CN114548280A - Fault diagnosis model training method, fault diagnosis method and electronic equipment - Google Patents
Fault diagnosis model training method, fault diagnosis method and electronic equipment Download PDFInfo
- Publication number
- CN114548280A CN114548280A CN202210167021.7A CN202210167021A CN114548280A CN 114548280 A CN114548280 A CN 114548280A CN 202210167021 A CN202210167021 A CN 202210167021A CN 114548280 A CN114548280 A CN 114548280A
- Authority
- CN
- China
- Prior art keywords
- data
- time sequence
- fault diagnosis
- characteristic
- sample
- Prior art date
- Legal status (The legal status 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 status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Abstract
The invention relates to the technical field of engineering machinery, in particular to a training and fault diagnosis method of a fault diagnosis model and electronic equipment, wherein the training method comprises the steps of obtaining sample data of an engine, wherein the sample data comprises first data in a preset time period before a fault and second data in normal operation; dividing each sample data by a preset time unit to obtain a time sequence; extracting at least two time sequence characteristics from each time sequence to obtain a first time sequence characteristic of each sample data; extracting a preset proportion based on the characteristic value of each time sequence characteristic in the first time sequence characteristic to obtain a second time sequence characteristic of each sample data; forming a sample timing feature based on the first timing feature and the second timing feature; and training an initial fault diagnosis model according to the sample time sequence characteristics, and determining a target fault diagnosis model. The target fault diagnosis model obtained by training the sample time sequence characteristics formed by the method has higher accuracy.
Description
Technical Field
The invention relates to the technical field of engineering machinery, in particular to a method for training a fault diagnosis model, a method for fault diagnosis and electronic equipment.
Background
At present, fault diagnosis methods for construction machinery mostly use fault code information of a sensor per se to identify faults, and the identification of the fault codes only has two states of 0 and 1. For the detection of the engine, data are collected in a normal operating state as well as in a fault state. The engine abnormality is not only affected by the nature of the engine itself, but also related to the load at the time of work. For example, when the load is large, the engine speed is low, and if the engine is abnormal, it is concluded based on the engine speed during this time. However, this conclusion is clearly erroneous.
Disclosure of Invention
In view of this, embodiments of the present invention provide a method for training a fault diagnosis model, a method for diagnosing a fault, and an electronic device, so as to solve the problem of low accuracy of engine fault diagnosis.
According to a first aspect, an embodiment of the present invention provides a method for training a fault diagnosis model, including:
acquiring sample data of an engine, wherein the sample data comprises first data in a preset time period before a fault and second data in normal operation;
dividing each sample data by a preset time unit to obtain a time sequence;
extracting at least two time sequence characteristics from each time sequence to obtain a first time sequence characteristic of each sample data;
extracting a preset proportion based on the characteristic value of each time sequence characteristic in the first time sequence characteristic to obtain a second time sequence characteristic of each sample data;
forming a sample timing feature based on the first timing feature and the second timing feature;
and training an initial fault diagnosis model according to the sample time sequence characteristics, and determining a target fault diagnosis model.
According to the training method of the fault diagnosis model provided by the embodiment of the invention, data in a preset time period before a fault is taken as a negative sample, rather than data when the fault occurs, so that the fault diagnosis model can learn characteristics before the fault; meanwhile, the second time sequence characteristic is extracted according to the preset proportion based on the size of the characteristic value, so that the abnormality of the engine caused by objective reasons can be avoided. Namely, the target fault diagnosis model obtained by training the sample time sequence characteristics formed by the method has higher accuracy.
With reference to the first aspect, in a first implementation manner of the first aspect, the obtaining sample data of the engine includes:
acquiring first original data in a preset time period before a fault and second original data in normal operation;
and respectively removing invalid data from the first original data and the second original data, and determining the first data and the second data.
According to the training method of the fault diagnosis model provided by the embodiment of the invention, invalid data is removed from the acquired original data, no-action data caused by idling or other reasons can be provided, the first data and the second data are both valid data, and the reliability of the subsequent sample time sequence characteristics is improved.
With reference to the first implementation manner of the first aspect, in a second implementation manner of the first aspect, the performing invalid data removal processing on the first original data and the second original data, respectively, to determine the first data and the second data includes:
acquiring an action code and an operation state corresponding to the engine;
and removing the invalid data from the first original data and the second original data respectively based on the action code and the running state, and determining the first data and the second data.
According to the training method of the fault diagnosis model provided by the embodiment of the invention, when invalid data is removed, the reliability of the removed invalid data can be ensured by combining the action code and the running state.
With reference to the first aspect, in a third implementation manner of the first aspect, the extracting a preset proportion based on the feature values of various time series features in the first time series feature to obtain a second time series feature of each sample data includes:
sorting based on the eigenvalue size of the various timing characteristics;
and respectively extracting corresponding time sequence characteristics from the sequencing result based on the preset proportion corresponding to each time sequence characteristic to obtain a second time sequence characteristic of each sample data.
With reference to the first aspect, in a fourth implementation manner of the first aspect, the forming a sample timing feature based on the first timing feature and the second timing feature includes:
respectively calculating an entropy value of a first rotation speed difference of the engine in a preset time period before a fault and a second rotation speed difference entropy value in normal operation based on the first data and the second data;
fusing the first time sequence feature in the preset time period before the fault, the second time sequence feature in the preset time period before the fault and the entropy value of the first rotation speed difference to obtain a sample time sequence feature in the preset time period before the fault;
and fusing the first time sequence characteristic in normal operation, the second time sequence characteristic in normal operation and the entropy value of the second rotation speed difference to obtain a sample time sequence characteristic in normal operation.
According to the fault diagnosis model training method provided by the embodiment of the invention, on the basis of the first time sequence characteristic and the second time sequence characteristic, the additional entropy of the rotating speed difference is added as a new characteristic, so that the sample time sequence characteristic can cover different fault types as much as possible.
With reference to the first aspect, in a fifth implementation manner of the first aspect, the training an initial fault diagnosis model according to the sample timing characteristics and determining a target fault diagnosis model includes:
inputting the sample time sequence characteristics into the initial fault diagnosis model to obtain a fault prediction result;
and updating the parameters of the initial fault diagnosis model based on the fault prediction result and the actual fault result corresponding to the sample time sequence characteristics so as to determine the target fault diagnosis model.
According to a second aspect, an embodiment of the present invention further provides a fault diagnosis method, including:
acquiring operating data of a target engine;
dividing the running data by a preset time unit to obtain a running time sequence;
extracting at least two time sequence characteristics of each running time sequence to obtain a first characteristic of the running data;
extracting a preset proportion based on the magnitude of the characteristic value of each time sequence characteristic in the first characteristic to obtain a second characteristic of each operating data;
forming a timing feature based on the first feature and the second feature;
inputting the time sequence characteristics into a target fault prediction model, and determining a fault diagnosis result, wherein the target fault prediction model is obtained by training according to the first aspect of the invention or the training method of the fault diagnosis model in any embodiment of the first aspect.
According to the fault diagnosis method provided by the embodiment of the invention, the fault diagnosis is carried out on the operation data of the target engine through the target fault prediction model with more accurate prediction, so that an accurate fault diagnosis result can be obtained.
According to a third aspect, an embodiment of the present invention provides an electronic device, including: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing therein computer instructions, and the processor executing the computer instructions to perform the method for training a fault diagnosis model according to the first aspect or any one of the embodiments of the first aspect, or to perform the method for fault diagnosis according to the second aspect.
According to a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the method for training a fault diagnosis model according to the first aspect or any one of the embodiments of the first aspect, or execute the method for fault diagnosis according to the second aspect.
According to a fifth aspect, an embodiment of the present invention provides a construction machine, including:
a body;
the electronic apparatus according to the third aspect of the present invention, wherein the electronic apparatus is provided in the main body.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.
FIG. 1 is a flow chart of a method of training a fault diagnosis model according to an embodiment of the invention;
FIG. 2 is a flow chart of a method of training a fault diagnosis model according to an embodiment of the invention;
FIG. 3 is a flow chart of a fault diagnosis method according to an embodiment of the present invention;
FIG. 4 is a block diagram of a training apparatus for a fault diagnosis model according to an embodiment of the present invention;
fig. 5 is a block diagram of the structure of a failure diagnosis apparatus according to an embodiment of the present invention;
fig. 6 is a schematic diagram of a hardware structure of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment of the invention provides engineering machinery, which comprises a body and electronic equipment arranged in the body. A target failure prediction model is operated in the electronic device and used for performing failure diagnosis based on operation data of the engine. The construction machine may be an excavator, a forklift, etc., and the specific form thereof is not limited in any way.
As an application scenario of the embodiment of the present invention, an engineering machine is taken as an excavator, an electronic device is taken as an external server, and a data acquisition device is arranged on the excavator and is used for acquiring operation data of an engine. The data acquisition device sends acquired operation data to an external server, the external server carries out fault diagnosis by using a target fault prediction model based on the operation data, and a diagnosis result is fed back to the excavator, so that the excavator can carry out corresponding processing or is displayed to an operator of the excavator.
As another application scenario of the embodiment of the present invention, taking an engineering machine as an excavator as an example, the electronic device is built in the excavator, and the excavator is provided with a data acquisition device for acquiring operation data of the engine. The data acquisition device sends acquired operation data to the electronic equipment, and the electronic equipment performs fault diagnosis by using a target fault prediction model based on the operation data so that the excavator can perform corresponding processing or display the fault diagnosis to an operator of the excavator.
In accordance with an embodiment of the present invention, there is provided a method of training a fault diagnosis model, and a fault diagnosis method embodiment, it is noted that the steps illustrated in the flowchart of the drawings may be performed in a computer system such as a set of computer executable instructions, and that while a logical order is illustrated in the flowchart, in some cases the steps illustrated or described may be performed in an order different than here.
In this embodiment, a method for training a fault diagnosis model is provided, which may be used in electronic devices, such as servers, computers, and the like, fig. 1 is a flowchart of a method for training a fault diagnosis model according to an embodiment of the present invention, and as shown in fig. 1, the flowchart includes the following steps:
and S11, acquiring sample data of the engine.
The sample data comprises first data in a preset time period before failure and second data in normal operation.
The preset time period before the fault is the previous preset time period when the fault occurs. The duration of the preset time period is set according to actual requirements. For example, it may be 20 hours, 30 hours, etc.
Specifically, the failure may not occur suddenly due to the failure of the engine, and it may be analyzed by the index data of the transmitter before the failure of the engine occurs. Therefore, the first data in the preset time period before the engine is in fault is used as negative sample data, and the second data in normal operation is used as positive sample data. The first data and the second data include, but are not limited to, the rotation speed, the temperature, and the like of the engine, and it is not limited to what index data is specifically adopted as the data of the engine, and the data may be specifically set according to actual requirements.
And S12, dividing each sample data by a preset time unit to obtain a time sequence.
For first data used for representing negative sample data, the electronic device divides the first data by using a preset time unit to obtain a time sequence corresponding to the first data, namely, first subdata of the preset time unit is obtained. For second data used for representing the positive sample data, the electronic device divides the second data by using a preset time unit to obtain a time sequence corresponding to the second data, namely, second subdata of the preset time unit is obtained.
The preset time units for dividing the first data and the second data may be the same or different in duration, and may be specifically set according to actual requirements.
And S13, respectively extracting at least two time sequence characteristics from each time sequence to obtain a first time sequence characteristic of each sample data.
Corresponding to the first data, the electronic device performs extraction of at least two time series features, for example, extraction of a mean, a variance, a standard deviation, or the like, for each time series, respectively. Likewise, the electronic device also performs extraction of at least two time series features, for example, extraction of a mean, a variance, a standard deviation, or the like, respectively for each time series, corresponding to the second data. Based on this, the electronic device may obtain the first timing characteristics of each sample data.
The first time-series feature is used to represent a feature obtained by extracting at least two time-series features from each time series. Specifically, the first data corresponds to a first timing characteristic, and the second data corresponds to a first timing characteristic.
And S14, extracting a preset proportion based on the characteristic value of each time sequence characteristic in the first time sequence characteristic to obtain a second time sequence characteristic of each sample data.
The manner of forming the corresponding second timing characteristic is similar whether the first timing characteristic corresponds to the first data or the first timing characteristic corresponds to the second data, and only the first timing characteristic corresponding to the first data is taken as an example for description.
The first timing characteristics corresponding to the first data include, for example, a mean, a variance, and a standard deviation. And respectively extracting the first N characteristic values with the maximum characteristic values according to the characteristic value sizes of the three time sequence characteristics to form a second time sequence characteristic. For example, for the mean value, the first N1 feature values with the largest feature value are extracted to form a mean value in the second time series feature; for the variance, extracting the first N2 eigenvalues with the largest eigenvalue to form the variance in the second time sequence characteristic; for the standard deviation, the first N3 feature values with the largest feature value are extracted, forming the standard deviation in the second time series feature. The specific values of N1, N2, and N3 may be set according to actual requirements, and are not limited herein.
S15, forming a sample timing characteristic based on the first timing characteristic and the second timing characteristic.
The electronic equipment splices the first time sequence characteristics and the second time sequence characteristics corresponding to the first data to form sample time sequence characteristics of the first data; and splicing the first time sequence characteristic and the second time sequence characteristic corresponding to the second data to form a sample time sequence characteristic of the second data.
Or, the electronic device may form a sample timing characteristic by combining other characteristics based on the timing characteristic obtained after splicing.
Details about this step will be described later.
And S16, training the initial fault diagnosis model according to the sample time sequence characteristics, and determining the target fault diagnosis model.
As described above, the first data is used to represent negative sample data, and the second data is used to represent positive sample data. The electronic equipment trains an initial fault diagnosis model by using the sample time sequence characteristics corresponding to the first data and the sample time sequence characteristics corresponding to the second data, and finally determines a target fault diagnosis model through multiple iterative updates. The input of the target fault diagnosis model is a time sequence characteristic, and the output is a fault diagnosis result. The failure diagnosis result may be normal or abnormal, or may be a probability of being normal or an abnormal probability.
In some optional implementations of this embodiment, the step S16 may include:
(1) and inputting the sample time sequence characteristics into the initial fault diagnosis model to obtain a fault prediction result.
(2) And updating the parameters of the initial fault diagnosis model based on the fault prediction result and the actual fault result corresponding to the sample time sequence characteristics so as to determine the target fault diagnosis model.
The model architecture of the initial fault diagnosis model may be a classification model, and the specific structure thereof is not limited in any way. And the electronic equipment inputs the sample time sequence characteristics into the initial fault diagnosis model to obtain a fault prediction result. And then, loss calculation is carried out by utilizing an actual fault result corresponding to the sample time sequence characteristics and a fault prediction structure, and parameters of the initial fault diagnosis model are updated. And finally determining a target fault diagnosis model through multiple iterative updating.
For example, the electronic device may select a random forest classification algorithm, which is one of machine learning algorithms, to model the sample timing characteristics; and meanwhile, screening the time sequence characteristics of the sample based on a genetic algorithm to determine a target fault diagnosis model.
According to the training method of the fault diagnosis model provided by the embodiment, data in a preset time period before a fault is taken as a negative sample, rather than data when the fault occurs, so that the fault diagnosis model can learn characteristics before the fault; meanwhile, the second time sequence characteristic is extracted according to the preset proportion based on the size of the characteristic value, so that the abnormality of the engine caused by objective reasons can be avoided. Namely, the target fault diagnosis model obtained by training the sample time sequence characteristics formed by the method has higher accuracy.
In this embodiment, a method for training a fault diagnosis model is provided, which may be used in electronic devices, such as servers, computers, and the like, fig. 2 is a flowchart of a method for training a fault diagnosis model according to an embodiment of the present invention, and as shown in fig. 2, the flowchart includes the following steps:
and S21, obtaining sample data of the engine.
The sample data comprises first data in a preset time period before failure and second data in normal operation.
Specifically, S21 includes:
s211, acquiring first original data in a preset time period before the fault and second original data in normal operation.
Raw data refers to data extracted from an engine log, which is data that has not been processed.
S212, performing invalid data removal processing on the first original data and the second original data, respectively, to determine the first data and the second data.
For an engine, the engine may be in a standby or off state during successive periods of time, as the raw data collected is data representative of the successive periods of time. Therefore, it is necessary to remove the invalid data from the original data to improve the reliability of the subsequent sample timing characteristics.
Specifically, the electronic device removes invalid data from first original data to obtain first data; and removing invalid data from the second original data to obtain second data. Wherein, for the judgment of invalid data, the electronic device can analyze the running state of the engine. For example, whether the data at the time is invalid data is analyzed in conjunction with the gear position of the engine at the corresponding time, and the like.
In some optional implementations of this embodiment, the step S212 may include:
(1) and acquiring the action code and the running state corresponding to the engine.
(2) And removing invalid data from the first original data and the second original data respectively based on the action codes and the running states, and determining the first data and the second data.
The action code is used to represent the action performed by the engine, e.g. acceleration, and is coded as 0000; decelerating and coding to 1111; shutdown, coding as 1010; and so on. That is, there is a unique motion code corresponding to each motion. The electronic equipment can determine whether the engine data at each time point is invalid data by extracting the action code and the running state corresponding to the engine at each time point.
After the electronic device determines the invalid data, the electronic device removes the invalid data from the corresponding original data, and thus the first data and the second data can be determined.
When invalid data is removed, the reliability of the removed invalid data can be ensured by combining action coding and the running state.
And S22, dividing each sample data by a preset time unit to obtain a time sequence.
Please refer to S12 in fig. 1, which is not repeated herein.
And S23, respectively extracting at least two time sequence characteristics from each time sequence to obtain a first time sequence characteristic of each sample data.
Please refer to S13 in fig. 1, which is not described herein again.
And S24, extracting a preset proportion based on the characteristic value of each time sequence characteristic in the first time sequence characteristic to obtain a second time sequence characteristic of each sample data.
Wherein the first timing characteristic comprises a mean value, a deviation value, and a standard deviation.
Specifically, S24 includes:
and S241, sorting based on the eigenvalue size of various time sequence characteristics.
And S242, respectively extracting corresponding time sequence characteristics from the sequencing result based on the preset proportions corresponding to the various time sequence characteristics to obtain second time sequence characteristics of the sample data.
As described above, for the first data, the electronic device sorts according to the magnitude of the characteristic values of various time sequence characteristics, and extracts corresponding time sequence characteristics from the sorting result according to different preset proportions, so as to obtain the second time sequence characteristics of each sample data. That is, the second timing characteristics are formed by respectively extracting corresponding timing characteristics from various timing characteristics according to respective corresponding preset proportions.
S25, forming a sample timing characteristic based on the first timing characteristic and the second timing characteristic.
Specifically, S25 includes:
and S251, respectively calculating the entropy value of a first rotating speed difference of the engine in a preset time period before the fault and the entropy value of a second rotating speed difference in normal operation based on the first data and the second data.
When constructing the sample time sequence characteristics, the entropy value of the corresponding rotation speed difference, that is, the chaos degree information in unit time is included in addition to the fusion of the first time sequence characteristics and the second time sequence characteristics.
And S252, fusing the first time sequence feature in the preset time period before the fault, the second time sequence feature in the preset time period before the fault and the entropy value of the first rotation speed difference to obtain a sample time sequence feature in the preset time period before the fault.
And S253, fusing the first time sequence characteristic in normal operation, the second time sequence characteristic in normal operation and the entropy value of the second rotation speed difference to obtain a sample time sequence characteristic in normal operation.
Whether the first data is within a preset time period before the fault or the second data is normally operated, the corresponding sample time sequence characteristics are fusion results of the corresponding first time sequence characteristics, the corresponding second time sequence characteristics and the entropy value of the rotating speed difference.
And S26, training the initial fault diagnosis model according to the sample time sequence characteristics, and determining the target fault diagnosis model.
Please refer to S16 in fig. 1, which is not described herein again.
According to the training method of the fault diagnosis model provided by the embodiment, invalid data is removed from the acquired original data, no-action data caused by idling or other reasons can be provided, the first data and the second data are both valid data, and the reliability of the time sequence characteristics of subsequent samples is improved. On the basis of the first time sequence characteristic and the second time sequence characteristic, an additional entropy value of the rotating speed difference is added to serve as a new characteristic, so that the sample time sequence characteristic can cover different fault types as much as possible.
The training method of the fault diagnosis model provided by the embodiment of the invention is a model establishing method for judging the normality and abnormality of engine indexes, and is characterized in that data is screened by recording the action information, the operation mode and the gear rotating speed information of the data to ensure that all the data are effective data under the condition of normal operation; extracting relevant mean values, deviation values, standard deviations, peak-to-peak characteristics of different proportions and chaos degree information in unit time from the information of engines of different fault models, and establishing a characteristic project; and finally, constructing an index fault diagnosis model of the excavator engine by a random forest algorithm which is one of machine learning methods, and judging the health state of the engine.
In this embodiment, a fault diagnosis method is provided, which may be used in an electronic device, such as a server, a computer, and the like, fig. 3 is a flowchart of a method for training a fault diagnosis model according to an embodiment of the present invention, as shown in fig. 3, the flowchart includes the following steps:
s31, the operation data of the target engine is acquired.
The operation data of the target engine can be acquired by a data acquisition device, can also be acquired by electronic equipment from the outside, and the like.
And S32, dividing the operation data by a preset time unit to obtain an operation time sequence.
This step is similar to S12 shown in fig. 1 and will not be described herein.
And S33, respectively extracting at least two time sequence characteristics of each running time sequence to obtain a first characteristic of the running data.
This step is similar to S13 shown in fig. 1 and will not be described herein.
And S34, extracting a preset proportion based on the characteristic value of each time sequence characteristic in the first characteristic to obtain a second characteristic of each operation data.
This step is similar to S14 shown in fig. 1 or S24 shown in fig. 2, and is not repeated here.
S35, forming a timing characteristic based on the first characteristic and the second characteristic.
This step is similar to S15 shown in fig. 1 or S25 shown in fig. 2, and is not repeated here.
And S36, inputting the time sequence characteristics into the target fault prediction model, and determining a fault diagnosis result.
The target fault prediction model is obtained by training according to the training method of the fault diagnosis model in any one of the above embodiments.
For the specific training process of the target failure prediction model, please refer to the above description, which is not repeated herein. And the electronic equipment inputs the time sequence characteristics into the target fault prediction model and outputs a fault diagnosis result.
According to the fault diagnosis method provided by the embodiment, the fault diagnosis is performed on the operation data of the target engine through the target fault prediction model with accurate prediction, and an accurate fault diagnosis result can be obtained.
The present embodiment further provides a training device for a fault diagnosis model and a fault diagnosis device, where the training device is used to implement the foregoing embodiments and preferred embodiments, and details of which have been already described are omitted. As used below, the term "module" may be a combination of software and/or hardware that implements a predetermined function. Although the means described in the embodiments below are preferably implemented in software, an implementation in hardware, or a combination of software and hardware is also possible and contemplated.
The present embodiment provides a training apparatus for a fault diagnosis model, as shown in fig. 4, including:
the first obtaining module 41 is configured to obtain sample data of the engine, where the sample data includes first data in a preset time period before a fault and second data in normal operation;
the first dividing module 42 is configured to divide each sample data by a preset time unit to obtain a time sequence;
a first extraction module 43, configured to extract at least two timing features from each time sequence, respectively, to obtain a first timing feature of each sample data;
a second extraction module 44, configured to extract a preset proportion based on the magnitude of a feature value of each time sequence feature in the first time sequence features, so as to obtain a second time sequence feature of each sample data;
a first timing characterization module 45 configured to form a sample timing characterization based on the first timing characterization and the second timing characterization;
and the training module 46 is configured to train an initial fault diagnosis model according to the sample timing characteristics, and determine a target fault diagnosis model.
The present embodiment provides a fault diagnosis apparatus, as shown in fig. 5, including:
a second acquisition module 51 for acquiring operating data of the target engine;
a second dividing module 52, configured to divide the running data by a preset time unit to obtain a running time sequence;
a third extraction module 53, configured to extract at least two timing characteristics from each running time sequence, respectively, to obtain a first characteristic of the running data;
a fourth extraction module 54, configured to perform extraction of a preset proportion based on the magnitude of a feature value of each time series feature in the first features, so as to obtain a second feature of each piece of operation data;
a second timing characteristics module 55 for forming timing characteristics based on the first characteristics and the second characteristics;
and a fault diagnosis module 56, configured to input the time sequence characteristics into a target fault prediction model, and determine a fault diagnosis result, where the target fault prediction model is obtained by training according to the first aspect of the present invention or the training method of the fault diagnosis model described in any embodiment of the first aspect.
The training device of the fault diagnosis model and the fault diagnosis device in this embodiment are presented in the form of functional units, where the units refer to ASIC circuits, processors and memories executing one or more software or fixed programs, and/or other devices that can provide the above-mentioned functions.
Further functional descriptions of the modules are the same as those of the corresponding embodiments, and are not repeated herein.
An embodiment of the present invention further provides an electronic device, which has the training apparatus for the fault diagnosis model shown in fig. 4 or the fault diagnosis apparatus shown in fig. 5.
Referring to fig. 6, fig. 6 is a schematic structural diagram of an electronic device according to an alternative embodiment of the present invention, and as shown in fig. 6, the electronic device may include: at least one processor 601, such as a CPU (Central Processing Unit), at least one communication interface 603, memory 604, and at least one communication bus 602. Wherein a communication bus 602 is used to enable the connection communication between these components. The communication interface 603 may include a Display (Display) and a Keyboard (Keyboard), and the optional communication interface 603 may also include a standard wired interface and a standard wireless interface. The Memory 604 may be a Random Access Memory (RAM) or a non-volatile Memory (non-volatile Memory), such as at least one disk Memory. The memory 604 may optionally be at least one storage device located remotely from the processor 601. Wherein the processor 601 may be in connection with the apparatus described in fig. 4 or fig. 5, the memory 604 stores an application program, and the processor 601 calls the program code stored in the memory 604 for performing any of the method steps described above.
The communication bus 602 may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The communication bus 602 may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown in FIG. 6, but this is not intended to represent only one bus or type of bus.
The memory 604 may include a volatile memory (RAM), such as a random-access memory (RAM); the memory may also include a non-volatile memory (english: non-volatile memory), such as a flash memory (english: flash memory), a hard disk (english: hard disk drive, abbreviated: HDD) or a solid-state drive (english: SSD); the memory 604 may also comprise a combination of the above types of memory.
The processor 601 may be a Central Processing Unit (CPU), a Network Processor (NP), or a combination of a CPU and an NP.
The processor 601 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a Programmable Logic Device (PLD), or a combination thereof. The PLD may be a Complex Programmable Logic Device (CPLD), a field-programmable gate array (FPGA), a General Array Logic (GAL), or any combination thereof.
Optionally, the memory 604 is also used for storing program instructions. The processor 601 may call program instructions to implement a training method of a fault diagnosis model as shown in any of the embodiments of the present application, or a fault diagnosis method.
Embodiments of the present invention further provide a non-transitory computer storage medium, where the computer storage medium stores computer-executable instructions, and the computer-executable instructions may execute the method for training a fault diagnosis model or the method for fault diagnosis in any of the above method embodiments. The storage medium may be a magnetic Disk, an optical Disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a Flash Memory (Flash Memory), a Hard Disk (Hard Disk Drive, abbreviated as HDD), a Solid State Drive (SSD), or the like; the storage medium may also comprise a combination of memories of the kind described above.
Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations fall within the scope defined by the appended claims.
Claims (10)
1. A method for training a fault diagnosis model, comprising:
acquiring sample data of an engine, wherein the sample data comprises first data in a preset time period before a fault and second data in normal operation;
dividing each sample data by a preset time unit to obtain a time sequence;
extracting at least two time sequence characteristics from each time sequence to obtain a first time sequence characteristic of each sample data;
extracting a preset proportion based on the characteristic value of each time sequence characteristic in the first time sequence characteristic to obtain a second time sequence characteristic of each sample data;
forming a sample timing feature based on the first timing feature and the second timing feature;
and training an initial fault diagnosis model according to the sample time sequence characteristics, and determining a target fault diagnosis model.
2. The method of claim 1, wherein said obtaining sample data for an engine comprises:
acquiring first original data in a preset time period before a fault and second original data in normal operation;
and respectively removing invalid data from the first original data and the second original data, and determining the first data and the second data.
3. The method according to claim 2, wherein the removing of the invalid data from the first original data and the second original data, respectively, and determining the first data and the second data comprises:
acquiring an action code and an operation state corresponding to the engine;
and removing the invalid data from the first original data and the second original data respectively based on the action code and the running state, and determining the first data and the second data.
4. The method of claim 1, wherein the first timing characteristic comprises a mean value, a deviation value and a standard deviation, and the extracting a predetermined ratio based on the characteristic values of various timing characteristics in the first timing characteristic to obtain the second timing characteristic of each sample data comprises:
sorting based on the eigenvalue size of the various timing characteristics;
and respectively extracting corresponding time sequence characteristics from the sequencing result based on the preset proportion corresponding to each time sequence characteristic to obtain a second time sequence characteristic of each sample data.
5. The method of claim 1, wherein forming a sample timing signature based on the first timing signature and the second timing signature comprises:
respectively calculating an entropy value of a first rotation speed difference of the engine in a preset time period before a fault and a second rotation speed difference entropy value in normal operation based on the first data and the second data;
fusing the first time sequence feature in the preset time period before the fault, the second time sequence feature in the preset time period before the fault and the entropy value of the first rotation speed difference to obtain a sample time sequence feature in the preset time period before the fault;
and fusing the first time sequence characteristic in normal operation, the second time sequence characteristic in normal operation and the entropy value of the second rotation speed difference to obtain a sample time sequence characteristic in normal operation.
6. The method of claim 1, wherein the training an initial fault diagnosis model according to the sample timing characteristics and determining a target fault diagnosis model comprises:
inputting the sample time sequence characteristics into the initial fault diagnosis model to obtain a fault prediction result;
and updating the parameters of the initial fault diagnosis model based on the fault prediction result and the actual fault result corresponding to the sample time sequence characteristics so as to determine the target fault diagnosis model.
7. A fault diagnosis method, comprising:
acquiring operating data of a target engine;
dividing the running data by a preset time unit to obtain a running time sequence;
extracting at least two time sequence characteristics of each running time sequence to obtain a first characteristic of the running data;
extracting a preset proportion based on the magnitude of the characteristic value of each time sequence characteristic in the first characteristic to obtain a second characteristic of each operating data;
forming a timing feature based on the first feature and the second feature;
inputting the time sequence characteristics into a target fault prediction model, and determining a fault diagnosis result, wherein the target fault prediction model is obtained by training according to the training method of the fault diagnosis model in any one of claims 1-6.
8. An electronic device, comprising:
a memory and a processor, the memory and the processor being communicatively connected to each other, the memory having stored therein computer instructions, the processor executing the computer instructions to perform the method of training a fault diagnosis model according to any one of claims 1 to 6, or to perform the method of fault diagnosis according to claim 7.
9. A computer-readable storage medium storing computer instructions for causing a computer to execute the method of training a fault diagnosis model according to any one of claims 1 to 6 or the method of fault diagnosis according to claim 7.
10. A work machine, comprising:
a body;
the electronic device of claim 8 disposed within the body.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210167021.7A CN114548280A (en) | 2022-02-23 | 2022-02-23 | Fault diagnosis model training method, fault diagnosis method and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210167021.7A CN114548280A (en) | 2022-02-23 | 2022-02-23 | Fault diagnosis model training method, fault diagnosis method and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114548280A true CN114548280A (en) | 2022-05-27 |
Family
ID=81678253
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210167021.7A Pending CN114548280A (en) | 2022-02-23 | 2022-02-23 | Fault diagnosis model training method, fault diagnosis method and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114548280A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114911788A (en) * | 2022-07-15 | 2022-08-16 | 中国长江三峡集团有限公司 | Data interpolation method and device and storage medium |
| CN115618206A (en) * | 2022-10-27 | 2023-01-17 | 圣名科技(广州)有限责任公司 | Interference data determination method and device, electronic equipment and storage medium |
-
2022
- 2022-02-23 CN CN202210167021.7A patent/CN114548280A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114911788A (en) * | 2022-07-15 | 2022-08-16 | 中国长江三峡集团有限公司 | Data interpolation method and device and storage medium |
| CN115618206A (en) * | 2022-10-27 | 2023-01-17 | 圣名科技(广州)有限责任公司 | Interference data determination method and device, electronic equipment and storage medium |
| CN115618206B (en) * | 2022-10-27 | 2023-07-07 | 圣名科技(广州)有限责任公司 | Interference data determining method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109048492B (en) | Tool wear state detection method, device and equipment based on convolutional neural network | |
| CN114548280A (en) | Fault diagnosis model training method, fault diagnosis method and electronic equipment | |
| CN110221145B (en) | Power equipment fault diagnosis method and device and terminal equipment | |
| CN111563606A (en) | Equipment predictive maintenance method and device | |
| CN109685805B (en) | Image segmentation method and device | |
| CN110837852A (en) | Fault diagnosis method and device for rolling mill gearbox and terminal equipment | |
| CN114943321A (en) | Fault prediction method, device and equipment for hard disk | |
| CN112565187A (en) | Power grid attack detection method, system, equipment and medium based on logistic regression | |
| CN110909826A (en) | Diagnosis monitoring method and device for energy equipment and electronic equipment | |
| CN115935286A (en) | Abnormal point detection method, device and terminal for railway bearing state monitoring data | |
| CN115033453A (en) | Abnormality detection method, apparatus, device, storage medium, and program | |
| CN111966045A (en) | Machine tool crash monitoring method and device, terminal equipment and storage medium | |
| CN115705413A (en) | Method and device for determining abnormal log | |
| CN113313304A (en) | Power grid accident abnormity analysis method and system based on big data decision tree | |
| CN113110961A (en) | Equipment abnormality detection method and device, computer equipment and readable storage medium | |
| KR20220068799A (en) | System for detecting error of automation equipment and method thereof | |
| CN111523322A (en) | Requirement document quality evaluation model training method and requirement document quality evaluation method | |
| CN110715799B (en) | Method and device for detecting mechanical state of circuit breaker and terminal equipment | |
| CN114742143A (en) | Safe training model construction method, device and system based on federal learning | |
| CN111860661B (en) | Data analysis method and device based on user behaviors, electronic equipment and medium | |
| CN114116845A (en) | Fault report generation method, device and storage medium | |
| CN114881112A (en) | System anomaly detection method, device, equipment and medium | |
| CN112804104A (en) | Early warning method, device, equipment and medium | |
| CN112379656A (en) | Processing method, device, equipment and medium for detecting abnormal data of industrial system | |
| CN116994609B (en) | Data analysis method and system applied to intelligent production line |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |