Method and system for fitting a target sensor for a vehicle, storage medium

Abstract

The present application relates to a method and system for fitting a target sensor for a vehicle, the target sensor for generating a target sensing signal, the method comprising: determining a characteristic of the target sensing signal generated by the target sensor with respect to a vehicle fault diagnosis; receiving a plurality of sensing signals from sensors within the vehicle; screening similar sensing signals from the plurality of sensing signals according to a degree of similarity with the target sensing signal; fitting the similar sensing signals based on a fitting algorithm to generate a fitted signal for fitting the target sensing signal; verifying an application performance of the fitted signal according to the characteristic of the vehicle fault diagnosis; and determining whether to adjust the fitting algorithm and / or a screening rule of the similar sensing signals according to the application performance.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of vehicle sensors, and in particular, to a method and system for fitting target sensors for vehicles, and a storage medium. BACKGROUND

[0002] During the development stage of a new vehicle, it can be necessary to add sensors (referred to herein as target sensors) to achieve the collection of certain information, and to send the collected information in the form of signals (referred to herein as target sensing signals) to the processing system of the vehicle for use in fault diagnosis and the like. However, the addition of sensors means additional costs in the mass-produced vehicle, and more signals will also occupy valuable processing resources of the vehicle. On the other hand, there can be some known or potential correlation or redundancy between the information collected by sensors, so that the target sensing signals of certain sensors can be fitted using the signals of other sensors.

[0003] In view of the above, the present application proposes a method for fitting target sensors for vehicles. SUMMARY

[0004] Embodiments of the present application provide a method and system for fitting target sensors for vehicles, and a storage medium, for achieving efficient fitting of target sensors.

[0005] According to an aspect of the present application, there is provided a method for fitting target sensors for vehicles, the target sensors being used to generate target sensing signals. The method comprises: determining characteristics of the target sensing signals generated by the target sensors with respect to vehicle fault diagnosis; receiving a plurality of sensing signals from sensors within the vehicle; screening similar sensing signals from the plurality of sensing signals according to the degree of similarity with the target sensing signals; fitting the similar sensing signals based on a fitting algorithm to generate a fitted signal used to fit the target sensing signals; verifying the application performance of the fitted signal according to the characteristics of the vehicle fault diagnosis; and determining whether to adjust the fitting algorithm and / or the screening rules of the similar sensing signals according to the application performance.

[0006] In some embodiments of the present application, optionally, determining the characteristics of the target sensing signals generated by the target sensors with respect to vehicle fault diagnosis comprises: collecting a first sensing signal in a fault state and a second sensing signal in a non-fault state by the target sensors; extracting time domain and / or frequency domain characteristics of the first sensing signal and the second sensing signal; and determining the time domain and / or frequency domain characteristics of the target sensing signals with respect to vehicle fault diagnosis according to the differences in the time domain and / or frequency domain characteristics of the first sensing signal and the second sensing signal.

[0007] In some embodiments of the present application, optionally, the target sensing signal has the same sampling frequency as a first signal in the plurality of sensing signals, and whether it is the similar sensing signal is determined according to the similarity of the first signal and the target sensing signal, wherein the similarity is determined by at least one of the following items: data correlation, timing shape similarity, frequency component similarity, timing data goodness of fit.

[0008] In some embodiments of the present application, optionally, the target sensing signal has different sampling frequencies from a second signal set in the plurality of sensing signals, and whether it is the similar sensing signal is determined according to the similarity of each of the second signal set and the target sensing signal, including: determining candidate signals from each of the second signal set and the target sensing signal according to the timing shape similarity; aligning each of the candidate signals with the target sensing signal using an interpolation algorithm; and determining whether the aligned candidate signal is the similar sensing signal according to the similarity with the target sensing signal, wherein the similarity is determined by at least one of the following items: data correlation, timing shape similarity, frequency component similarity, timing data goodness of fit.

[0009] In some embodiments of the present application, optionally, a fitting model is constructed by replacing the convolution layer of the deep residual network with a regression deep residual network of fully connected layers, wherein the fitting model is used to implement the fitting algorithm; a Base Model is constructed based on the Stacking strategy to construct a fitting model for implementing the fitting algorithm; and / or a regression analysis is performed by empirical mode decomposition to construct a fitting model for implementing the fitting algorithm.

[0010] In some embodiments of the present application, optionally, verifying the application performance of the fitting signal according to the characteristics of vehicle fault diagnosis includes: extracting the characteristics of the fitting signal about the vehicle fault diagnosis to diagnose the fault of a vehicle known to have a fault; and determining the application performance according to the fault diagnosis result, wherein the application performance includes finding a fault or not finding a fault.

[0011] In some embodiments of the present application, optionally, the fitting algorithm and the screening rule of the similar sensing signal are solidified for fitting the target sensing signal in the case that the application performance is finding a fault; and the fitting algorithm and / or the screening rule of the similar sensing signal are adjusted in the case that the application performance is not finding a fault.

[0012] According to another aspect of the present application, there is provided a system for fitting a target sensor for a vehicle, the target sensor being configured to generate a target sensing signal. The system comprises: an extraction unit configured to determine characteristics of the target sensing signal generated by the target sensor with respect to a vehicle fault diagnosis; a receiving unit configured to receive a plurality of sensing signals from sensors within the vehicle; and a fitting unit configured to: screen similar sensing signals from the plurality of sensing signals according to a degree of similarity with the target sensing signal; fit the similar sensing signals based on a fitting algorithm to generate a fitted signal for fitting the target sensing signal; a verification unit configured to verify an application performance of the fitted signal according to characteristics of the vehicle fault diagnosis; and an adjustment unit configured to determine whether to adjust the fitting algorithm and / or a screening rule of the similar sensing signals according to the application performance.

[0013] In some embodiments of the present application, optionally, the extraction unit is configured to: collect a first sensing signal in a fault state and a second sensing signal in a non-fault state by the target sensor; extract time domain and / or frequency domain characteristics of the first sensing signal and the second sensing signal; and determine time domain and / or frequency domain characteristics of the target sensing signal with respect to the vehicle fault diagnosis according to a difference between the time domain and / or frequency domain characteristics of the first sensing signal and the second sensing signal.

[0014] In some embodiments of the present application, optionally, the target sensing signal has a same sampling frequency as a first signal of the plurality of sensing signals, and the fitting unit is configured to determine whether the first signal is the similar sensing signal according to a degree of similarity of the first signal with the target sensing signal, wherein the degree of similarity is determined by at least one of: data correlation, timing shape similarity, frequency component similarity, timing data goodness of fit.

[0015] In some embodiments of the present application, optionally, the target sensing signal has a different sampling frequency from a second set of signals of the plurality of sensing signals, and the fitting unit is configured to determine whether each of the second set of signals is the similar sensing signal according to a degree of similarity of each of the second set of signals with the target sensing signal: determine candidate signals from each of the second set of signals according to a timing shape similarity of each of the second set of signals with the target sensing signal; align each of the candidate signals with the target sensing signal using an interpolation algorithm; and determine whether the aligned candidate signals are the similar sensing signals according to a degree of similarity with the target sensing signal, wherein the degree of similarity is determined by at least one of: data correlation, timing shape similarity, frequency component similarity, timing data goodness of fit.

[0016] In some embodiments of the present application, optionally, the verifying unit is configured to extract the feature of the fitting signal about the vehicle fault diagnosis to diagnose the vehicle with known existing faults; and determine the application performance according to the fault diagnosis result, wherein the application performance includes finding faults or not finding faults.

[0017] In some embodiments of the present application, optionally, the adjusting unit is configured to solidify the fitting algorithm and the screening rule of the similar sensing signal for fitting the target sensing signal in the case that the application performance is finding faults; and adjust the fitting algorithm and / or the screening rule of the similar sensing signal in the case that the application performance is not finding faults.

[0018] According to another aspect of the present application, a computer readable storage medium is provided, and the computer readable storage medium stores instructions, and the instructions, when executed by a processor, cause the processor to perform any one of the methods described above.

[0019] The method and system for fitting a target sensor for a vehicle, and the storage medium according to some embodiments of the present application can achieve accurate fitting of the target sensor without increasing the target sensor, and the fitted signal can be used for fault detection, etc., so that the development cost of the vehicle can be reduced. BRIEF DESCRIPTION OF DRAWINGS

[0020] The above and other objects and advantages of the present application will become more apparent by describing in detail the following embodiments thereof in connection with the accompanying drawings, in which like reference numerals refer to like elements throughout the several views.

[0021] Figure 1 A method for fitting a target sensor for a vehicle according to one embodiment of the present application is shown;

[0022] Figure 2 A system for fitting a target sensor for a vehicle according to one embodiment of the present application is shown. DETAILED DESCRIPTION

[0023] For brevity and illustrative purposes, the principles of the present application are described herein primarily in reference to exemplary embodiments thereof. However, those skilled in the art will readily recognize that the same principles can be equivalently applied to all types of methods and systems for fitting a target sensor for a vehicle, and storage media, and can be implemented therein, any such variations not departing from the true spirit and scope of the present application.

[0024] One aspect of the present application provides a method for fitting a target sensor for a vehicle, wherein the target sensor is used to generate a target sensing signal. As Figure 1As shown, the method 10 of fitting a target sensor for a vehicle (hereinafter referred to as method 10) includes the following steps: determining, in step S102, a feature of a target sensing signal generated by a target sensor with respect to vehicle fault diagnosis; receiving, in step S104, a plurality of sensing signals from sensors in the vehicle; screening, in step S106, similar sensing signals from the plurality of sensing signals according to the degree of similarity with the target sensing signal; fitting, in step S108, the similar sensing signals based on a fitting algorithm to generate a fitted signal for fitting the target sensing signal; verifying, in step S110, the application performance of the fitted signal according to the feature of the fitted signal with respect to vehicle fault diagnosis; and determining, in step S112, whether to adjust the fitting algorithm and / or the screening rule of the similar sensing signals according to the application performance. The above steps of the method 10 can achieve the fitting of the target sensing signal that should be generated by the target sensor.

[0025] The method 10 determines, in step S102, a feature of a target sensing signal generated by a target sensor with respect to vehicle fault diagnosis (abbreviated as "extracting fault features" in the figure). Specifically, it can include the following processes: collecting, by the target sensor, a first sensing signal in a fault state and a second sensing signal in a non-fault state; extracting time domain and / or frequency domain features of the first sensing signal and the second sensing signal; and determining time domain and / or frequency domain features of the target sensing signal with respect to vehicle fault diagnosis according to the difference between the time domain and / or frequency domain features of the first sensing signal and the second sensing signal.

[0026] In some examples, in step S102, the data features for fault diagnosis and positioning in the target sensor signal (e.g., the signal generated by the target sensor installed in the actual vehicle as a sample) can be determined. Taking fault diagnosis of a motor-mounted vibration sensor (target sensor) as an example, fault diagnosis and positioning can be achieved by the following steps: (a) collecting high-frequency vibration signals of normal and fault parts, and the frequency response range of the vibration signal needs to be determined in combination with component mechanism analysis; (b) extracting signal features, for vibration signals, time domain common feature indicators include: waveform indicators, pulse indicators, kurtosis indicators, margin indicators, peak-to-peak values, etc.; frequency domain indicators include center of gravity frequency, mean square frequency, root mean square frequency, frequency variance, frequency standard deviation, etc. Time domain features are usually used to reflect the state of the equipment and are used for fault monitoring and trend forecasting; while frequency domain features are usually used to diagnose fault types, causes and positions; (c) abnormality judgment, according to the extracted signal features, compare the normal and fault parts to determine the distinguishing features and deviation thresholds; (d) fault diagnosis, combined with the equipment inherent frequency information determined by mechanism analysis, determine the time domain and frequency domain performances under different fault modes.

[0027] The above steps can be used to extract target sensing signals that can be used to determine a specific fault, and the target sensing signals generated by the target sensor can be used to determine whether such a fault exists. According to the data characteristics of the target sensor signals used for anomaly determination and fault diagnosis, the extraction ideas of the key signals can be used as a reference. In addition, after fitting the target sensor signals from the received sensing signals, the effectiveness of the fitted signals (finding a fault or not finding a fault) can be determined according to the above steps (b), (c), and (d).

[0028] The method 10 receives a plurality of sensing signals from in-vehicle sensors in step S104 (simplified as "receiving sensing signals" in the figure). The in-vehicle sensors referred to herein are existing sensors in the vehicle, or sensors that are essential in the development of the vehicle according to conventional experience (for example, motor speed sensors, rain sensors, etc.), and the redundancy between the signals generated by these sensors can not be significant.

[0029] Although the existing sensors can not directly generate the required target sensing signals, since the information collected by the sensors can be associated (or the signals generated have information redundancy), the sensing signals generated by these existing sensors can be used to fit the required target sensing signals. In some examples, some existing sensors can generate more than one sensing signal, and in other examples, some existing sensors can only generate one corresponding sensing signal. These sensing signals can be arranged in the form of a signal matrix, and of course, other forms can be used to package the received sensing signals.

[0030] The method 10 screens similar sensing signals from the plurality of sensing signals according to the similarity to the target sensing signals in step S106 (simplified as "screening similar sensing signals" in the figure). The sensing signals received in step S104 can not be suitable for fitting the target sensing signals, so screening of these sensing signals is required. Specifically, the retention of which sensing signals can be evaluated according to the similarity (for example, data correlation, timing shape similarity, frequency component similarity, timing data fitting degree, or a combination thereof) of the received sensing signals to the target sensing signals in terms of time domain characteristics and / or frequency domain characteristics. For example, if the target sensing signal is related to the vibration of the vehicle, the light intensity sensing signal received can have a relatively low mathematical similarity to the target sensing signal, and thus is not suitable for fitting the target sensing signal. In addition, since the sampling frequencies of the received sensing signals and the target sensing signals can be the same or different, in some examples, the similarity of the sensing signals to the target sensing signals needs to be determined according to the sampling frequencies of the sensing signals.

[0031] In some embodiments of the present application, in the case that the target sensing signal has the same sampling frequency as a certain signal (e.g., the first signal) in the plurality of sensing signals, it can be determined whether the first signal is a similar sensing signal according to the similarity between the first signal and the target sensing signal. The similarity between the two can be determined by comprehensively considering the following centralized mathematical metrics: data correlation, timing shape similarity, frequency component similarity, and timing data goodness of fit between the first signal and the target sensing signal.

[0032] For data correlation evaluation, the data correlation between the first signal and the target sensing signal can be determined according to the Spearman algorithm or the MIC (The Maximal Information Coefficient) algorithm. For timing shape similarity evaluation, it can be determined by an EMD (Empirical Mode Decomposition) algorithm combined with a dynamic time warping (DTW) algorithm. Frequency component similarity evaluation between the two can be performed by an EMD algorithm combined with a spectral coherence algorithm. In addition, timing data goodness of fit evaluation between the two can be mainly determined by a machine learning algorithm, or a machine learning algorithm combined with stepwise regression. Those skilled in the art can understand after reading the present application that the above specific algorithms can be implemented by referring to the existing technologies in the field, and this will not be repeated here.

[0033] In some embodiments of the present application, in the case that the target sensing signal has a different sampling frequency from a certain signal (e.g., the second signal set) in the plurality of sensing signals, it can be determined whether each signal in the second signal set is a similar sensing signal according to the similarity between each signal in the second signal set and the target sensing signal. Specifically, it can be determined whether it is a similar sensing signal by the following method: first, determine the candidate signals from the second signal set according to the timing shape similarity between each signal in the second signal set and the target sensing signal. This step mainly uses the property that timing shape similarity is not sensitive to signal sampling frequency to preliminarily screen out obviously irrelevant signals. Secondly, align each candidate signal with the target sensing signal by using an interpolation algorithm. Finally, it can be determined whether the aligned candidate signal is a similar sensing signal according to the similarity with the target sensing signal, wherein the similarity is determined by at least one of the following items: data correlation, timing shape similarity, frequency component similarity, and timing data goodness of fit.

[0034] The method 10 fits the similar sensing signals based on a fitting algorithm to generate a fitted signal for fitting the target sensing signal in step S108 (simplified as "generate fitted signal based on fitting algorithm" in the figure). The fitting algorithm is also referred to as a fitting model or an algorithm model herein. In some embodiments of the present application, the fitting algorithm is constructed in the following way: (1) a regression deep residual network (ResNetfor regression) is implemented by replacing the convolutional layers of a deep residual network (ResNet) with fully connected layers, wherein the regression deep residual network can include ten or more network levels; (2) based on a Stacking strategy, several linear regression and nonlinear regression algorithms are selected, the Base Model and the predicted values of the Base Model are obtained after tuning, and the next step of training is performed in two ways: ① the predicted values are directly input into the Meta Model for training; ② the predicted values and the actual data are input into the Meta Model for training. According to the prediction effect, multiple Base Models can be constructed; (3) linear regression analysis is performed by combining empirical mode decomposition (EMD) with methods (1) and (2). Specifically, the corresponding algorithm model can be selected according to the comparison of MSE (Mean squared error) and fault diagnosis effect. The working principles of the regression deep residual network, the Stacking strategy and the empirical mode decomposition are not described herein again, and the related processes can be implemented according to the existing algorithms after reading the present application by the person skilled in the art.

[0035] In addition, the method 10 verifies the application performance of the fitted signal in step S110 (simplified as "verify performance of fitted signal" in the figure); and adjusts the fitting algorithm and / or the screening rule of the similar sensing signals according to the application performance in step S112 (simplified as "adjust fitting algorithm and screening rule" in the figure). The fitted signal needs to be further tested in practice, so as to adjust the fitting algorithm (for example, adjust part of the parameters of the model) and the screening rule of the similar sensing signals (for example, add part of the sensor signals) according to the results of the practice.

[0036] In some embodiments of the present application, verifying the application performance of the fitted signal includes: extracting features of the fitted signal related to vehicle fault diagnosis, performing fault diagnosis on vehicles known to have faults (the extraction process can refer to the extraction process in (b), (c) and (d) in step S102), and determining the application performance according to the fault diagnosis results, wherein the application performance includes finding faults or not finding faults. In some embodiments, the fitting algorithm and the screening rule of the similar sensing signal are solidified for fitting the target sensing signal in the case of finding faults. In addition, the fitting algorithm and / or the screening rule of the similar sensing signal can be adjusted in the case of not finding faults in order to optimize the generated fitted signal in the next cycle.

[0037] Another aspect of the present application provides a system for fitting a target sensor for a vehicle, the target sensor being used to generate a target sensing signal. As shown in Figure 2 The system 20 for fitting a target sensor for a vehicle (hereinafter referred to as system 20) includes an extraction unit 202, a receiving unit 204, a fitting unit 206, a verification unit 208 and an adjustment unit 210. The specific working principle of each unit in the system 20 can be in accordance with the corresponding steps of the method 10 described above, and each unit can implement the fitting of the target sensor according to the specific process described above. The content described above is incorporated herein, and will not be repeated here due to the limited space.

[0038] The extraction unit 202 of the system 20 can determine the features of the target sensing signal generated by the target sensor related to vehicle fault diagnosis. Specifically, the extraction unit 202 can collect a first sensing signal in a fault state and a second sensing signal in a non-fault state through the target sensor; extract the time domain and / or frequency domain features of the first sensing signal and the second sensing signal; and determine the time domain and / or frequency domain features of the target sensing signal related to vehicle fault diagnosis according to the difference between the time domain and / or frequency domain features of the first sensing signal and the second sensing signal.

[0039] The receiving unit 204 of the system 20 can receive a plurality of sensing signals from sensors in the vehicle, the fitting unit 206 can screen similar sensing signals from the plurality of sensing signals according to the degree of similarity with the target sensing signal, and the fitting unit 206 can fit the similar sensing signals based on a fitting algorithm to generate a fitted signal for fitting the target sensing signal.

[0040] In addition, as shown in Figure 2 The system 20 further includes a verification unit 208 and an adjustment unit 210. The verification unit 208 can verify the application performance of the fitted signal. The adjustment unit 210 can adjust the fitting algorithm and / or the screening rule of the similar sensing signal according to the application performance.

[0041] In some embodiments of the present application, in the case that the sampling frequency of the target sensing signal is the same as that of the first signal in the plurality of sensing signals, the fitting unit 206 can determine whether the first signal is a similar sensing signal according to the similarity of the target sensing signal and the first signal, wherein the similarity is determined by at least one of the following items: data correlation, timing shape similarity, frequency component similarity, timing data fitting goodness.

[0042] In some embodiments of the present application, in the case that the sampling frequency of the target sensing signal is different from that of the second signal set in the plurality of sensing signals, the fitting unit 206 can determine whether each of the second signal set is a similar sensing signal according to the similarity of the target sensing signal and each of the second signal set: determine a candidate signal from the timing shape similarity of each of the second signal set and the target sensing signal; align each of the candidate signals with the target sensing signal by using an interpolation algorithm; and determine whether the aligned candidate signal is a similar sensing signal according to the similarity of the target sensing signal and the aligned candidate signal, wherein the similarity is determined by at least one of the following items: data correlation, timing shape similarity, frequency component similarity, timing data fitting goodness.

[0043] In some embodiments of the present application, the verification unit 208 can verify the fitting signal according to the characteristics of the vehicle fault diagnosis, and determine the application performance according to the fault diagnosis result, wherein the application performance includes finding a fault or not finding a fault.

[0044] In some embodiments of the present application, the adjustment unit 210 can solidify the fitting algorithm and the screening rule of the similar sensing signal for fitting the target sensing signal in the case that the application performance is finding a fault, and adjust the fitting algorithm and / or the screening rule of the similar sensing signal in the case that the application performance is not finding a fault.

[0045] According to another aspect of the present application, a computer readable storage medium is provided, wherein instructions are stored, when the instructions are executed by a processor, the processor executes any one of the methods for fitting a target sensor for a vehicle as described above. The computer readable medium in the present application includes various types of computer storage media, which can be any available medium that can be accessed by a general or special purpose computer. For example, the computer readable medium can include RAM, ROM, EPROM, EEPROM, floppy disks, optical disks, etc. 2PROM, a register, a hard disk, a removable disk, a CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other non-transitory medium of carriage or storage which can be used to carry or store desired program elements in the form of instructions or data structures and that can be accessed by a general purpose or special purpose computer, or a general purpose or special purpose processor. Disk typically magnetically copy data while discs optically copy data using a laser. Combinations such as optical storage in combination with a disk storage device should also be included within the computer readable media. An exemplary storage medium is affixed to a processor so as to read from and / or write to the storage medium. In the alternative, the storage medium can be integral to the processor. The processor and the storage medium can reside in an ASIC. The ASIC can reside in a user terminal. In the alternative, the processor and the storage medium can reside as discrete components in a user terminal.

[0046] The specific implementation described above can only be a preferable embodiment of the present application, but the protection scope of the present application is not limited to this. Those skilled in the art can think of other feasible changes or replacements according to the technical range disclosed by the present application, and such changes or replacements are all included in the protection scope of the present application. The embodiments and the features in the embodiments of the present application can also be combined with each other without conflict. The protection scope of the present application is defined by the claims.

Claims

1. A method of fitting a target sensor for a vehicle, characterized by, The target sensor is configured to generate a target sensing signal, and the method comprises: determining a feature of the target sensing signal generated by the target sensor with respect to vehicle fault diagnosis; receiving a plurality of sensing signals from sensors in the vehicle; screening similar sensing signals from the plurality of sensing signals according to a similarity degree with the target sensing signal; fitting the similar sensing signals based on a fitting algorithm to generate a fitted signal for fitting the target sensing signal; verifying an application performance of the fitted signal according to a feature of the fitted signal with respect to vehicle fault diagnosis; and determining whether to adjust the fitting algorithm and / or the screening rule of the similar sensing signals according to the application performance, wherein verifying the application performance of the fitted signal according to a feature of the fitted signal with respect to vehicle fault diagnosis comprises: extracting the feature of the fitted signal with respect to vehicle fault diagnosis to diagnose a fault of a vehicle known to have a fault; and determining the application performance according to a fault diagnosis result, wherein the application performance comprises finding a fault or not finding a fault.

2. The method of claim 1, wherein, Determining a feature of the target sensing signal generated by the target sensor with respect to vehicle fault diagnosis comprises: collecting a first sensing signal in a fault state and a second sensing signal in a non-fault state by the target sensor; extracting time domain and / or frequency domain features of the first sensing signal and the second sensing signal; and determining a time domain and / or frequency domain feature of the target sensing signal with respect to vehicle fault diagnosis according to a difference between the time domain and / or frequency domain features of the first sensing signal and the second sensing signal.

3. The method of claim 1 or 2, wherein, The target sensing signal has a same sampling frequency as a first signal in the plurality of sensing signals, and whether the first signal is the similar sensing signal is determined according to a similarity degree between the first signal and the target sensing signal, wherein the similarity degree is determined by at least one of the following items: data correlation, timing shape similarity, frequency component similarity, and timing data fitting goodness.

4. The method of claim 1 or 2, wherein, The target sensing signal has a different sampling frequency from a second signal set in the plurality of sensing signals, and whether each signal in the second signal set is the similar sensing signal is determined according to a similarity degree between each signal in the second signal set and the target sensing signal, comprising: determining a candidate signal from each signal in the second signal set according to a timing shape similarity between each signal in the second signal set and the target sensing signal; aligning each candidate signal with the target sensing signal by using an interpolation algorithm; and determining whether the aligned candidate signal is the similar sensing signal according to a similarity degree between the aligned candidate signal and the target sensing signal, wherein the similarity degree is determined by at least one of the following items: data correlation, timing shape similarity, frequency component similarity, and timing data fitting goodness.

5. The method of claim 1 or 2, wherein, The fitting algorithm is constructed in the following way: a fitting model for implementing the fitting algorithm is constructed by replacing convolution layers of a deep residual network with fully connected layers of a regression deep residual network; a Base Model is constructed based on a Stacking strategy to construct a fitting model for implementing the fitting algorithm; and / or performing regression analysis by empirical mode decomposition to construct a fitting model for implementing the fitting algorithm.

6. The method of claim 1, wherein, in case of a finding of a fault, solidifying the fitting algorithm and the screening rule of the similar sensing signals for fitting the target sensing signal; and in case of a non-finding of a fault, adjusting the fitting algorithm and / or the screening rule of the similar sensing signals.

7. A system for fitting a target sensor for a vehicle, characterized by the target sensor for generating a target sensing signal, the system comprising: an extraction unit configured to determine a feature of the target sensing signal generated by the target sensor with respect to a vehicle fault diagnosis; a receiving unit configured to receive a plurality of sensing signals from sensors within a vehicle; and a fitting unit configured to: screen similar sensing signals from the plurality of sensing signals according to a degree of similarity to the target sensing signal; fit the similar sensing signals based on a fitting algorithm to generate a fitted signal for fitting the target sensing signal; a verification unit configured to verify an application performance of the fitted signal according to a feature of the fitted signal with respect to the vehicle fault diagnosis; and an adjustment unit configured to determine whether to adjust the fitting algorithm and / or the screening rule of the similar sensing signals according to the application performance, wherein the verification unit is further configured to extract the feature of the fitted signal with respect to the vehicle fault diagnosis to perform a fault diagnosis on a vehicle known to have a fault; and determine the application performance according to a result of the fault diagnosis, wherein the application performance comprises a finding of a fault, a non-finding of a fault.

8. The system of claim 7, the extraction unit is configured to: collect a first sensing signal in a fault state and a second sensing signal in a non-fault state by the target sensor; extract time domain and / or frequency domain features of the first sensing signal, the second sensing signal; and determine a time domain and / or frequency domain feature of the target sensing signal with respect to the vehicle fault diagnosis according to a difference between the time domain and / or frequency domain features of the first sensing signal, the second sensing signal.

9. The system of claim 7 or 8, wherein, the target sensing signal has a same sampling frequency as a first signal of the plurality of sensing signals, and the fitting unit is configured to determine whether the first signal is the similar sensing signal according to a degree of similarity of the first signal to the target sensing signal, wherein the degree of similarity is determined by at least one of: data correlation, timing shape similarity, frequency component similarity, timing data goodness of fit.

10. The system of claim 7 or 8, wherein, the target sensing signal has a different sampling frequency from a second set of signals of the plurality of sensing signals, and the fitting unit is configured to determine whether each of the second set of signals is the similar sensing signal according to a degree of similarity of each of the second set of signals to the target sensing signal: determine a candidate signal from each of the second set of signals according to a timing shape similarity of each of the second set of signals to the target sensing signal; align each of the candidate signals to the target sensing signal using an interpolation algorithm; and ​ determining whether the aligned candidate signal is the similar sensing signal according to a similarity degree with the target sensing signal, wherein the similarity degree is determined by at least one of the following items: data correlation, timing shape similarity, frequency component similarity, timing data goodness of fit.

11. The system of claim 7, wherein, the adjusting unit is configured to solidify the fitting algorithm and the screening rule of the similar sensing signal for fitting the target sensing signal in case that the application behaves as a fault is found; and adjusting the fitting algorithm and / or the screening rule of the similar sensing signal in case that the application behaves as a fault is not found.

12. A computer-readable storage medium having stored therein instructions, the computer-readable storage medium comprising: The instructions, when executed by a processor, cause the processor to perform the method of fitting a vehicle in-target sensor as claimed in any one of claims 1-6.

Images