CN117270495A - Fault prediction method, device, storage medium and vehicle - Google Patents

Abstract

The application discloses a fault prediction method, a device, a storage medium and a vehicle, wherein real vehicle operation data and prediction operation data of the vehicle are obtained, and the prediction operation data are data which are generated by a pre-trained generator and are similar to vehicle operation data of the vehicle in a normal operation state; comparing the real vehicle operation data with the predicted operation data; and when the difference degree of the actual vehicle operation data and the predicted operation data meets a preset difference condition, determining that the actual vehicle operation data is abnormal actual vehicle data, and determining a fault risk area corresponding to the vehicle. Because the pre-trained generator predicts the running data of the vehicle according to the normal running state of the vehicle, when the degree of difference between the real running data of the vehicle and the predicted running data acquired at the same time is large, the fact that the running state of the real vehicle deviates from the normal running state is large can be indicated, and thus the abnormal situation of the vehicle during running is captured in time.

Description

Fault prediction method, device, storage medium and vehicle

Technical Field

The present disclosure relates to the field of automotive technologies, and in particular, to a fault prediction method, a device, a storage medium, and a vehicle.

Background

The state of the vehicle affects the driving safety of the user at all times, and therefore, prediction and diagnosis of vehicle faults are increasingly important. Generally, a technician sets some thresholds or rules based on experience, and performs fault prediction and diagnosis on a vehicle to determine whether the operating state of the vehicle is normal. However, when some atypical or unknown faults occur, it is still difficult to make efficient predictions and diagnoses. Based on this, it is necessary to develop a method of fault prediction to improve efficiency and accuracy of fault prediction to protect driving safety of a user.

Disclosure of Invention

The application provides a fault prediction method, a device, a storage medium and a vehicle, which can solve the technical problems of untimely and inaccurate fault prediction in the related technology.

In a first aspect, an embodiment of the present application provides a fault prediction method, including:

acquiring real vehicle operation data and forecast operation data of a vehicle, wherein the forecast operation data is data similar to the vehicle operation data of the vehicle in a normal operation state, which is generated by a pre-trained generator;

comparing the actual vehicle operation data with the predicted operation data to obtain the degree of difference between the actual vehicle operation data and the predicted operation data;

And when the difference degree meets a preset difference condition, determining that the real vehicle operation data is abnormal real vehicle data, and determining a fault risk area corresponding to the vehicle according to the abnormal real vehicle data.

In a second aspect, an embodiment of the present application provides a failure prediction apparatus, including:

the data acquisition module is used for acquiring real vehicle operation data and forecast operation data of the vehicle, wherein the forecast operation data is data similar to the vehicle operation data of the vehicle in a normal operation state and generated by a pre-trained generator;

the anomaly comparison module is used for comparing the real vehicle operation data with the predicted operation data to obtain the degree of difference between the real vehicle operation data and the predicted operation data;

and the fault determining module is used for determining that the real vehicle operation data are abnormal real vehicle data when the difference degree meets a preset difference condition, and determining a fault risk area corresponding to the vehicle according to the abnormal real vehicle data.

In a third aspect, embodiments of the present application provide a computer storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform the steps of the method described above.

In a fourth aspect, embodiments of the present application provide a vehicle comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the computer program being adapted to be loaded by the processor and to perform the steps of the method described above.

The technical scheme provided by some embodiments of the present application has the beneficial effects that at least includes:

the application provides a fault prediction method, which is used for acquiring real vehicle operation data and prediction operation data of a vehicle, wherein the prediction operation data is data similar to vehicle operation data of the vehicle in a normal operation state and generated by a pre-trained generator; comparing the real vehicle operation data with the predicted operation data to obtain the difference degree between the real vehicle operation data and the predicted operation data; and when the difference degree meets the preset difference condition, determining that the real vehicle operation data is abnormal real vehicle data, and determining a fault risk area corresponding to the vehicle according to the abnormal real vehicle data. Because the pre-trained generator predicts the running data of the vehicle according to the normal running state of the vehicle, when the difference degree between the real running data of the vehicle and the predicted running data acquired at the same time is large, the fact that the running state of the real vehicle deviates greatly from the normal running state can be explained, so that the abnormal situation of the vehicle in running is captured in time, the fault risk area corresponding to the vehicle is judged according to the abnormal real vehicle data, and the prediction and judgment of the fault risk of the vehicle are accurately and timely realized.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is an exemplary system architecture diagram of a fault prediction method provided in an embodiment of the present application;

fig. 2 is a schematic flow chart of a fault prediction method provided in an embodiment of the present application;

fig. 3 is a schematic flow chart of a fault prediction method provided in an embodiment of the present application;

fig. 4 is a schematic diagram of a model training flow of a fault prediction method according to an embodiment of the present application;

fig. 5 is a schematic flow chart of a fault prediction method provided in an embodiment of the present application;

FIG. 6 is a schematic diagram of a logic framework of a fault prediction method according to an embodiment of the present disclosure;

fig. 7 is a block diagram of a fault prediction device according to an embodiment of the present application;

fig. 8 is a schematic structural diagram of a vehicle according to an embodiment of the present application.

Detailed Description

In order to make the features and advantages of the present application more comprehensible, the following description will be given in detail with reference to the accompanying drawings in which embodiments of the present application are shown, and it is apparent that the described embodiments are merely some but not all embodiments of the present application. All other embodiments, which can be made by those skilled in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.

When the following description refers to the accompanying drawings, the same numbers in different drawings refer to the same or similar elements, unless otherwise indicated. The implementations described in the following exemplary examples are not representative of all implementations consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present application as detailed in the accompanying claims.

With the rapid development of automobile development technology, various functional systems are mounted in a vehicle to provide comfortable automobile experience for users, and the running state of each functional module not only can influence the realization of the own functions, but also can influence the overall driving safety of the vehicle. For example, an electric power steering system (Electric Power Steering, abbreviated as EPS) is a power steering system that directly relies on a motor to provide an assist torque, and an electric power steering machine directly provides a steering assist force, so that a driver can operate an automobile more easily, and the electric power steering system has advantages of energy saving, environmental protection and the like compared with a conventional hydraulic power steering system HPS (Hydraulic Power Steering), and is widely used in various electric automobiles at present. However, if the electric power steering system fails, it may cause steering difficulty of the vehicle, and in severe cases, may even threaten driving safety of the vehicle.

Considering that the state of the vehicle affects the driving safety of the user, in order to protect the driving safety of the user, vehicle operation data can be collected in real time in the vehicle to monitor the operation state of the vehicle. Generally, a technician sets some thresholds or rules based on experience, and performs fault prediction and diagnosis on a vehicle to determine whether the operating state of the vehicle is normal. However, for some atypical or unknown faults, the human experience may be relatively sparse, and it may be difficult to effectively predict and diagnose such faults based on the human experience.

Therefore, the embodiment of the application provides a fault prediction method to solve the technical problems of untimely and inaccurate fault prediction.

Referring to fig. 1, fig. 1 is an exemplary system architecture diagram of a fault prediction method according to an embodiment of the present application.

As shown in fig. 1, the system architecture may include a vehicle 101, a network 102, and a server 103. Network 102 is used to provide a medium for communication links between vehicle 101 and server 103. Network 102 may include various types of wired or wireless communication links, such as: the wired communication link includes an optical fiber, a twisted pair wire, or a coaxial cable, and the Wireless communication link includes a bluetooth communication link, a Wireless-Fidelity (Wi-Fi) communication link, a microwave communication link, or the like.

The vehicle 101 may interact with the server 103 via the network 102 to receive messages from the server 103 or to send messages to the server 103, or the vehicle 101 may interact with the server 103 via the network 102 to receive messages or data sent by other users to the server 103. The vehicle 101 may be hardware or software. When the vehicle 101 is hardware, it may be a variety of electronic devices including, but not limited to, a smart watch, a smart phone, a tablet computer, a laptop portable computer, a desktop computer, and the like. When the vehicle 101 is software, it may be installed in the above-listed electronic device, which may be implemented as a plurality of software or software modules (for example, to provide distributed services), or may be implemented as a single software or software module, which is not specifically limited herein.

In the embodiment of the present application, the vehicle 101 first acquires real vehicle operation data and predicted operation data, where the predicted operation data is data similar to vehicle operation data of the vehicle in a normal operation state, which is generated by a pre-trained generator; then, the vehicle 101 compares the actual vehicle operation data with the predicted operation data to obtain the degree of difference between the actual vehicle operation data and the predicted operation data; finally, the vehicle 101 judges according to the difference degree, and when the difference degree meets the preset difference condition, the real vehicle operation data is determined to be abnormal real vehicle data, and the fault risk area corresponding to the vehicle is determined according to the abnormal real vehicle data.

The server 103 may be a business server providing various services. The server 103 may be hardware or software. When the server 103 is hardware, it may be implemented as a distributed server cluster composed of a plurality of servers, or may be implemented as a single server. When the server 103 is software, it may be implemented as a plurality of software or software modules (for example, to provide a distributed service), or may be implemented as a single software or software module, which is not specifically limited herein.

Alternatively, the system architecture may not include the server 103, in other words, the server 103 may be an optional device in the embodiment of the present disclosure, that is, the method provided in the embodiment of the present disclosure may be applied to a system structure including only the vehicle 101, which is not limited in the embodiment of the present application.

It should be appreciated that the number of vehicles, networks, and servers in FIG. 1 is merely illustrative, and that any number of vehicles, networks, and servers may be used as desired for an implementation.

Referring to fig. 2, fig. 2 is a flow chart of a fault prediction method according to an embodiment of the present application. The execution body of the embodiment of the present application may be a vehicle that executes the failure prediction, a processor in the vehicle that executes the failure prediction method, or a server connected to the vehicle that executes the failure prediction method. For convenience of description, a specific implementation procedure of the failure prediction method will be described below taking an example in which the execution subject is a processor in a vehicle.

As shown in fig. 2, the fault prediction method may at least include:

s202, acquiring real vehicle operation data and prediction operation data of a vehicle, wherein the prediction operation data is data similar to vehicle operation data of the vehicle in a normal operation state, and the data is generated by a pre-trained generator.

Alternatively, existing methods of vehicle fault prediction based on rules or classical machine learning have limited accuracy of prediction for some atypical or unknown faults. In addition, a rule threshold value or classical machine learning is set, a rule between conventional data and abnormal data is required to be captured by means of a large amount of sample data, in actual conditions, when the sample data is collected, the duty ratio of the conventional data is far more than that of the abnormal data, the threshold value is set to be difficult and unstable in accuracy due to too sparse abnormal data quantity, the abnormal data and the conventional data are mixed together, all the abnormal data are difficult to accurately find and label the abnormal data in a manual labeling mode, only the sample data collected by a real vehicle can be directly used when a model is trained, the model can learn the conventional data more fully in the training process, the abnormal data is not learned sufficiently, and the effect of fault prediction is affected.

Alternatively, to save labor costs and improve reliability of the failure prediction results, a neural network model may be used to monitor the vehicle operating state. Considering that the model learns the distribution of the regular data more than the difference between the regular data and the abnormal data when the model is trained, the model can master the operation data rule of the vehicle in the normal operation state after convergence. If the model is a generating network, the network characteristics of the generating network are utilized to generate data similar to the characteristics of the sample data according to the learned sample data, then the generating network processes the sample data, and since the conventional data in the sample data is far more than the abnormal data, the generating network can output data similar to the conventional data after more distributions of the conventional data are learned, namely, the vehicle running data of the vehicle in a normal running state can be predicted. When the vehicle is in an abnormal running state, the real vehicle running data can deviate from the running data rule of the normal running state, so that the real vehicle running data and the predicted running data predicted by the generating network have larger deviation, and whether the running state of the vehicle is normal or not can be judged based on the deviation degree.

In the embodiment of the application, the pre-trained generator needs to be trained in advance first, and along with the convergence of the training of the pre-trained generator, the predicted running data follows the distribution rule of the conventional data and can be very similar to the real conventional data. The pre-trained generator is deployed in the vehicle and can be used for generating predicted running data of the vehicle according to the normal running state of the vehicle in the running process of the vehicle, meanwhile, real running data of the vehicle is also collected, the real running data are real running data of the vehicle, the current real running state of the vehicle is reflected, and whether the current real running state of the vehicle is the normal state can be determined based on the predicted running data and the real running data.

S204, comparing the real vehicle operation data with the predicted operation data to obtain the degree of difference between the real vehicle operation data and the predicted operation data.

Optionally, after the actual vehicle operation data and the predicted operation data of the vehicle are obtained as described above, the actual vehicle operation data and the predicted operation data are compared, and the degree of difference between the actual vehicle operation data and the predicted operation data is obtained. The predicted running data are data which are similar to the running data of the vehicle in the normal running state and are generated by the pre-trained generator, the running data can be represented if the vehicle is in the normal running state, the real running data represent the current running state of the vehicle, the difference degree between the real running data and the predicted running data is used for explaining the difference degree between the current running state and the normal running state of the vehicle, based on the difference degree, whether the vehicle is at fault risk or not is judged later, and timely and fast fault response is realized.

And S206, when the degree of difference meets a preset difference condition, determining that the real vehicle operation data is abnormal real vehicle data, and determining a fault risk area corresponding to the vehicle according to the abnormal real vehicle data.

Optionally, in the normal running state of the vehicle, the generated real vehicle data and the predicted running data may have small differences, the fluctuation of the differences in a small range is in the normal error range, and if the degree of the differences between the real vehicle running data and the predicted running data is large, the running state of the vehicle is obviously abnormal. In order to accurately judge the abnormal running state of the vehicle, a difference condition can be preset, so that when the difference degree meets the preset difference condition, the real vehicle running data is determined to be abnormal real vehicle data.

Further, the abnormal real vehicle data are further analyzed, the accurate source of the abnormality can be determined, the continuous development of the abnormal situation can possibly lead to the occurrence of the fault of the vehicle, and the driving safety is threatened, so that the fault risk area corresponding to the vehicle can be determined according to the abnormal real vehicle data, the prediction and the judgment of the fault risk of the vehicle are accurately realized, and the follow-up timely response to the fault situation is facilitated.

In the embodiment of the application, a fault prediction method is provided, real vehicle operation data and prediction operation data of a vehicle are obtained, and the prediction operation data is data similar to vehicle operation data of the vehicle in a normal operation state and generated by a pre-trained generator; comparing the real vehicle operation data with the predicted operation data to obtain the difference degree between the real vehicle operation data and the predicted operation data; and when the difference degree meets the preset difference condition, determining that the real vehicle operation data is abnormal real vehicle data, and determining a fault risk area corresponding to the vehicle according to the abnormal real vehicle data. Because the pre-trained generator predicts the running data of the vehicle according to the normal running state of the vehicle, when the difference degree between the real running data of the vehicle and the predicted running data acquired at the same time is large, the fact that the running state of the real vehicle deviates greatly from the normal running state can be explained, so that the abnormal situation of the vehicle in running is captured in time, the fault risk area corresponding to the vehicle is judged according to the abnormal real vehicle data, and the prediction and judgment of the fault risk of the vehicle are accurately and timely realized.

Referring to fig. 3, fig. 3 is a flow chart of a fault prediction method according to an embodiment of the present application.

As shown in fig. 3, the fault prediction method may at least include:

s302, determining sample original operation data, wherein the sample original operation data is operation data generated in a real environment of a sample vehicle acquired in advance.

Optionally, as can be seen from the description of the foregoing embodiment, a pre-trained generator capable of generating the predicted running data of the vehicle according to the normal running state of the vehicle is deployed in the vehicle, and the current running state of the vehicle is determined by comparing the predicted running data generated by the pre-trained generator with the actual running data generated by the vehicle at the same time. Then prior to this, the pre-trained producers need to be trained to converge so that they learn the distribution of regular operational data. In the training of the pre-trained generator, the execution subject for training may be a vehicle that executes the failure prediction, a processor in the vehicle that executes the failure prediction method, or a server connected to the vehicle that executes the failure prediction method, which is not limited in this embodiment of the present application. If training of the pre-trained generator is completed in the vehicle, the vehicle can directly use the converged pre-trained generator to detect real vehicle data, and if training of the pre-trained generator is completed in the server, the pre-trained generator is required to be deployed into the vehicle for use after being converged.

Optionally, before training starts, first, sample raw operation data is determined, where the sample raw operation data is operation data generated by a sample vehicle collected in advance in a real environment. However, when data are collected, the types of operation data are more, the distribution of various types of data is different, and the units are different, so that data preprocessing is needed, and after the data are normalized, the model can capture the relation among various types of operation data more accurately, thereby realizing more accurate fitting. In the embodiment of the application, the sample real vehicle running data of the sample vehicle running in the real environment can be collected in advance, and the sample real vehicle running data is subjected to data preprocessing to obtain the sample original running data.

Alternatively, when collecting the actual running data of the sample vehicle, relevant data may be collected from various sensors and systems of the sample vehicle, including steering angle, steering torque, vehicle speed, motor torque, battery voltage, and battery current, etc., and these data are recorded during the running of the vehicle in real time, reflecting the running state of the vehicle. After the sample real vehicle running data is collected, data preprocessing is carried out, wherein the data preprocessing comprises at least one of null value eliminating processing, abnormal value eliminating processing and numerical value normalizing processing. The collected data may have missing or outliers and therefore require pre-processing to improve the effectiveness of subsequent model learning. In addition, there may be dimensional differences between different types of data, where different features are represented in different units or orders of magnitude, resulting in some features being weighted more heavily when computed in the model, thereby affecting the performance of the model. Through the data normalization processing, all the features can be scaled to the same scale, so that the contribution of the features to the results in the model is equivalent, the dimensional difference between the data is reduced, and the model can capture the relation between the fields more accurately.

Optionally, referring to fig. 4, fig. 4 is a schematic diagram of a model training flow of a fault prediction method according to an embodiment of the present application. As shown in fig. 4, after the model training process starts, first, sample real vehicle running data in the sample vehicle is collected, and then data preprocessing is performed on the sample real vehicle running data, so as to obtain sample original running data, so as to improve the effect of subsequent model learning.

S304, performing countermeasure training between a preset generator and a preset discriminator based on the original operation data of the sample, and determining that the preset generator after convergence is a pre-trained generator.

Alternatively, for a general generative network, the model is trained solely by sample data, the loss function is calculated by the output of the model, the model is called by using the loss function, the training is usually supervised training, the real sample data needs to be marked, the manual marking cost is high for large data volume, and the constraint accuracy of the conventional loss function on the model is limited. The pre-trained generator may be trained using the Generated Antagonism Network (GANs) as a basis. The generation of the antagonism network consists of two parts: a preset Generator (G) and a preset Discriminator (D), the preset Generator being responsible for generating data, the preset Discriminator being responsible for determining whether the generated data is real data, the preset Generator and the preset Discriminator being mutually opposed during training, and finally enabling the preset Generator to generate data very similar to the real data, and the preset Discriminator being capable of accurately distinguishing the real data from the generated data.

In this embodiment, please continue to refer to fig. 4, the countermeasure training between the preset generator and the preset arbiter may be performed based on the sample raw operation data, and the preset generator after convergence is determined to be a pre-trained generator. Specifically, in the process of countermeasure training, initial parameter adjustment is carried out on the preset generator, the fitting target of the preset generator is set to generate data which is as similar as possible to the original running data of the sample, so that the preset discriminator is confused, the data generated by the preset generator can be finally considered to be real data by the preset discriminator, after the sample false running data are generated based on the characteristics of a generating network, the sample false running data and the sample original running data are input into the preset discriminator, the reality discrimination result of the preset discriminator output aiming at the sample false running data is determined, the preset generator and the preset discriminator are trained based on the reality discrimination result, and at the moment, the training target of the preset discriminator is more accurate to distinguish the real data and the false data generated by the preset generator, so that the preset generator and the preset discriminator gradually converge in the countermeasure training process, and the converged preset generator is determined to be a pre-trained generator. Along with the training convergence of the preset identifier, the constraint precision of the preset generator is higher and higher, and finally the preset generator can generate data very similar to real data. After the preset generator converges, the pre-trained generator is stored so as to facilitate the deployment of the following vehicles, and then the predicted running data generated by the following pre-trained generator can accurately represent the normal running state of the vehicles.

S306, acquiring real vehicle operation data and prediction operation data of the vehicle, wherein the prediction operation data is data similar to vehicle operation data of the vehicle in a normal operation state, and the data is generated by a pre-trained generator.

S308, comparing the real vehicle operation data with the predicted operation data to obtain the degree of difference between the real vehicle operation data and the predicted operation data.

And S310, when the degree of difference meets a preset difference condition, determining that the real vehicle operation data is abnormal real vehicle data, and determining a fault risk area corresponding to the vehicle according to the abnormal real vehicle data.

For steps S306-S310, please refer to the detailed description in steps S202-S206, and the detailed description is omitted here.

In the embodiment of the application, a fault prediction method is provided, sample real vehicle operation data of a sample vehicle in a real environment is collected in advance, the sample real vehicle operation data is subjected to data preprocessing to obtain sample original operation data, and the data processing enables a model to capture relations among various operation data more accurately, so that more accurate fitting is realized; the method can perform countermeasure training between the preset generator and the preset arbiter based on the original running data of the sample, the preset generator after convergence is determined to be a pre-trained generator, along with the convergence of training of the preset arbiter, the constraint precision of the preset generator is higher and higher, and finally the pre-trained generator can generate data very similar to real data.

Referring to fig. 5, fig. 5 is a flow chart of a fault prediction method according to an embodiment of the present application.

As shown in fig. 5, the fault prediction method may at least include:

s502, generating predictive regular operation data through a pre-trained generator.

Optionally, considering that the abnormal real vehicle data may be caused by data fluctuation of the vehicle itself, the abnormal real vehicle data may not necessarily cause a vehicle fault, so in order to further improve the accuracy of fault prediction of the vehicle, after determining that the real vehicle running data is the abnormal real vehicle data, the abnormal real vehicle data may be checked for a second time, to determine whether the abnormal real vehicle data may cause a vehicle fault, and after determining that the risk of the fault does exist, fault response is performed, so that the accuracy and stability of fault prediction are improved, and the user experience is improved.

Alternatively, when the abnormal operation data is checked, the relationship between the abnormal operation data and the vehicle failure may be judged by the failure prediction model, so as to determine whether the abnormal operation data may cause the vehicle to fail. Then the fault prediction model first needs to be trained so that it learns the correlation between the abnormal operation data and the fault and then predicts new data using the learned knowledge of the relationship in the actual scenario. In the training of the failure prediction model, the execution subject for training may be a vehicle that executes the failure prediction, a processor in the vehicle that executes the failure prediction method, or a server connected to the vehicle that executes the failure prediction method, which is not limited in this embodiment of the present application. If the training of the fault prediction model is completed in the vehicle, the vehicle can directly use the converged fault prediction model to detect real vehicle data, and if the training of the fault prediction model is completed in the server, the vehicle needs to be deployed into the vehicle for use after the fault prediction model is converged.

Further, referring to fig. 4, when the fault prediction model is trained, the sample data used for training should be abnormal data with a mark, and because the pre-trained generator is trained based on the sample original operation data, and the conventional data in the sample original operation data is far more than the abnormal data, the distribution and the data rule of the conventional data are fully learned after the pre-trained generator converges, at this time, the pre-trained generator can generate predicted conventional operation data similar to the conventional operation data, the predicted conventional operation data generated by the pre-trained generator is compared with each sample data in the sample original operation data, the sample abnormal operation data with the difference degree meeting the preset difference condition with the predicted conventional operation data can be screened out, the screening marking process does not need manual marking, the manual marking cost is saved, and the sample abnormal operation data can be marked with the mark of the abnormal data for training of the fault prediction model.

S504, determining sample abnormal operation data in the sample original operation data, wherein the difference degree of the sample abnormal operation data and the predicted regular operation data meets a preset difference condition.

Optionally, after the pre-trained generator generates the predicted regular operation data through the description of the above embodiment, comparing the sample original operation data with the predicted regular operation data, it may be determined that, in the sample original operation data, the data whose degree of difference from the predicted regular operation data satisfies the preset difference condition is the sample abnormal operation data.

S506, training a preset prediction model based on the sample abnormal operation data, and determining that the converged preset prediction model is a fault prediction model.

Optionally, after obtaining the sample abnormal operation data, training the preset prediction model based on the sample abnormal operation data, and determining that the converged preset prediction model is a fault prediction model. In a possible implementation manner, the preset prediction model may be based on an Xgboost model, which is called Extreme Gradient Boosting, and is a gradient lifting model, xgboost performs second-order taylor expansion on a loss function, optimizes the loss function by using second-order derivative information of the loss function, and performs greedy selection according to whether the loss function is reduced or not to split nodes or not, so that various types of data can be efficiently processed based on the Xgboost model.

Optionally, if the sample abnormal operation data is data carrying a standard fault label, then referring to fig. 4, during training, determining a predicted fault label output by a preset prediction model for the sample abnormal operation data, calculating a prediction loss of the preset prediction model based on the standard fault label and the predicted fault label corresponding to the sample abnormal operation data, training the preset prediction model through the prediction loss, and determining that the converged preset prediction model is the fault prediction model.

S508, acquiring real vehicle operation data and forecast operation data of the vehicle, wherein the forecast operation data is data similar to vehicle operation data of the vehicle in a normal operation state, which is generated by a pre-trained generator.

S510, comparing the real vehicle operation data with the predicted operation data to obtain the degree of difference between the real vehicle operation data and the predicted operation data.

For steps S508-S510, please refer to the detailed description in steps S202-S204, and the detailed description is omitted here.

S512, when the difference degree meets the preset difference condition, determining that the real vehicle operation data is abnormal real vehicle data.

Optionally, referring to fig. 6, fig. 6 is a schematic diagram of a logic framework of a fault prediction method according to an embodiment of the present application. As shown in fig. 6, in an actual scene, a pre-trained generator and a fault prediction model are deployed in a vehicle, after real vehicle operation data of the vehicle are obtained, data preprocessing is performed, then the pre-trained generator is input, and the pre-trained generator is compared with predicted operation data generated by the pre-trained generator to obtain the difference degree between the real vehicle operation data and the predicted operation data; judging whether a preset difference condition is met or not according to the difference degree, and determining that the real vehicle operation data is abnormal real vehicle data when the difference degree meets the preset difference condition.

S514, inputting the abnormal real vehicle data into a fault prediction model, and determining a state prediction result output by the fault prediction model.

Optionally, referring to fig. 6, after determining that the real vehicle operation data is abnormal real vehicle data, performing secondary verification on the abnormal real vehicle data through the fault prediction model, and judging whether the abnormal real vehicle data can cause a vehicle fault, that is, inputting the abnormal real vehicle data into the fault prediction model, and determining a state prediction result output by the fault prediction model.

And S516, if the state prediction result is a fault state, determining that the area corresponding to the abnormal real vehicle data is a fault risk area of the vehicle.

Optionally, if the state prediction result is a fault state, it is indicated that the abnormality at this time may cause the vehicle to send a fault, and the area corresponding to the abnormal real vehicle data is the fault risk area of the vehicle. After determining that the fault risk does exist, the alarm prompt can be carried out on the abnormal situation and the fault situation, including but not limited to playing alarm information in a vehicle central control display screen, highlighting an indicator lamp corresponding to the fault risk area on an instrument panel and playing a voice prompt through a vehicle-mounted sound box, so that a user or a maintainer can timely take measures, abnormal and potential faults of the vehicle can be timely found and predicted, the safety and the service life of the electric vehicle are improved, the precision and the stability of fault prediction are improved, and the user experience is improved.

In the embodiment of the application, a fault prediction method is provided, the converged pre-trained generator is used for generating predicted normal operation data, and the data which meets the preset difference condition with the difference degree of the predicted normal operation data is determined to be sample abnormal operation data, so that manual labeling is not needed in the screening and labeling process, and the manual labeling cost is saved; training a preset prediction model based on sample abnormal operation data, determining the converged preset prediction model as a fault prediction model, performing secondary verification on abnormal real vehicle data by using the fault prediction model, judging whether the abnormal real vehicle data can cause vehicle faults or not, and improving the precision and stability of fault prediction; abnormality and potential faults of steering assistance are found and predicted in advance, so that a user can take measures in time, and the safety and the service life of the vehicle are improved.

Referring to fig. 7, fig. 7 is a block diagram of a fault prediction device according to an embodiment of the present application.

As shown in fig. 7, the failure prediction apparatus 700 includes:

the data acquisition module 710 is configured to acquire real vehicle operation data and predicted operation data of the vehicle, where the predicted operation data is data similar to vehicle operation data of the vehicle in a normal operation state, the data being generated by a pre-trained generator;

The anomaly comparison module 720 is configured to compare the real vehicle operation data with the predicted operation data to obtain a degree of difference between the real vehicle operation data and the predicted operation data;

the fault determining module 730 is configured to determine that the real vehicle operation data is abnormal real vehicle data when the degree of difference satisfies a preset difference condition, and determine a fault risk area corresponding to the vehicle according to the abnormal real vehicle data.

Optionally, the fault prediction apparatus 700 further includes: the generator training module is used for determining sample original operation data, wherein the sample original operation data is operation data generated in a real environment by a pre-collected sample vehicle; and performing countermeasure training between a preset generator and a preset discriminator based on the sample original operation data, and determining that the converged preset generator is a pre-trained generator.

Optionally, the generator training module is further configured to generate sample spurious operation data of the vehicle based on a preset generator; inputting the sample false operation data and the sample original operation data into a preset discriminator, and determining an authenticity discrimination result output by the preset discriminator aiming at the sample false operation data; training a preset generator and a preset discriminator based on the authenticity discrimination result, and determining that the converged preset generator is a pre-trained generator.

Optionally, the fault determining module 730 is further configured to input the abnormal real vehicle data into the fault prediction model, and determine a state prediction result output by the fault prediction model; if the state prediction result is a fault state, determining the region corresponding to the abnormal real vehicle data as a fault risk region of the vehicle.

Optionally, the fault prediction apparatus 700 further includes: the fault prediction model training module is used for generating prediction conventional operation data through a pre-trained generator; determining sample abnormal operation data in the sample original operation data, wherein the difference degree of the sample abnormal operation data and the predicted normal operation data meets a preset difference condition; training a preset prediction model based on the sample abnormal operation data, and determining the converged preset prediction model as a fault prediction model.

Optionally, the fault prediction model training module is further configured to determine a prediction fault label of the preset prediction model output for the sample abnormal operation data; calculating the prediction loss of a preset prediction model based on a standard fault label and a prediction fault label corresponding to the sample abnormal operation data; and training a preset prediction model through prediction loss, and determining the converged preset prediction model as a fault prediction model.

Optionally, the data acquisition module 710 is further configured to acquire sample real vehicle operation data when the sample vehicle is operating in the real environment, and perform data preprocessing on the sample real vehicle operation data to obtain sample original operation data; the data preprocessing comprises at least one of null value eliminating processing, abnormal value eliminating processing and numerical value normalizing processing.

In an embodiment of the present application, a fault prediction device is provided, where a data acquisition module is configured to acquire real vehicle operation data and predicted operation data of a vehicle, where the predicted operation data is data similar to vehicle operation data of the vehicle in a normal operation state, which is generated by a pre-trained generator; the anomaly comparison module is used for comparing the real vehicle operation data with the predicted operation data to obtain the degree of difference between the real vehicle operation data and the predicted operation data; the fault determining module is used for determining that the real vehicle operation data are abnormal real vehicle data when the difference degree meets the preset difference condition, and determining a fault risk area corresponding to the vehicle according to the abnormal real vehicle data. Because the pre-trained generator predicts the running data of the vehicle according to the normal running state of the vehicle, when the difference degree between the real running data of the vehicle and the predicted running data acquired at the same time is large, the fact that the running state of the real vehicle deviates greatly from the normal running state can be explained, so that the abnormal situation of the vehicle in running is captured in time, the fault risk area corresponding to the vehicle is judged according to the abnormal real vehicle data, and the prediction and judgment of the fault risk of the vehicle are accurately and timely realized.

Embodiments of the present application also provide a computer storage medium having stored thereon a plurality of instructions adapted to be loaded by a processor and to perform the steps of the method according to any of the embodiments described above.

Referring to fig. 8, fig. 8 is a schematic structural diagram of a vehicle according to an embodiment of the present application. As shown in fig. 8, a vehicle 800 may include: at least one vehicle processor 801, at least one network interface 804, a user interface 803, memory 805, at least one communication bus 802.

Wherein a communication bus 802 is used to enable connected communication between these components.

The user interface 803 may include a Display screen (Display) and a Camera (Camera), and the optional user interface 803 may further include a standard wired interface and a wireless interface.

The network interface 804 may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface), among others.

Wherein the vehicle processor 801 may include one or more processing cores. The vehicle processor 801 connects various parts within the overall vehicle 800 using various interfaces and lines, performs various functions of the vehicle 800 and processes data by executing or executing instructions, programs, code sets, or instruction sets stored in the memory 805, and invoking data stored in the memory 805. Alternatively, the vehicle processor 801 may be implemented in hardware in at least one of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), programmable logic array (Programmable Logic Array, PLA). The vehicle processor 801 may integrate one or a combination of several of a central processing unit (Central Processing Unit, CPU), an image processor (Graphics Processing Unit, GPU), and a modem, etc. The CPU mainly processes an operating system, a user interface, an application program and the like; the GPU is used for rendering and drawing the content required to be displayed by the display screen; the modem is used to handle wireless communications. It will be appreciated that the modem may not be integrated into the vehicle processor 801 and may be implemented on a single chip.

The Memory 805 may include a random access Memory (Random Access Memory, RAM) or a Read-Only Memory (ROM). Optionally, the memory 805 includes a non-transitory computer readable medium (non-transitory computer-readable storage medium). Memory 805 may be used to store instructions, programs, code, sets of codes, or instruction sets. The memory 805 may include a stored program area and a stored data area, wherein the stored program area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playing function, an image playing function, etc.), instructions for implementing the above-described respective method embodiments, etc.; the storage data area may store data or the like referred to in the above respective method embodiments. The memory 805 may also optionally be at least one storage device located remotely from the vehicle processor 801. As shown in fig. 8, an operating system, a network communication module, a user interface module, and a failure prediction program may be included in the memory 805, which is a type of computer storage medium.

In the vehicle 800 shown in fig. 8, the user interface 803 is mainly used for providing an input interface for a user, and acquiring data input by the user; and the vehicle processor 801 may be configured to call a failure prediction program stored in the memory 805 and specifically perform the following operations:

Acquiring real vehicle operation data and forecast operation data of a vehicle, wherein the forecast operation data is data similar to the vehicle operation data of the vehicle in a normal operation state, which is generated by a pre-trained generator;

comparing the real vehicle operation data with the predicted operation data to obtain the difference degree between the real vehicle operation data and the predicted operation data;

and when the difference degree meets the preset difference condition, determining that the real vehicle operation data is abnormal real vehicle data, and determining a fault risk area corresponding to the vehicle according to the abnormal real vehicle data.

In some embodiments, the vehicle processor 801, prior to performing the acquisition of real vehicle operation data and the prediction of operation data for the vehicle, further specifically performs the steps of: determining sample original operation data, wherein the sample original operation data is operation data generated in a real environment of a sample vehicle acquired in advance; and performing countermeasure training between a preset generator and a preset discriminator based on the sample original operation data, and determining that the converged preset generator is a pre-trained generator.

In some embodiments, the vehicle processor 801, when performing the countermeasure training between the preset generator and the preset arbiter based on the sample raw operation data, determines that the converged preset generator is a pre-trained generator, specifically performs the following steps: generating sample false operation data of the vehicle based on a preset generator; inputting the sample false operation data and the sample original operation data into a preset discriminator, and determining an authenticity discrimination result output by the preset discriminator aiming at the sample false operation data; training a preset generator and a preset discriminator based on the authenticity discrimination result, and determining that the converged preset generator is a pre-trained generator.

In some embodiments, the vehicle processor 801, when executing the determination of the failure risk area corresponding to the vehicle based on the abnormal real vehicle data, specifically executes the following steps: inputting abnormal real vehicle data into a fault prediction model, and determining a state prediction result output by the fault prediction model; if the state prediction result is a fault state, determining the region corresponding to the abnormal real vehicle data as a fault risk region of the vehicle.

In some embodiments, the vehicle processor 801, after performing the determination that the converged preset generator is a pre-trained generator, further specifically performs the following steps: generating predictive conventional operational data by a pre-trained generator; determining sample abnormal operation data in the sample original operation data, wherein the difference degree of the sample abnormal operation data and the predicted normal operation data meets a preset difference condition; training a preset prediction model based on the sample abnormal operation data, and determining the converged preset prediction model as a fault prediction model.

In some embodiments, when the vehicle processor 801 performs training on the preset prediction model based on the sample abnormal operation data and determines that the converged preset prediction model is a failure prediction model, the following steps are specifically performed: determining a predictive failure tag of a preset predictive model for outputting sample abnormal operation data; calculating the prediction loss of a preset prediction model based on a standard fault label and a prediction fault label corresponding to the sample abnormal operation data; and training a preset prediction model through prediction loss, and determining the converged preset prediction model as a fault prediction model.

In some embodiments, the vehicle processor 801, when executing the determination of the sample raw operational data, specifically performs the following steps: collecting sample real vehicle operation data of a sample vehicle when the sample vehicle operates in a real environment, and carrying out data preprocessing on the sample real vehicle operation data to obtain sample original operation data; the data preprocessing comprises at least one of null value eliminating processing, abnormal value eliminating processing and numerical value normalizing processing.

In the several embodiments provided in this application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of modules is merely a logical function division, and there may be additional divisions of actual implementation, e.g., multiple modules or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or modules, which may be in electrical, mechanical, or other forms.

The modules illustrated as separate components may or may not be physically separate, and components shown as modules may or may not be physical modules, i.e., may be located in one place, or may be distributed over a plurality of network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In the above embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented in software, may be implemented in whole or in part in the form of a computer program product. The computer program product described above includes one or more computer instructions. When the computer program instructions described above are loaded and executed on a computer, the processes or functions described in accordance with the embodiments of the present specification are all or partially produced. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions may be stored in or transmitted across a computer-readable storage medium. The computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by a wired (e.g., coaxial cable, fiber optic, digital subscriber line (Digital Subscriber Line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.). The computer readable storage media may be any available media that can be accessed by a computer or a data storage device such as a server, data center, or the like that contains an integration of one or more available media. The usable medium may be a magnetic medium (e.g., a floppy Disk, a hard Disk, a magnetic tape), an optical medium (e.g., a digital versatile Disk (Digital Versatile Disc, DVD)), or a semiconductor medium (e.g., a Solid State Disk (SSD)), or the like.

It should be noted that, for the sake of simplicity of description, the foregoing method embodiments are all expressed as a series of combinations of actions, but it should be understood by those skilled in the art that the present application is not limited by the order of actions described, as some steps may be performed in other order or simultaneously in accordance with the present application. Further, those skilled in the art will also appreciate that the embodiments described in the specification are all preferred embodiments, and that the acts and modules referred to are not necessarily all necessary for the present application.

In the foregoing embodiments, the descriptions of the embodiments are emphasized, and for parts of one embodiment that are not described in detail, reference may be made to the related descriptions of other embodiments.

The foregoing describes a fault prediction method, apparatus, storage medium, and vehicle provided in the present application, and those skilled in the art, based on the concepts of the embodiments of the present application, may change in terms of specific implementation and application scope, so that the disclosure should not be interpreted as limiting the present application.

Claims (10)

1. A method of fault prediction, the method comprising:

acquiring real vehicle operation data and forecast operation data of a vehicle, wherein the forecast operation data is data similar to the vehicle operation data of the vehicle in a normal operation state, which is generated by a pre-trained generator;

Comparing the actual vehicle operation data with the predicted operation data to obtain the degree of difference between the actual vehicle operation data and the predicted operation data;

and when the difference degree meets a preset difference condition, determining that the real vehicle operation data is abnormal real vehicle data, and determining a fault risk area corresponding to the vehicle according to the abnormal real vehicle data.

2. The method of claim 1, wherein prior to obtaining real vehicle operation data and predicted operation data of the vehicle, further comprising:

determining sample original operation data, wherein the sample original operation data is operation data generated in a real environment of a sample vehicle acquired in advance;

and performing countermeasure training between a preset generator and a preset discriminator based on the sample original operation data, and determining that the preset generator after convergence is a pre-trained generator.

3. The method of claim 2, wherein the performing the countermeasure training between a preset generator and a preset arbiter based on the sample raw operation data, determining that the preset generator after convergence is a pre-trained generator, comprises:

Generating sample spurious operation data of the vehicle based on a preset generator;

inputting the sample false operation data and the sample original operation data into a preset discriminator, and determining an authenticity discrimination result output by the preset discriminator aiming at the sample false operation data;

training the preset generator and the preset discriminator based on the authenticity discrimination result, and determining that the converged preset generator is a pre-trained generator.

4. The method of claim 2, wherein the determining the fault risk region corresponding to the vehicle from the anomalous real vehicle data comprises:

inputting the abnormal real vehicle data into a fault prediction model, and determining a state prediction result output by the fault prediction model;

and if the state prediction result is a fault state, determining the region corresponding to the abnormal real vehicle data as a fault risk region of the vehicle.

5. The method of claim 4, wherein after the determining that the converged preset generator is a pre-trained generator, further comprising:

generating predictive conventional operational data by the pre-trained generator;

Determining sample abnormal operation data in the sample original operation data, wherein the difference degree of the sample abnormal operation data and the predicted regular operation data meets a preset difference condition;

training a preset prediction model based on the sample abnormal operation data, and determining that the converged preset prediction model is a fault prediction model.

6. The method of claim 5, wherein training a predetermined prediction model based on the sample abnormal operation data, determining that the predetermined prediction model after convergence is a failure prediction model, comprises:

determining a predictive failure tag of a preset predictive model output aiming at the sample abnormal operation data;

calculating the prediction loss of the preset prediction model based on the standard fault label corresponding to the sample abnormal operation data and the prediction fault label;

training the preset prediction model through the prediction loss, and determining the converged preset prediction model as a fault prediction model.

7. The method of claim 2, wherein the determining sample raw operational data comprises:

collecting sample real vehicle operation data of a sample vehicle when the sample vehicle operates in a real environment, and carrying out data preprocessing on the sample real vehicle operation data to obtain sample original operation data;

The data preprocessing comprises at least one of null value eliminating processing, abnormal value eliminating processing and numerical value normalizing processing.

8. A fault prediction device, the device comprising:

the data acquisition module is used for acquiring real vehicle operation data and forecast operation data of the vehicle, wherein the forecast operation data is data similar to the vehicle operation data of the vehicle in a normal operation state and generated by a pre-trained generator;

the anomaly comparison module is used for comparing the real vehicle operation data with the predicted operation data to obtain the degree of difference between the real vehicle operation data and the predicted operation data;

and the fault determining module is used for determining that the real vehicle operation data are abnormal real vehicle data when the difference degree meets a preset difference condition, and determining a fault risk area corresponding to the vehicle according to the abnormal real vehicle data.

9. A computer storage medium storing a plurality of instructions adapted to be loaded by a processor and to perform the steps of the method according to any one of claims 1 to 7.

10. A vehicle, characterized in that it is capable of performing the steps of the method according to any one of claims 1-7.