驱动器三相电流采样值、电机驱动器降额状态；和/或，
所述冷却系统的参数包括如下至少一种：冷却液流量、冷却液温度、油泵目标转速、油泵状态、油泵实际转速、油泵供电电压、油泵的功率半导体器件温度、油泵电流；和/或，
所述电机的参数包括如下至少一种：电磁频率、电机工作模式命令、电机控制器工作状态、电机目标转矩命令、电机当前转矩、电机目标转速命令、电机当前转速、电机温度、电机直轴电压、电机直轴电流给定值、电机直轴电流反馈值、电机交轴电压、电机交轴电流给定值、电机交轴电流反馈值、电机当前校验转矩；和/或，
所述电池的参数包括如下至少一种：额定输出电压、电池容量、最大输出电流。
第二方面，本申请实施例还提供一种用于电机驱动器的故障预警装置，其特征在于，包括：
获取模块，用于获取电机驱动器运行时的运行态数据；
推理模块，用于将所述运行态数据输入到预设的人工智能AI诊断模型，通过所述AI诊断模型输出所述电机驱动器的第一异常度参数，其中，所述AI诊断模型用于根据所述运行态数据提取数据特征，并通过所述数据特征进行决策推理；
预警模块，用于根据所述第一异常度参数对所述电机驱动器进行故障预警。
在一种可能的实现方式中，所述获取模块，还用于获取车辆处于停止阶段时对所述电机驱动器进行低频采样得到的静态数据，所述电机驱动器安装在所述车辆上；
所述推理模块，还用于将所述静态数据输入到所述AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第二异常度参数；
所述预警模块，还用于根据所述第二异常度参数对所述电机驱动器进行故障预警。
在一种可能的实现方式中，所述车辆处于停止阶段，至少包括如下一种：
所述车辆处于充电阶段、所述车辆处于打火阶段、所述车辆处于熄火阶段。
在一种可能的实现方式中，所述预警模块，用于通过所述AI诊断模型对预设的异常条件进行动态更新，以得到更新后的异常条件；当所述第一异常度参数满足所述更新后的异常条件时，对所述电机驱动器进行故障预警。
在一种可能的实现方式中，所述装置还包括：数据处理模块，用于所述获取模块获取电机驱动器运行时的运行态数据之后，对所述运行态数据进行清洗处理，得到清洗处理后的运行态数据；
所述推理模块，用于将所述清洗处理后的运行态数据输入到所述AI诊断模型。
在一种可能的实现方式中，所述装置具体为云端服务器；所述装置还包括：训练模块，
所述获取模块，还用于获取同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
所述训练模块，用于使用所述同类型的多个车辆的训练样本数据和所述多个不同类型的车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
在一种可能的实现方式中，所述装置具体为云端服务器；所述装置还包括：训练模块，
所述获取模块，还用于获取一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
所述训练模块，用于使用所述一个车辆的训练样本数据或者所述同类型的多个车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
在一种可能的实现方式中，所述装置具体为云端服务器；所述预警模块，用于根据所述第一异常度参数生成故障预警信息；向车辆终端设备发送所述故障预警信息。
在一种可能的实现方式中，所述装置具体为车辆终端设备；
当云端服务器向所述车辆终端设备发送所述AI诊断模型时，所述获取模块，还用于接收所述云端服务器发送的AI诊断模型。
在一种可能的实现方式中，所述装置具体为云端服务器；所述装置还包括：训练模块，
所述获取模块，还用于获取多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据；
所述训练模块，用于对所述多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据，进行数据关联关系提取，以得到预训练模型，所述预训练模型用于表示提取到的数据关联关系；向车辆终端设备发送所述预训练模型。
在一种可能的实现方式中，所述装置具体为车辆终端设备；所述装置还包括：训练模 块，
所述训练模块，用于接收来自云端服务器的预训练模型，其中，所述预训练模块用于表示所述云端服务器对多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据进行提取后得到的数据关联关系；根据预设的标签数据对所述预训练模型进行模型参数寻优处理，以得到所述预设的AI诊断模型。
在一种可能的实现方式中，所述运行态数据包括：所述电机驱动器运行时默认采集的动态数据；或者，
所述运行态数据包括：通过控制器局域网络CAN总线采集的动态数据。
在一种可能的实现方式中，所述运行态数据包括：车辆处于运行阶段进行高频采样得到的动态数据，所述电机驱动器安装在所述车辆上。
在一种可能的实现方式中，所述运行态数据至少包括如下一种：电机驱动器的参数、冷却系统的参数、电机的参数、电池的参数。
在一种可能的实现方式中，所述电机驱动器的参数包括如下至少一种：直流母线电压、直流母线电流、低压电源电压、车辆行驶状态命令、车辆状态、电机驱动器温度、转子位置电角度、电机驱动器三相电流有效值、电机驱动器三相电流、电机驱动器三相电流采样值、电机驱动器降额状态；和/或，
所述冷却系统的参数包括如下至少一种：冷却液流量、冷却液温度、油泵目标转速、油泵状态、油泵实际转速、油泵供电电压、油泵的功率半导体器件温度、油泵电流；和/或，
所述电机的参数包括如下至少一种：电磁频率、电机工作模式命令、电机控制器工作状态、电机目标转矩命令、电机当前转矩、电机目标转速命令、电机当前转速、电机温度、电机直轴电压、电机直轴电流给定值、电机直轴电流反馈值、电机交轴电压、电机交轴电流给定值、电机交轴电流反馈值、电机当前校验转矩；和/或，
所述电池的参数包括如下至少一种：额定输出电压、电池容量、最大输出电流。
在本申请的第二方面中，用于电机驱动器的故障预警装置的组成模块还可以执行前述第一方面以及各种可能的实现方式中所描述的步骤，详见前述对第一方面以及各种可能的实现方式中的说明。
第三方面，本申请实施例还提供一种用于电机驱动器的故障预警方法，所述方法包括：
获取车辆处于停止阶段时对所述电机驱动器进行低频采样得到的静态数据，所述电机驱动器安装在所述车辆上；
将所述静态数据输入到所述AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第二异常度参数；
根据所述第二异常度参数对所述电机驱动器进行故障预警。
在一种可能的实现方式中，所述车辆处于停止阶段，至少包括如下一种：
所述车辆处于充电阶段、所述车辆处于打火阶段、所述车辆处于熄火阶段。
在一种可能的实现方式中，所述根据所述第二异常度参数对所述电机驱动器进行故障预警，包括：
通过所述AI诊断模型对预设的异常条件进行动态更新，以得到更新后的异常条件；
当所述第二异常度参数满足所述更新后的异常条件时，对所述电机驱动器进行故障预警。
在一种可能的实现方式中，所述获取车辆处于停止阶段时对所述电机驱动器进行低频采样得到的静态数据之后，所述方法还包括：
对所述静态数据进行清洗处理，得到清洗处理后的静态数据；
所述将所述静态数据输入到预设的人工智能AI诊断模型包括：
将所述清洗处理后的静态数据输入到所述AI诊断模型。
在一种可能的实现方式中，所述方法还包括：
云端服务器获取同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
所述云端服务器使用所述同类型的多个车辆的训练样本数据和所述多个不同类型的车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
在一种可能的实现方式中，所述方法还包括：
云端服务器获取一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
所述云端服务器使用所述一个车辆的训练样本数据或者所述同类型的多个车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
在一种可能的实现方式中，当云端服务器通过所述AI诊断模型输出所述电机驱动器的第一异常度参数时，所述根据所述第二异常度参数对所述电机驱动器进行故障预警，包括：
所述云端服务器根据所述第二异常度参数生成故障预警信息；
所述云端服务器向车辆终端设备发送所述故障预警信息。
在一种可能的实现方式中，当云端服务器向车辆终端设备发送所述AI诊断模型时，所述方法还包括：所述车辆终端设备接收所述云端服务器发送的AI诊断模型；
所述将所述静态数据输入到预设的人工智能AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第二异常度参数，包括：
所述车辆终端设备将所述静态数据输入到所述AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第二异常度参数。
在一种可能的实现方式中，所述方法还包括：
云端服务器获取多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据；
所述云端服务器对所述多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据，进行数据关联关系提取，以得到预训练模型，所述预训练模型用于表示提取到的数据关联关系；
所述云端服务器向车辆终端设备发送所述预训练模型。
在一种可能的实现方式中，所述方法，还包括：
所述车辆终端设备接收来自所述云端服务器的预训练模型；
所述车辆终端设备根据预设的标签数据对所述预训练模型进行模型参数寻优处理，以得到所述预设的AI诊断模型。
第四方面，本申请实施例还提供一种用于电机驱动器的故障预警装置，所述装置包括：
获取模块，用于获取车辆处于停止阶段时对所述电机驱动器进行低频采样得到的静态数据，所述电机驱动器安装在所述车辆上；
推理模块，用于将所述静态数据输入到所述AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第二异常度参数；
预警模块，用于根据所述第二异常度参数对所述电机驱动器进行故障预警。
在一种可能的实现方式中，所述车辆处于停止阶段，至少包括如下一种：
所述车辆处于充电阶段、所述车辆处于打火阶段、所述车辆处于熄火阶段。
在一种可能的实现方式中，所述预警模块，用于通过所述AI诊断模型对预设的异常条件进行动态更新，以得到更新后的异常条件；当所述第二异常度参数满足所述更新后的异常条件时，对所述电机驱动器进行故障预警。
在一种可能的实现方式中，所述装置还包括：数据处理模块，用于所述获取模块获取车辆处于停止阶段时对所述电机驱动器进行低频采样得到的静态数据之后，对所述静态数据进行清洗处理，得到清洗处理后的静态数据；
所述推理模块，用于将所述清洗处理后的静态数据输入到所述AI诊断模型。
在一种可能的实现方式中，所述装置具体为云端服务器；所述装置还包括：训练模块，
所述获取模块，还用于获取同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
所述训练模块，用于使用所述同类型的多个车辆的训练样本数据和所述多个不同类型的车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
在一种可能的实现方式中，所述装置具体为云端服务器；所述装置还包括：训练模块，
所述获取模块，还用于获取一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
所述训练模块，用于使用所述一个车辆的训练样本数据或者所述同类型的多个车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
在一种可能的实现方式中，所述装置具体为云端服务器；所述预警模块，用于根据所述第二异常度参数生成故障预警信息；向车辆终端设备发送所述故障预警信息。
在一种可能的实现方式中，所述装置具体为车辆终端设备；
当云端服务器向所述车辆终端设备发送所述AI诊断模型时，所述获取模块，还用于接收所述云端服务器发送的AI诊断模型。
在一种可能的实现方式中，所述装置具体为云端服务器；所述装置还包括：训练模块，
所述获取模块，还用于获取多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据；
所述训练模块，用于对所述多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据，进行数据关联关系提取，以得到预训练模型，所述预训练模型用于表示提取到的数据关联关系；向车辆终端设备发送所述预训练模型。
在一种可能的实现方式中，所述装置具体为车辆终端设备；所述装置还包括：训练模块，
所述训练模块，用于接收来自云端服务器的预训练模型，其中，所述预训练模块用于表示所述云端服务器对多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据进行提取后得到的数据关联关系；根据预设的标签数据对所述预训练模型进行模型参数寻优处理，以得到所述预设的AI诊断模型。
第五方面，本申请实施例提供了一种计算机可读存储介质，所述计算机可读存储介质中存储有指令，当其在计算机上运行时，使得计算机执行上述第一方面或第三方面所述的方法。
第六方面，本申请实施例提供了一种包含指令的计算机程序产品，当其在计算机上运行时，使得计算机执行上述第一方面或第三方面所述的方法。
第七方面，本申请实施例提供一种通信装置，该通信装置可以包括云端服务器、或者车辆终端设备或者芯片等实体，所述通信装置包括：处理器、存储器；所述存储器用于存储指令；所述处理器用于执行所述存储器中的所述指令，使得所述通信装置执行如前述第一方面或第三方面中任一项所述的方法。
第八方面，本申请提供了一种芯片系统，该芯片系统包括处理器，用于支持故障预警装置实现上述第一方面或第三方面中所涉及的功能，例如，发送或处理上述方法中所涉及的数据和/或信息。在一种可能的设计中，所述芯片系统还包括存储器，所述存储器，用于保存故障预警装置必要的程序指令和数据。该芯片系统，可以由芯片构成，也可以包括芯片和其他分立器件。
第九方面，本申请提供了一种云端服务器，该云端服务器包括处理器，用于支持故障预警装置实现上述第一方面或第三方面中所涉及的功能，例如，发送或处理上述方法中所涉及的数据和/或信息。在一种可能的设计中，所述云端服务器还包括存储器，所述存储器，用于保存故障预警装置必要的程序指令和数据。
第十方面，本申请提供了一种车辆终端设备，该芯片系统包括处理器，用于支持故障预警装置实现上述第一方面或第三方面中所涉及的功能，例如，发送或处理上述方法中所涉及的数据和/或信息。在一种可能的设计中，所述车辆终端设备还包括存储器，所述存储器，用于保存故障预警装置必要的程序指令和数据。
第十一方面，本申请实施例还提供一种故障预警系统，该故障预警系统可以包括如前述第九方面所述的云端服务器和如第十方面所述的车辆终端设备。
第十二方面，本申请实施例提供一种车联网装置，该车联网装置可以包括车联网服务器、路侧单元、车联网通信装置或者芯片等实体，所述车联网装置包括：处理器。可选的，所述车联网装置还包括：存储器；所述存储器用于存储指令；所述处理器用于执行所述存储器中的所述指令，使得所述车联网装置执行如前述第一方面或第三方面中任一项所述的方法。
从以上技术方案可以看出，本申请实施例具有以下优点：
在本申请实施例中，首先获取电机驱动器运行时的运行态数据，然后将运行态数据输入到预设的AI诊断模型，通过AI诊断模型输出电机驱动器的第一异常度参数，其中，AI 诊断模型用于根据运行态数据提取数据特征，并通过数据特征进行决策推理；根据第一异常度参数对电机驱动器进行故障预警。由于本申请实施例中电机驱动器运行时车辆处于运行状态，根据电机驱动器运行时的运行态数据可以对电机驱动器进行故障预警，电机驱动器的运行态数据是电机驱动器运行时采集到的数据，运行态数据能够反映电机驱动器的真实运行状态，基于该运行态数据进行故障预警，可以实现对电机驱动器失效的提前预测，实现对电机驱动器的故障预警。
附图说明
图1为本申请实施例提供的一种用于电机驱动器的故障预警方法的流程方框示意图；
图2为本申请实施例提供的一种用于电机驱动器的故障预警方法的流程方框示意图；
图3为本申请实施例提供的一种用于电机驱动器的故障预警方法的流程方框示意图；
图4为本申请实施例提供的一种用于电机驱动器的故障预警方法的流程方框示意图；
图5为本申请实施例提供的一种用于电机驱动器的故障预警方法的流程方框示意图；
图6为本申请实施例提供的一种用于电机驱动器的故障预警方法的流程方框示意图；
图7为本申请实施例提供的一种云端服务器和车辆终端设备的组成结构示意图；
图8为本申请实施例提供的一种AI诊断模型的执行流程示意图；
图9为本申请实施例提供的一种AI诊断模型的执行流程示意图；
图10为本申请实施例提供的一种运行态数据和静态数据的采集流程示意图；
图11a为本申请实施例提供的一种用于电机驱动器的故障预警装置的组成结构示意图；
图11b为本申请实施例提供的一种用于电机驱动器的故障预警装置的组成结构示意图；
图11c为本申请实施例提供的一种用于电机驱动器的故障预警装置的组成结构示意图；
图12为本申请实施例提供的一种用于电机驱动器的故障预警装置的组成结构示意图。
具体实施方式
本申请实施例提供了一种用于电机驱动器的故障预警方法和装置，用于实现对电机驱动器的故障预警。
下面结合附图，对本申请的实施例进行描述。
本申请的说明书和权利要求书及上述附图中的术语“第一”、“第二”等是用于区别类似的对象，而不必用于描述特定的顺序或先后次序。应该理解这样使用的术语在适当情况下可以互换，这仅仅是描述本申请的实施例中对相同属性的对象在描述时所采用的区分方式。此外，术语“包括”和“具有”以及他们的任何变形，意图在于覆盖不排他的包含，以便包含一系列单元的过程、方法、系统、产品或设备不必限于那些单元，而是可包括没有清楚地列出的或对于这些过程、方法、产品或设备固有的其它单元。
本申请实施例提供一种用于电机驱动器的故障预警方法，可以对电机驱动器进行故障预警，提高对电机驱动器的故障预警精确度。该电机驱动器安装在车辆上，电机驱动器是 车辆上的电子控制设备，该车辆可以是新能源汽车、智能汽车、车辆终端设备等，电机驱动器也可以称为电机控制单元(motor control unit，MCU)。电机驱动器中可以包括一个或多个的功率半导体器件，功率半导体器件是指实现电路开关功能的半导体器件，例如功率半导体器件可以是开关管，本申请实施例以各个开关管为金属氧化物半导体场效应管(metal-oxide-semiconductor field-effect transistor，MOSFET)进行示例性说明，应当理解的是，各个开关管还可以是绝缘栅双极型晶体管(insulated gate bipolar transistor，IGBT)等其他半导体器件，例如开关管还可以是二极管(Diode)等，后续实施例中开关管主要应用于IGBT进行举例说明。本申请实施例中，电机驱动器通过控制电机的旋转角度和运转速度，实现对电机的控制。例如，电机驱动器既可通过继电器或功率晶体管驱动，也可利用可控硅或功率型MOS场效应管驱动，电机驱动器的控制要求可以有多种，例如对电机的工作电流、电压，电机的调速，直流电机的正反转控制等。
本申请实施例中电机驱动器运行时车辆处于运行状态，根据电机驱动器运行时的运行态数据可以对电机驱动器进行故障预警，由于电机驱动器的运行态数据是电机驱动器运行时采集到的数据，运行态数据能够反映电机驱动器的真实运行状态，基于该运行态数据进行故障预警，可以实现对电机驱动器失效的提前预测，实现对电机驱动器的故障预警。
本申请实施例提供的用于电机驱动器的故障预警方法由用于电机驱动器的故障预警装置(后续简称为故障预警装置)实现，该故障预警装置中可以存储预先训练完成的人工智能(artificial intelligence，AI)诊断模型，该AI诊断模型可以是使用AI诊断算法进行模型训练后得到的诊断模型，该AI诊断模型具有数据特征的提取能力，并基于该数据特征进行决策推理，输出电机驱动器的异常度参数，通过异常度参数的分析可以确定出电机驱动器是否存在故障，从而对电机驱动器进行故障预警。
本申请实施例提供的故障预警装置具有多种实现方式，该故障预警装置可以是独立于车辆的服务器，或者与车辆集成的终端设备，此处不做限定。例如该故障预警装置可以是云端服务器，该云端服务器又可以称为云端平台或者云平台。云端服务器中可以存储预先训练完成的AI诊断模型，从而云端服务器可以在AI诊断模型完成决策推理之后，向车辆终端设备通知推理结果。又如，故障预警装置还可以是车辆终端设备，该车辆终端设备可以是车辆的控制器，或者车辆终端设备可以是集成在车辆内的终端设备，车辆终端设备又可以称为端侧终端设备，或者端侧智能设备，车辆终端设备可以从云端服务器获取到推理结果，从而车辆终端设备可以对电机驱动器进行故障预警。又如，故障预警装置可以是车辆终端设备，车辆终端设备可以从云端服务器获取预训练模型，然后车辆终端设备根据该预训练模型生成AI诊断模型，从而车辆终端设备可以使用该AI诊断模型对电机驱动器进行故障预警。本申请实施例对故障预警装置的实现方式不做限定。
首先请参阅图1所示，本申请实施例提供一种用于电机驱动器的故障预警方法，主要包括如下步骤：
101、获取电机驱动器运行时的运行态数据。
在本申请实施例中，电机驱动器安装在车辆中，车辆运行过程中需要电机驱动器运行，故障预警装置可以获取电机驱动器运行时的运行态数据，该运行态数据能够反映电机驱动器的真实运行状态，本申请实施例对电机驱动器的运行态数据的具体采集过程以及该运行 态数据包括的数据项和数据参量的采集范围不做限定，例如可以根据电机驱动器的类型以及功能、车辆运行状态等综合确定运行态数据，后续实施例中对运行态数据进行详细说明。
例如故障预警装置可以是云端服务器，则车辆终端设备可以采集电机驱动器运行时的运行态数据，然后车辆终端设备可以通过通信接口或者无线网络向云端服务器发送电机驱动器运行时的运行态数据。又如，故障预警装置可以是车辆终端设备，车辆终端设备中设置有数据测量模块，该数据测量模块可以采集电机驱动器运行的运行态数据。
在本申请的一些实施例中，运行态数据包括：电机驱动器运行时默认采集的动态数据；或者，
运行态数据包括：通过控制器局域网络(controller area network，CAN)总线采集的动态数据。
其中，运行态数据包括电机驱动器运行时默认采集的动态数据，在电机驱动器运行时会默认采集动态数据，默认采集的动态数据除了用于车辆的原有控制功能，本申请实施例中可以将默认采集的动态数据用于电机驱动器的故障预警。另外，运行态数据包括通过CAN总线采集的动态数据，电机控制器连接CAN总线，因此可以通过CAN总线进行动态数据采集，以得到上述的运行态数据。本申请实施例不局限于上述两种数据采集方式，通过上述数据采集方式，都不需要对车辆以及电机驱动器进行单板硬件升级，就可以获取到电机驱动器运行时的运行态数据，提高了运行态数据的采集效率。本申请实施例对电机驱动器的运行态数据的具体采集过程以及该运行态数据包括的数据项和数据参量的采集范围不做限定。
在本申请的一些实施例中，运行态数据包括：车辆处于运行阶段进行高频采样得到的动态数据，电机驱动器安装在车辆上。
其中，车辆具有多种状态，例如车辆处于运行阶段，该运行阶段是指车辆在打火启动之后的阶段，例如运行阶段是指车辆处于行驶阶段。在车辆处于运行阶段时进行高频采样得到的动态数据可以是前述的运行态数据，后续实施例中对于高频采样的数据项以及数据量进行举例说明。其中，高频采样是指数据采集的频率大于预设的频率阈值，本申请实施例中对于频率阈值的具体取值不做限定。
在本申请的一些实施例中，运行态数据至少包括如下一种：电机驱动器的参数、冷却系统的参数、电机的参数、电池的参数。
其中，本申请实施例中电机驱动器运行时的运行态数据具有多种实现方式，例如在电机驱动器运行时可以针对电机驱动器进行数据采集以得到电机驱动器的参数，或者在电机驱动器运行时可以针对冷却系统进行数据采集以得到冷却系统的参数，或者在电机驱动器运行时可以针对电机进行数据采集以得到电机的参数，或者在电机驱动器运行时可以针对电池进行数据采集以电池的参数。本申请实施例中对于电机驱动器的参数、冷却系统的参数、电机的参数、电池的参数中具体的参数名称和参数内容不做限定。
需要说明的是，本申请实施例中通过数据采集可以得到电机驱动器的参数、冷却系统的参数、电机的参数、电池的参数。由于电机驱动器的参数、冷却系统的参数、电机的参数、电池的参数都是电机驱动器运行时采集到的数据，因此电机驱动器的参数、冷却系统的参数、电机的参数、电池的参数中的任意一项都能够反映电机驱动器的真实运行状态， 基于上述参数进行故障预警，可以实现对电机驱动器失效的提前预测，实现对电机驱动器的故障预警。
进一步的，在本申请的一些实施例中，电机驱动器的参数包括如下至少一种：直流母线电压、直流母线电流、低压电源电压、车辆行驶状态命令、车辆状态、电机驱动器温度、转子位置电角度、电机驱动器三相电流有效值、电机驱动器三相电流、电机驱动器三相电流采样值、电机驱动器降额状态；和/或，
冷却系统的参数包括如下至少一种：冷却液流量、冷却液温度、油泵目标转速、油泵状态、油泵实际转速、油泵供电电压、油泵的功率半导体器件温度、油泵电流；和/或，
电机的参数包括如下至少一种：电磁频率、电机工作模式命令、电机控制器工作状态、电机目标转矩命令、电机当前转矩、电机目标转速命令、电机当前转速、电机温度、电机直轴电压、电机直轴电流给定值、电机直轴电流反馈值、电机交轴电压、电机交轴电流给定值、电机交轴电流反馈值、电机当前校验转矩；和/或，
电池的参数包括如下至少一种：额定输出电压、电池容量、最大输出电流。
其中，在电机驱动器的参数中，电机驱动器的三相电流可以是U、V和W相的电流。
需要说明的是，电机驱动器的参数、冷却系统的参数、电机的参数、电池的参数的说明详见后续实施例。
102、将运行态数据输入到预设的人工智能AI诊断模型，通过AI诊断模型输出电机驱动器的第一异常度参数，其中，AI诊断模型用于根据运行态数据提取数据特征，并通过数据特征进行决策推理。
在本申请实施例中，故障诊断装置可以预先训练完成AI诊断模型，故障诊断装置在获取到运行态数据之后，使用该AI诊断模型进行决策推理。具体的，运行态数据输入到AI诊断模型中，该AI诊断模型根据运行态数据提取数据特征，该数据特征由AI诊断模型根据输入的运行态数据进行特征提前得到，该数据特征是AI诊断模型进行决策推理的依据。该AI诊断模型可以是机器学习模型，例如该AI诊断模型可以是经过预先训练得到的神经网络模型，或者AI诊断模型也可以其它的机器学习模型，例如可以是线性回归模型、决策树模型等，AI诊断模型根据提取到的数据特征进行决策推理。本申请实施例中对于数据特征所包括的数据内容不做限定。AI诊断模型进行决策推理之后，可以输出电机驱动器的第一异常度参数，该第一异常度参数是为了区分后续实施例中出现的其它异常度参数(例如第二异常度参数)而定义的异常度参数。其中，第一异常度参数是用于衡量电机驱动器的异常情况的参数，该第一异常度参数也可以称为第一异常因子分数、第一异常度评分等。
需要说明的是，故障诊断装置执行前述步骤102，具体的，该故障诊断装置可以是云端服务器，或者该故障诊断装置可以是车辆终端设备。
在本申请的一些实施例中，步骤101获取电机驱动器运行时的运行态数据之后，本申请实施例提供的方法还可以包括如下步骤：
对运行态数据进行清洗处理，得到清洗处理后的运行态数据。
其中，故障诊断装置在获取到多个运行态数据之后，可以先对该运行态数据进行清洗处理，例如可以先对运行态数据中符合清洗条件的数据进行剔除。该清洗条件可以是运行态数据不完整，或者运行态数据的某项指标超标等。
在对运行态数据进行清洗处理之后，前述步骤102将运行态数据输入到预设的AI诊断模型包括：
将清洗处理后的运行态数据输入到AI诊断模型。
其中，故障诊断装置获取到清洗处理后的运行态数据之后，将清洗处理后的运行态数据输入到AI诊断模型中，通过AI诊断模型对清洗处理后的运行态数据进行数据特征的提取，通过对运行态数据进行清洗处理之后再输入到AI诊断模型，可以提高AI诊断模型的推理效率。
103、根据第一异常度参数对电机驱动器进行故障预警。
在本申请实施例中，通过AI诊断模型输出第一异常度参数之后，根据该第一异常度参数可以衡量电机驱动器的工作情况，以识别该电机驱动器是否出现异常情况，当电机驱动器出现异常情况时，确定该电机驱动器存在故障，对该电机驱动器进行故障预警，以实现对电机驱动器失效的提前预测。其中，故障预警是指在故障发生前一定时间提前预测到故障将在未来发生，并发出报警信号。本申请实施例提供的方法可应用于汽车电驱动系统，提前诊断汽车电驱动系统因为电机驱动器造成的失效并进行预警，让用户在故障发生之前进行维护，避免驾驶过程中车辆出现抛锚等情况。
在本申请的一些实施例中，步骤103根据第一异常度参数对电机驱动器进行故障预警，包括：
通过AI诊断模型对预设的异常条件进行动态更新，以得到更新后的异常条件；
当第一异常度参数满足更新后的异常条件时，对电机驱动器进行故障预警。
其中，故障诊断装置可以预设异常条件，并且该异常条件可以通过AI诊断模型进行动态更新，以保证该异常条件能够用于故障判断，保证故障判断的准确率。在对异常条件进行动态更新之后，可以判断第一异常度是否满足动态更新后的异常条件，当第一异常度参数满足更新后的异常条件时，说明该电机驱动器可能存在故障，此时对电机驱动器进行故障预警。本申请实施例中通过对异常条件进行动态更新，可以进一步提高故障预警的准确度。
通过前述实施例的举例说明可知，首先获取电机驱动器运行时的运行态数据，然后将运行态数据输入到预设的AI诊断模型，通过AI诊断模型输出电机驱动器的第一异常度参数，其中，AI诊断模型用于根据运行态数据提取数据特征，并通过数据特征进行决策推理；根据第一异常度参数对电机驱动器进行故障预警。由于本申请实施例中电机驱动器运行时车辆处于运行状态，根据电机驱动器运行时的运行态数据可以对电机驱动器进行故障预警，电机驱动器的运行态数据是电机驱动器运行时采集到的数据，运行态数据能够反映电机驱动器的真实运行状态，基于该运行态数据进行故障预警，可以实现对电机驱动器失效的提前预测，实现对电机驱动器的故障预警。
本申请前述实施例中说明了通过电机驱动器的运行态数据进行故障预警，接下来介绍本申请实施例提供的根据车辆停止时的静态数据进行故障预警的方案，请参阅图2所示，本申请实施例提供一种用于电机驱动器的故障预警方法，主要包括如下步骤：
201、获取车辆处于停止阶段时对电机驱动器进行低频采样得到的静态数据，电机驱动器安装在车辆上。
其中，车辆具有多种状态，例如车辆处于停止阶段，该停止阶段又可以称为静止阶段。
在车辆处于停止阶段时对电机驱动器进行低频采样得到静态数据，后续实施例中对于低频采样的数据项以及数据量进行举例说明。其中，低频采样是指数据采集的频率低于预设的频率阈值，本申请实施例中对于频率阈值的取值不做限定。
例如，该故障诊断装置可以是车辆终端设备，在车辆处于停止阶段时，车辆终端设备可以对电机驱动器进行低频采样以得到静态数据，该静态数据可以用于AI诊断模型的决策推理。又如，该故障诊断装置可以是云端服务器，车辆终端设备在车辆处于停止阶段时，车辆终端设备可以对电机驱动器进行低频采样以得到静态数据，车辆终端设备可以向云端服务器发送该静态数据，从而云端服务器可以从车辆终端设备接收到该静态数据。
在本申请的一些实施例中，车辆处于停止阶段，至少包括如下一种：车辆处于充电阶段、车辆处于打火阶段、车辆处于熄火阶段。
其中，在车辆充电阶段、车辆打火阶段、车辆熄火阶段时进行静态数据的采样，从而故障诊断装置可以根据该静态数据提供断网情况下的紧急响应功能。
202、将静态数据输入到AI诊断模型中，通过AI诊断模型输出电机驱动器的第二异常度参数。
在本申请实施例中，故障诊断装置可以预先训练完成AI诊断模型，故障诊断装置在获取到静态数据之后，使用该AI诊断模型进行决策推理。具体的，静态数据输入到AI诊断模型中，该AI诊断模型根据静态数据提取数据特征，该数据特征由AI诊断模型根据输入的静态数据进行特征提前得到，该数据特征是AI诊断模型进行决策推理的依据。该AI诊断模型可以是机器学习模型，例如该AI诊断模型可以是经过预先训练得到的神经网络模型，或者AI诊断模型也可以其它的机器学习模型，例如可以是线性回归模型、决策树模型等，AI诊断模型根据提取到的数据特征进行决策推理。本申请实施例中对于数据特征所包括的数据内容不做限定。AI诊断模型进行决策推理之后，可以输出电机驱动器的第二异常度参数。其中，第二异常度参数是用于衡量电机驱动器的异常情况的参数，该第二异常度参数也可以称为第二异常因子分数、第二异常度评分等。
203、根据第二异常度参数对电机驱动器进行故障预警。
在本申请实施例中，通过AI诊断模型输出第二异常度参数之后，根据该第二异常度参数可以衡量电机驱动器的工作情况，以识别该电机驱动器是否出现异常情况，当电机驱动器出现异常情况时，确定该电机驱动器存在故障，对该电机驱动器进行故障预警，以实现对电机驱动器失效的提前预测。本申请实施例提供的方法可应用于汽车电驱动系统，提前诊断汽车电驱动系统因为电机驱动器造成的失效并进行预警，让用户在故障发生之前进行维护，避免驾驶过程中车辆出现抛锚等情况。
需要说明的是，本申请实施例提供的上述步骤201至步骤203可以是独立于图1所示的实施例，该步骤201至步骤203也可以在图1所示的步骤101至步骤103之后执行，或者该步骤201至步骤203也可以在图1所示的步骤101至步骤103之前执行，此处不做限定。例如故障诊断装置可以获取到前述的静态数据和运行态数据，将静态数据和运行态数据都输入AI诊断模型，AI诊断模型根据静态数据和运行态数据提取到数据特征，再根据该数据特征进行决策推理，输出第三异常度参数，最后根据该第三异常度参数对电机驱动 器进行故障预警。
通过前述实施例的举例说明可知，本申请实施例中车辆处于停止阶段时获取电机驱动器的静态数据，根据该静态数据可以对电机驱动器进行故障预警，电机驱动器的静态数据是车辆停止时对电机驱动器采集到的数据，静态数据能够反映电机驱动器的真实静止状态，基于该静态数据进行故障预警，可以实现对电机驱动器失效的提前预测，实现对电机驱动器的故障预警。
前述图1和图2中以故障诊断装置执行用于电机驱动器的故障预警方法进行示例说明，接下来介绍本申请实施例中云端服务器和车辆终端设备之间的一种交互流程，如图3所示，本申请实施例提供一种用于电机驱动器的故障预警方法，主要包括如下步骤：
301、云端服务器获取同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，电机驱动器安装在车辆上。
其中，云端服务器可以和多个车辆终端设备进行交互，每个车辆中可以安装一个或多个电机驱动器。云端服务器可以获取到多个训练样本数据，训练样本数据可用于模型训练。例如多个训练样本数据可以是同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，该类型是指车辆的类型。对于训练样本数据的数据类型和数据内容不做限定。
302、云端服务器使用同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到预设的AI诊断模型。
其中，云端服务器获取到同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，云端服务器可以预先定义训练任务为故障预警任务，云端服务器使用同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据进行模型训练，本申请实施例中对于模型的具体训练过程不再详细说明，在模型训练完成之后可以得到预设的AI诊断模型。例如云端服务器先使用同类型的多个车辆的训练样本数据进行模型训练，然后再对不同类型车辆的训练样本数据进行泛化，最后可以完成模型的训练，输出预设的AI诊断模型。
303、云端服务器获取车辆终端设备发送的电机驱动器运行时的运行态数据。
304、云端服务器将运行态数据输入到预设的人工智能AI诊断模型，通过AI诊断模型输出电机驱动器的第一异常度参数，其中，AI诊断模型用于根据运行态数据提取数据特征，并通过数据特征进行决策推理。
其中，本申请实施例中可以由云端服务器使用AI诊断模型进行决策推理，通过AI诊断模型输出电机驱动器的第一异常度参数。步骤303至步骤304的实现方式与前述实施例中的步骤101至步骤102的实现方式类似，详见前述的实施例说明，此处不做限定。
305、云端服务器根据第一异常度参数生成故障预警信息。
306、云端服务器向车辆终端设备发送故障预警信息。
其中，云端服务器通过AI诊断模型输出第一异常度参数之后，云端服务器根据第一异常度参数生成故障预警信息，该故障预警信息可以是云端服务器根据第一异常度参数进行推理决策的结果，云端服务器向车辆终端设备发送该故障预警信息，以使得车辆终端设备可以根据故障预警信息对电机驱动器进行故障预警。
307、车辆终端设备接收来自云端服务器的故障预警信息。
308、车辆终端设备根据故障预警信息对电机驱动器进行故障预警。
需要说明的是，本申请实施例中，云端服务器使用了多个训练样本数据，该多个训练样本数据可以包括同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，从而得到的AI诊断模型是AI全局模型，该模型是面向具体的故障预警任务进行的，该模型基于所有采集的车辆终端设备侧数据训练，基于该模型并给出每个车辆终端设备的端侧推理结果，作为决策策略下发给车辆终端设备，车辆终端设备不做训练推理，直接执行云端服务器的推理结果，因此可以通过云端服务器和车辆终端设备的交互，完成对电机驱动器的故障预警。
通过前述实施例的举例说明可知，本申请实施例中云端服务器具有模型训练能力和决策推理能力，车辆终端设备可以按照云端服务器的决策结果进行故障预警，通过云端服务器和车辆终端设备的交互，可以实现对电机驱动器的故障预警。
如图4所示，本申请实施例提供一种用于电机驱动器的故障预警方法，主要包括如下步骤：
401、云端服务器获取同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，电机驱动器安装在车辆上。
402、云端服务器使用同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到预设的AI诊断模型。
上述步骤401至步骤402的实现方式与前述实施例中的步骤301至步骤302的实现方式类似，详见前述的实施例说明，此处不做限定。
403、云端服务器向车辆终端设备发送AI诊断模型。
在本申请实施例中，云端服务器在完成AI诊断模型的训练之后，云端服务器可以向车辆终端设备发送训练好的AI诊断模型。
404、车辆终端设备接收云端服务器发送的AI诊断模型。
在本申请实施例中，车辆终端设备不具备模型训练能力，车辆终端设备可以从云端服务器接收AI诊断模型，从而车辆终端设备可以使用该AI诊断模型进行决策推理，具体的，车辆终端设备可以执行后续步骤405至407。
405、车辆终端设备获取电机驱动器运行时的运行态数据。
406、车辆终端设备将运行态数据输入到预设的AI诊断模型，通过AI诊断模型输出电机驱动器的第一异常度参数，其中，AI诊断模型用于根据运行态数据提取数据特征，并通过数据特征进行决策推理。
407、车辆终端设备根据第一异常度参数对电机驱动器进行故障预警。
其中，步骤405至步骤407的实现方式与前述实施例中的步骤101至步骤103的实现方式类似，详见前述的实施例说明，此处不做限定。
需要说明的是，本申请实施例中，云端服务器使用了多个训练样本数据，该多个训练样本数据可以包括同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，从而得到的AI诊断模型是AI全局模型，该模型是面向具体的故障预警任务进行的，该模型基于所有采集的车辆终端设备侧数据训练，云端服务器将训练完的AI诊断模型发送 给车辆终端设备，车辆终端设备基于该模型并给出每个车辆终端设备的端侧推理结果，车辆终端设备不做训练，直接使用AI诊断模型进行推理，执行推理结果，因此可以通过云端服务器和车辆终端设备的交互，完成对电机驱动器的故障预警。
通过前述实施例的举例说明可知，本申请实施例中云端服务器具有模型训练能力，车辆终端设备具有决策推理能力，车辆终端设备可以接收云端服务器的AI诊断模型，从而车辆终端设备可以根据AI诊断模型输出决策结果，对电机驱动器进行故障预警，通过云端服务器和车辆终端设备的交互，可以实现对电机驱动器的故障预警。
如图5所示，本申请实施例提供一种用于电机驱动器的故障预警方法，主要包括如下步骤：
501、云端服务器获取一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，电机驱动器安装在车辆上。
其中，云端服务器可以和一个车辆终端设备或者同类型的多个车辆终端设备进行交互，每个车辆中可以安装一个或多个电机驱动器。云端服务器可以获取到多个训练样本数据，训练样本数据可用于模型训练。例如多个训练样本数据可以是一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，该类型是指车辆的类型。对于训练样本数据的数据类型和数据内容不做限定。
502、云端服务器使用一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到预设的AI诊断模型。
其中，云端服务器获取到一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，云端服务器可以预先定义训练任务为故障预警任务，云端服务器使用一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据进行模型训练，本申请实施例中对于模型的具体训练过程不再详细说明，在模型训练完成之后可以得到预设的AI诊断模型。例如云端服务器先使用一个车辆的训练样本数据进行模型训练，然后再对同类型多个车辆的训练样本数据进行泛化，最后可以完成模型的训练，输出预设的AI诊断模型。
503、云端服务器获取车辆终端设备发送的电机驱动器运行时的运行态数据。
504、云端服务器将运行态数据输入到预设的人工智能AI诊断模型，通过AI诊断模型输出电机驱动器的第一异常度参数，其中，AI诊断模型用于根据运行态数据提取数据特征，并通过数据特征进行决策推理。
505、云端服务器根据第一异常度参数生成故障预警信息。
506、云端服务器向车辆终端设备发送故障预警信息。
507、车辆终端设备接收来自云端服务器的故障预警信息。
508、车辆终端设备根据故障预警信息对电机驱动器进行故障预警。
其中，步骤503至步骤508的实现方式与前述实施例中的步骤303至步骤308的实现方式类似，详见前述的实施例说明，此处不做限定。
需要说明的是，本申请实施例中，云端服务器使用了多个训练样本数据，该多个训练样本数据可以包括一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，从而得到的AI诊断模型是AI局部模型，该模型是面向具体的故障预警任务进行的，该AI局部模型可能是某个型号的车辆终端设备甚至是某个车辆终端设备，基于该模型并给出每个端 侧推理结果，作为决策策略下发给车辆终端设备，车辆终端设备不做训练推理，直接执行云端服务器的推理结果，因此可以通过云端服务器和车辆终端设备的交互，完成对电机驱动器的故障预警。
通过前述实施例的举例说明可知，本申请实施例中云端服务器具有模型训练能力和决策推理能力，车辆终端设备可以按照云端服务器的决策结果进行故障预警，通过云端服务器和车辆终端设备的交互，可以实现对电机驱动器的故障预警。
如图6所示，本申请实施例提供一种用于电机驱动器的故障预警方法，主要包括如下步骤：
601、云端服务器获取多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据。
其中，云端服务器可以和不同类型的车辆终端设备、或者一个车辆终端设备或者同类型的多个车辆终端设备进行交互，每个车辆中可以安装一个或多个电机驱动器。云端服务器可以获取到多个训练样本数据，训练样本数据可用于数据关联关系的提取。例如多个训练样本数据可以是多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，该类型是指车辆的类型。对于训练样本数据的数据类型和数据内容不做限定。
602、云端服务器对多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据，进行数据关联关系提取，以得到预训练模型，预训练模型用于表示提取到的数据关联关系。
其中，云端服务器不面向具体的任务或功能，对多个不同类型的车辆的训练样本数据进行数据关联关系提取，或者对一个车辆的训练样本数据进行数据关联关系提取，或者对同类型的多个车辆的训练样本数据进行数据关联关系提取，以得到预训练模型。数据关联关系可以是多个训练样本数据中各个数据维度及其特征在一定时序及工况中的关系。
603、云端服务器向车辆终端设备发送预训练模型。
在本申请实施例中，云端服务器在得到预训练模型之后，云端服务器可以向车辆终端设备发送预训练模型。
604、车辆终端设备接收来自云端服务器的预训练模型。
605、车辆终端设备根据预设的标签数据对预训练模型进行模型参数寻优处理，以得到预设的AI诊断模型。
在本申请实施例中，云端服务器不需要进行AI诊断模型的训练，云端服务器只需要提取数据关联关系，得到预训练模型之后，向车辆终端设备发送该预训练模型，车辆终端设备根据预设的标签数据对预训练模型进行模型参数寻优处理，模型参数寻优处理也可以称为微调，即车辆终端设备可以进行模型训练，以得到预设的AI诊断模型。例如，车辆终端设备在接收到预训练模型之后，车辆终端设备根据故障预警任务，结合预设的标签数据对预训练模型进行微调，在低资源消耗下实现故障预警任务。
606、车辆终端设备获取电机驱动器运行时的运行态数据。
607、车辆终端设备将运行态数据输入到预设的AI诊断模型，通过AI诊断模型输出电机驱动器的第一异常度参数，其中，AI诊断模型用于根据运行态数据提取数据特征，并通 过数据特征进行决策推理。
608、车辆终端设备根据第一异常度参数对电机驱动器进行故障预警。
其中，步骤606至步骤608的实现方式与前述实施例中的步骤101至步骤103的实现方式类似，详见前述的实施例说明，此处不做限定。
需要说明的是，本申请实施例中，云端服务器使用了多个训练样本数据，云端服务器可以获取预训练模型，将该预训练模型下发到车辆终端设备，作为车辆终端设备的训练初始化参数，车辆终端设备基于采集的少量数据进行模型参数寻优，即微调或迁移学习过程，从而可以到AI诊断模型，车辆终端设备基于该模型并给出每个车辆终端设备的端侧推理结果，车辆终端设备直接使用AI诊断模型进行推理，执行推理结果，因此可以通过云端服务器和车辆终端设备的交互，完成对电机驱动器的故障预警。
通过前述实施例的举例说明可知，本申请实施例中云端服务器向车辆终端设备发送预训练模型，车辆终端设备根据该预训练模型生成AI诊断模型，车辆终端设备具有决策推理能力，从而车辆终端设备可以根据AI诊断模型输出决策结果，对电机驱动器进行故障预警，通过云端服务器和车辆终端设备的交互，可以实现对电机驱动器的故障预警。
为便于更好的理解和实施本申请实施例的上述方案，下面举例相应的应用场景来进行具体说明。
本申请实施例提出一种用于对电机驱动器的故障预警方法，本申请实施例不仅适用于电机驱动器，还适用于所有用到功率半导体模组的电力电子部件，比如充电机、直流/直流变换器(direct-current/direct-current converter，DC/DC)等。本申请实施例提供的技术方案可以实现电机驱动器整机级的智能诊断。接下来以电机驱动器具体为功率半导体器件为例进行说明，在车辆运行态采集功率半导体器件的运行态数据，通过AI诊断模型判断数据特征，并对数据特征进行异常度评分，将异常度超标的功率半导体器件识别出来，以实现对单个功率半导体器件失效的提前预测。
请参阅如图7所示，为本申请实施例提供的一种云端服务器和车辆终端设备的组成结构示意图。其中，云端服务器也可以称为云端，云端服务器包括如下至少一个模块：数据处理模块、数据存储模块、模型训练模块和云端推理模块。车辆终端设备又可以称为端侧或者板端，车辆终端设备包括如下至少一个模块：数据测量模块、微调模块、策略执行模块和端侧推理模块。
其中，数据测量模块用于采集功率半导体器件的训练样本数据。
数据存储模块用于存储端侧上报的海量数据，便于后续处理。
数据处理模块用于对海量数据进行清洗处理，便于后续AI诊断模型的训练。
模型训练模块用于进行故障预警或者状态计算，后续以故障预警场景进行示例说明。云端服务器上的模型训练模块具有如下三种实现方式：
1.AI诊断模型为通用模型。所有车辆共用同一个模型，该模型是面向具体的任务或功能的，比如故障预警任务或者容量估计任务等，模型可以直接用于这些场景的结果推理。提取训练样本车中的特征信息(例如故障或者容量相关特征)，训练一个模型，云端推理模块可以基于该模型在云服务器上对每台车的数据进行推理预测以得到推理结果，将推理结果(例如故障预警等)作为策略，下发到车辆终端设备，车辆终端设备中的策略执行模块 按照云端服务器的策略进行控制处理。
2.AI诊断模型为局部模型。一类车共用一个模型，或者一个车单独使用一个模型，局部模型与前述的通用模型类似，这些模型也是面向具体任务的。云端服务器上基于某一类车或者每台车的数据进行提取特征，并训练AI诊断模型，在云端服务器上基于这些模型对相应的车进行推理预测以得到推理结果，将推理结果(例如故障预警等)作为策略，下发到车辆终端设备，车辆终端设备中的策略执行模块按照云端服务器的策略进行控制处理。
3.通过预训练模型得到AI诊断模型。预训练模型是不面向具体的任务或功能，从海量数据中学习端侧各个数据维度及其特征在一定时序及工况中的关系，形成预训练模型。车辆终端设备的微调模块可以结合一定的标注数据对预训练模型进行微调处理，以得到AI诊断模型，在低资源消耗下实现故障预警、容量估计等任务。以车载领域为例，云端服务器储存了大量的时序、工况、驾驶行为、车辆位置、总电流、总电压、温度、电机状态、告警类型及等级、故障状态等数据，预训练模型通过自监督学习这些维度及其衍生特征之间的关系，车辆终端设备在做类似故障预警等任务时，只需在预训练模型上结合自身有标签的告警数据进行微调，即可得到该车面向故障预警任务的AI诊断模型，这个模型既有通过大数据学习得到的通用信息，又有自身的独有信息。其中，通用信息是指云端服务器从多个运行态车辆中学习到的各种测量值之间的数据分布，通用信息反映的是大数据条件下这些测量数据之间的相互关系。而独有信息是指车辆终端设备的测量值，结合预训练模型的通用信息进行微调或者迁移学习之后就可以得到端侧独有的AI诊断模型。
云端推理模块执行云端服务器的推理功能，基于云端训练的通用模型或者局部模型，对诸如故障预警或者容量估计等任务给出推理结果。
微调模块执行端侧的微调(即迁移学习)功能，基于云端的预训练模型，结合车辆终端设备的数据进行微调，得到面向特定任务(如故障预警任务)的AI诊断模型。
端侧推理模块执行端侧的推理功能，可以支持两种情况下的推理：一种是云端服务器的通用模型或者局部模型轻量化后下沉到车辆终端设备，车辆终端设备根据云端服务器发送的AI诊断模型进行推理。另一种是车辆终端设备在基于云端的预训练模型及微调(迁移学习)得到AI诊断模型后进行推理。
策略执行模块用于对安全预警(如故障预警任务)或者状态估算(如容量估计任务)的推理结果采取措施进行处理。
通过上述举例说明可知，本申请实施例中可以采取云端的训练推理和端侧的训练推理，或者仅通过云端进行推理，或者仅通过端侧进行推理，对于具体的实现方式不做限定。
本实施例采用电机驱动器的默认运行态数据作为AI诊断模型的输入，计算出电机驱动器的异常因子分数，通过异常因子分数来判断电机驱动器在未来发生故障的风险等级，针对高风险等级的产品在真实故障发生前发出预警信号，以便用户提前检测维修，避免汽车行驶过程中发生损坏。
如图8所示，为本申请实施例提供的一种AI诊断模型的执行流程示意图。例如AI诊断模型的输入数据包括：母线电压V_dc、三相电流I_phase、IGBT温度T_jav、冷却液流量L、冷却液温度T_fluid、电磁频率f_em。
具体的，输入的运行态数据如下表1所示，这些数据为MCU运行时默认采样的数据， 可以直接从CAN总线上获取。对于不同的车辆，采集到的运行态数据的规格可能不同，此处仅为示例，不做限制，只要CAN总线上可以采集到的数据内容都可作为AI诊断模型的输入数据。
表1为采集的运行态数据的部分项目举例，测试项目可以增加，跟需要检测的失效模式对应：
采集数据项目
直流母线电压
直流母线电流
驱动电机工作模式命令
驱动电机当前转矩
驱动电机当前转速
驱动电机温度
Ud(电机控制d轴的电压)
Uq(电机控制q轴的电压)
MCU IGBT温度(U相)
MCU IGBT温度(V相)
MCU IGBT温度(W相)
U相电流采样值
V相电流采样值
W相电流采样值
AI诊断模型根据上述输入数据进行决策推理，输出第一异常度参数。对于AI诊断模型的决策推理过程，详见前述图7所示的模型推理过程的说明，此处不再赘述。
在本申请的另一些实施例中，在电机驱动器中采用额外的采样传感器或电路，提取特定的关键数据作为AI诊断模型的输入数据，计算出功率半导体器件的异常因子分数，通过异常因子分数来判断功率半导体器件在未来发生故障的风险等级，针对高风险等级的产品在真实故障发生前发出预警信号，以便用户提前检测维修，避免汽车行驶过程中发生损坏。
如图9所示，为AI诊断模型的一种执行流程示意图，AI诊断模型的输入数据可以包括：击穿电压Brv_ce、CE漏电流I_ces、GE漏电流I_ges、阈值电压Vth、IGBT饱和压降V_cesat、二极管导通压降V_f、IGBT寄生电容Cies，Coes，Cres、模组开关损耗Eon，Eoff，Err。本申请实施例中不再对上述输入数据的具体含义进行详细介绍。
本申请实施例中采用的测试电路具有多种实现方式，例如该测试电路可以是Iges测试电路、Vth测试电路、Bvces和Ices测试电路、Ices低端测试电路、Vcesat和Vf测试电路。本申请实施例中不限定测试电路的类型。
AI诊断模型的输入数据针对IGBT和MOS两种功率半导体器件分别如下表2和表3所示，这些数据不是MCU运行时默认采样的数据，可以在MCU中额外增加传感器或者采样电路实现数据采集。
表2为IGBT模块的测试部分项目，真实测试项目可以增加，测试项跟需要检测的失效模式对应：
Figure PCTCN2021087861-appb-000001
表3为MOSFET的部分测试项，真实项目可以增加，测试项跟需要检测的失效模式对应：
Figure PCTCN2021087861-appb-000002
如图10所示，为本申请实施例提供的一种运行态数据和静态数据的采集流程示意图。本申请实施例中，汽车运行态对功率半导体器件的行驶数据进行动态采样，在云平台上进行基于AI诊断模型的决策推理。本实施例中的数据采集过程需要功率半导体器件在运行条件下完成，即功率半导体器件整机处在高频开关工作模式的条件下进行。比如汽车运行阶段进行关键数据采样。
在汽车充电、汽车打火、汽车熄火阶段进行关键数据采样，主要包括如下两个阶段：第一阶段，将采集到的数据传入云平台，在云平台上进行基于大数据的AI诊断模型的决策推理，得到推理结果；第二阶段，云平台将推理结果发送给车辆终端设备，车辆终端设备进行数据预处理压缩，并提供断网情况下的紧急响应功能。本实施例中的数据采集工作需 要功率半导体在静态条件下，即功率半导体器件整机未处在低频开关工作模式的条件下进行，比如汽车充电、汽车打火、汽车熄火阶段，进行如上关键数据采样。
通过前述的举例说明可知，本申请实施例中云端和/或端侧具备AI诊断模型的训练和推理能力。本申请实施例通过AI诊断模型判断数据特征，并对特征进行异常度评分，将异常度超标的电机驱动器识别出来，以实现对电机驱动器失效的提前预测。提前诊断汽车电驱动系统因为电机驱动器造成的失效并进行预警，让用户在故障发生之前进行维护，避免驾驶过程中抛锚。另外，端侧在算力受限约束下具备本地训练及决策推理能力，故障预测等模型的精度以及推理的实时性、可靠性都有提升。
需要说明的是，对于前述的各方法实施例，为了简单描述，故将其都表述为一系列的动作组合，但是本领域技术人员应该知悉，本申请并不受所描述的动作顺序的限制，因为依据本申请，某些步骤可以采用其他顺序或者同时进行。其次，本领域技术人员也应该知悉，说明书中所描述的实施例均属于优选实施例，所涉及的动作和模块并不一定是本申请所必须的。
为便于更好的实施本申请实施例的上述方案，下面还提供用于实施上述方案的相关装置。
请参阅图11a所示，本申请实施例提供的一种用于电机驱动器的故障预警装置1100，可以包括：获取模块1101、推理模块1102、预警模块1103，其中，
获取模块，用于获取电机驱动器运行时的运行态数据；
推理模块，用于将所述运行态数据输入到预设的人工智能AI诊断模型，通过所述AI诊断模型输出所述电机驱动器的第一异常度参数，其中，所述AI诊断模型用于根据所述运行态数据提取数据特征，并通过所述数据特征进行决策推理；
预警模块，用于根据所述第一异常度参数对所述电机驱动器进行故障预警。
在本申请的一些实施例中，所述获取模块，还用于获取车辆处于停止阶段时对所述电机驱动器进行低频采样得到的静态数据，所述电机驱动器安装在所述车辆上；
所述推理模块，还用于将所述静态数据输入到所述AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第二异常度参数；
所述预警模块，还用于根据所述第二异常度参数对所述电机驱动器进行故障预警。
在本申请的一些实施例中，所述车辆处于停止阶段，至少包括如下一种：
所述车辆处于充电阶段、所述车辆处于打火阶段、所述车辆处于熄火阶段。
在本申请的一些实施例中，所述预警模块，用于通过所述AI诊断模型对预设的异常条件进行动态更新，以得到更新后的异常条件；当所述第一异常度参数满足所述更新后的异常条件时，对所述电机驱动器进行故障预警。
在本申请的一些实施例中，如图11b所示，所述装置还包括：数据处理模块1104，用于所述获取模块获取电机驱动器运行时的运行态数据之后，对所述运行态数据进行清洗处理，得到清洗处理后的运行态数据；
所述推理模块，用于将所述清洗处理后的运行态数据输入到所述AI诊断模型。
在本申请的一些实施例中，所述装置具体为云端服务器；如图11c所示，所述装置还包括：训练模块1105，
所述获取模块，还用于获取同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
所述训练模块，用于使用所述同类型的多个车辆的训练样本数据和所述多个不同类型的车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
在本申请的一些实施例中，所述装置具体为云端服务器；所述装置还包括：训练模块，
所述获取模块，还用于获取一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
所述训练模块，用于使用所述一个车辆的训练样本数据或者所述同类型的多个车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
在本申请的一些实施例中，所述装置具体为云端服务器；所述预警模块，用于根据所述第一异常度参数生成故障预警信息；向车辆终端设备发送所述故障预警信息。
在本申请的一些实施例中，所述装置具体为车辆终端设备；
当云端服务器向所述车辆终端设备发送所述AI诊断模型时，所述获取模块，还用于接收所述云端服务器发送的AI诊断模型。
在本申请的一些实施例中，所述装置具体为云端服务器；所述装置还包括：训练模块，
所述获取模块，还用于获取多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据；
所述训练模块，用于对所述多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据，进行数据关联关系提取，以得到预训练模型，所述预训练模型用于表示提取到的数据关联关系；向车辆终端设备发送所述预训练模型。
在本申请的一些实施例中，所述装置具体为车辆终端设备；所述装置还包括：训练模块，
所述训练模块，用于接收来自云端服务器的预训练模型，其中，所述预训练模块用于表示所述云端服务器对多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据进行提取后得到的数据关联关系；根据预设的标签数据对所述预训练模型进行模型参数寻优处理，以得到所述预设的AI诊断模型。
在本申请的一些实施例中，所述运行态数据包括：所述电机驱动器运行时默认采集的动态数据；或者，
所述运行态数据包括：通过控制器局域网络CAN总线采集的动态数据。
在本申请的一些实施例中，所述运行态数据包括：车辆处于运行阶段进行高频采样得到的动态数据，所述电机驱动器安装在所述车辆上。
在本申请的一些实施例中，所述运行态数据至少包括如下一种：电机驱动器的参数、冷却系统的参数、电机的参数、电池的参数。
在本申请的一些实施例中，所述电机驱动器的参数包括如下至少一种：直流母线电压、直流母线电流、低压电源电压、车辆行驶状态命令、车辆状态、电机驱动器温度、转子位置电角度、电机驱动器三相电流有效值、电机驱动器三相电流、电机驱动器三相电流采样 值、电机驱动器降额状态；和/或，
所述冷却系统的参数包括如下至少一种：冷却液流量、冷却液温度、油泵目标转速、油泵状态、油泵实际转速、油泵供电电压、油泵的功率半导体器件温度、油泵电流；和/或，
所述电机的参数包括如下至少一种：电磁频率、电机工作模式命令、电机控制器工作状态、电机目标转矩命令、电机当前转矩、电机目标转速命令、电机当前转速、电机温度、电机直轴电压、电机直轴电流给定值、电机直轴电流反馈值、电机交轴电压、电机交轴电流给定值、电机交轴电流反馈值、电机当前校验转矩；和/或，
所述电池的参数包括如下至少一种：额定输出电压、电池容量、最大输出电流。
通过前述实施例的举例说明可知，首先获取电机驱动器运行时的运行态数据，然后将运行态数据输入到预设的AI诊断模型，通过AI诊断模型输出电机驱动器的第一异常度参数，其中，AI诊断模型用于根据运行态数据提取数据特征，并通过数据特征进行决策推理；根据第一异常度参数对电机驱动器进行故障预警。由于本申请实施例中电机驱动器运行时车辆处于运行状态，根据电机驱动器运行时的运行态数据可以对电机驱动器进行故障预警，电机驱动器的运行态数据是电机驱动器运行时采集到的数据，运行态数据能够反映电机驱动器的真实运行状态，基于该运行态数据进行故障预警，可以实现对电机驱动器失效的提前预测，实现对电机驱动器的故障预警。
本申请实施例还提供一种用于电机驱动器的故障预警装置，所述装置包括：
获取模块，用于获取车辆处于停止阶段时对所述电机驱动器进行低频采样得到的静态数据，所述电机驱动器安装在所述车辆上；
推理模块，用于将所述静态数据输入到所述AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第二异常度参数；
预警模块，用于根据所述第二异常度参数对所述电机驱动器进行故障预警。
在一种可能的实现方式中，所述车辆处于停止阶段，至少包括如下一种：
所述车辆处于充电阶段、所述车辆处于打火阶段、所述车辆处于熄火阶段。
在一种可能的实现方式中，所述预警模块，用于通过所述AI诊断模型对预设的异常条件进行动态更新，以得到更新后的异常条件；当所述第二异常度参数满足所述更新后的异常条件时，对所述电机驱动器进行故障预警。
在一种可能的实现方式中，所述装置还包括：数据处理模块，用于所述获取模块获取车辆处于停止阶段时对所述电机驱动器进行低频采样得到的静态数据之后，对所述静态数据进行清洗处理，得到清洗处理后的静态数据；
所述推理模块，用于将所述清洗处理后的静态数据输入到所述AI诊断模型。
在一种可能的实现方式中，所述装置具体为云端服务器；所述装置还包括：训练模块，
所述获取模块，还用于获取同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
所述训练模块，用于使用所述同类型的多个车辆的训练样本数据和所述多个不同类型的车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
在一种可能的实现方式中，所述装置具体为云端服务器；所述装置还包括：训练模块，
所述获取模块，还用于获取一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
所述训练模块，用于使用所述一个车辆的训练样本数据或者所述同类型的多个车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
在一种可能的实现方式中，所述装置具体为云端服务器；所述预警模块，用于根据所述第二异常度参数生成故障预警信息；向车辆终端设备发送所述故障预警信息。
在一种可能的实现方式中，所述装置具体为车辆终端设备；
当云端服务器向所述车辆终端设备发送所述AI诊断模型时，所述获取模块，还用于接收所述云端服务器发送的AI诊断模型。
在一种可能的实现方式中，所述装置具体为云端服务器；所述装置还包括：训练模块，
所述获取模块，还用于获取多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据；
所述训练模块，用于对所述多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据，进行数据关联关系提取，以得到预训练模型，所述预训练模型用于表示提取到的数据关联关系；向车辆终端设备发送所述预训练模型。
在一种可能的实现方式中，所述装置具体为车辆终端设备；所述装置还包括：训练模块，
所述训练模块，用于接收来自云端服务器的预训练模型，其中，所述预训练模块用于表示所述云端服务器对多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据进行提取后得到的数据关联关系；根据预设的标签数据对所述预训练模型进行模型参数寻优处理，以得到所述预设的AI诊断模型。
需要说明的是，上述装置各模块/单元之间的信息交互、执行过程等内容，由于与本申请方法实施例基于同一构思，其带来的技术效果与本申请方法实施例相同，具体内容可参见本申请前述所示的方法实施例中的叙述，此处不再赘述。
本申请实施例还提供一种计算机存储介质，其中，该计算机存储介质存储有程序，该程序执行包括上述方法实施例中记载的部分或全部步骤。
接下来介绍本申请实施例提供的另一种用于电机驱动器的故障预警装置，请参阅图12所示，用于电机驱动器的故障预警装置1200包括：
接收器1201、发射器1202、处理器1203和存储器1204(其中用于电机驱动器的故障预警装置1200中的处理器1203的数量可以一个或多个，图12中以一个处理器为例)。在本申请的一些实施例中，接收器1201、发射器1202、处理器1203和存储器1204可通过总线或其它方式连接，其中，图12中以通过总线连接为例。
存储器1204可以包括只读存储器和随机存取存储器，并向处理器1203提供指令和数据。存储器1204的一部分还可以包括非易失性随机存取存储器(non-volatile random access memory，NVRAM)。存储器1204存储有操作系统和操作指令、可执行模块或者数据结构，或者它们的子集，或者它们的扩展集，其中，操作指令可包括各种操作指令，用于 实现各种操作。操作系统可包括各种系统程序，用于实现各种基础业务以及处理基于硬件的任务。
处理器1203控制用于电机驱动器的故障预警装置的操作，处理器1203还可以称为中央处理单元(central processing unit，CPU)。具体的应用中，用于电机驱动器的故障预警装置的各个组件通过总线系统耦合在一起，其中总线系统除包括数据总线之外，还可以包括电源总线、控制总线和状态信号总线等。但是为了清楚说明起见，在图中将各种总线都称为总线系统。
上述本申请实施例揭示的方法可以应用于处理器1203中，或者由处理器1203实现。处理器1203可以是一种集成电路芯片，具有信号的处理能力。在实现过程中，上述方法的各步骤可以通过处理器1203中的硬件的集成逻辑电路或者软件形式的指令完成。上述的处理器1203可以是通用处理器、数字信号处理器(digital signal processing，DSP)、专用集成电路(application specific integrated circuit，ASIC)、现场可编程门阵列(field-programmable gate array，FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。可以实现或者执行本申请实施例中的公开的各方法、步骤及逻辑框图。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。结合本申请实施例所公开的方法的步骤可以直接体现为硬件译码处理器执行完成，或者用译码处理器中的硬件及软件模块组合执行完成。软件模块可以位于随机存储器，闪存、只读存储器，可编程只读存储器或者电可擦写可编程存储器、寄存器等本领域成熟的存储介质中。该存储介质位于存储器1204，处理器1203读取存储器1204中的信息，结合其硬件完成上述方法的步骤。
接收器1201可用于接收输入的数字或字符信息，以及产生与用于电机驱动器的故障预警装置的相关设置以及功能控制有关的信号输入，发射器1202可包括显示屏等显示设备，发射器1202可用于通过外接接口输出数字或字符信息。
本申请实施例中，处理器1203，用于执行如下步骤图1至图6所示的方法步骤。
在另一种可能的设计中，当用于电机驱动器的故障预警装置为终端内的芯片时，芯片包括：处理单元和通信单元，所述处理单元例如可以是处理器，所述通信单元例如可以是输入/输出接口、管脚或电路等。该处理单元可执行存储单元存储的计算机执行指令，以使该终端内的芯片执行上述第一方面任意一项的方法。可选地，所述存储单元为所述芯片内的存储单元，如寄存器、缓存等，所述存储单元还可以是所述终端内的位于所述芯片外部的存储单元，如只读存储器(read-only memory，ROM)或可存储静态信息和指令的其他类型的静态存储设备，随机存取存储器(random access memory，RAM)等。
其中，上述任一处提到的处理器，可以是一个通用中央处理器，微处理器，ASIC，或一个或多个用于控制上述第一方面方法的程序执行的集成电路。
另外需说明的是，以上所描述的装置实施例仅仅是示意性的，其中所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。另外，本申请提供的装置实施例附图中，模块之间的连接关系表示它们之间具有通信连接，具体可以实现为一条 或多条通信总线或信号线。
通过以上的实施方式的描述，所属领域的技术人员可以清楚地了解到本申请可借助软件加必需的通用硬件的方式来实现，当然也可以通过专用硬件包括专用集成电路、专用CPU、专用存储器、专用元器件等来实现。一般情况下，凡由计算机程序完成的功能都可以很容易地用相应的硬件来实现，而且，用来实现同一功能的具体硬件结构也可以是多种多样的，例如模拟电路、数字电路或专用电路等。但是，对本申请而言更多情况下软件程序实现是更佳的实施方式。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品存储在可读取的存储介质中，如计算机的软盘、U盘、移动硬盘、ROM、RAM、磁碟或者光盘等，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行本申请各个实施例所述的方法。
在上述实施例中，可以全部或部分地通过软件、硬件、固件或者其任意组合来实现。当使用软件实现时，可以全部或部分地以计算机程序产品的形式实现。
所述计算机程序产品包括一个或多个计算机指令。在计算机上加载和执行所述计算机程序指令时，全部或部分地产生按照本申请实施例所述的流程或功能。所述计算机可以是通用计算机、专用计算机、计算机网络、或者其他可编程装置。所述计算机指令可以存储在计算机可读存储介质中，或者从一个计算机可读存储介质向另一计算机可读存储介质传输，例如，所述计算机指令可以从一个网站站点、计算机、服务器或数据中心通过有线(例如同轴电缆、光纤、数字用户线(DSL))或无线(例如红外、无线、微波等)方式向另一个网站站点、计算机、服务器或数据中心进行传输。所述计算机可读存储介质可以是计算机能够存储的任何可用介质或者是包含一个或多个可用介质集成的服务器、数据中心等数据存储设备。所述可用介质可以是磁性介质，(例如，软盘、硬盘、磁带)、光介质(例如，DVD)、或者半导体介质(例如固态硬盘(Solid State Disk，SSD))等。

Claims (45)

  1. 一种用于电机驱动器的故障预警方法，其特征在于，包括：
    获取电机驱动器运行时的运行态数据；
    将所述运行态数据输入到预设的人工智能AI诊断模型，通过所述AI诊断模型输出所述电机驱动器的第一异常度参数，其中，所述AI诊断模型用于根据所述运行态数据提取数据特征，并通过所述数据特征进行决策推理；
    根据所述第一异常度参数对所述电机驱动器进行故障预警。
  2. 根据权利要求1所述的方法，其特征在于，所述方法还包括：
    获取车辆处于停止阶段时对所述电机驱动器进行低频采样得到的静态数据，所述电机驱动器安装在所述车辆上；
    将所述静态数据输入到所述AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第二异常度参数；
    根据所述第二异常度参数对所述电机驱动器进行故障预警。
  3. 根据权利要求1或2所述的方法，其特征在于，所述车辆处于停止阶段，至少包括如下一种：
    所述车辆处于充电阶段、所述车辆处于打火阶段、所述车辆处于熄火阶段。
  4. 根据权利要求1至3中任一项所述的方法，其特征在于，所述根据所述第一异常度参数对所述电机驱动器进行故障预警，包括：
    通过所述AI诊断模型对预设的异常条件进行动态更新，以得到更新后的异常条件；
    当所述第一异常度参数满足所述更新后的异常条件时，对所述电机驱动器进行故障预警。
  5. 根据权利要求1至4中任一项所述的方法，其特征在于，所述获取电机驱动器运行时的运行态数据之后，所述方法还包括：
    对所述运行态数据进行清洗处理，得到清洗处理后的运行态数据；
    所述将所述运行态数据输入到预设的人工智能AI诊断模型包括：
    将所述清洗处理后的运行态数据输入到所述AI诊断模型。
  6. 根据权利要求1至5中任一项所述的方法，其特征在于，所述方法还包括：
    云端服务器获取同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
    所述云端服务器使用所述同类型的多个车辆的训练样本数据和所述多个不同类型的车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
  7. 根据权利要求1至5中任一项所述的方法，其特征在于，所述方法还包括：
    云端服务器获取一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
    所述云端服务器使用所述一个车辆的训练样本数据或者所述同类型的多个车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
  8. 根据权利要求1至5中任一项所述的方法，其特征在于，当云端服务器通过所述AI诊断模型输出所述电机驱动器的第一异常度参数时，所述根据所述第一异常度参数对所述 电机驱动器进行故障预警，包括：
    所述云端服务器根据所述第一异常度参数生成故障预警信息；
    所述云端服务器向车辆终端设备发送所述故障预警信息。
  9. 根据权利要求1至5中任一项所述的方法，其特征在于，当云端服务器向车辆终端设备发送所述AI诊断模型时，所述方法还包括：所述车辆终端设备接收所述云端服务器发送的AI诊断模型；
    所述将所述运行态数据输入到预设的人工智能AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第一异常度参数，包括：
    所述车辆终端设备将所述运行态数据输入到所述AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第一异常度参数。
  10. 根据权利要求1至5中任一项所述的方法，其特征在于，所述方法还包括：
    云端服务器获取多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据；
    所述云端服务器对所述多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据，进行数据关联关系提取，以得到预训练模型，所述预训练模型用于表示提取到的数据关联关系；
    所述云端服务器向车辆终端设备发送所述预训练模型。
  11. 根据权利要求10所述的方法，其特征在于，所述方法，还包括：
    所述车辆终端设备接收来自所述云端服务器的预训练模型；
    所述车辆终端设备根据预设的标签数据对所述预训练模型进行模型参数寻优处理，以得到所述预设的AI诊断模型。
  12. 根据权利要求1至11中任一项所述的方法，其特征在于，所述运行态数据包括：所述电机驱动器运行时默认采集的动态数据；或者，
    所述运行态数据包括：通过控制器局域网络CAN总线采集的动态数据。
  13. 根据权利要求1至11中任一项所述的方法，其特征在于，所述运行态数据包括：车辆处于运行阶段进行高频采样得到的动态数据，所述电机驱动器安装在所述车辆上。
  14. 根据权利要求1至13中任一项所述的方法，其特征在于，所述运行态数据至少包括如下一种：电机驱动器的参数、冷却系统的参数、电机的参数、电池的参数。
  15. 根据权利要求14所述的方法，其特征在于，所述电机驱动器的参数包括如下至少一种：直流母线电压、直流母线电流、低压电源电压、车辆行驶状态命令、车辆状态、电机驱动器温度、转子位置电角度、电机驱动器三相电流有效值、电机驱动器三相电流、电机驱动器三相电流采样值、电机驱动器降额状态；和/或，
    所述冷却系统的参数包括如下至少一种：冷却液流量、冷却液温度、油泵目标转速、油泵状态、油泵实际转速、油泵供电电压、油泵的功率半导体器件温度、油泵电流；和/或，
    所述电机的参数包括如下至少一种：电磁频率、电机工作模式命令、电机控制器工作状态、电机目标转矩命令、电机当前转矩、电机目标转速命令、电机当前转速、电机温度、电机直轴电压、电机直轴电流给定值、电机直轴电流反馈值、电机交轴电压、电机交轴电 流给定值、电机交轴电流反馈值、电机当前校验转矩；和/或，
    所述电池的参数包括如下至少一种：额定输出电压、电池容量、最大输出电流。
  16. 一种用于电机驱动器的故障预警装置，其特征在于，包括：
    获取模块，用于获取电机驱动器运行时的运行态数据；
    推理模块，用于将所述运行态数据输入到预设的人工智能AI诊断模型，通过所述AI诊断模型输出所述电机驱动器的第一异常度参数，其中，所述AI诊断模型用于根据所述运行态数据提取数据特征，并通过所述数据特征进行决策推理；
    预警模块，用于根据所述第一异常度参数对所述电机驱动器进行故障预警。
  17. 根据权利要求16所述的装置，其特征在于，所述获取模块，还用于获取车辆处于停止阶段时对所述电机驱动器进行低频采样得到的静态数据，所述电机驱动器安装在所述车辆上；
    所述推理模块，还用于将所述静态数据输入到所述AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第二异常度参数；
    所述预警模块，还用于根据所述第二异常度参数对所述电机驱动器进行故障预警。
  18. 根据权利要求16或17所述的装置，其特征在于，所述车辆处于停止阶段，至少包括如下一种：
    所述车辆处于充电阶段、所述车辆处于打火阶段、所述车辆处于熄火阶段。
  19. 根据权利要求16至18中任一项所述的装置，其特征在于，所述预警模块，用于通过所述AI诊断模型对预设的异常条件进行动态更新，以得到更新后的异常条件；当所述第一异常度参数满足所述更新后的异常条件时，对所述电机驱动器进行故障预警。
  20. 根据权利要求16至19中任一项所述的装置，其特征在于，所述装置还包括：数据处理模块，用于所述获取模块获取电机驱动器运行时的运行态数据之后，对所述运行态数据进行清洗处理，得到清洗处理后的运行态数据；
    所述推理模块，用于将所述清洗处理后的运行态数据输入到所述AI诊断模型。
  21. 根据权利要求16至20中任一项所述的装置，其特征在于，所述装置具体为云端服务器；所述装置还包括：训练模块，
    所述获取模块，还用于获取同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
    所述训练模块，用于使用所述同类型的多个车辆的训练样本数据和所述多个不同类型的车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
  22. 根据权利要求16至20中任一项所述的装置，其特征在于，所述装置具体为云端服务器；所述装置还包括：训练模块，
    所述获取模块，还用于获取一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
    所述训练模块，用于使用所述一个车辆的训练样本数据或者所述同类型的多个车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
  23. 根据权利要求16至20中任一项所述的装置，其特征在于，所述装置具体为云端服 务器；所述预警模块，用于根据所述第一异常度参数生成故障预警信息；向车辆终端设备发送所述故障预警信息。
  24. 根据权利要求16至20中任一项所述的装置，其特征在于，所述装置具体为车辆终端设备；
    当云端服务器向所述车辆终端设备发送所述AI诊断模型时，所述获取模块，还用于接收所述云端服务器发送的AI诊断模型。
  25. 根据权利要求16至20中任一项所述的装置，其特征在于，所述装置具体为云端服务器；所述装置还包括：训练模块，
    所述获取模块，还用于获取多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据；
    所述训练模块，用于对所述多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据，进行数据关联关系提取，以得到预训练模型，所述预训练模型用于表示提取到的数据关联关系；向车辆终端设备发送所述预训练模型。
  26. 根据权利要求16至20中任一项所述的装置，其特征在于，所述装置具体为车辆终端设备；所述装置还包括：训练模块，
    所述训练模块，用于接收来自云端服务器的预训练模型，其中，所述预训练模块用于表示所述云端服务器对多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据进行提取后得到的数据关联关系；根据预设的标签数据对所述预训练模型进行模型参数寻优处理，以得到所述预设的AI诊断模型。
  27. 根据权利要求16至26中任一项所述的装置，其特征在于，所述运行态数据包括：所述电机驱动器运行时默认采集的动态数据；或者，
    所述运行态数据包括：通过控制器局域网络CAN总线采集的动态数据。
  28. 根据权利要求16至26中任一项所述的装置，其特征在于，所述运行态数据包括：车辆处于运行阶段进行高频采样得到的动态数据，所述电机驱动器安装在所述车辆上。
  29. 根据权利要求16至28中任一项所述的装置，其特征在于，所述运行态数据至少包括如下一种：电机驱动器的参数、冷却系统的参数、电机的参数、电池的参数。
  30. 根据权利要求29所述的装置，其特征在于，所述电机驱动器的参数包括如下至少一种：直流母线电压、直流母线电流、低压电源电压、车辆行驶状态命令、车辆状态、电机驱动器温度、转子位置电角度、电机驱动器三相电流有效值、电机驱动器三相电流、电机驱动器三相电流采样值、电机驱动器降额状态；和/或，
    所述冷却系统的参数包括如下至少一种：冷却液流量、冷却液温度、油泵目标转速、油泵状态、油泵实际转速、油泵供电电压、油泵的功率半导体器件温度、油泵电流；和/或，
    所述电机的参数包括如下至少一种：电磁频率、电机工作模式命令、电机控制器工作状态、电机目标转矩命令、电机当前转矩、电机目标转速命令、电机当前转速、电机温度、电机直轴电压、电机直轴电流给定值、电机直轴电流反馈值、电机交轴电压、电机交轴电流给定值、电机交轴电流反馈值、电机当前校验转矩；和/或，
    所述电池的参数包括如下至少一种：额定输出电压、电池容量、最大输出电流。
  31. 一种用于电机驱动器的故障预警方法，其特征在于，所述方法包括：
    获取车辆处于停止阶段时对所述电机驱动器进行低频采样得到的静态数据，所述电机驱动器安装在所述车辆上；
    将所述静态数据输入到所述AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第二异常度参数；
    根据所述第二异常度参数对所述电机驱动器进行故障预警。
  32. 根据权利要求31所述的方法，其特征在于，所述车辆处于停止阶段，至少包括如下一种：
    所述车辆处于充电阶段、所述车辆处于打火阶段、所述车辆处于熄火阶段。
  33. 根据权利要求31或32所述的方法，其特征在于，所述根据所述第二异常度参数对所述电机驱动器进行故障预警，包括：
    通过所述AI诊断模型对预设的异常条件进行动态更新，以得到更新后的异常条件；
    当所述第二异常度参数满足所述更新后的异常条件时，对所述电机驱动器进行故障预警。
  34. 根据权利要求31至33中任一项所述的方法，其特征在于，所述获取车辆处于停止阶段时对所述电机驱动器进行低频采样得到的静态数据之后，所述方法还包括：
    对所述静态数据进行清洗处理，得到清洗处理后的静态数据；
    所述将所述静态数据输入到预设的人工智能AI诊断模型包括：
    将所述清洗处理后的静态数据输入到所述AI诊断模型。
  35. 根据权利要求31至34中任一项所述的方法，其特征在于，所述方法还包括：
    云端服务器获取同类型的多个车辆的训练样本数据和多个不同类型的车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
    所述云端服务器使用所述同类型的多个车辆的训练样本数据和所述多个不同类型的车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
  36. 根据权利要求31至34中任一项所述的方法，其特征在于，所述方法还包括：
    云端服务器获取一个车辆的训练样本数据或者同类型的多个车辆的训练样本数据，所述电机驱动器安装在所述车辆上；
    所述云端服务器使用所述一个车辆的训练样本数据或者所述同类型的多个车辆的训练样本数据，进行基于故障预警任务的模型训练，以得到所述预设的AI诊断模型。
  37. 根据权利要求31至34中任一项所述的方法，其特征在于，当云端服务器通过所述AI诊断模型输出所述电机驱动器的第一异常度参数时，所述根据所述第二异常度参数对所述电机驱动器进行故障预警，包括：
    所述云端服务器根据所述第二异常度参数生成故障预警信息；
    所述云端服务器向车辆终端设备发送所述故障预警信息。
  38. 根据权利要求31至34中任一项所述的方法，其特征在于，当云端服务器向车辆终端设备发送所述AI诊断模型时，所述方法还包括：所述车辆终端设备接收所述云端服务器发送的AI诊断模型；
    所述将所述静态数据输入到预设的人工智能AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第二异常度参数，包括：
    所述车辆终端设备将所述静态数据输入到所述AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第二异常度参数。
  39. 根据权利要求31至34中任一项所述的方法，其特征在于，所述方法还包括：
    云端服务器获取多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据；
    所述云端服务器对所述多个不同类型的车辆的训练样本数据，或者一个车辆的训练样本数据，或者同类型的多个车辆的训练样本数据，进行数据关联关系提取，以得到预训练模型，所述预训练模型用于表示提取到的数据关联关系；
    所述云端服务器向车辆终端设备发送所述预训练模型。
  40. 根据权利要求31至34中任一项所述的方法，其特征在于，所述方法，还包括：
    所述车辆终端设备接收来自所述云端服务器的预训练模型；
    所述车辆终端设备根据预设的标签数据对所述预训练模型进行模型参数寻优处理，以得到所述预设的AI诊断模型。
  41. 一种用于电机驱动器的故障预警装置，其特征在于，所述装置包括：
    获取模块，用于获取车辆处于停止阶段时对所述电机驱动器进行低频采样得到的静态数据，所述电机驱动器安装在所述车辆上；
    推理模块，用于将所述静态数据输入到所述AI诊断模型中，通过所述AI诊断模型输出所述电机驱动器的第二异常度参数；
    预警模块，用于根据所述第二异常度参数对所述电机驱动器进行故障预警。
  42. 一种用于电机驱动器的故障预警装置，其特征在于，包括：
    存储器，存储有可执行的程序指令；和，
    处理器，所述处理器用于与所述存储器耦合，读取并执行所述存储器中的指令，以使所述装置实现如权利要求1至15，或者31至40中任一所述的方法。
  43. 根据权利要求42所述的装置，其特征在于，所述装置为云端服务器、云端服务器中的芯片、车辆终端设备、或者车辆终端设备中的芯片。
  44. 一种计算机可读存储介质，包括指令，当其在计算机上运行时，使得计算机执行如权利要求1至15，或者31至40中任一所述的方法。
  45. 一种包含指令的计算机程序产品，当其在计算机上运行时，使得计算机执行如权利要求1至15，或者31至40中任意一项所述的方法。
PCT/CN2021/087861 2021-04-16 2021-04-16 一种用于电机驱动器的故障预警方法和装置 WO2022217597A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN202180002804.3A CN115398439A (zh) 2021-04-16 2021-04-16 一种用于电机驱动器的故障预警方法和装置
PCT/CN2021/087861 WO2022217597A1 (zh) 2021-04-16 2021-04-16 一种用于电机驱动器的故障预警方法和装置

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2021/087861 WO2022217597A1 (zh) 2021-04-16 2021-04-16 一种用于电机驱动器的故障预警方法和装置

Publications (1)

Publication Number Publication Date
WO2022217597A1 true WO2022217597A1 (zh) 2022-10-20

Family

ID=83639430

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/087861 WO2022217597A1 (zh) 2021-04-16 2021-04-16 一种用于电机驱动器的故障预警方法和装置

Country Status (2)

Country Link
CN (1) CN115398439A (zh)
WO (1) WO2022217597A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116016120A (zh) * 2023-01-05 2023-04-25 中国联合网络通信集团有限公司 故障处理方法、终端设备和可读存储介质
CN116466689B (zh) * 2023-06-19 2023-09-05 广汽埃安新能源汽车股份有限公司 故障诊断方法及装置

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000206215A (ja) * 1999-01-20 2000-07-28 Honda Motor Co Ltd 車両用電源診断装置
CN106612094A (zh) * 2015-10-26 2017-05-03 发那科株式会社 机器学习装置及方法以及寿命预测装置及电动机驱动装置
CN110490248A (zh) * 2019-08-16 2019-11-22 集美大学 一种电力电子变换器故障诊断方法、终端设备及存储介质
CN110874079A (zh) * 2018-08-30 2020-03-10 Abb瑞士股份有限公司 用于监测电驱动器的状况的方法和系统

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2000206215A (ja) * 1999-01-20 2000-07-28 Honda Motor Co Ltd 車両用電源診断装置
CN106612094A (zh) * 2015-10-26 2017-05-03 发那科株式会社 机器学习装置及方法以及寿命预测装置及电动机驱动装置
CN110874079A (zh) * 2018-08-30 2020-03-10 Abb瑞士股份有限公司 用于监测电驱动器的状况的方法和系统
CN110490248A (zh) * 2019-08-16 2019-11-22 集美大学 一种电力电子变换器故障诊断方法、终端设备及存储介质

Also Published As

Publication number Publication date
CN115398439A (zh) 2022-11-25

Similar Documents

Publication Publication Date Title
WO2022217597A1 (zh) 一种用于电机驱动器的故障预警方法和装置
CN108614167B (zh) 一种功率组件故障记录诊断系统及方法
CN104965148A (zh) 一种电机驱动系统中逆变器功率管开路故障实时检测方法
CN111830409A (zh) 一种基于深度神经网络的电机热故障诊断方法及装置
US20210261003A1 (en) Monitoring and Identifying Sensor Failure in an Electric Drive System
CN106610481A (zh) 诊断环境友好车辆的电流传感器的装置和方法
US11156666B2 (en) System and methods for fault detection
Hussain et al. Fault Detection and Identification Using Deep Learning Algorithms in InductionMotors.
Li et al. Open circuit fault diagnosis strategy of PMSM drive system based on grey prediction theory for industrial robot
CN107415936B (zh) 电池预热方法、装置和混合动力控制设备
Liu et al. Fault diagnosis of new energy vehicles based on improved machine learning
CN113646988A (zh) 连接盒、电动汽车的测试台架、控制方法及装置
CN103939374B (zh) 一种有刷电子风扇驱动保护系统及自诊断方法
CN204857891U (zh) 一种基于互联网的电池管理诊断控制系统
CN105204492B (zh) 一种元器件故障诊断方法、控制器及空调
US20170232855A1 (en) High voltage battery charger and methods of operation
WO2024198109A1 (zh) 一种基于gpr-rc模型的电机温度估算方法及装置
CN108267684A (zh) 一种变频器故障诊断方法
Zhang et al. Artificial Intelligence Technique-Based EV Powertrain Condition Monitoring and Fault Diagnosis: A Review
CN108123419B (zh) 一种过电流保护方法、装置、芯片和控制电路
CN113665365B (zh) 电动汽车运行模式的控制方法、装置、电动汽车及介质
Ginart et al. Self-healing from a PHM perspective
Xu et al. Fault propagation analysis of IGBT fault in CRH5 traction system based on signed directed graph
CN106950058A (zh) 基于负载转矩特征分析法的齿轮故障诊断装置
CN105811371A (zh) 直流电机控制系统故障诊断方法及装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21936472

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 22.02.2024)