Method for detecting input and output signals of vehicle control unit of electric vehicle
By collecting and processing signals in the electric vehicle vehicle controller, cross-domain consistency verification and fault diagnosis of deep learning models, the problem of insufficient comprehensive and accurate detection in the existing technology is solved, and the control of electric vehicle vehicle is realized Efficient detection and prediction of input and output signals of the device.
Patent Information
- Application Number
- CN202411266495.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-11
- Publication Date
- 2024-11-12
AI Technical Summary
The existing input and output signal detection methods of electric vehicle complete vehicle controllers cannot fully and accurately identify signal abnormalities, especially small deviations or potential faults, and lack real-time and prediction capabilities, making it difficult to adapt to complex vehicle operating environments.
The Hall effect sensor, current sensor, and pressure sensor are used to collect signals, and converted into virtual signals through digital signal processing technology, and comprehensive analysis of the time domain, frequency domain and spatial domain are carried out, combining fast Fourier transform and multi-node synchronous acquisition technology. Perform cross-domain consistency verification and use deep learning models for troubleshooting and prediction.
It realizes a comprehensive analysis of the signal, improves the accuracy and adaptability of detection, can identify abnormalities in a timely manner, reduces the probability of failure, and provides early warnings to adapt to complex vehicle operating conditions.
Smart Images

Figure QLYQS_7 
Figure QLYQS_8 
Figure QLYQS_9
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of electric vehicles, and in particular to a method for detecting input and output signals of a whole vehicle controller of an electric vehicle. Background Art
[0002] An electric vehicle is a vehicle that uses electricity as a power source. It drives the wheels through an electric motor, which is usually powered by one or more battery packs. Compared with traditional internal combustion engine vehicles, electric vehicles do not rely on fossil fuels such as gasoline or diesel, and can significantly reduce tail gas emissions. Electric vehicles include pure electric vehicles, plug-in hybrid vehicles, and hydrogen fuel cell vehicles. The electric vehicle controller is one of the core control systems of electric vehicles. It is responsible for coordinating and controlling the operation of various subsystems of electric vehicles, such as battery management systems, motor controllers, charging systems, and on-board electrical systems. The vehicle controller ensures the safety, economy, and driving comfort of the vehicle by collecting and analyzing various signals of the vehicle. It acts as the brain of the electric vehicle and is responsible for the formulation and execution of the overall control strategy. The main purpose of detecting the input and output signals of the electric vehicle controller is to ensure the safety and reliability of the vehicle. Since the vehicle controller plays a vital role in the operation of electric vehicles, any abnormality in the signal may cause the failure of the control system, which in turn affects the normal operation of the vehicle. For example, erroneous signals may lead to misoperation of the motor, improper battery management, etc., which may bring safety hazards. Therefore, regular testing of these signals can detect potential problems early, prevent failures, and improve the overall performance of the vehicle and user experience.
[0003] The existing detection methods of input and output signals of electric vehicle controllers mainly use sensors and data acquisition cards to collect analog signals in the vehicle control system and conduct real-time monitoring. For digital signals, data is collected through communication buses (such as CAN buses), and special equipment is used for monitoring and analysis, or the collected signal data is recorded and analyzed offline through special software to find abnormalities. Or by comparing the real-time signal with the preset standard signal, it is determined whether there is a deviation, and then whether the system is normal. These existing detection methods can often only detect significant signal abnormalities, and cannot effectively identify smaller deviations or potential faults, and may miss early signs of faults. In addition, the existing methods can usually only analyze within a single domain (such as time domain or frequency domain), and cannot perform consistency verification across domains (time domain, frequency domain, space domain), which may lead to incomplete identification of signal abnormalities. In addition, most of the existing methods are post-analysis, with poor real-time performance, lack of fault prediction ability, and cannot warn potential problems in advance. At the same time, traditional methods mainly rely on fixed standard signals for comparison, cannot flexibly respond to signal changes under different working conditions, and are difficult to adapt to complex vehicle operating environments.
[0004] Therefore, the present invention proposes a method for detecting input and output signals of an electric vehicle controller to solve the above-mentioned problems. Summary of the invention
[0005] In view of the deficiencies in the prior art, the present invention provides a method for detecting input and output signals of a whole vehicle controller of an electric vehicle, which solves the problem that the prior detection method is not comprehensive and accurate enough.
[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for detecting input and output signals of an electric vehicle controller, specifically comprising the following steps: S1. Physical signal acquisition Through the Hall effect sensor, current sensor, and pressure sensor, they are arranged at the input and output ports of the controller, and analog and digital signals are collected to ensure that each key signal has a corresponding collection channel, and then the signal acquisition module is installed; S2. Virtual signal processing Digital signal processing technology is used to convert physical signals into virtual signals. The virtual signal formula is as follows: in, Represents a virtual signal, represents the signal frequency, represents the noise component, represents the amplitude of the signal, Indicates the initial phase of the signal; S3. Time Domain Analysis Perform transient response analysis on virtual signals, including rise time ( )、Fall time( )、Pulse Width( ), the calculation formulas are: ; S4. Fast Fourier Transform Perform FFT processing on the signal to obtain the signal spectrum. The calculation formula of FFT is: in, is the spectrum intensity, is the number of data points; S5. Analysis of signal propagation characteristics Use multi-node synchronous acquisition technology to analyze the propagation characteristics of signals at different nodes and calculate signal delay ( ) and phase difference ( ), the calculation formula is as follows: ; S6. Cross-domain consistency verification Comprehensively compare the analysis results of the time domain, frequency domain, and space domain to ensure that the signal behaves consistently in different domains, including the consistency of signal characteristics in the time domain, frequency domain, and space domain. If cross-domain inconsistencies are found, record and mark them as potential faults. S7. Comparison between virtual signal and physical signal According to the results of cross-domain consistency verification, the collected physical signal is compared with the standard virtual signal in the virtual signal library to detect the deviation of the signal. The deviation is calculated as follows: ; S8. Fault diagnosis and prediction The deep learning model LSTM is trained using historical data to identify abnormal signal patterns, and feature mapping is obtained through supervised learning. Based on real-time and historical data, the prediction model regression model is used to predict the probability of failure and provide early warning. Based on the detection results, the virtual signal library and analysis parameters are dynamically adjusted to perform closed-loop feedback corrections.
[0007] Preferably, in the step S1 of physical signal acquisition, the signal acquisition module is preferably a 16-bit AD sampling card with a sampling rate ≥ 10kHz, and a low-noise amplifier is used to amplify the signal during acquisition, and a low-pass filter is used to remove high-frequency noise.
[0008] Preferably, the virtual signal in the virtual signal processing in step S2 not only retains the core features of the original signal, but can also simulate signal changes under a variety of different working conditions. According to the results of the virtualization processing, a virtual signal library containing signal models under a variety of working conditions needs to be established. The signal library can be used as a reference benchmark in subsequent detection processes for signal matching and anomaly analysis.
[0009] Preferably, the time domain analysis in step S3 is used to detect transient response characteristics of the signal, such as rise time, settling time, and pulse response, and can be compared with standard virtual signals in the virtual signal library to determine whether there is an abnormality.
[0010] Preferably, the fast Fourier transform in step S4 is to convert the virtual signal into the frequency domain, analyze the spectrum characteristics of the signal, and focus on the abnormal phenomena of harmonics and noise peaks in the spectrum to detect possible electromagnetic interference or hardware failure in the control system.
[0011] Preferably, the main purpose of the cross-domain consistency verification in step S6 is to ensure that the signal performs consistently in different analysis domains, time domains, frequency domains, and space domains. If the same signal performs inconsistently in different domains, it means that the signal is abnormal or interfered in a certain domain.
[0012] Preferably, the step S7 of comparing the virtual signal with the physical signal is to compare the physical signal collected in real time with the standard virtual signal in the virtual signal library to check whether there is deviation or abnormal change in the signal. This process can identify abnormal signals caused by environmental changes or hardware aging. The standard virtual signal is derived from the signal model established through multiple experiments and data analysis. These models represent the typical signal characteristics of the system under normal working conditions.
[0013] The present invention provides a method for detecting input and output signals of an electric vehicle controller. It has the following beneficial effects: The present invention provides a detection method for input and output signals of an electric vehicle controller. The method converts physical signals into virtual signals through digital signal processing technology. Without affecting the actual system operation, the method can simulate signal changes under various working conditions. The established virtual signal library contains signal models under various working conditions. These models are used as benchmarks for subsequent detection and comparison. The method not only retains the core features of the original signal, but also can flexibly cope with complex vehicle operation states, thereby improving the coverage and adaptability of detection. In addition, the method of the present invention can comprehensively analyze the signal in the time domain, frequency domain and space domain, deeply analyze the different characteristics of the signal through transient response analysis, fast Fourier transform and multi-node synchronous acquisition technology, and at the same time, through cross-domain consistency verification, ensure the consistency of the performance of the signal in different analysis domains, thereby more comprehensively identifying the anomalies in the signal and avoiding the possible missed detection problem of single domain analysis. In addition, the method of the present invention allows the real-time collected physical signal to be compared with the standard virtual signal in the virtual signal library in real time. By analyzing the deviation of the signal in real time, the anomaly can be discovered in time and the probability of failure can be reduced. DETAILED DESCRIPTION
[0014] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.
[0015] The embodiment of the present invention provides a method for detecting input and output signals of an electric vehicle controller, which specifically includes the following steps: S1. Physical signal acquisition Through the Hall effect sensor, current sensor, and pressure sensor, they are arranged at the input and output ports of the controller, and analog and digital signals are collected to ensure that each key signal has a corresponding collection channel, and then the signal acquisition module is installed; S2. Virtual signal processing Digital signal processing technology is used to convert physical signals into virtual signals. The virtual signal formula is as follows: in, Represents a virtual signal, represents the signal frequency, represents the noise component, represents the amplitude of the signal, Indicates the initial phase of the signal; S3. Time Domain Analysis Perform transient response analysis on virtual signals, including rise time ( )、Fall time( )、Pulse Width( ), the calculation formulas are: ; S4. Fast Fourier Transform Perform FFT processing on the signal to obtain the signal spectrum. The calculation formula of FFT is: in, is the spectrum intensity, is the number of data points; S5. Analysis of signal propagation characteristics Use multi-node synchronous acquisition technology to analyze the propagation characteristics of signals at different nodes and calculate signal delay ( ) and phase difference ( ), the calculation formula is as follows: ; S6. Cross-domain consistency verification Comprehensively compare the analysis results of the time domain, frequency domain, and space domain to ensure that the signal behaves consistently in different domains, including the consistency of signal characteristics in the time domain, frequency domain, and space domain. If cross-domain inconsistencies are found, record and mark them as potential faults. S7. Comparison between virtual signal and physical signal According to the results of cross-domain consistency verification, the collected physical signal is compared with the standard virtual signal in the virtual signal library to detect the deviation of the signal. The deviation is calculated as follows: ; S8. Fault diagnosis and prediction The deep learning model LSTM is trained using historical data to identify abnormal signal patterns, and feature mapping is obtained through supervised learning. Based on real-time and historical data, the prediction model regression model is used to predict the probability of failure and provide early warning. Based on the detection results, the virtual signal library and analysis parameters are dynamically adjusted to perform closed-loop feedback corrections.
[0016] In step S1, the signal acquisition module in the physical signal acquisition is preferably a 16-bit AD sampling card with a sampling rate ≥ 10kHz, and a low-noise amplifier is used to amplify the signal during acquisition, and a low-pass filter is used to remove high-frequency noise.
[0017] In step S2 virtual signal processing, the virtual signal not only retains the core features of the original signal, but also can simulate signal changes under a variety of different working conditions. According to the results of the virtualization processing, a virtual signal library containing signal models under a variety of working conditions needs to be established. This signal library can be used as a reference benchmark in subsequent detection processes for signal matching and anomaly analysis.
[0018] Step S3 time domain analysis is used to detect the transient response characteristics of the signal, such as rise time, settling time, and pulse response, and can be compared with the standard virtual signal in the virtual signal library to determine whether there is an abnormality.
[0019] Step S4 Fast Fourier Transform is to convert the virtual signal into the frequency domain, analyze the spectrum characteristics of the signal, pay attention to the harmonics and noise peak abnormalities in the spectrum, so as to detect possible electromagnetic interference or hardware failure in the control system.
[0020] The main purpose of the cross-domain consistency verification in step S6 is to ensure that the signal performs consistently in different analysis domains, time domain, frequency domain, and space domain. If the same signal performs inconsistently in different domains, it means that the signal is abnormal or interfered in a certain domain.
[0021] Verification process: Data collection: Collect relevant characteristic data of virtual signals in the time domain, frequency domain and space domain. These data include transient response characteristics in the time domain (such as rise time and pulse width), spectrum characteristics in the frequency domain (such as harmonics and noise peaks), and signal propagation characteristics in the space domain (such as phase difference and delay).
[0022] Characteristic comparison: The characteristic data in the above domains are comprehensively compared. By comparing, the difference between the signals in different domains can be identified. For example, if the harmonic detected in the frequency domain does not cause a corresponding transient response change in the time domain, it may indicate that the harmonic is an abnormal phenomenon.
[0023] Consistency check: The comparison results are checked for consistency through the set threshold value. If the check result shows that the signal is consistent in each domain, the signal is marked as normal; if the check result shows that the signal is significantly different (exceeding the threshold value), the system will mark the signal as a potential fault.
[0024] Fault marking and recording: Mark and record detected inconsistent signals as a basis for further analysis and diagnosis.
[0025] Step S7 compares the virtual signal with the physical signal to check whether there is any deviation or abnormal change in the signal by comparing the physical signal collected in real time with the standard virtual signal in the virtual signal library. This process can identify abnormal signals caused by environmental changes or hardware aging. The standard virtual signal comes from the signal model established through multiple experiments and data analysis. These models represent the typical signal characteristics of the system under normal working conditions.
[0026] Signal deviation indicates the difference in value from the standard virtual signal, and abnormal signal indicates the signal whose deviation exceeds the normal range.
[0027] Application of Deep Learning Models: Long Short-Term Memory (LSTM) is a deep learning model suitable for time series data. It can predict future signal change trends by learning characteristic patterns in historical data. In fault diagnosis, the LSTM model can identify abnormal patterns in signals, such as signal fluctuations and abnormal frequency growth. The historically collected signal data set is used as the training set and input into the LSTM model after preprocessing. Through supervised learning, the model automatically adjusts parameters and learns the difference between normal and abnormal signal patterns. In actual operation, the LSTM model predicts possible future development trends based on real-time input signal data. If the model predicts that the characteristics of the signal will deviate from the normal range, it will issue an early warning.
[0028] Application of regression model: The regression model is used to quantitatively predict the probability of a fault. Based on the abnormal signal pattern and historical fault data identified by the LSTM model, the regression model can calculate the probability value of a specific fault occurring in the current state. Based on the fault probability calculated by the regression model, the system automatically generates maintenance recommendations. For example, if the frequency of a signal is abnormal and it is predicted that it may cause motor damage, it will be recommended to focus on checking the hardware of that part during the next maintenance.
[0029] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A method for detecting input and output signals of an electric vehicle controller, characterized in that: The specific steps include: S1. Physical signal acquisition Through the Hall effect sensor, current sensor, and pressure sensor, they are arranged at the input and output ports of the controller, and analog and digital signals are collected to ensure that each key signal has a corresponding collection channel, and then the signal acquisition module is installed; S2. Virtual signal processing Digital signal processing technology is used to convert physical signals into virtual signals. The virtual signal formula is as follows: in, Represents a virtual signal, represents the signal frequency, represents the noise component, represents the amplitude of the signal, Indicates the initial phase of the signal; S3. Time Domain Analysis Perform transient response analysis on virtual signals, including rise time ( )、Fall time( )、Pulse Width( ), the calculation formulas are: ; S4. Fast Fourier Transform Perform FFT processing on the signal to obtain the signal spectrum. The calculation formula of FFT is: in, is the spectrum intensity, is the number of data points; S5. Analysis of signal propagation characteristics Use multi-node synchronous acquisition technology to analyze the propagation characteristics of signals at different nodes and calculate signal delay ( ) and phase difference ( ), the calculation formula is as follows: ; S6. Cross-domain consistency verification Comprehensively compare the analysis results of the time domain, frequency domain, and space domain to ensure that the signal behaves consistently in different domains, including the consistency of signal characteristics in the time domain, frequency domain, and space domain. If cross-domain inconsistencies are found, record and mark them as potential faults. S7. Comparison between virtual signal and physical signal According to the results of cross-domain consistency verification, the collected physical signal is compared with the standard virtual signal in the virtual signal library to detect the deviation of the signal. The deviation is calculated as follows: ; S8. Fault diagnosis and prediction The deep learning model LSTM is trained using historical data to identify abnormal signal patterns, and feature mapping is obtained through supervised learning. Based on real-time and historical data, the prediction model regression model is used to predict the probability of failure and provide early warning. Based on the detection results, the virtual signal library and analysis parameters are dynamically adjusted to perform closed-loop feedback corrections.
2. The method for detecting input and output signals of an electric vehicle controller according to claim 1, characterized in that: In the physical signal acquisition in step S1, the signal acquisition module is preferably a 16-bit AD sampling card with a sampling rate ≥ 10kHz, and a low-noise amplifier is used to amplify the signal during acquisition, and a low-pass filter is used to remove high-frequency noise.
3. The method for detecting input and output signals of an electric vehicle controller according to claim 1, characterized in that: In the virtual signal processing in step S2, the virtual signal not only retains the core features of the original signal, but also can simulate the signal changes under a variety of different working conditions. According to the results of the virtualization processing, a virtual signal library containing signal models under a variety of working conditions needs to be established. The signal library can be used as a reference benchmark in subsequent detection processes for signal matching and anomaly analysis.
4. The method for detecting input and output signals of an electric vehicle controller according to claim 1, characterized in that: The time domain analysis in step S3 is used to detect the transient response characteristics of the signal, such as rise time, settling time, and pulse response, and can be compared with the standard virtual signal in the virtual signal library to determine whether there is an abnormality.
5. The method for detecting input and output signals of an electric vehicle controller according to claim 1, characterized in that: The fast Fourier transform in step S4 is to convert the virtual signal into the frequency domain, analyze the spectrum characteristics of the signal, pay attention to the abnormal phenomena of harmonics and noise peaks in the spectrum, so as to detect possible electromagnetic interference or hardware failure in the control system.
6. The method for detecting input and output signals of an electric vehicle controller according to claim 1, characterized in that: The main purpose of the cross-domain consistency verification in step S6 is to ensure that the signal performs consistently in different analysis domains, time domains, frequency domains, and space domains. If the same signal performs inconsistently in different domains, it means that the signal is abnormal or interfered in a certain domain.
7. The method for detecting input and output signals of an electric vehicle controller according to claim 1, characterized in that: The step S7 of comparing the virtual signal with the physical signal is to compare the physical signal collected in real time with the standard virtual signal in the virtual signal library to check whether there is any deviation or abnormal change in the signal. This process can identify abnormal signals caused by environmental changes or hardware aging. The standard virtual signal comes from the signal model established through multiple experiments and data analysis. These models represent the typical signal characteristics of the system under normal working conditions.
