WO2022262073A1 - 通过样本掺杂来训练目标设备的设备模型的方法及系统 - Google Patents

通过样本掺杂来训练目标设备的设备模型的方法及系统 Download PDF

Info

Publication number
WO2022262073A1
WO2022262073A1 PCT/CN2021/108158 CN2021108158W WO2022262073A1 WO 2022262073 A1 WO2022262073 A1 WO 2022262073A1 CN 2021108158 W CN2021108158 W CN 2021108158W WO 2022262073 A1 WO2022262073 A1 WO 2022262073A1
Authority
WO
WIPO (PCT)
Prior art keywords
sample
signal
model
sample set
doping
Prior art date
Application number
PCT/CN2021/108158
Other languages
English (en)
French (fr)
Inventor
郭春林
郭尔富
Original Assignee
华北电力大学
北京大地纵横科技有限公司
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 华北电力大学, 北京大地纵横科技有限公司 filed Critical 华北电力大学
Publication of WO2022262073A1 publication Critical patent/WO2022262073A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
    • 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
    • G01—MEASURING; TESTING
    • G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M99/00—Subject matter not provided for in other groups of this subclass
    • G01M99/008—Subject matter not provided for in other groups of this subclass by doing functionality tests
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
    • G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing

Definitions

  • the present invention relates to the technical field of artificial intelligence, and more specifically, to a method and system for training a device model of a target device through sample doping.
  • the data quality of the sample signal/sample data affects the training effect of the model.
  • the number or rate of failures of some types of equipment during operation is relatively low, so the sample signal/sample data of this type of equipment in normal operation has a relatively large amount of data, while abnormal operation or failure
  • the data size of the sample signal/sample data is small. In this case, abnormal sample signals/sample data with a small amount of data cannot meet the requirements of model training or testing.
  • the present invention proposes a method and system for training a device model of a target device through sample doping, so as to solve the situation that the amount of sample signal/sample data is small during abnormal operation or failure.
  • the target model is trained based on the normal sample set, the abnormal sample set and the preset training algorithm, so as to obtain the trained target model.
  • the at least one sample signal includes: vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness.
  • At least one sample signal among the at least one sample signal is a vibration/acoustic emission signal collected by a sensor close to the device casing.
  • At least one sample signal of the at least one sample signal is a sound signal collected outside the device.
  • a method for testing a device model of a target device through sample doping comprising:
  • the target model is tested based on the normal sample set and the abnormal sample set, so as to determine the performance index of the target model based on the test results.
  • selecting means for responding to the received model training request, selecting a target device from a plurality of devices based on the training request, and determining a device model associated with the target device;
  • determining means configured to acquire attribute information associated with the device model, and determine at least one sample signal associated with the device model based on the attribute information
  • An obtaining device configured to perform signal collection or signal simulation on the at least one sample signal when the target device is in a normal operating state, so as to obtain a normal sample set including the at least one sample signal;
  • a processing device configured to select a doping device from a plurality of devices based on the attribute information, and perform signal acquisition or signal simulation on at least one sample signal when the doping device is in a predetermined operating state, so as to obtain a doping sample set;
  • the training device is used to train the target model based on the normal sample set, the abnormal sample set and a preset training algorithm, so as to obtain the trained target model.
  • At least one sample signal among the at least one sample signal is a vibration/acoustic emission signal collected by a sensor close to the device casing.
  • At least one sample signal of the at least one sample signal is a sound signal collected outside the device.
  • selecting means configured to select a target device associated with the device model from a plurality of devices based on the test request in response to the received test request for the device model;
  • determining means configured to acquire attribute information associated with the device model, and determine at least one sample signal associated with the device model based on the attribute information
  • a doping device configured to use at least one sample signal in the doped sample set to perform sample signal doping on at least one corresponding sample signal in the normal sample set, so as to obtain an abnormal sample set corresponding to the normal sample set;
  • the testing device is used for testing the target model based on the normal sample set and the abnormal sample set, so as to determine the performance index of the target model based on the test result.
  • the normal samples are generated by collecting or simulating signals under normal operating conditions of the target device.
  • the types of sample signals include but are not limited to vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image, brightness, and can also be a combination of various signals.
  • the abnormal sample is generated by superimposing a certain amount of impurity components on the signal of the normal sample.
  • adding a certain amount of impurity components in normal samples means adding a certain proportion of impurity components in normal samples according to amplitude, amplitude square, energy value, energy value square (peak value or average value).
  • Superimposing a certain amount of impurity components in normal samples means superimposing a certain amount of impurity components in normal samples with a certain time difference (the time difference is determined according to correlation or mutual information analysis. 1) Correlation between normal sample signal and impurity component signal The time when the sex or mutual information is maximum is used as the reference, and a fixed offset time or offset phase is added. 2) Add a fixed offset time or offset phase based on the average value at the moment of maximum correlation or mutual information between this type of normal sample signal and this type of impurity component signal. ).
  • Adding a certain amount of impurity components in the normal sample is adding the same signal produced by other objects in the normal sample.
  • adding a certain amount of impurity components in the normal sample is to increase the signal of the target equipment in the abnormal state (including the signal in the fault state, hidden danger state, abnormal operation state, abnormal input state and other operating states) in the normal sample.
  • a doping-based model evaluation method is provided, and the model is used for diagnosing, judging or identifying the state of a target device. in:
  • Generate normal samples by collecting or simulating signals under normal operating conditions of the target device.
  • Abnormal samples are generated by adding a certain amount of impurity components to normal samples.
  • the model is tested with the normal samples and abnormal samples, and then the performance of the model is evaluated according to the test results.
  • adding a certain amount of impurity components in the normal sample is to increase a certain proportion of impurity components in the normal sample according to the amplitude, the square of the amplitude, the energy value, and the square of the energy value.
  • the accuracy of the model is also evaluated by reducing the amount of impurity components in abnormal samples.
  • Adding a certain amount of impurity components in the normal sample is adding the same signal produced by other objects in the normal sample.
  • the increase of a certain amount of impurity components in the normal sample is to increase the signal of the abnormal state of the target device in the normal sample.
  • FIG. 1 is a flow chart of a method for training a device model of a target device through sample doping according to an embodiment of the present invention
  • FIG. 2 is a flowchart of a method for testing a device model of a target device through sample doping according to an embodiment of the present invention
  • FIG. 3 is a schematic structural diagram of a system for training a device model of a target device through sample doping according to an embodiment of the present invention
  • FIG. 4 is a schematic structural diagram of a system for testing a device model of a target device through sample doping according to an embodiment of the present invention.
  • FIG. 1 is a flowchart of a method 100 for training a device model of a target device through sample doping according to an embodiment of the present invention.
  • Method 100 starts at step 101 .
  • a target device in response to a received model training request, a target device is selected from a plurality of devices based on the training request, and a device model associated with the target device is determined.
  • a device model associated with the target device is determined.
  • various types and/or various sizes of equipment are widely used in various locations, production links, monitoring links, etc. For this reason, if it is necessary to determine the operating status of the target device, or obtain parameters of the target device, etc., it is necessary to determine a model or device model associated with the target device.
  • the model of each different type of equipment or the equipment model can be used to determine the operating state of the equipment, obtain the operating parameters of the equipment, and so on.
  • a model training request needs to be generated and sent to the processing device for training or testing the model or the device model.
  • the model training request includes the name, location, identifier, etc. of the target device.
  • the processing device selects a target device from the plurality of devices based on the training request. For example, the processing device extracts the name, location, identifier, etc. of the target device from the model training request, and determines the target device based on the name, location, and/or identifier, etc. of the target device.
  • the target device After the target device is determined, it is necessary to select a device model that needs to be trained and is associated with the target device from models or device models associated with multiple different devices. To this end, after the name, location, and identifier of the target device are determined, the name, location, and identifier of the target device can be used to search in the model library to determine the device model associated with the target device.
  • each model or device model has attribute information, and the attribute information is used to describe various attributes of the model or device model.
  • the various attributes are, for example: input parameters, output parameters, model type, model effect, model accuracy, device type, device name, device identifier, and the like.
  • the equipment model has various attributes, and for example, the input parameters, output parameters, model type, model role, model accuracy, etc. of the equipment model can be determined through the attribute information of the equipment model.
  • step 102 attribute information associated with the device model is acquired, and at least one sample signal associated with the device model is determined based on the attribute information.
  • the input parameters, output parameters, model type, model function, model accuracy, etc. of the equipment model can be determined through the attribute information of the equipment model.
  • the target device involved in the device model may be determined by parsing the attribute information of the device model.
  • a target device can be any type of device.
  • the attribute information of the device model may also include information of various sample signals associated with the target device.
  • the device identifier or device name of the target device can be used to search in the sample signal information library to obtain at least one sample signal associated with the target device Information.
  • the at least one sample signal includes: vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness.
  • one or more sample signals of vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness can be used to characterize, train, test, describe target device. It should be understood that this application is only described by taking vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness as examples, and those skilled in the art should understand Note that this application can use any reasonable sample signal. In practical scenarios, various types of sensors can be used to acquire any one of vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image, and brightness.
  • step 103 when the target device is in a normal operating state, signal acquisition or signal simulation is performed on the at least one sample signal, so as to obtain a normal sample set including the at least one sample signal.
  • the normal sample set is one of the data collected by sensors such as vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness when the target device is running normally.
  • a sample signal composed of one or more sample signals or data.
  • the present application performs signal acquisition or signal simulation on each sample signal in at least one sample signal, so as to obtain the at least two A collection of normal samples for a sample signal.
  • the at least two sample signals are vibration signal, sound signal and voltage signal
  • the normal sample set includes a plurality of samples arranged in order of sample sampling time, wherein each sample includes vibration signal, sound signal and voltage signal , and each sample has a sampling time. That is, each sample in the set of normal samples is a signal group or signal set having a sample time and including each sample signal at that sample time.
  • the normal sample set can include at least two sample subsets, and each sample subset is a vibration sample signal subset, an acoustic emission sample signal subset, a sound sample signal subset or an electric field strength sample signal subset, etc. It should be understood that the division of signal subsets is only for data storage or data display.
  • each sample includes each of at least two sample signals.
  • a plurality of sample signal groups are included in the normal sample set, and each sample signal group includes a single vibration sample signal, a single acoustic emission sample signal and a single sound sample signal, for example, each sample signal group is ⁇ vibration sample signal, acoustic sample signal Transmit sample signal, sound sample signal>. It should be appreciated that each sample signal group can be considered as a sample in the set of normal samples.
  • At least one of the at least one sample signal/each of the at least one sample signal is a vibration/acoustic emission signal collected by a sensor that is closely attached to a device casing of the target device.
  • at least one of the at least one sample signal/each of the at least one sample signal is a sound signal collected outside the target device.
  • the sensor may be placed close to the casing of the device or the target device, placed outside the device or the target device, or placed inside the device or the object.
  • a doping device is selected from a plurality of devices based on the attribute information, and when the doping device is in a predetermined operating state, signal acquisition or signal simulation is performed on at least one sample signal, so as to obtain a doping sample set.
  • the doping device is different from the target device, and the predetermined operating state is a normal operating state.
  • the doping device is different from the target device, and the predetermined operating state is an abnormal operating state.
  • the doping device is the same as the target device, and the predetermined operating state is an abnormal operating state.
  • the doping device is the same as the target device, and the predetermined operating state is a normal operating state.
  • the selection of doping equipment from a plurality of equipment based on the attribute information includes: based on input parameters, output parameters, model type, model function, model accuracy, equipment type, equipment name and/or equipment identifier, etc. in the attribute information Select a doping device from multiple devices.
  • Performing signal acquisition or signal simulation on at least one sample signal to obtain a doped sample set includes: vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and At least one sample signal in brightness is subjected to signal acquisition or signal simulation, so as to obtain a set of doped samples.
  • the normal running state includes but not limited to: fault state, hidden danger state, abnormal operation state and abnormal input state.
  • step 105 at least one sample signal in the doped sample set is used to perform sample signal doping on corresponding at least one sample signal in the normal sample set, so as to obtain an abnormal sample set corresponding to the normal sample set.
  • equipment with a stable operating state has fewer failures or a lower failure rate in actual operation, so the sample signal/sample data volume of this type of equipment is relatively large for normal operation, while abnormal operation Or the data amount of the sample signal/sample data at the time of failure is small. In this case, it is often difficult to obtain enough abnormal sample signals.
  • the present application performs sample signal doping on at least one sample signal in the normal sample set according to a preset doping method.
  • using at least one sample signal in the doped sample set to perform sample signal doping on the corresponding at least one sample signal in the normal sample set includes:
  • using at least one sample signal in the doped sample set to perform sample signal doping on the corresponding at least one sample signal in the normal sample set includes:
  • the characteristic parameters include one or more of the following: signal-to-noise ratio, ratio of odd and even harmonic amplitudes, frequency complexity, proportion of main frequency, proportion of fundamental frequency, current correlation, center of gravity of spectrum, signal amplitude value level and 50Hz frequency amplitude.
  • the preset training algorithm may be any reasonable training algorithm in the field of artificial intelligence, deep learning, and machine learning algorithm.
  • using at least one sample signal in the doped sample set to perform sample signal doping on corresponding at least one sample signal in the normal sample set includes:
  • the time difference is based on the moment when the correlation or mutual information of at least one sample signal in the adulterated sample set and the corresponding at least one sample signal in the normal sample set is maximum or minimum, and the predetermined offset time or offset is added phase.
  • using at least one sample signal in the doped sample set to perform sample signal doping on corresponding at least one sample signal in the normal sample set includes:
  • the energy value or the square of the energy value increases or decreases the corresponding ratio, resulting in a change in the value range of the corresponding ratio, the amplitude square, the energy value or the energy value square.
  • the ratio of the sample signal doped by the sample signal to the sample signal not doped by the sample signal in the abnormal sample set becomes smaller, and an adjusted abnormal sample set is obtained, based on the normal sample set , the adjusted abnormal sample set, and a preset training algorithm to train the device model, so as to obtain a trained high-precision device model.
  • the preset training algorithm may be any reasonable training algorithm in the field of artificial intelligence, deep learning, and machine learning algorithm.
  • the device model is trained based on the normal sample set, the abnormal sample set and a preset training algorithm, so as to obtain a trained device model. It also includes obtaining an adjusted abnormal sample set by increasing the type of adulterated sample signals in the abnormal sample set, and training the device model based on the normal sample set, the adjusted abnormal sample set, and a preset training algorithm, thereby Increase the fitness of the trained device model.
  • the preset training algorithm may be any reasonable training algorithm in the field of artificial intelligence, deep learning, and machine learning algorithm.
  • FIG. 2 is a flowchart of a method 200 for testing a device model of a target device by sample doping according to an embodiment of the present invention.
  • Method 200 starts at step 201 .
  • a target device associated with the device model is selected from a plurality of devices based on the test request.
  • various types and/or various sizes of equipment are widely used in various locations, production links, monitoring links, etc. For this reason, if it is necessary to determine the operating status of the target device, or obtain parameters of the target device, etc., it is necessary to determine a model or device model associated with the target device.
  • the model of each different type of equipment or the equipment model can be used to determine the operating state of the equipment, obtain the operating parameters of the equipment, and so on.
  • a model training request needs to be generated and sent to the processing device for training or testing the model or the device model.
  • the test request includes the name, identifier, etc. of the device model.
  • the processing device selects a target device associated with the device model from the plurality of devices based on the testing request. For example, the processing device extracts the name, identifier, etc. of the device model from the model test request, and determines the target device based on the name, location, and/or identifier, etc. of the target device.
  • each model or device model has attribute information, and the attribute information is used to describe various attributes of the model or device model.
  • the various attributes are, for example: input parameters, output parameters, model type, model effect, model accuracy, device type, device name, device identifier, and the like.
  • the equipment model has various attributes, and for example, the input parameters, output parameters, model type, model role, model accuracy, etc. of the equipment model can be determined through the attribute information of the equipment model.
  • step 202 attribute information associated with the device model is acquired, and at least one sample signal associated with the device model is determined based on the attribute information.
  • the input parameters, output parameters, model type, model function, model accuracy, etc. of the equipment model can be determined through the attribute information of the equipment model.
  • the target device involved in the device model may be determined by parsing the attribute information of the device model.
  • a target device can be any type of device.
  • the attribute information of the device model may also include information of various sample signals associated with the target device.
  • the device identifier or device name of the target device can be used to search in the sample signal information library to obtain at least one sample signal associated with the target device Information.
  • the at least one sample signal includes: vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness.
  • one or more sample signals of vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness can be used to characterize, train, test, describe target device. It should be understood that this application is only described by taking vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness as examples, and those skilled in the art should understand Note that this application can use any reasonable sample signal. In practical scenarios, various types of sensors can be used to acquire any one of vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image, and brightness.
  • the normal sample set is one of the data collected by sensors such as vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness when the target device is running normally.
  • a sample signal composed of one or more sample signals or data.
  • the present application performs signal acquisition or signal simulation on each sample signal in at least one sample signal, so as to obtain the at least two A collection of normal samples for a sample signal.
  • the selection of doping equipment from a plurality of equipment based on the attribute information includes: based on input parameters, output parameters, model type, model function, model accuracy, equipment type, equipment name and/or equipment identifier, etc. in the attribute information Select a doping device from multiple devices.
  • Performing signal acquisition or signal simulation on at least one sample signal to obtain a doped sample set includes: vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and At least one sample signal in brightness is subjected to signal acquisition or signal simulation, so as to obtain a set of doped samples.
  • the normal running state includes but not limited to: fault state, hidden danger state, abnormal operation state and abnormal input state.
  • using at least one sample signal in the doped sample set to perform sample signal doping on the corresponding at least one sample signal in the normal sample set includes:
  • the preset training algorithm may be any reasonable training algorithm in the field of artificial intelligence, deep learning, and machine learning algorithm.
  • It also includes, by reducing the modification amount, the difference between the sample signal doped by the sample signal and the sample signal not doped by the sample signal in the abnormal sample set becomes smaller, and the adjusted abnormal sample set is obtained, based on the normal
  • the sample set and the adjusted abnormal sample set test the device model, so as to determine the performance index of the device model based on the test result.
  • using at least one sample signal in the doped sample set to perform sample signal doping on corresponding at least one sample signal in the normal sample set includes:
  • the time difference is based on the moment when the correlation or mutual information of at least one sample signal in the adulterated sample set and the corresponding at least one sample signal in the normal sample set is maximum or minimum, and the predetermined offset time or offset is added phase.
  • the energy value or the square of the energy value increases or decreases the corresponding ratio, resulting in a change in the value range of the corresponding ratio, the amplitude square, the energy value or the energy value square.
  • the ratio of the sample signal doped by the sample signal to the sample signal not doped by the sample signal in the abnormal sample set becomes smaller, and an adjusted abnormal sample set is obtained, based on the normal sample set , the adjusted abnormal sample set, and a preset training algorithm to train the device model, so as to obtain a trained high-precision device model.
  • the device model is tested based on the normal sample set and the abnormal sample set, so as to determine the performance index of the device model based on the test result. It also includes obtaining an adjusted abnormal sample set by increasing the type of adulterated sample signal in the abnormal sample set, and testing the device model based on the normal sample set and the adjusted abnormal sample set, so as to determine the performance of the device model based on the test results. Performance.
  • the equipment model is tested based on the normal sample set and the abnormal sample set, so as to determine the performance index of the equipment model based on the test results. Diagnose, judge or identify the result status of the target device, and obtain the verification status corresponding to the normal sample set and/or the abnormal sample set. Determining a correct ratio of the result status based on the verification status, determining a test result based on the correct ratio, and determining a performance index of the device model based on the test result.
  • test result when the test result is greater than or equal to 97%, it is determined that the performance index of the device model is high precision, and when the test result is less than 97% and greater than or equal to 90%, it is determined that the performance index of the device model is medium precision, and When the test result is less than 90%, it is determined that the performance index of the device model is low precision.
  • FIG. 3 is a schematic structural diagram of a system 300 for training a device model of a target device through sample doping according to an embodiment of the present invention.
  • the system 300 includes: a selection device 301 , a determination device 302 , an acquisition device 303 , a processing device 304 , a doping device 305 and a training device 306 .
  • the selecting unit 301 is configured to, in response to the received model training request, select a target device from multiple devices based on the training request, and determine a device model associated with the target device.
  • various types and/or various sizes of equipment are widely used in various locations, production links, monitoring links, etc. For this reason, if it is necessary to determine the operating status of the target device, or obtain parameters of the target device, etc., it is necessary to determine a model or device model associated with the target device.
  • the model of each different type of equipment or the equipment model can be used to determine the operating state of the equipment, obtain the operating parameters of the equipment, and so on.
  • the target device After the target device is determined, it is necessary to select a device model that needs to be trained and is associated with the target device from models or device models associated with multiple different devices. To this end, after the name, location, and identifier of the target device are determined, the name, location, and identifier of the target device can be used to search in the model library to determine the device model associated with the target device.
  • the determining means 302 is configured to acquire attribute information associated with the equipment model, and determine at least one sample signal associated with the equipment model based on the attribute information.
  • the input parameters, output parameters, model type, model function, model accuracy, etc. of the equipment model can be determined through the attribute information of the equipment model.
  • the target device involved in the device model may be determined by parsing the attribute information of the device model.
  • a target device can be any type of device.
  • the attribute information of the device model may also include information of various sample signals associated with the target device.
  • the device identifier or device name of the target device can be used to search in the sample signal information library to obtain at least one sample signal associated with the target device Information.
  • the at least one sample signal includes: vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness.
  • one or more sample signals of vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness can be used to characterize, train, test, describe target device. It should be understood that this application is only described by taking vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness as examples, and those skilled in the art should understand Note that this application can use any reasonable sample signal. In practical scenarios, various types of sensors can be used to acquire any one of vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image, and brightness.
  • the normal sample set is one of the data collected by sensors such as vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness when the target device is running normally.
  • a sample signal composed of one or more sample signals or data.
  • the present application performs signal acquisition or signal simulation on each sample signal in at least one sample signal, so as to obtain the at least two A collection of normal samples for a sample signal.
  • the at least two sample signals are vibration signal, sound signal and voltage signal
  • the normal sample set includes a plurality of samples arranged in order of sample sampling time, wherein each sample includes vibration signal, sound signal and voltage signal , and each sample has a sampling time. That is, each sample in the set of normal samples is a signal group or signal set having a sample time and including each sample signal at that sample time.
  • the normal sample set can include at least two sample subsets, and each sample subset is a vibration sample signal subset, an acoustic emission sample signal subset, a sound sample signal subset or an electric field strength sample signal subset, etc. It should be understood that the division of signal subsets is only for data storage or data display.
  • each sample includes each of at least two sample signals.
  • a plurality of sample signal groups are included in the normal sample set, and each sample signal group includes a single vibration sample signal, a single acoustic emission sample signal and a single sound sample signal, for example, each sample signal group is ⁇ vibration sample signal, acoustic sample signal Transmit sample signal, sound sample signal>. It should be appreciated that each sample signal group can be considered as a sample in the set of normal samples.
  • At least one of the at least one sample signal/each of the at least one sample signal is a vibration/acoustic emission signal collected by a sensor that is closely attached to a device casing of the target device.
  • at least one of the at least one sample signal/each of the at least one sample signal is a sound signal collected outside the target device.
  • the sensor can be placed close to the casing of the device or target device, placed outside the device or target device, or placed inside the device or object.
  • the processing device 304 is configured to select a doping device from a plurality of devices based on the attribute information, and when the doping device is in a predetermined operating state, perform signal acquisition or signal simulation on at least one sample signal, so as to obtain a doping sample set .
  • the doping device is different from the target device, and the predetermined operating state is a normal operating state.
  • the doping device is different from the target device, and the predetermined operating state is an abnormal operating state.
  • the doping device is the same as the target device, and the predetermined operating state is an abnormal operating state.
  • the doping device is the same as the target device, and the predetermined operating state is a normal operating state.
  • the selection of doping equipment from a plurality of equipment based on the attribute information includes: based on input parameters, output parameters, model type, model function, model accuracy, equipment type, equipment name and/or equipment identifier, etc. in the attribute information Select a doping device from multiple devices.
  • Performing signal acquisition or signal simulation on at least one sample signal to obtain a doped sample set includes: vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and At least one sample signal in brightness is subjected to signal acquisition or signal simulation, so as to obtain a set of doped samples.
  • the normal running state includes but not limited to: fault state, hidden danger state, abnormal operation state and abnormal input state.
  • the doping device 305 is configured to use at least one sample signal in the doped sample set to perform sample signal doping on corresponding at least one sample signal in the normal sample set, so as to obtain an abnormal sample set corresponding to the normal sample set.
  • equipment with a stable operating state has fewer failures or a lower failure rate in actual operation, so the sample signal/sample data volume of this type of equipment is relatively large for normal operation, while abnormal operation Or the data amount of the sample signal/sample data at the time of failure is small. In this case, it is often difficult to obtain enough abnormal sample signals.
  • the present application performs sample signal doping on at least one sample signal in the normal sample set according to a preset doping method.
  • using at least one sample signal in the doped sample set to perform sample signal doping on the corresponding at least one sample signal in the normal sample set includes:
  • using at least one sample signal in the doped sample set to perform sample signal doping on the corresponding at least one sample signal in the normal sample set includes:
  • the characteristic parameters include one or more of the following: signal-to-noise ratio, ratio of odd and even harmonic amplitudes, frequency complexity, proportion of main frequency, proportion of fundamental frequency, current correlation, center of gravity of spectrum, signal amplitude value level and 50Hz frequency amplitude.
  • It also includes obtaining an adjusted abnormal sample set by reducing the doping ratio a or doping amplitude b, and training the device model based on the normal sample set, the adjusted abnormal sample set and a preset training algorithm, In this way, a trained high-precision device model is obtained.
  • It also includes, by reducing the modification amount, the difference between the sample signal doped by the sample signal and the sample signal not doped by the sample signal in the abnormal sample set becomes smaller, and the adjusted abnormal sample set is obtained, based on the normal
  • the sample set, the adjusted abnormal sample set, and the preset training algorithm train the device model, so as to obtain a trained high-precision device model.
  • using at least one sample signal in the doped sample set to perform sample signal doping on corresponding at least one sample signal in the normal sample set includes:
  • At least one sample signal in the doped sample set is used to perform sample signal doping on the corresponding at least one sample signal in the normal sample set
  • the time difference is based on the moment when the correlation or mutual information of at least one sample signal in the adulterated sample set and the corresponding at least one sample signal in the normal sample set is maximum or minimum, and the predetermined offset time or offset is added phase.
  • performing sample signal doping on corresponding at least one sample signal in the normal sample set by using at least one sample signal in the doped sample set includes:
  • the energy value or the square of the energy value increases or decreases the corresponding ratio, resulting in a change in the value range of the corresponding ratio, the amplitude square, the energy value or the energy value square.
  • the ratio of the sample signal doped by the sample signal to the sample signal not doped by the sample signal in the abnormal sample set becomes smaller, and an adjusted abnormal sample set is obtained, based on the normal sample set , the adjusted abnormal sample set, and a preset training algorithm to train the device model, so as to obtain a trained high-precision device model.
  • the training device 306 is configured to train the device model based on the normal sample set, the abnormal sample set and a preset training algorithm, so as to obtain a trained device model. It also includes obtaining an adjusted abnormal sample set by increasing the type of adulterated sample signals in the abnormal sample set, and training the device model based on the normal sample set, the adjusted abnormal sample set, and a preset training algorithm, thereby Increase the fitness of the trained device model.
  • FIG. 4 is a schematic structural diagram of a system 400 for training a device model of a target device through sample doping according to an embodiment of the present invention.
  • the system 400 includes: a selection device 401 , a determination device 402 , an acquisition device 403 , a processing device 404 , a doping device 405 and a testing device 406 .
  • the selecting means 401 is configured to, in response to the received test request for the device model, select a target device associated with the device model from multiple devices based on the test request.
  • various types and/or various sizes of equipment are widely used in various locations, production links, monitoring links, etc. For this reason, if it is necessary to determine the operating status of the target device, or obtain parameters of the target device, etc., it is necessary to determine a model or device model associated with the target device.
  • the model of each different type of equipment or the equipment model can be used to determine the operating state of the equipment, obtain the operating parameters of the equipment, and so on.
  • a model training request needs to be generated and sent to the processing device for training or testing the model or the device model.
  • the test request includes the name, identifier, etc. of the device model.
  • the processing device selects a target device associated with the device model from the plurality of devices based on the testing request. For example, the processing device extracts the device model's name, identifier, etc. from the model testing request, and determines the target device based on the target device's name, location, and/or identifier, etc.
  • the determining unit 402 is configured to acquire attribute information associated with the equipment model, and determine at least one sample signal associated with the equipment model based on the attribute information.
  • the input parameters, output parameters, model type, model function, model accuracy, etc. of the equipment model can be determined through the attribute information of the equipment model.
  • the target device involved in the device model may be determined by parsing the attribute information of the device model.
  • a target device can be any type of device.
  • the attribute information of the device model may also include information of various sample signals associated with the target device.
  • the device identifier or device name of the target device can be used to search in the sample signal information library to obtain at least one sample signal associated with the target device Information.
  • the at least one sample signal includes: vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness.
  • one or more sample signals of vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness can be used to characterize, train, test, describe target device. It should be understood that this application is only described by taking vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness as examples, and those skilled in the art should understand Note that this application can use any reasonable sample signal. In practical scenarios, various types of sensors can be used to acquire any one of vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image, and brightness.
  • the obtaining unit 403 is configured to perform signal collection or signal simulation on at least one sample signal when the target device is in a normal operating state, so as to obtain a normal sample set including the at least one sample signal.
  • the normal sample set is one of the data collected by sensors such as vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and brightness when the target device is running normally.
  • a sample signal composed of one or more sample signals or data.
  • the present application performs signal acquisition or signal simulation on each sample signal in at least one sample signal, so as to obtain the at least two A collection of normal samples for a sample signal.
  • the at least two sample signals are vibration signal, sound signal and voltage signal
  • the normal sample set includes a plurality of samples arranged in order of sample sampling time, wherein each sample includes vibration signal, sound signal and voltage signal , and each sample has a sampling time. That is, each sample in the set of normal samples is a signal group or signal set having a sample time and including each sample signal at that sample time.
  • the normal sample set can include at least two sample subsets, and each sample subset is a vibration sample signal subset, an acoustic emission sample signal subset, a sound sample signal subset or an electric field strength sample signal subset, etc. It should be understood that the division of signal subsets is only for data storage or data display.
  • each sample includes each of at least two sample signals.
  • a plurality of sample signal groups are included in the normal sample set, and each sample signal group includes a single vibration sample signal, a single acoustic emission sample signal and a single sound sample signal, for example, each sample signal group is ⁇ vibration sample signal, acoustic sample signal Transmit sample signal, sound sample signal>. It should be appreciated that each sample signal group can be considered as a sample in the set of normal samples.
  • At least one of the at least one sample signal/each of the at least one sample signal is a vibration/acoustic emission signal collected by a sensor that is closely attached to a device casing of the target device.
  • at least one of the at least one sample signal/each of the at least one sample signal is a sound signal collected outside the target device.
  • the sensor may be placed close to the casing of the device or the target device, placed outside the device or the target device, or placed inside the device or the object.
  • the processing device 404 is configured to select a doping device from a plurality of devices based on the attribute information, and when the doping device is in a predetermined operating state, perform signal acquisition or signal simulation on at least one sample signal, so as to obtain a doping sample set .
  • the doping device is different from the target device, and the predetermined operating state is a normal operating state.
  • the doping device is different from the target device, and the predetermined operating state is an abnormal operating state.
  • the doping device is the same as the target device, and the predetermined operating state is an abnormal operating state.
  • the doping device is the same as the target device, and the predetermined operating state is a normal operating state.
  • the selection of doping equipment from a plurality of equipment based on the attribute information includes: based on input parameters, output parameters, model type, model function, model accuracy, equipment type, equipment name and/or equipment identifier, etc. in the attribute information Select a doping device from multiple devices.
  • Performing signal acquisition or signal simulation on at least one sample signal to obtain a doped sample set includes: vibration, sound, velocity, displacement, stress, pressure, voltage, current, power, electric field strength, magnetic field strength, temperature, image and At least one sample signal in brightness is subjected to signal acquisition or signal simulation, so as to obtain a set of doped samples.
  • the normal running state includes but not limited to: fault state, hidden danger state, abnormal operation state and abnormal input state.
  • the doping device 405 is configured to use at least one sample signal in the doped sample set to perform sample signal doping on at least one corresponding sample signal in the normal sample set, so as to obtain an abnormal sample set corresponding to the normal sample set.
  • equipment with a stable operating state has fewer failures or a lower failure rate in actual operation, so the sample signal/sample data volume of this type of equipment is relatively large for normal operation, while abnormal operation Or the data amount of the sample signal/sample data at the time of failure is small. In this case, it is often difficult to obtain enough abnormal sample signals.
  • the present application performs sample signal doping on at least one sample signal in the normal sample set according to a preset doping method.
  • using at least one sample signal in the doped sample set to perform sample signal doping on the corresponding at least one sample signal in the normal sample set includes:
  • Performing sample signal doping on corresponding at least one sample signal in the normal sample set by using at least one sample signal in the doped sample set includes:
  • the characteristic parameters include one or more of the following: signal-to-noise ratio, ratio of odd and even harmonic amplitudes, frequency complexity, proportion of main frequency, proportion of fundamental frequency, current correlation, center of gravity of spectrum, signal amplitude value level and 50Hz frequency amplitude.
  • It also includes obtaining an adjusted abnormal sample set by reducing the doping ratio a or doping amplitude b, and training the device model based on the normal sample set, the adjusted abnormal sample set and a preset training algorithm, In this way, a trained high-precision device model is obtained.
  • It also includes, by reducing the modification amount, the difference between the sample signal doped by the sample signal and the sample signal not doped by the sample signal in the abnormal sample set becomes smaller, and the adjusted abnormal sample set is obtained, based on the normal
  • the sample set and the adjusted abnormal sample set test the device model, so as to determine the performance index of the device model based on the test result.
  • the ratio of the sample signal doped by the sample signal to the sample signal not doped by the sample signal in the abnormal sample set becomes smaller, and an adjusted abnormal sample set is obtained, based on the normal sample set and the adjusted abnormal sample set to test the device model, so as to determine the performance index of the device model based on the test result.
  • using at least one sample signal in the doped sample set to perform sample signal doping on corresponding at least one sample signal in the normal sample set includes:
  • At least one sample signal in the doped sample set is used to perform sample signal doping on the corresponding at least one sample signal in the normal sample set
  • the time difference is based on the moment when the correlation or mutual information of at least one sample signal in the adulterated sample set and the corresponding at least one sample signal in the normal sample set is maximum or minimum, and the predetermined offset time or offset is added phase.
  • the energy value or the square of the energy value increases or decreases the corresponding ratio, resulting in a change in the value range of the corresponding ratio, the amplitude square, the energy value or the energy value square.
  • the equipment model is tested based on the normal sample set and the abnormal sample set, so as to determine the performance index of the equipment model based on the test results. Diagnose, judge or identify the result status of the target device, and obtain the verification status corresponding to the normal sample set and/or the abnormal sample set. Determining a correct ratio of the result status based on the verification status, determining a test result based on the correct ratio, and determining a performance index of the device model based on the test result.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • Computing Systems (AREA)
  • Medical Informatics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Testing Or Measuring Of Semiconductors Or The Like (AREA)
  • Investigating Or Analyzing Materials By The Use Of Ultrasonic Waves (AREA)

Abstract

一种通过样本掺杂来训练目标设备的设备模型的方法及系统，所述方法包括：获取与设备模型相关联的属性信息，基于所述属性信息确定与设备模型相关联的至少一种样本信号（102）；在所述目标设备处于正常运行状态时，对所述至少一种样本信号进行信号采集或信号仿真，从而获取包括所述至少一种样本信号的正常样本集合（103）；基于所述属性信息从多个设备中选择掺杂设备，在掺杂设备处于预定运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合（104）；利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，从而获得与正常样本集合相对应的异常样本集合（105）；基于正常样本集合、异常样本集合以及预先设定的训练算法对设备模型进行训练，从而获得经过训练的设备模型（106）。

Description

通过样本掺杂来训练目标设备的设备模型的方法及系统 技术领域
本发明涉及人工智能技术领域，并且更具体地，涉及一种通过样本掺杂来训练目标设备的设备模型的方法及系统。
背景技术
目前，随着人工智能技术的发展，大量机器学习算法不断涌现。机器学习算法特别是深度学习近年来取得了极大的成功，而数据才是使机器学习成为可能的关键因素。技术人员可以使用简单的算法实现机器学习，但是没有好的数据无法对算法进行优化。
由此可知，在基于机器学习的模型训练中，样本信号/样本数据的数据质量影响着模型的训练效果。然而，在实际情况中，部分类型的设备在运行中出现故障的次数或比率较低，因此这种类型的设备的正常运行的样本信号/样本数据的数据量比较大，而异常运行或故障时的样本信号/样本数据的数据量较小。在这种情况下，较小数据量的异常样本信号/样本数据无法满足模型训练或测试的需求。
发明内容
为了解决现有技术中的问题，本发明提出通过样本掺杂来训练目标设备的设备模型的方法及系统，从而解决异常运行或故障时的样本信号/样本数据的数据量较小的情况。
根据本发明的一个方面，提供一种通过样本掺杂来训练目标设备的设备模型的方法，所述方法包括：
响应于接收到的模型训练请求，基于训练请求从多个设备中选择目标设备，并确定与目标设备相关联的设备模型；
获取与设备模型相关联的属性信息，基于所述属性信息确定与设备模型相关联的至少一种样本信号；
在所述目标设备处于正常运行状态时，对所述至少一种样本信号进行信号采集或信号仿真，从而获取包括所述至少一种样本信号的正常样本集 合；
基于所述属性信息从多个设备中选择掺杂设备，在掺杂设备处于预定运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合；
利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，从而获得与正常样本集合相对应的异常样本集合；
基于正常样本集合、异常样本集合以及预先设定的训练算法对目标模型进行训练，从而获得经过训练的目标模型。
所述至少一种样本信号包括：振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度。
所述至少一种样本信号之中的至少一种样本信号是通过紧贴在设备外壳的传感器所采集的振动/声发射信号。
所述至少一种样本信号之中的至少一种样本信号是在设备外部采集的声音信号。
根据本发明的再一个方面，提供一种通过样本掺杂来测试目标设备的设备模型的方法，所述方法包括：
响应于接收到的针对设备模型的测试请求，基于测试请求从多个设备中选择与设备模型相关联的目标设备；
获取与设备模型相关联的属性信息，基于所述属性信息确定与设备模型相关联的至少一种样本信号；
在所述目标设备处于正常运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获取包括所述至少一种样本信号的正常样本集合；
基于所述属性信息从多个设备中选择掺杂设备，在掺杂设备处于预定运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合；
利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，从而获得与正常样本集合相对应的异常样本集合；
基于正常样本集合和异常样本集合对目标模型进行测试，从而基于测 试结果确定目标模型的性能指标。
根据本发明的再一个方面，提供一种通过样本掺杂来训练目标设备的设备模型的系统，所述系统包括：
选择装置，用于响应于接收到的模型训练请求，基于训练请求从多个设备中选择目标设备，并确定与目标设备相关联的设备模型；
确定装置，用于获取与设备模型相关联的属性信息，基于所述属性信息确定与设备模型相关联的至少一种样本信号；
获取装置，用于在所述目标设备处于正常运行状态时，对所述至少一种样本信号进行信号采集或信号仿真，从而获取包括所述至少一种样本信号的正常样本集合；
处理装置，用于基于所述属性信息从多个设备中选择掺杂设备，在掺杂设备处于预定运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合；
掺杂装置，用于利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，从而获得与正常样本集合相对应的异常样本集合；
训练装置，用于基于正常样本集合、异常样本集合以及预先设定的训练算法对目标模型进行训练，从而获得经过训练的目标模型。
所述至少一种样本信号包括：振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度。
所述至少一种样本信号之中的至少一种样本信号是通过紧贴在设备外壳的传感器所采集的振动/声发射信号。
所述至少一种样本信号之中的至少一种样本信号是在设备外部采集的声音信号。
根据本发明的再一个方面，提供一种通过样本掺杂来测试目标设备的设备模型的系统，所述系统包括：
选择装置，用于响应于接收到的针对设备模型的测试请求，基于测试请求从多个设备中选择与设备模型相关联的目标设备；
确定装置，用于获取与设备模型相关联的属性信息，基于所述属性信息确定与设备模型相关联的至少一种样本信号；
获取装置，用于在所述目标设备处于正常运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获取包括所述至少一种样本信号的正常样本集合；
处理装置，用于基于所述属性信息从多个设备中选择掺杂设备，在掺杂设备处于预定运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合；
掺杂装置，用于利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，从而获得与正常样本集合相对应的异常样本集合；
测试装置，用于基于正常样本集合和异常样本集合对目标模型进行测试，从而基于测试结果确定目标模型的性能指标。
根据本发明的再一个方面，提供一种基于掺杂的模型训练方法，首先生成大量正常样本和异常样本，然后再使用样本和设定算法训练一个模型，用于对目标设备的状态进行诊断、判别或识别。其中，
所述正常样本通过采集或者仿真目标设备正常运行情况下的信号生成。(样本信号的种类包括但不限于振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像、亮度，还可以是多种信号的组合)
所述异常样本通过在正常样本的信号中叠加一定量的杂质成分生成。
其中在正常样本中增加一定量的杂质成分是按照幅值、幅值平方、能量值、能量值平方(的峰值或者平均值)在正常样本中增加一定比例的杂质成分。
其中在正常样本中叠加一定量的杂质成分是一定的时间差在正常样本中叠加一定量的杂质成分(所述时间差根据相关性或者互信息分析来确定。1)以正常样本信号和杂质成分信号相关性或互信息最大的时刻为基准，增加一个固定的偏移时间或偏移相位。2)以该类正常样本信号和该类杂质成分信号相关性或互信息最大时刻的平均值为基准，增加一个固定的偏移时间或偏移相位。)。
其中还通过减小异常样本中杂质成分的量，从而获得更高精度的模型。
其中还通过增加异常样本的类型，从而获得更高适应性的模型。
其中在正常样本中增加一定量的杂质成分是在正常样本中增加其它对象产生的同种信号。
其中在正常样本中增加一定量的杂质成分是在正常样本中增加目标设备在异常状态下的信号(包括在故障状态、隐患状态、异常操作状态、异常输入状态和其它运行状态下的信号)。
根据本发明的再一方面，提供一种基于掺杂的模型评价方法，所述模型用于对目标设备的状态进行诊断、判别或识别。其中：
通过采集或者仿真目标设备正常运行情况下的信号生成正常样本。
通过在正常样本中增加一定量的杂质成分生成异常样本。
用所述正常样本、异常样本对所述模型进行测试，再根据测试结果评价模型的性能。
其中在正常样本中增加一定量的杂质成分是按照幅值、幅值平方、能量值、能量值平方在正常样本中增加一定比例的杂质成分。
其中还通过减小异常样本中杂质成分的量，从而评价模型的精度。
其中还通过增加异常样本的类型，从而评价模型的适应性。
其中在正常样本中增加一定量的杂质成分是在正常样本中增加其它对象产生的同种信号。
其中在正常样本中增加一定量的杂质成分是在正常样本中增加目标设备在异常状态下的信号。
附图说明
通过参考下面的附图，可以更为完整地理解本发明的示例性实施方式：
图1为根据本发明实施方式的通过样本掺杂来训练目标设备的设备模型的方法的流程图；
图2为根据本发明实施方式的通过样本掺杂来测试目标设备的设备模型的方法的流程图；
图3为根据本发明实施方式的通过样本掺杂来训练目标设备的设备模型的系统的结构示意图；
图4为根据本发明实施方式的通过样本掺杂来测试目标设备的设备模型的系统的结构示意图。
具体实施方式
图1为根据本发明实施方式的通过样本掺杂来训练目标设备的设备模型的方法100的流程图。方法100从步骤101处开始。
在步骤101，响应于接收到的模型训练请求，基于训练请求从多个设备中选择目标设备，并确定与目标设备相关联的设备模型。在工业生产或设备运行的实际场景中，各种类型和/或各种尺寸的设备被广泛应用于各个位置、生产环节、监控环节等。为此，如果需要确定目标设备的运行状态，或获取目标设备的参数等，需要确定与目标设备相关联的模型或设备模型。通常，每个不同类型的设备的模型或设备模型可以用于确定设备的运行状态、获取设备的运行参数等。为此，在需要对目标设备的设备模型进行训练或测试时，需要生成模型训练请求并将模型训练请求发送给用于对模型或设备模型进行训练或测试的处理设备。所述模型训练请求中包括目标设备的名称、位置、标识符等。响应于接收到的模型训练请求，处理设备基于训练请求从多个设备中选择目标设备。例如，处理设备从模型训练请求中提取目标设备的名称、位置、标识符等，并基于目标设备的名称、位置和/或标识符等确定目标设备。
在确定了目标设备后，需要从与多个不同设备各自相关联的模型或设备模型中选择需要进行训练与目标设备相关联的设备模型。为此，在确定目标设备的名称、位置、标识符等后，可以利用目标设备的名称、位置、标识符在模型库中进行检索，以确定与目标设备相关联的设备模型。
通常，每个模型或设备模型均具有属性信息，并且属性信息用于描述模型或设备模型的多种属性。多种属性例如是：输入参数、输出参数、模型类型、模型作用、模型准确度、设备类型、设备名称、设备标识符等。为此，设备模型具有多种属性，并且例如，通过设备模型的属性信息可以确定设备模型的输入参数、输出参数、模型类型、模型作用、模型准确度等。
在步骤102，获取与设备模型相关联的属性信息，基于所述属性信息确定与设备模型相关联的至少一种样本信号。如上所述，通过设备模型的属性信息可以确定设备模型的输入参数、输出参数、模型类型、模型作用、 模型准确度等。并且进一步地，可以通过对设备模型的属性信息进行解析来确定设备模型所涉及的目标设备。目标设备可以是任意类型的设备。此外，设备模型的属性信息中还可以包括与目标设备相关联的多种样本信号的信息。可替换地，在确定了与目标设备相关联的设备模型之后，利用目标设备的设备标识符或设备名称可以在样本信号信息库中进行检索，以获取与目标设备相关联的至少一种样本信号的信息。
其中至少一种样本信号包括：振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度。通常，可以使用振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的一种或多种样本信号来表征、训练、测试、描述目标设备。应当了解的是，本申请仅是以振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度为例进行描述，所属领域技术人员应当了解的是，本申请可以使用任何合理的样本信号。在实际场景中，可以使用各种类型的传感器来获取振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的任意一个。
在步骤103，在所述目标设备处于正常运行状态时，对所述至少一种样本信号进行信号采集或信号仿真，从而获取包括所述至少一种样本信号的正常样本集合。
正常样本集合是目标设备在正常运行时，利用传感器所采集的诸如振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的一种或多种的样本信号或数据所构成的样本信号。在本申请中，为了使经过训练或测试的设备模型的模型准确度更高，本申请对至少一种样本信号中的每种样本信号进行信号采集或信号仿真，从而获取包括所述至少两种样本信号的正常样本集合。
举例来说，至少两种样本信号为振动信号、声音信号和电压信号，正常样本集合中的包括以样本的采样时间顺序排列的多个样本，其中每个样本包括振动信号、声音信号和电压信号，并且每个样本具有采样时间。即，正常样本集合中的每个样本为具有采样时间的并且包括在所述采样时间处的每种样本信号的信号组或信号集。在对样本数据进行存储时，可以使正 常样本集合中包括至少两个样本子集，每个样本子集为振动样本信号子集、声发射样本信号子集、声音样本信号子集或电场强度样本信号子集等。应当了解到是，信号子集的划分方式仅是为了数据存储或数据展示。实际上，每个样本包括至少两种样本信号中的每种样本信号。可替换地，正常样本集合中包括多个样本信号组，每个样本信号组包括单个振动样本信号、单个声发射样本信号和单个声音样本信号，例如每个样本信号组为<振动样本信号、声发射样本信号、声音样本信号>。应当了解的是，每个样本信号组可以被认为是正常样本集合中的一个样本。
其中至少一种样本信号之中的至少一种样本信号/至少一种样本信号之中的每种样本信号是通过紧贴在目标设备的设备外壳的传感器所采集的振动/声发射信号。其中至少一种样本信号之中的至少一种样本信号/至少一种样本信号之中的每种样本信号是在目标设备的设备外部采集的声音信号。在实际情况中，可以将传感器设置为紧贴在设备或目标表设备的外壳处、将传感器设置在设备或目标设备的外部或将传感器设置在设备或对象的内部。
在步骤104，基于所述属性信息从多个设备中选择掺杂设备，在掺杂设备处于预定运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合。优选地，掺杂设备与目标设备不同，并且所述预定运行状态为正常运行状态。或者，掺杂设备与目标设备不同，并且所述预定运行状态为异常运行状态。可替换地，掺杂设备与目标设备相同，并且所述预定运行状态为异常运行状态。或者掺杂设备与目标设备相同，并且所述预定运行状态为正常运行状态。
其中基于所述属性信息从多个设备中选择掺杂设备包括：基于属性信息中的输入参数、输出参数、模型类型、模型作用、模型准确度、设备类型、设备名称和/或设备标识符等从多个设备中选择掺杂设备。对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合包括：对振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合。
其中常运行状态包括但不限于：故障状态、隐患状态、异常操作状态 和异常输入状态。
在步骤105，利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，从而获得与正常样本集合相对应的异常样本集合。在实际情况中，运行状态稳定的设备在实际运行中出现故障的次数较少或故障比率较低，因此这种类型的设备的正常运行的样本信号/样本数据的数据量比较大，而异常运行或故障时的样本信号/样本数据的数据量较小。在这种情况下，通常难以获得足够的异常样本信号。为此，本申请根据预设的掺杂方式对正常样本集合中的至少一种样本信号进行样本信号掺杂。
优选地，利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂包括：
按照幅值、幅值平方、能量值或能量值平方的峰值或者平均值，将正常样本集合和掺杂样本集合中的样本信号归一化；
确定掺杂比例a，其中0<a<1，将正常样本集合中的样本信号减去掺杂比例a对应的信号量，再加上掺杂样本集合中的样本信号乘以a得到的信号量，从而获得掺杂样本；
或者，按照幅值、幅值平方、能量值或能量值平方的峰值或者平均值，将正常样本集合和掺杂样本集合中的样本信号归一化；
按照幅值、幅值平方、能量值或能量值平方的峰值或者平均值确定掺杂幅度b，将正常样本集合中的样本信号减去幅度b对应的信号量，加上掺杂样本集合中的样本信号幅度b对应的信号量，从而获得掺杂样本。
优选地，利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂包括：
利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，使得掺杂后的特征参数达到设定值。
所述特征参数包括以下内容中的一个或多个：信噪比、奇偶次谐波幅值之比、频率复杂度、主频占比、基频占比、电流相关性、频谱重心、信号幅值水平和50Hz频率幅值。
还包括，通过减小所述掺杂比例a或者掺杂幅度b，获得经过调整的异常样本集合，基于正常样本集合、经过调整的异常样本集合以及预先设 定的训练算法对设备模型进行训练，从而获得经过训练的高精度设备模型。其中，预先设定的训练算法可以是人工智能领域、深度学习领域、机器学习算法领域中任意的合理的训练算法。
还包括，通过减小所述修改量，使得异常样本集合中经过样本信号掺杂的样本信号与未经过样本信号掺杂的样本信号的差值变小，获得经过调整的异常样本集合，基于正常样本集合、经过调整的异常样本集合以及预先设定的训练算法对设备模型进行训练，从而获得经过训练的高精度设备模型。其中，预先设定的训练算法可以是人工智能领域、深度学习领域、机器学习算法领域中任意的合理的训练算法。
优选地，所述利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂包括：
根据预定的时间差，利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，
其中时间差是以掺杂样本集合中的至少一种样本信号和正常样本集合中相应的至少一种样本信号的相关性或互信息最大或最小的时刻为基准，增加预定的偏移时间或偏移相位。
优选地，所述利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂包括：
利用掺杂样本集合中的至少一种样本信号与正常样本集合中相应的至少一种样本信号进行样本信号叠加，以使得正常样本集合中相应的至少一种样本信号的幅值、幅值平方、能量值或能量值平方，升高或降低相应的比率，导致升高或降低相应的比率的幅值、幅值平方、能量值或能量值平方的所属取值区间发生变化。
还包括，通过减小所述比率，使得异常样本集合中经过样本信号掺杂的样本信号与未经过样本信号掺杂的样本信号的比率变小，获得经过调整的异常样本集合，基于正常样本集合、经过调整的异常样本集合以及预先设定的训练算法对设备模型进行训练，从而获得经过训练的高精度设备模型。其中，预先设定的训练算法可以是人工智能领域、深度学习领域、机器学习算法领域中任意的合理的训练算法。
在步骤106，基于正常样本集合、异常样本集合以及预先设定的训练 算法对设备模型进行训练，从而获得经过训练的设备模型。还包括，通过增加异常样本集合中掺杂样本信号的类型，来获得经过调整的异常样本集合，基于正常样本集合、经过调整的异常样本集合以及预先设定的训练算法对设备模型进行训练，从而增加经过训练的设备模型的适应性。其中，预先设定的训练算法可以是人工智能领域、深度学习领域、机器学习算法领域中任意的合理的训练算法。
图2为根据本发明实施方式的通过样本掺杂来测试目标设备的设备模型的方法200的流程图。方法200从步骤201处开始。
在步骤201，响应于接收到的针对设备模型的测试请求，基于测试请求从多个设备中选择与设备模型相关联的目标设备。在工业生产或设备运行的实际场景中，各种类型和/或各种尺寸的设备被广泛应用于各个位置、生产环节、监控环节等。为此，如果需要确定目标设备的运行状态，或获取目标设备的参数等，需要确定与目标设备相关联的模型或设备模型。通常，每个不同类型的设备的模型或设备模型可以用于确定设备的运行状态、获取设备的运行参数等。为此，在需要对目标设备的设备模型进行训练或测试时，需要生成模型训练请求并将模型训练请求发送给用于对模型或设备模型进行训练或测试的处理设备。测试请求中包括设备模型的名称、标识符等。响应于接收到的模型测试请求，处理设备基于测试请求从多个设备中选择与设备模型相关联的目标设备。例如，处理设备从模型测试请求中提取设备模型的名称、标识符等，并基于目标设备的名称、位置和/或标识符等确定目标设备。
通常，每个模型或设备模型均具有属性信息，并且属性信息用于描述模型或设备模型的多种属性。多种属性例如是：输入参数、输出参数、模型类型、模型作用、模型准确度、设备类型、设备名称、设备标识符等。为此，设备模型具有多种属性，并且例如，通过设备模型的属性信息可以确定设备模型的输入参数、输出参数、模型类型、模型作用、模型准确度等。
在步骤202，获取与设备模型相关联的属性信息，基于所述属性信息确定与设备模型相关联的至少一种样本信号。如上所述，通过设备模型的属性信息可以确定设备模型的输入参数、输出参数、模型类型、模型作用、 模型准确度等。并且进一步地，可以通过对设备模型的属性信息进行解析来确定设备模型所涉及的目标设备。目标设备可以是任意类型的设备。此外，设备模型的属性信息中还可以包括与目标设备相关联的多种样本信号的信息。可替换地，在确定了与目标设备相关联的设备模型之后，利用目标设备的设备标识符或设备名称可以在样本信号信息库中进行检索，以获取与目标设备相关联的至少一种样本信号的信息。
所述至少一种样本信号包括：振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度。通常，可以使用振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的一种或多种样本信号来表征、训练、测试、描述目标设备。应当了解的是，本申请仅是以振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度为例进行描述，所属领域技术人员应当了解的是，本申请可以使用任何合理的样本信号。在实际场景中，可以使用各种类型的传感器来获取振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的任意一个。
在步骤203，在所述目标设备处于正常运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获取包括所述至少一种样本信号的正常样本集合。
正常样本集合是目标设备在正常运行时，利用传感器所采集的诸如振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的一种或多种的样本信号或数据所构成的样本信号。在本申请中，为了使经过训练或测试的设备模型的模型准确度更高，本申请对至少一种样本信号中的每种样本信号进行信号采集或信号仿真，从而获取包括所述至少两种样本信号的正常样本集合。
举例来说，至少两种样本信号为振动信号、声音信号和电压信号，正常样本集合中的包括以样本的采样时间顺序排列的多个样本，其中每个样本包括振动信号、声音信号和电压信号，并且每个样本具有采样时间。即，正常样本集合中的每个样本为具有采样时间的并且包括在所述采样时间处的每种样本信号的信号组或信号集。在对样本数据进行存储时，可以使正 常样本集合中包括至少两个样本子集，每个样本子集为振动样本信号子集、声发射样本信号子集、声音样本信号子集或电场强度样本信号子集等。应当了解到是，信号子集的划分方式仅是为了数据存储或数据展示。实际上，每个样本包括至少两种样本信号中的每种样本信号。可替换地，正常样本集合中包括多个样本信号组，每个样本信号组包括单个振动样本信号、单个声发射样本信号和单个声音样本信号，例如每个样本信号组为<振动样本信号、声发射样本信号、声音样本信号>。应当了解的是，每个样本信号组可以被认为是正常样本集合中的一个样本。
其中至少一种样本信号之中的至少一种样本信号/至少一种样本信号之中的每种样本信号是通过紧贴在目标设备的设备外壳的传感器所采集的振动/声发射信号。其中至少一种样本信号之中的至少一种样本信号/至少一种样本信号之中的每种样本信号是在目标设备的设备外部采集的声音信号。在实际情况中，可以将传感器设置为紧贴在设备或目标表设备的外壳处、将传感器设置在设备或目标设备的外部或将传感器设置在设备或对象的内部。
在步骤204，基于所述属性信息从多个设备中选择掺杂设备，在掺杂设备处于预定运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合。优选地，掺杂设备与目标设备不同，并且所述预定运行状态为正常运行状态。或者，掺杂设备与目标设备不同，并且所述预定运行状态为异常运行状态。可替换地，掺杂设备与目标设备相同，并且所述预定运行状态为异常运行状态。或者掺杂设备与目标设备相同，并且所述预定运行状态为正常运行状态。
其中基于所述属性信息从多个设备中选择掺杂设备包括：基于属性信息中的输入参数、输出参数、模型类型、模型作用、模型准确度、设备类型、设备名称和/或设备标识符等从多个设备中选择掺杂设备。对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合包括：对振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合。
其中常运行状态包括但不限于：故障状态、隐患状态、异常操作状态 和异常输入状态。
在步骤205，利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，从而获得与正常样本集合相对应的异常样本集合。在实际情况中，运行状态稳定的设备在实际运行中出现故障的次数较少或故障比率较低，因此这种类型的设备的正常运行的样本信号/样本数据的数据量比较大，而异常运行或故障时的样本信号/样本数据的数据量较小。在这种情况下，通常难以获得足够的异常样本信号。为此，本申请根据预设的掺杂方式对正常样本集合中的至少一种样本信号进行样本信号掺杂。
优选地，利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂包括：
按照幅值、幅值平方、能量值或能量值平方的峰值或者平均值，将正常样本集合和掺杂样本集合中的样本信号归一化；
确定掺杂比例a，其中0<a<1，将正常样本集合中的样本信号减去掺杂比例a对应的信号量，再加上掺杂样本集合中的样本信号乘以a得到的信号量，从而获得掺杂样本；
或者，按照幅值、幅值平方、能量值或能量值平方的峰值或者平均值，将正常样本集合和掺杂样本集合中的样本信号归一化；
按照幅值、幅值平方、能量值或能量值平方的峰值或者平均值确定掺杂幅度b，将正常样本集合中的样本信号减去幅度b对应的信号量，加上掺杂样本集合中的样本信号幅度b对应的信号量，从而获得掺杂样本。
利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂包括：
利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，使得掺杂后的特征参数达到设定值。
所述特征参数包括以下内容中的一个或多个：信噪比、奇偶次谐波幅值之比、频率复杂度、主频占比、基频占比、电流相关性、频谱重心、信号幅值水平和50Hz频率幅值。
还包括，通过减小所述掺杂比例a或者掺杂幅度b，获得经过调整的异常样本集合，基于正常样本集合、经过调整的异常样本集合以及预先设 定的训练算法对设备模型进行训练，从而获得经过训练的高精度设备模型。其中，预先设定的训练算法可以是人工智能领域、深度学习领域、机器学习算法领域中任意的合理的训练算法。
还包括，通过减小所述修改量，使得异常样本集合中经过样本信号掺杂的样本信号与未经过样本信号掺杂的样本信号的差值变小，获得经过调整的异常样本集合，基于正常样本集合和经过调整的异常样本集合对设备模型进行测试，从而基于测试结果确定设备模型的性能指标。
还包括，通过减小所述比率，使得异常样本集合中经过样本信号掺杂的样本信号与未经过样本信号掺杂的样本信号的比率变小，获得经过调整的异常样本集合，基于正常样本集合和经过调整的异常样本集合对设备模型进行测试，从而基于测试结果确定设备模型的性能指标。
优选地，所述利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂包括：
根据预定的时间差，利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，
其中时间差是以掺杂样本集合中的至少一种样本信号和正常样本集合中相应的至少一种样本信号的相关性或互信息最大或最小的时刻为基准，增加预定的偏移时间或偏移相位。
利用掺杂样本集合中的至少一种样本信号与正常样本集合中相应的至少一种样本信号进行样本信号叠加，以使得正常样本集合中相应的至少一种样本信号的幅值、幅值平方、能量值或能量值平方，升高或降低相应的比率，导致升高或降低相应的比率的幅值、幅值平方、能量值或能量值平方的所属取值区间发生变化。
还包括，通过减小所述比率，使得异常样本集合中经过样本信号掺杂的样本信号与未经过样本信号掺杂的样本信号的比率变小，获得经过调整的异常样本集合，基于正常样本集合、经过调整的异常样本集合以及预先设定的训练算法对设备模型进行训练，从而获得经过训练的高精度设备模型。
在步骤206，基于正常样本集合和异常样本集合对设备模型进行测试，从而基于测试结果确定设备模型的性能指标。还包括，通过增加异常样本 集合中掺杂样本信号的类型，来获得经过调整的异常样本集合，基于正常样本集合和经过调整的异常样本集合对设备模型进行测试，从而基于测试结果确定设备模型的性能指标。
基于正常样本集合和异常样本集合对设备模型进行测试，从而基于测试结果确定设备模型的性能指标包括：将正常样本集合和异常样本集合分别或依次输入经过训练的设备模型，使得经过训练的设备模型诊断、判别或识别目标设备的结果状态，获取与正常样本集合和/或异常样本集相对应的验证状态。确定基于验证状态确定结果状态的正确比率，基于正确比率确定测试结果并基于测试结果确定设备模型的性能指标。例如，在将正常样本集合和异常样本集合分别或依次输入经过训练的设备模型后，经过训练的设备模型诊断、判别或识别目标设备的结果状态100次，基于验证状态确定100次结果状态中诊断、判别或识别正确的次数为99次，那么基于验证状态确定结果状态的正确比率为99/100＝99％，那么测试结果为99％。
根据一个实施方式，当测试结果大于或等于97％时，确定设备模型的性能指标为高精度，当测试结果小于97％并且大于或等于90％时，确定设备模型的性能指标为中精度，以及当测试结果小于90％时，确定设备模型的性能指标为低精度。
图3为根据本发明实施方式的通过样本掺杂来训练目标设备的设备模型的系统300的结构示意图。系统300包括：选择装置301、确定装置302、获取装置303、处理装置304、掺杂装置305以及训练装置306。
选择装置301，用于响应于接收到的模型训练请求，基于训练请求从多个设备中选择目标设备，并确定与目标设备相关联的设备模型。在工业生产或设备运行的实际场景中，各种类型和/或各种尺寸的设备被广泛应用于各个位置、生产环节、监控环节等。为此，如果需要确定目标设备的运行状态，或获取目标设备的参数等，需要确定与目标设备相关联的模型或设备模型。通常，每个不同类型的设备的模型或设备模型可以用于确定设备的运行状态、获取设备的运行参数等。为此，在需要对目标设备的设备模型进行训练或测试时，需要生成模型训练请求并将模型训练请求发送给用于对模型或设备模型进行训练或测试的处理设备。所述模型训练请求中包括目标设备的名称、位置、标识符等。响应于接收到的模型训练请求， 处理设备基于训练请求从多个设备中选择目标设备。例如，处理设备从模型训练请求中提取目标设备的名称、位置、标识符等，并基于目标设备的名称、位置和/或标识符等确定目标设备。
在确定了目标设备后，需要从与多个不同设备各自相关联的模型或设备模型中选择需要进行训练与目标设备相关联的设备模型。为此，在确定目标设备的名称、位置、标识符等后，可以利用目标设备的名称、位置、标识符在模型库中进行检索，以确定与目标设备相关联的设备模型。
通常，每个模型或设备模型均具有属性信息，并且属性信息用于描述模型或设备模型的多种属性。多种属性例如是：输入参数、输出参数、模型类型、模型作用、模型准确度、设备类型、设备名称、设备标识符等。为此，设备模型具有多种属性，并且例如，通过设备模型的属性信息可以确定设备模型的输入参数、输出参数、模型类型、模型作用、模型准确度等。
确定装置302，用于获取与设备模型相关联的属性信息，基于所述属性信息确定与设备模型相关联的至少一种样本信号。如上所述，通过设备模型的属性信息可以确定设备模型的输入参数、输出参数、模型类型、模型作用、模型准确度等。并且进一步地，可以通过对设备模型的属性信息进行解析来确定设备模型所涉及的目标设备。目标设备可以是任意类型的设备。此外，设备模型的属性信息中还可以包括与目标设备相关联的多种样本信号的信息。可替换地，在确定了与目标设备相关联的设备模型之后，利用目标设备的设备标识符或设备名称可以在样本信号信息库中进行检索，以获取与目标设备相关联的至少一种样本信号的信息。
其中至少一种样本信号包括：振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度。通常，可以使用振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的一种或多种样本信号来表征、训练、测试、描述目标设备。应当了解的是，本申请仅是以振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度为例进行描述，所属领域技术人员应当了解的是，本申请可以使用任何合理的样本信号。在实际场景中，可以使用各种类型的传感器来 获取振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的任意一个。
获取装置303，用于在所述目标设备处于正常运行状态时，对所述至少一种样本信号进行信号采集或信号仿真，从而获取包括所述至少一种样本信号的正常样本集合。
正常样本集合是目标设备在正常运行时，利用传感器所采集的诸如振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的一种或多种的样本信号或数据所构成的样本信号。在本申请中，为了使经过训练或测试的设备模型的模型准确度更高，本申请对至少一种样本信号中的每种样本信号进行信号采集或信号仿真，从而获取包括所述至少两种样本信号的正常样本集合。
举例来说，至少两种样本信号为振动信号、声音信号和电压信号，正常样本集合中的包括以样本的采样时间顺序排列的多个样本，其中每个样本包括振动信号、声音信号和电压信号，并且每个样本具有采样时间。即，正常样本集合中的每个样本为具有采样时间的并且包括在所述采样时间处的每种样本信号的信号组或信号集。在对样本数据进行存储时，可以使正常样本集合中包括至少两个样本子集，每个样本子集为振动样本信号子集、声发射样本信号子集、声音样本信号子集或电场强度样本信号子集等。应当了解到是，信号子集的划分方式仅是为了数据存储或数据展示。实际上，每个样本包括至少两种样本信号中的每种样本信号。可替换地，正常样本集合中包括多个样本信号组，每个样本信号组包括单个振动样本信号、单个声发射样本信号和单个声音样本信号，例如每个样本信号组为<振动样本信号、声发射样本信号、声音样本信号>。应当了解的是，每个样本信号组可以被认为是正常样本集合中的一个样本。
其中至少一种样本信号之中的至少一种样本信号/至少一种样本信号之中的每种样本信号是通过紧贴在目标设备的设备外壳的传感器所采集的振动/声发射信号。其中至少一种样本信号之中的至少一种样本信号/至少一种样本信号之中的每种样本信号是在目标设备的设备外部采集的声音信号。在实际情况中，可以将传感器设置为紧贴在设备或目标表设备的外壳处、将传感器设置在设备或目标设备的外部或将传感器设置在设备或对象 的内部。
处理装置304，用于基于所述属性信息从多个设备中选择掺杂设备，在掺杂设备处于预定运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合。优选地，掺杂设备与目标设备不同，并且所述预定运行状态为正常运行状态。或者，掺杂设备与目标设备不同，并且所述预定运行状态为异常运行状态。可替换地，掺杂设备与目标设备相同，并且所述预定运行状态为异常运行状态。或者掺杂设备与目标设备相同，并且所述预定运行状态为正常运行状态。
其中基于所述属性信息从多个设备中选择掺杂设备包括：基于属性信息中的输入参数、输出参数、模型类型、模型作用、模型准确度、设备类型、设备名称和/或设备标识符等从多个设备中选择掺杂设备。对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合包括：对振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合。
其中常运行状态包括但不限于：故障状态、隐患状态、异常操作状态和异常输入状态。
掺杂装置305，用于利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，从而获得与正常样本集合相对应的异常样本集合。在实际情况中，运行状态稳定的设备在实际运行中出现故障的次数较少或故障比率较低，因此这种类型的设备的正常运行的样本信号/样本数据的数据量比较大，而异常运行或故障时的样本信号/样本数据的数据量较小。在这种情况下，通常难以获得足够的异常样本信号。为此，本申请根据预设的掺杂方式对正常样本集合中的至少一种样本信号进行样本信号掺杂。
优选地，利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂包括：
按照幅值、幅值平方、能量值或能量值平方的峰值或者平均值，将正常样本集合和掺杂样本集合中的样本信号归一化；
确定掺杂比例a(0<a<1)，将正常样本集合中的样本信号减去掺杂比 例a对应的信号量，再加上掺杂样本集合中的样本信号乘以a得到的信号量，从而获得掺杂样本；
或者，按照幅值、幅值平方、能量值或能量值平方的峰值或者平均值，将正常样本集合和掺杂样本集合中的样本信号归一化；
按照幅值、幅值平方、能量值或能量值平方的峰值或者平均值确定掺杂幅度b，将正常样本集合中的样本信号减去幅度b对应的信号量，加上掺杂样本集合中的样本信号幅度b对应的信号量，从而获得掺杂样本。
优选地，利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂包括：
利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，使得掺杂后的特征参数达到设定值。
所述特征参数包括以下内容中的一个或多个：信噪比、奇偶次谐波幅值之比、频率复杂度、主频占比、基频占比、电流相关性、频谱重心、信号幅值水平和50Hz频率幅值。
还包括，通过减小所述掺杂比例a或者掺杂幅度b，获得经过调整的异常样本集合，基于正常样本集合、经过调整的异常样本集合以及预先设定的训练算法对设备模型进行训练，从而获得经过训练的高精度设备模型。
还包括，通过减小所述修改量，使得异常样本集合中经过样本信号掺杂的样本信号与未经过样本信号掺杂的样本信号的差值变小，获得经过调整的异常样本集合，基于正常样本集合、经过调整的异常样本集合以及预先设定的训练算法对设备模型进行训练，从而获得经过训练的高精度设备模型。
优选地，所述利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂包括：
根据预定的时间差，利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，
其中时间差是以掺杂样本集合中的至少一种样本信号和正常样本集合中相应的至少一种样本信号的相关性或互信息最大或最小的时刻为基准，增加预定的偏移时间或偏移相位。
优选地，所述利用掺杂样本集合中的至少一种样本信号对正常样本集 合中相应的至少一种样本信号进行样本信号掺杂包括：
利用掺杂样本集合中的至少一种样本信号与正常样本集合中相应的至少一种样本信号进行样本信号叠加，以使得正常样本集合中相应的至少一种样本信号的幅值、幅值平方、能量值或能量值平方，升高或降低相应的比率，导致升高或降低相应的比率的幅值、幅值平方、能量值或能量值平方的所属取值区间发生变化。
还包括，通过减小所述比率，使得异常样本集合中经过样本信号掺杂的样本信号与未经过样本信号掺杂的样本信号的比率变小，获得经过调整的异常样本集合，基于正常样本集合、经过调整的异常样本集合以及预先设定的训练算法对设备模型进行训练，从而获得经过训练的高精度设备模型。
训练装置306，用于基于正常样本集合、异常样本集合以及预先设定的训练算法对设备模型进行训练，从而获得经过训练的设备模型。还包括，通过增加异常样本集合中掺杂样本信号的类型，来获得经过调整的异常样本集合，基于正常样本集合、经过调整的异常样本集合以及预先设定的训练算法对设备模型进行训练，从而增加经过训练的设备模型的适应性。
图4为根据本发明实施方式的通过样本掺杂来训练目标设备的设备模型的系统400的结构示意图。系统400包括：选择装置401、确定装置402、获取装置403、处理装置404、掺杂装置405以及测试装置406。
选择装置401，用于响应于接收到的针对设备模型的测试请求，基于测试请求从多个设备中选择与设备模型相关联的目标设备。在工业生产或设备运行的实际场景中，各种类型和/或各种尺寸的设备被广泛应用于各个位置、生产环节、监控环节等。为此，如果需要确定目标设备的运行状态，或获取目标设备的参数等，需要确定与目标设备相关联的模型或设备模型。通常，每个不同类型的设备的模型或设备模型可以用于确定设备的运行状态、获取设备的运行参数等。为此，在需要对目标设备的设备模型进行训练或测试时，需要生成模型训练请求并将模型训练请求发送给用于对模型或设备模型进行训练或测试的处理设备。测试请求中包括设备模型的名称、标识符等。响应于接收到的模型测试请求，处理设备基于测试请求从多个设备中选择与设备模型相关联的目标设备。例如，处理设备从模型测试请 求中提取设备模型的名称、标识符等，并基于目标设备的名称、位置和/或标识符等确定目标设备。
通常，每个模型或设备模型均具有属性信息，并且属性信息用于描述模型或设备模型的多种属性。多种属性例如是：输入参数、输出参数、模型类型、模型作用、模型准确度、设备类型、设备名称、设备标识符等。为此，设备模型具有多种属性，并且例如，通过设备模型的属性信息可以确定设备模型的输入参数、输出参数、模型类型、模型作用、模型准确度等。
确定装置402，用于获取与设备模型相关联的属性信息，基于所述属性信息确定与设备模型相关联的至少一种样本信号。如上所述，通过设备模型的属性信息可以确定设备模型的输入参数、输出参数、模型类型、模型作用、模型准确度等。并且进一步地，可以通过对设备模型的属性信息进行解析来确定设备模型所涉及的目标设备。目标设备可以是任意类型的设备。此外，设备模型的属性信息中还可以包括与目标设备相关联的多种样本信号的信息。可替换地，在确定了与目标设备相关联的设备模型之后，利用目标设备的设备标识符或设备名称可以在样本信号信息库中进行检索，以获取与目标设备相关联的至少一种样本信号的信息。
所述至少一种样本信号包括：振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度。通常，可以使用振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的一种或多种样本信号来表征、训练、测试、描述目标设备。应当了解的是，本申请仅是以振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度为例进行描述，所属领域技术人员应当了解的是，本申请可以使用任何合理的样本信号。在实际场景中，可以使用各种类型的传感器来获取振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的任意一个。
获取装置403，用于在所述目标设备处于正常运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获取包括所述至少一种样本信号的正常样本集合。正常样本集合是目标设备在正常运行时，利用传感器 所采集的诸如振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的一种或多种的样本信号或数据所构成的样本信号。在本申请中，为了使经过训练或测试的设备模型的模型准确度更高，本申请对至少一种样本信号中的每种样本信号进行信号采集或信号仿真，从而获取包括所述至少两种样本信号的正常样本集合。
举例来说，至少两种样本信号为振动信号、声音信号和电压信号，正常样本集合中的包括以样本的采样时间顺序排列的多个样本，其中每个样本包括振动信号、声音信号和电压信号，并且每个样本具有采样时间。即，正常样本集合中的每个样本为具有采样时间的并且包括在所述采样时间处的每种样本信号的信号组或信号集。在对样本数据进行存储时，可以使正常样本集合中包括至少两个样本子集，每个样本子集为振动样本信号子集、声发射样本信号子集、声音样本信号子集或电场强度样本信号子集等。应当了解到是，信号子集的划分方式仅是为了数据存储或数据展示。实际上，每个样本包括至少两种样本信号中的每种样本信号。可替换地，正常样本集合中包括多个样本信号组，每个样本信号组包括单个振动样本信号、单个声发射样本信号和单个声音样本信号，例如每个样本信号组为<振动样本信号、声发射样本信号、声音样本信号>。应当了解的是，每个样本信号组可以被认为是正常样本集合中的一个样本。
其中至少一种样本信号之中的至少一种样本信号/至少一种样本信号之中的每种样本信号是通过紧贴在目标设备的设备外壳的传感器所采集的振动/声发射信号。其中至少一种样本信号之中的至少一种样本信号/至少一种样本信号之中的每种样本信号是在目标设备的设备外部采集的声音信号。在实际情况中，可以将传感器设置为紧贴在设备或目标表设备的外壳处、将传感器设置在设备或目标设备的外部或将传感器设置在设备或对象的内部。
处理装置404，用于基于所述属性信息从多个设备中选择掺杂设备，在掺杂设备处于预定运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合。优选地，掺杂设备与目标设备不同，并且所述预定运行状态为正常运行状态。或者，掺杂设备与目标设备不同，并且所述预定运行状态为异常运行状态。可替换地，掺杂设备与目标设备 相同，并且所述预定运行状态为异常运行状态。或者掺杂设备与目标设备相同，并且所述预定运行状态为正常运行状态。
其中基于所述属性信息从多个设备中选择掺杂设备包括：基于属性信息中的输入参数、输出参数、模型类型、模型作用、模型准确度、设备类型、设备名称和/或设备标识符等从多个设备中选择掺杂设备。对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合包括：对振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度中的至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合。
其中常运行状态包括但不限于：故障状态、隐患状态、异常操作状态和异常输入状态。
掺杂装置405，用于利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，从而获得与正常样本集合相对应的异常样本集合。在实际情况中，运行状态稳定的设备在实际运行中出现故障的次数较少或故障比率较低，因此这种类型的设备的正常运行的样本信号/样本数据的数据量比较大，而异常运行或故障时的样本信号/样本数据的数据量较小。在这种情况下，通常难以获得足够的异常样本信号。为此，本申请根据预设的掺杂方式对正常样本集合中的至少一种样本信号进行样本信号掺杂。
优选地，利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂包括：
按照幅值、幅值平方、能量值或能量值平方的峰值或者平均值，将正常样本集合和掺杂样本集合中的样本信号归一化；
确定掺杂比例a(0<a<1)，将正常样本集合中的样本信号减去掺杂比例a对应的信号量，再加上掺杂样本集合中的样本信号乘以a得到的信号量，从而获得掺杂样本；
或者，按照幅值、幅值平方、能量值或能量值平方的峰值或者平均值，将正常样本集合和掺杂样本集合中的样本信号归一化；
按照幅值、幅值平方、能量值或能量值平方的峰值或者平均值确定掺杂幅度b，将正常样本集合中的样本信号减去幅度b对应的信号量，加上 掺杂样本集合中的样本信号幅度b对应的信号量，从而获得掺杂样本。
利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂包括：
利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，使得掺杂后的特征参数达到设定值。
所述特征参数包括以下内容中的一个或多个：信噪比、奇偶次谐波幅值之比、频率复杂度、主频占比、基频占比、电流相关性、频谱重心、信号幅值水平和50Hz频率幅值。
还包括，通过减小所述掺杂比例a或者掺杂幅度b，获得经过调整的异常样本集合，基于正常样本集合、经过调整的异常样本集合以及预先设定的训练算法对设备模型进行训练，从而获得经过训练的高精度设备模型。
还包括，通过减小所述修改量，使得异常样本集合中经过样本信号掺杂的样本信号与未经过样本信号掺杂的样本信号的差值变小，获得经过调整的异常样本集合，基于正常样本集合和经过调整的异常样本集合对设备模型进行测试，从而基于测试结果确定设备模型的性能指标。
还包括，通过减小所述比率，使得异常样本集合中经过样本信号掺杂的样本信号与未经过样本信号掺杂的样本信号的比率变小，获得经过调整的异常样本集合，基于正常样本集合和经过调整的异常样本集合对设备模型进行测试，从而基于测试结果确定设备模型的性能指标。
优选地，所述利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂包括：
根据预定的时间差，利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，
其中时间差是以掺杂样本集合中的至少一种样本信号和正常样本集合中相应的至少一种样本信号的相关性或互信息最大或最小的时刻为基准，增加预定的偏移时间或偏移相位。
利用掺杂样本集合中的至少一种样本信号与正常样本集合中相应的至少一种样本信号进行样本信号叠加，以使得正常样本集合中相应的至少一种样本信号的幅值、幅值平方、能量值或能量值平方，升高或降低相应的比率，导致升高或降低相应的比率的幅值、幅值平方、能量值或能量值平 方的所属取值区间发生变化。
还包括，通过减小所述比率，使得异常样本集合中经过样本信号掺杂的样本信号与未经过样本信号掺杂的样本信号的比率变小，获得经过调整的异常样本集合，基于正常样本集合、经过调整的异常样本集合以及预先设定的训练算法对设备模型进行训练，从而获得经过训练的高精度设备模型。
测试装置406，用于基于正常样本集合和异常样本集合对设备模型进行测试，从而基于测试结果确定设备模型的性能指标。还包括，通过增加异常样本集合中掺杂样本信号的类型，来获得经过调整的异常样本集合，基于正常样本集合和经过调整的异常样本集合对设备模型进行测试，从而基于测试结果确定设备模型的性能指标。
基于正常样本集合和异常样本集合对设备模型进行测试，从而基于测试结果确定设备模型的性能指标包括：将正常样本集合和异常样本集合分别或依次输入经过训练的设备模型，使得经过训练的设备模型诊断、判别或识别目标设备的结果状态，获取与正常样本集合和/或异常样本集相对应的验证状态。确定基于验证状态确定结果状态的正确比率，基于正确比率确定测试结果并基于测试结果确定设备模型的性能指标。例如，在将正常样本集合和异常样本集合分别或依次输入经过训练的设备模型后，经过训练的设备模型诊断、判别或识别目标设备的结果状态100次，基于验证状态确定100次结果状态中诊断、判别或识别正确的次数为99次，那么基于验证状态确定结果状态的正确比率为99/100＝99％，那么测试结果为99％。
根据一个实施方式，当测试结果大于或等于97％时，确定设备模型的性能指标为高精度，当测试结果小于97％并且大于或等于90％时，确定设备模型的性能指标为中精度，以及当测试结果小于90％时，确定设备模型的性能指标为低精度。

Claims (10)

  1. 一种通过样本掺杂来训练目标设备的设备模型的方法，所述方法包括：
    响应于接收到的模型训练请求，基于训练请求从多个设备中选择目标设备，并确定与目标设备相关联的设备模型；
    获取与设备模型相关联的属性信息，基于所述属性信息确定与设备模型相关联的至少一种样本信号；
    在所述目标设备处于正常运行状态时，对所述至少一种样本信号进行信号采集或信号仿真，从而获取包括所述至少一种样本信号的正常样本集合；
    基于所述属性信息从多个设备中选择掺杂设备，在掺杂设备处于预定运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合；
    利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，从而获得与正常样本集合相对应的异常样本集合；
    基于正常样本集合、异常样本集合以及预先设定的训练算法对目标模型进行训练，从而获得经过训练的目标模型。
  2. 根据权利要求1所述的方法，所述至少一种样本信号包括：振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度。
  3. 根据权利要求1所述的方法，所述至少一种样本信号之中的至少一种样本信号是通过紧贴在设备外壳的传感器所采集的振动/声发射信号。
  4. 根据权利要求1所述的方法，所述至少一种样本信号之中的至少一种样本信号是在设备外部采集的声音信号。
  5. 一种通过样本掺杂来测试目标设备的设备模型的方法，所述方法包括：
    响应于接收到的针对设备模型的测试请求，基于测试请求从多个设备中选择与设备模型相关联的目标设备；
    获取与设备模型相关联的属性信息，基于所述属性信息确定与设备模型相关联的至少一种样本信号；
    在所述目标设备处于正常运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获取包括所述至少一种样本信号的正常样本集合；
    基于所述属性信息从多个设备中选择掺杂设备，在掺杂设备处于预定运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合；
    利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，从而获得与正常样本集合相对应的异常样本集合；
    基于正常样本集合和异常样本集合对目标模型进行测试，从而基于测试结果确定目标模型的性能指标。
  6. 一种通过样本掺杂来训练目标设备的设备模型的系统，所述系统包括：
    选择装置，用于响应于接收到的模型训练请求，基于训练请求从多个设备中选择目标设备，并确定与目标设备相关联的设备模型；
    确定装置，用于获取与设备模型相关联的属性信息，基于所述属性信息确定与设备模型相关联的至少一种样本信号；
    获取装置，用于在所述目标设备处于正常运行状态时，对所述至少一种样本信号进行信号采集或信号仿真，从而获取包括所述至少一种样本信号的正常样本集合；
    处理装置，用于基于所述属性信息从多个设备中选择掺杂设备，在掺杂设备处于预定运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合；
    掺杂装置，用于利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，从而获得与正常样本 集合相对应的异常样本集合；
    训练装置，用于基于正常样本集合、异常样本集合以及预先设定的训练算法对目标模型进行训练，从而获得经过训练的目标模型。
  7. 根据权利要求6所述的系统，所述至少一种样本信号包括：振动、声音、速度、位移、应力、压力、电压、电流、功率、电场强度、磁场强度、温度、图像和亮度。
  8. 根据权利要求6所述的系统，所述至少一种样本信号之中的至少一种样本信号是通过紧贴在设备外壳的传感器所采集的振动/声发射信号。
  9. 根据权利要求6所述的系统，所述至少一种样本信号之中的至少一种样本信号是在设备外部采集的声音信号。
  10. 一种通过样本掺杂来测试目标设备的设备模型的系统，所述系统包括：
    选择装置，用于响应于接收到的针对设备模型的测试请求，基于测试请求从多个设备中选择与设备模型相关联的目标设备；
    确定装置，用于获取与设备模型相关联的属性信息，基于所述属性信息确定与设备模型相关联的至少一种样本信号；
    获取装置，用于在所述目标设备处于正常运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获取包括所述至少一种样本信号的正常样本集合；
    处理装置，用于基于所述属性信息从多个设备中选择掺杂设备，在掺杂设备处于预定运行状态时，对至少一种样本信号进行信号采集或信号仿真，从而获得掺杂样本集合；
    掺杂装置，用于利用掺杂样本集合中的至少一种样本信号对正常样本集合中相应的至少一种样本信号进行样本信号掺杂，从而获得与正常样本集合相对应的异常样本集合；
    测试装置，用于基于正常样本集合和异常样本集合对目标模型进行测试，从而基于测试结果确定目标模型的性能指标。
PCT/CN2021/108158 2021-06-16 2021-07-23 通过样本掺杂来训练目标设备的设备模型的方法及系统 WO2022262073A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202110663862.2A CN113537289B (zh) 2021-06-16 2021-06-16 通过样本掺杂来训练目标设备的设备模型的方法及系统
CN202110663862.2 2021-06-16

Publications (1)

Publication Number Publication Date
WO2022262073A1 true WO2022262073A1 (zh) 2022-12-22

Family

ID=78096026

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/108158 WO2022262073A1 (zh) 2021-06-16 2021-07-23 通过样本掺杂来训练目标设备的设备模型的方法及系统

Country Status (2)

Country Link
CN (1) CN113537289B (zh)
WO (1) WO2022262073A1 (zh)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190065989A1 (en) * 2017-08-30 2019-02-28 Intel Corporation Constrained sample selection for training models
CN112116002A (zh) * 2020-09-18 2020-12-22 北京旋极信息技术股份有限公司 一种检测模型的确定方法、验证方法和装置
CN112231971A (zh) * 2020-09-26 2021-01-15 浙江大学 基于相对整体趋势扩散故障样本生成的高炉故障诊断方法
CN112581719A (zh) * 2020-11-05 2021-03-30 清华大学 基于时序生成对抗网络的半导体封装过程预警方法和装置
CN112733872A (zh) * 2020-08-26 2021-04-30 南京航空航天大学 基于动态半径支持向量数据描述的航空发动机故障检测方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190065989A1 (en) * 2017-08-30 2019-02-28 Intel Corporation Constrained sample selection for training models
CN112733872A (zh) * 2020-08-26 2021-04-30 南京航空航天大学 基于动态半径支持向量数据描述的航空发动机故障检测方法
CN112116002A (zh) * 2020-09-18 2020-12-22 北京旋极信息技术股份有限公司 一种检测模型的确定方法、验证方法和装置
CN112231971A (zh) * 2020-09-26 2021-01-15 浙江大学 基于相对整体趋势扩散故障样本生成的高炉故障诊断方法
CN112581719A (zh) * 2020-11-05 2021-03-30 清华大学 基于时序生成对抗网络的半导体封装过程预警方法和装置

Also Published As

Publication number Publication date
CN113537289A (zh) 2021-10-22
CN113537289B (zh) 2024-05-28

Similar Documents

Publication Publication Date Title
KR102189269B1 (ko) 컨볼루션 신경망을 이용한 유도 모터의 결함 진단방법 및 시스템
KR102106775B1 (ko) 딥러닝을 이용한 전지 진단 방법
US11494643B2 (en) Noise data artificial intelligence apparatus and pre-conditioning method for identifying source of problematic noise
CN107271170B (zh) 一种机械设备故障类型的自动诊断方法及系统
CN101299004B (zh) 一种基于确定学习理论的振动故障诊断方法
CN101959115B (zh) 用于检测、定位和分类被测设备的缺陷的方法和装置
CN109061495A (zh) 一种混合储能电池组故障诊断方法
CN116417013B (zh) 水下推进器故障诊断方法及系统
TWI759027B (zh) 電池性能評鑑方法及電池性能評鑑裝置
CN109991956B (zh) 一种液体火箭发动机稳态故障预测方法
JP3333494B2 (ja) 装置の異常診断方法
WO2022262072A1 (zh) 基于样本信号的修改对目标模型进行训练的方法及系统
WO2022262073A1 (zh) 通过样本掺杂来训练目标设备的设备模型的方法及系统
CN112927720B (zh) 一种音频异常的检测方法及装置
KR101046748B1 (ko) 발전기 고정자 권선의 전용 진동 진단 방법 및 분석 시스템
CN115588439B (zh) 一种基于深度学习的声纹采集装置的故障检测方法及装置
JP2002323371A (ja) 音響診断装置及び音響診断方法
CN113593536B (zh) 一种检测语音识别准确率的装置和系统
CN110824293A (zh) 一种基于狼群算法的多特征融合参数的电网故障诊断方法
CN111948286A (zh) 一种基于超声波和深度学习的应力检测方法、装置及设备
CN117741321B (zh) 移动式储充系统故障诊断方法及系统
Martinod et al. Estimation of combustion engine technical state by multidimensional analysis using SVD method
CN110263828A (zh) 一种油浸式电抗器故障诊断方法
CN110032758A (zh) 计算电信号的能量的方法、装置和计算机存储介质
KR102131242B1 (ko) 진동 센싱 장치 및 이를 포함하는 센싱 시스템

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: 21945655

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21945655

Country of ref document: EP

Kind code of ref document: A1