Integrated ECU power-on burning test monitoring system and method

Abstract

The invention relates to the technical field of automobile electronic testing, and discloses an integrated ECU power-on burning test monitoring system and method. The system comprises a power supply intelligent regulation and control module, a burning verification module, a test diagnosis module, an abnormity monitoring module and a feedback optimization module. The power supply intelligent regulation and control module realizes stable power supply output through a dynamic voltage frequency regulation algorithm and a temperature self-adaptive algorithm; the burning verification module is used for ensuring the burning integrity by using a block redundancy check and increment verification algorithm; the test diagnosis module diagnoses faults based on the multi-dimensional test data and the decision tree model; the abnormity monitoring module monitors abnormity in real time by using real-time streaming data processing and an isolated forest algorithm; and the feedback optimization module optimizes test diagnosis through incremental learning and a Bayesian optimization algorithm. According to the system and the method, comprehensive and accurate monitoring and optimization of the ECU are realized, and the performance and the reliability of the ECU are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of automotive electronics testing, and particularly to an integrated ECU power-on programming test monitoring system and method. Background Art

[0002] In the field of automotive electronics, the ECU (Engine Control Unit) is the core control component of an automobile, and its performance directly affects the safety, stability, and functionality of the entire vehicle. With the increasing trend of automotive intelligence and electrification, the functions of the ECU are becoming more and more complex, and accurate and efficient testing and monitoring of it have become a key requirement in the industry.

[0003] Early ECU test monitoring technologies were relatively simple. In the ECU power-on process, traditional power supply methods were difficult to adapt to the complex and variable power requirements of the ECU. For example, when the instantaneous current demand of the ECU increases, the traditional power supply cannot quickly adjust the output, easily leading to voltage fluctuations, affecting the normal startup of the ECU, and even possibly damaging the chip. Moreover, traditional heat dissipation measures are often in a fixed mode, without considering real-time temperature changes, resulting in energy waste or insufficient heat dissipation.

[0004] The programming process also has many problems. In the past, the data verification method for programming was single, and when the data volume was large, it was difficult to ensure the verification accuracy. If there are errors in the programmed data, subsequent ECU operations may experience logical confusion, functional abnormalities, etc., increasing the risk of vehicle failures. At the same time, there is no effective error handling mechanism during the programming process. Once a data transmission error occurs, only manual intervention or reprogramming can be performed, with extremely low efficiency.

[0005] In terms of ECU function testing, traditional testing methods mostly analyze based on single-dimensional data, making it difficult to comprehensively and accurately judge faults. For example, only detecting the communication function and ignoring factors such as signal amplitude error and power consumption may miss potential fault hazards. Moreover, fault diagnosis relies on manual experience and simple threshold judgments, resulting in a high misjudgment rate in the face of complex fault modes.

[0006] For the monitoring of the ECU operating state, traditional technologies can only passively handle after a fault occurs and cannot give early warnings. For example, using a regular detection method, sudden abnormalities during the operation process cannot be captured in a timely manner, making it difficult to take timely measures to avoid the expansion of faults, posing a serious threat to the safe driving of the vehicle. Moreover, each test monitoring link is independent, lacking a data interaction and collaborative optimization mechanism, unable to form a closed-loop feedback, and difficult to meet the rapid development needs of the ECU.

[0007] With the continuous improvement of the requirements for ECU performance and reliability in the automotive industry, the drawbacks of these traditional technologies have become increasingly prominent. There is an urgent need for an integrated and intelligent ECU power-on programming test and monitoring system and method to improve the overall performance and reliability of automotive electronic systems and meet the development needs of the modern automotive industry. Summary of the Invention

[0008] The purpose of the present invention is to provide an integrated ECU power-on programming test and monitoring system and method to solve the problems raised in the above background technology.

[0009] To achieve the above object, the present invention provides the following technical solution: An integrated ECU power-on programming test and monitoring system, the system includes: A power intelligent regulation module, used to perform real-time tracking and processing on voltage and current data during the ECU power-on process through a dynamic voltage and frequency adjustment algorithm to generate dynamic power parameters; and perform heat dissipation strategy optimization processing on the dynamic power parameters by combining a temperature adaptive algorithm with real-time temperature monitoring data to generate a stable power output; A programming verification module, used to perform segmented hash calculation processing on the programming data stream through a block redundancy check algorithm to obtain a data verification sequence; and perform integrity matching processing on the data verification sequence by combining an incremental verification algorithm with a preset reference hash value to generate a programming integrity report; A test diagnosis module, used to perform feature extraction processing on ECU function parameters based on a multi-dimensional test data set to obtain test feature vectors; and perform fault mode matching processing on the test feature vectors by using a decision tree classification model to generate a diagnostic result set; An anomaly monitoring module, used to perform sliding window sampling processing on ECU operation status data through real-time stream data processing technology to obtain status time series data; and perform anomaly probability calculation processing on the status time series data by using an isolation forest algorithm to generate real-time anomaly alarms; A feedback optimization module, used to perform correlation analysis processing on the diagnostic result set and real-time anomaly alarms through an incremental learning algorithm to obtain optimized feature parameters; and perform dynamic adjustment processing on the threshold of the test diagnosis module by combining a Bayesian optimization algorithm with the optimized feature parameters to generate a closed-loop test strategy.

[0010] Preferably, the power intelligent regulation module includes: Perform dynamic modeling processing on the voltage rise slope and current peak value during the ECU power-on stage through a dynamic voltage and frequency adjustment algorithm to generate a power regulation model; Perform gradient descent optimization processing on the cooling fan speed in the power regulation model based on real-time temperature sensor data to obtain temperature adaptive parameters; Compare the temperature adaptive parameters with a preset power stability threshold to generate a heat dissipation strategy instruction.

[0011] Preferably, the function formula of the dynamic voltage and frequency adjustment algorithm is as follows: , In the formula, is the optimized output voltage, is the voltage regulation weight coefficient, is the power change rate, is the current compensation factor, is the real-time current peak value, is the reference current threshold.

[0012] Preferably, the programming verification module includes: The programming data stream is segmented into data blocks according to a fixed byte length through the block redundancy check algorithm, and double hash calculation processing is performed on each data block to generate local check codes; The incremental verification algorithm is used to compare the local check codes with the cloud reference hash library block by block to screen out abnormal data block indexes; Based on the abnormal data block index, trigger the data retransmission protocol until all data blocks pass the integrity verification.

[0013] Preferably, the test and diagnosis module includes: The communication response delay, signal amplitude error and power consumption data of the ECU are standardized through a multi-dimensional test data set to generate a normalized test matrix; The decision tree classification model is used to perform feature weight allocation processing on the normalized test matrix to construct a fault decision tree; Based on the fault decision tree, perform path matching processing on the real-time test data and output the fault type and confidence level.

[0014] Preferably, the anomaly monitoring module includes: The ECU operation state data is segmented by time series through the sliding window technique to generate window data segments; The isolation forest algorithm is used to calculate the anomaly scores of the window data segments to obtain the anomaly probabilities of each data segment; Based on the dynamic threshold segmentation method, classify the anomaly probabilities and trigger a hierarchical alarm instruction.

[0015] Preferably, the formula for calculating the anomaly score of the isolation forest algorithm is as follows: , In the formula, is the data point 's anomaly score, is the data point 's expected path length in the isolation tree, is the normalization factor, and [[ID=]] is the number of samples.

[0016] Preferably, the feedback optimization module includes: Performing feature correlation processing on historical diagnosis results and real-time anomaly alerts through an incremental learning algorithm to generate an optimized weight matrix; Using the Bayesian optimization algorithm to perform posterior probability update processing on the fault determination threshold of the test diagnosis module to generate dynamically adjusted parameters; Embedding the dynamically adjusted parameters into the decision-making process of the test diagnosis module to achieve closed-loop policy iteration.

[0017] Preferably, the system further includes: A visualization module, which is used to convert the dynamic power supply parameters of the power intelligent regulation module, the burn-in integrity report of the burn-in verification module, the diagnosis result set of the test diagnosis module, and the real-time anomaly alerts of the anomaly monitoring module into visualization elements through a data mapping algorithm; using a graphics rendering engine to perform layout and rendering processing on the visualization elements in combination with a preset visualization template to generate a multi-dimensional visualization interface.

[0018] Preferably, the present invention further includes an integrated ECU power-on burn-in test monitoring method, and the method includes the following steps: Step 1: Performing real-time tracking processing on the voltage and current data during the ECU power-on process through a dynamic voltage frequency adjustment algorithm to generate dynamic power supply parameters, and then using a temperature adaptive algorithm to combine real-time temperature monitoring data to perform heat dissipation strategy optimization processing on the dynamic power supply parameters to generate a stable power output; Step 2: Performing segmented hash calculation processing on the burn-in data stream through a block redundancy check algorithm to obtain a data check sequence, and then using an incremental verification algorithm to combine a preset reference hash value to perform integrity matching processing on the data check sequence to generate a burn-in integrity report; Step 3: Performing feature extraction processing on the ECU function parameters based on a multi-dimensional test data set to obtain test feature vectors, and then using a decision tree classification model to perform fault mode matching processing on the test feature vectors to generate a diagnosis result set; Step 4: Performing sliding window sampling processing on the ECU operating state data through real-time stream data processing technology to obtain state time series data, and then using an isolation forest algorithm to perform anomaly probability calculation processing on the state time series data to generate real-time anomaly alerts; Step 5: Performing correlation analysis processing on the diagnosis result set and the real-time anomaly alerts through an incremental learning algorithm to obtain optimized feature parameters, and finally using the Bayesian optimization algorithm to combine the optimized feature parameters to perform dynamic adjustment processing on the threshold of the test diagnosis module to generate a closed-loop test strategy.

[0019] Compared with the prior art, the beneficial effects of the present invention are as follows: In terms of intelligent power regulation, by using the dynamic voltage and frequency adjustment algorithm to track and process the voltage and current data in real time during the ECU power-on process, it can accurately match the power requirements of different stages of the ECU. For example, at the moment when the ECU starts, the algorithm can quickly adjust the voltage rise slope to avoid voltage impact on the chip; at the same time, it monitors the current peak value in real time to prevent components from being burned by excessive current. Combining the temperature adaptive algorithm and real-time temperature monitoring data to optimize the heat dissipation strategy, automatically adjusting the speed of the cooling fan according to the ambient temperature and the working temperature of the ECU, etc., effectively reducing the risk of system failures caused by overheating, improving the stability and reliability of the power supply, and thus extending the service life of the ECU.

[0020] The programming verification module adopts the block redundancy verification algorithm and the incremental verification algorithm, which greatly improves the accuracy and integrity of the programmed data. The programmed data stream is processed in blocks and double hashing calculations are performed to generate accurate local verification codes for each block of data, so that even if some data is incorrect, it can be accurately located. Comparing with the cloud benchmark hash library block by block, timely screening out the indexes of abnormal data blocks and triggering the retransmission protocol to ensure that the data programmed into the ECU is completely correct, avoiding abnormal ECU functions caused by data errors, and ensuring the stable operation of the vehicle control system.

[0021] The test and diagnosis module deeply analyzes the ECU function parameters based on a multi-dimensional test data set. Comprehensively collecting data such as communication response delay, signal amplitude error, and power consumption and standardizing them, the constructed fault decision tree can accurately assign feature weights to achieve accurate diagnosis of ECU faults. It can not only quickly determine the type of fault, but also give the fault confidence level, providing detailed and reliable fault information for maintenance personnel, greatly shortening the fault troubleshooting time, improving the maintenance efficiency, and reducing the maintenance cost.

[0022] The anomaly monitoring module uses real-time stream data processing technology and the isolation forest algorithm to monitor the running state of the ECU in real time and efficiently. The sliding window sampling is used to obtain the state time series data to timely capture the subtle changes during operation; the isolation forest algorithm accurately calculates the anomaly probability, and combines the dynamic threshold segmentation method to issue alarms at different levels, enabling operators to discover potential problems and take measures in the first time, effectively preventing serious consequences caused by ECU failures and ensuring vehicle driving safety.

[0023] The feedback optimization module realizes the self-optimization and iteration of the system through the incremental learning algorithm and the Bayesian optimization algorithm. It conducts correlation analysis on the diagnostic result set and real-time anomaly alarms, extracts optimized characteristic parameters, and dynamically adjusts the thresholds of the test diagnostic module, enabling the system to adapt to different working conditions and individual differences of ECUs, continuously improving the accuracy and adaptability of test diagnosis, and continuously enhancing the system performance.

[0024] The visualization module converts the key information of each module into intuitive visualization elements and generates a multi-dimensional visualization interface. Operators can clearly grasp the operating parameters, programming status, fault diagnosis results, and anomaly alarm situations of the ECU at a glance, facilitating real-time monitoring and management, and improving work efficiency and decision-making accuracy. Description of the Drawings

[0025] Figure 1 It is the working principle diagram of the integrated ECU power-on programming test and monitoring system described in the present invention; Figure 2 It is the flowchart of data processing and verification of the programming verification module; Figure 3 It is the flowchart of fault diagnosis of the test diagnostic module; Figure 4 It is the flowchart of real-time monitoring of the anomaly monitoring module. Detailed Embodiments

[0026] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0027] Please refer to Figures 1-4 , the present invention provides an integrated ECU power-on programming test and monitoring system, which mainly consists of a power intelligent regulation module, a programming verification module, a test diagnostic module, an anomaly monitoring module, and a feedback optimization module. Each module works in coordination to achieve comprehensive monitoring and optimization of the ECU power-on, programming, testing, and operating states. The overall implementation solution is as follows: Power intelligent regulation module: During the ECU power-on stage, this module uses the dynamic voltage and frequency adjustment algorithm to perform real-time tracking and processing on voltage and current data. By monitoring the real-time changes in voltage and current, it continuously adjusts relevant parameters to generate dynamic power parameters. At the same time, in combination with the temperature adaptive algorithm and real-time temperature monitoring data, it optimizes the heat dissipation strategy for the dynamic power parameters. For example, when the monitored temperature rises, it automatically adjusts the speed of the cooling fan, etc., so as to generate a stable power output and ensure that the ECU can obtain a stable power supply environment during the power-on process.

[0028] Programming verification module: The programming verification module processes the programming data stream through the block redundancy verification algorithm. It segments the programming data stream according to certain rules, performs hash calculation on each segment to obtain a data verification sequence. Then, using the incremental verification algorithm, it performs integrity matching between the obtained data verification sequence and the preset reference hash value. If data incompleteness or errors are found during the matching process, corresponding processing is carried out in a timely manner, and finally a programming integrity report is generated to ensure the accuracy of the programming data.

[0029] Test and diagnosis module: Based on a multi-dimensional test data set, it extracts the feature parameters of the ECU's functional parameters. These functional parameters cover multiple aspects such as communication response delay, signal amplitude error, power consumption, etc. After obtaining the test feature vector, it uses the decision tree classification model to analyze it, matches the test feature vector with various fault modes, and then generates a diagnosis result set to determine whether the ECU has a fault and the type of the fault.

[0030] Abnormal monitoring module: The abnormal monitoring module uses real-time stream data processing technology to perform sliding window sampling on the ECU operation status data. The collected data is divided into small window data segments in chronological order to form status time series data. Then, the isolation forest algorithm is used to calculate the abnormal probability of these status time series data. Once the calculated abnormal probability exceeds the set threshold, a real-time abnormal alarm is generated to timely detect abnormal situations during the ECU operation process.

[0031] Feedback optimization module: This module performs correlation analysis on the diagnosis result set and real-time abnormal alarm through the incremental learning algorithm. It mines potential optimization information from a large amount of historical data and real-time data to obtain optimization feature parameters. Then, using the Bayesian optimization algorithm, in combination with these optimization feature parameters, it dynamically adjusts the threshold of the test and diagnosis module to generate a closed-loop test strategy, continuously improving the system's test and diagnosis performance.

[0032] The present invention will be further described below in conjunction with Embodiments 1 to 6: Embodiment 1: In this embodiment, the specific implementation of the power intelligent regulation module is elaborated in detail. The power intelligent regulation module mainly uses the dynamic voltage and frequency adjustment algorithm and the temperature adaptive algorithm to ensure stable power supply during the ECU power-on process and optimize the heat dissipation strategy.

[0033] The dynamic voltage and frequency adjustment algorithm plays a key role in the ECU power-on stage. In this stage, it is necessary to perform dynamic modeling on the voltage rise slope and current peak to generate a power regulation model. Specifically, through the dynamic voltage and frequency adjustment algorithm, the voltage rise slope and current peak during the ECU power-on stage are monitored and analyzed in real time to determine their change rules, thereby constructing a power regulation model that can accurately describe the dynamic characteristics of the power supply.

[0034] The function formula of the dynamic voltage and frequency adjustment algorithm is: . Among them, represents the optimized output voltage, which is the voltage value finally output to the ECU after algorithm adjustment and directly affects the power supply stability of the ECU; is the voltage adjustment weight coefficient, and its value is set according to the characteristics of the ECU and different working scenarios, and is used to adjust the influence degree of different power change rates on the output voltage; represents the power change rate, which reflects the change of power with time during the ECU power-on process and is an important dynamic parameter in the algorithm; is the current compensation factor, which is used to compensate the influence of current on the output voltage and ensure the stability of the output voltage during current fluctuations; is the real-time current peak value, that is, the maximum current value that appears at a certain moment during the ECU power-on process. The accurate monitoring and processing of this parameter are crucial to ensure power supply safety and stability; is the reference current threshold value, which is a preset reference value used for comparison and calculation with the real-time current peak value.

[0035] Another key step of this module is to perform gradient descent optimization on the rotation speed of the cooling fan in the power regulation model based on the real-time temperature sensor data. The real-time temperature sensor is installed near the ECU or the power supply, and it collects temperature data in real time and transmits it to the power intelligent regulation module. Using the gradient descent optimization algorithm, with the rotation speed of the cooling fan as the optimization variable and the goal of reducing the system temperature, the rotation speed of the cooling fan is continuously adjusted. The specific process is to calculate the gradient direction according to the difference between the current temperature and the target temperature, and gradually adjust the rotation speed of the cooling fan according to the principle of gradient descent, so that the system temperature gradually approaches the target temperature. In this way, the temperature adaptive parameter is obtained, which reflects the optimal rotation speed of the cooling fan under different temperature conditions.

[0036] Compare the temperature adaptation parameter with a preset power supply stability threshold to generate a heat dissipation strategy instruction. The preset power supply stability threshold is determined according to the design requirements of the ECU and the power supply, and is used to measure whether the temperature of the system is within a safe and stable range. When the temperature adaptation parameter (i.e., the temperature adjustment effect corresponding to the current heat dissipation fan speed) exceeds the preset power supply stability threshold, it indicates that the system temperature is too high and further heat dissipation is required. At this time, the generated heat dissipation strategy instruction may be to increase the speed of the heat dissipation fan; conversely, if the temperature adaptation parameter is lower than the preset threshold, it indicates that the heat dissipation effect is good, and the speed of the heat dissipation fan can be appropriately reduced to reduce power consumption and noise. In this way, the optimization of the heat dissipation strategy is achieved, ensuring the stable operation of the ECU during the power-on process.

[0037] Embodiment 2: During the programming process, the programming verification module first divides the programming data stream into data blocks according to a fixed byte length through a block redundancy check algorithm. The selection of the fixed byte length is usually determined comprehensively based on factors such as the characteristics of the programming data, the read and write characteristics of the storage device, and the accuracy requirements of the verification. For example, in some scenarios with high programming speed requirements and large data volumes, a larger fixed byte length, such as 512 bytes or 1024 bytes, can be selected to reduce the number of blocks and improve the verification efficiency; while in cases with extremely high requirements for data accuracy, a smaller fixed byte length, such as 64 bytes or 128 bytes, may be selected to enhance the verification fineness.

[0038] Perform double hash calculation processing on each data block to generate a local check code. Double hash calculation is a method to enhance the reliability of data verification. It calculates the same data block using two different hash algorithms. For example, first use the MD5 algorithm to perform the first hash calculation on the data block to obtain a hash value; then use the SHA-256 algorithm to perform the second hash calculation on the same data block to obtain another hash value. Combine these two hash values to form the local check code of the data block. This double hash calculation method can effectively reduce the probability of hash collisions and improve the accuracy of verification.

[0039] The incremental verification algorithm is used to compare the local check codes with the cloud-based reference hash library block by block, and filter out the abnormal data block indexes. The cloud-based reference hash library pre-stores the hash values of the correctly burned data, which are obtained by performing the same hash calculation on the standard data before data burning. When performing the comparison, the incremental verification algorithm does not compare the check codes of all data blocks at once, but block by block. When it is found that the local check code of a certain data block is inconsistent with the corresponding hash value in the cloud-based reference hash library, the index of this data block is recorded, and this index is the abnormal data block index. Through this block-by-block comparison and incremental verification method, the data blocks with problems during the burning process can be quickly located, improving the efficiency and accuracy of verification.

[0040] Based on the abnormal data block index, trigger the data retransmission protocol until all data blocks pass the integrity verification. Once the abnormal data block index is determined, the system will immediately trigger the data retransmission protocol. The data retransmission protocol stipulates rules such as the method, number of times, and timeout handling of data retransmission. For example, the system will send a retransmission request to the burning device, requesting to retransmit the data block with the abnormal data block index. During the retransmission process, a retransmission count limit may be set, such as a maximum of 3 retransmissions. If the check code of this data block passes the comparison with the cloud-based reference hash library within the specified number of retransmissions, it is considered that the data block is burned successfully; if it still fails the verification after exceeding the retransmission count, error information may be recorded and the operator may be prompted for manual intervention. By continuously retransmitting the abnormal data blocks until all data blocks pass the integrity verification, the accuracy and integrity of the burned data are ensured.

[0041] Example 3: The test and diagnosis module is used to determine whether the ECU has a fault and the type of the fault. The working process and implementation details of this module are introduced in detail in this example.

[0042] The test and diagnosis module first standardizes the communication response delay, signal amplitude error, and power consumption data of the ECU through a multi-dimensional test data set to generate a normalized test matrix. The multi-dimensional test data set covers various performance data of the ECU under different working conditions. The communication response delay reflects the speed and efficiency of the ECU when communicating with other devices, the signal amplitude error reflects the accuracy of the ECU's output signal, and the power consumption data shows the energy consumption of the ECU.

[0043] When performing the standardization process, different standardization methods are adopted for different types of data. For the communication response delay data, assuming its original data is , first calculate its mean and standard deviation , and then use the formula to perform standardization processing on each data point to obtain the standardized communication response delay data For the signal amplitude error data and power consumption data, a similar standardization method is adopted to obtain the standardized signal amplitude error data and power consumption data These standardized data are arranged in columns to form a normalized test matrix .

[0044] The decision tree classification model is used to perform feature weight assignment processing on the normalized test matrix to construct a fault decision tree. The decision tree classification model is a classification algorithm based on a tree structure, which classifies data according to the features of the data. When constructing the fault decision tree, first calculate the information gain for each feature in the normalized test matrix (i.e., the standardized columns corresponding to communication response delay, signal amplitude error, and power consumption data). The calculation formula for information gain is , where represents the information gain of feature on the data set , is the information entropy of the data set , is the conditional information entropy of the data set under the condition given by feature . The calculation formula for information entropy is , where is the number of samples in the data set , is the number of samples in the data set that belong to class . By calculating the information gain, the importance of each feature for classification is determined. Features with large information gain are assigned higher weights, and features with small information gain are assigned lower weights. Then, according to the feature weights and the distribution of the data, the fault decision tree is constructed recursively.

[0045] Perform path matching processing on real-time test data based on the fault decision tree, and output the fault type and confidence level. When real-time test data is input, it is processed according to the standardization method used when constructing the fault decision tree to obtain standardized real-time test data. Starting from the root node of the fault decision tree, match downward along the branches of the decision tree according to the characteristic values of the real-time test data. For example, if the root node is divided according to the communication response delay characteristic, when the standardized value of the communication response delay of the real-time test data is greater than a certain threshold, enter a branch of the decision tree; otherwise, enter another branch. And so on until reaching the leaf node. The leaf nodes correspond to different fault types, and at the same time, according to factors such as the proportion of the sample quantity of the leaf node in the training data, calculate the confidence level of the fault type. For example, if the fault type corresponding to a certain leaf node appears more frequently in the training data, then the output confidence level is higher; conversely, the confidence level is lower. In this way, the judgment of the ECU fault type and the evaluation of the confidence level are realized.

[0046] Embodiment 4: The anomaly monitoring module performs time-series segmentation processing on the ECU operating state data through the sliding window technique to generate window data segments. The sliding window technique is a commonly used method in time-series data processing, which divides the time-series data according to a fixed window size. The selection of the window size is determined according to the characteristics of the ECU operating state data and the monitoring accuracy requirements. For example, for some operating state data with rapid changes, such as voltage fluctuation data or frequency change data, a smaller window size, such as 10 data points or 20 data points, can be selected to capture data changes more timely; for some data with relatively slow changes, such as temperature change data, a larger window size, such as 50 data points or 100 data points, can be selected.

[0047] In practical applications, assume that the ECU operating state data is , and the window size is . First, starting from the starting position of the data, select the first data points as the first window data segment; then slide the window one data point to the right and select as the second window data segment; and so on until sliding to the end of the data to obtain a series of window data segments. In this way, the continuous time-series data is divided into multiple window data segments of fixed length, which is convenient for subsequent analysis and processing.

[0048] Use the isolation forest algorithm to calculate the anomaly score of the window data segment to obtain the anomaly probability of each data segment. The isolation forest algorithm is an anomaly detection algorithm based on the isolation idea, which isolates anomaly points by constructing isolation trees. The anomaly score calculation formula of the isolation forest algorithm is Among them, is the anomaly score of the data point , and this score reflects the anomaly degree of the data point . The higher the score, the more likely the data point is an outlier; is the expected path length of the data point in the isolation tree. The smaller the expected path length, the easier it is for the data point to be isolated, and the more likely it is an outlier; is the normalization factor, which is a constant calculated according to the sample size . It is used to normalize the expected path length so that the anomaly scores under different sample sizes are comparable. The calculation formula of is , where is the harmonic number, ;

[0049] Classify the anomaly probability based on the dynamic threshold segmentation method and trigger the hierarchical alarm instruction. The dynamic threshold segmentation method is a method that dynamically adjusts the threshold according to the data distribution and historical data. First, calculate the anomaly score values of different quantiles based on the anomaly scores of a large number of historical window data segments. For example, calculate the 25th quantile , the 50th quantile and the 75th quantile . Then, determine different threshold intervals according to these quantiles. When the anomaly probability of the window data segment is greater than a certain threshold, trigger the corresponding level of alarm instruction. For example, when the anomaly probability is greater than , trigger the high-level alarm instruction, indicating that there is a serious anomaly in the ECU operating state; when the anomaly probability is greater than and less than or equal to , trigger the medium-level alarm instruction, indicating that there is a certain degree of anomaly in the ECU operating state; when the anomaly probability is greater than and less than or equal to , trigger the low-level alarm instruction, indicating that there may be a slight anomaly in the ECU operating state. Through this hierarchical alarm mechanism, operators can take corresponding measures in a timely manner according to the alarm level to ensure the normal operation of the ECU.

[0050] Example 5: The feedback optimization module continuously optimizes the system through the incremental learning algorithm and the Bayesian optimization algorithm to enhance the accuracy and reliability of the test diagnosis.

[0051] When performing feature correlation processing on historical diagnostic results and real-time anomaly alerts, the incremental learning algorithm plays an important role. The system collects a large number of past diagnostic results, which record the fault types, fault characteristics, and corresponding test data of different ECUs under various test scenarios. At the same time, real-time anomaly alerts reflect the abnormal conditions during the current operation of the ECU. The incremental learning algorithm analyzes these historical diagnostic results and real-time anomaly alerts to extract key feature information. For example, for a specific model of ECU, it is found in the historical diagnostic results that when the communication response delay exceeds a certain threshold and the power consumption data also shows abnormal fluctuations, there is a high probability of a communication module failure; if similar abnormal changes in communication response delay and power consumption data appear in the real-time anomaly alerts, the incremental learning algorithm will correlate these features. By continuously accumulating and analyzing new historical diagnostic results and real-time anomaly alerts, the algorithm gradually generates an optimized weight matrix. Each element in this matrix represents the correlation strength and importance between different features. For example, for different features such as communication response delay, signal amplitude error, and power consumption data, their roles in determining whether the ECU has a fault and the type of fault are different, and the optimized weight matrix will clearly reflect these differences.

[0052] The Bayesian optimization algorithm is used to perform posterior probability update processing on the fault determination threshold of the test diagnosis module to generate dynamic adjustment parameters. The Bayesian optimization algorithm is based on Bayes' theorem and continuously updates the estimation of unknown parameters based on the known prior information and observed data. In this system, the fault determination threshold is a key indicator for the test diagnosis module to judge whether the ECU has a fault. For example, when judging whether the communication response delay of the ECU is normal, a threshold range is set. If the actual communication response delay exceeds this range, it is considered that there may be a fault. The Bayesian optimization algorithm adjusts these thresholds according to the information provided by historical diagnostic results and real-time anomaly alerts. It first determines an initial fault determination threshold as the prior probability based on the existing data and experience. As new test data and anomaly alerts appear, the algorithm uses Bayes' formula to update the prior probability by combining this new information to obtain the posterior probability. By analyzing the posterior probability, it determines how to adjust the fault determination threshold, thereby generating dynamic adjustment parameters. Suppose the initial fault determination threshold for the communication response delay is set at 50 milliseconds, but as the test data accumulates, it is found that when the threshold is adjusted to 40 milliseconds, the communication module fault can be more accurately judged. The Bayesian optimization algorithm will generate corresponding dynamic adjustment parameters to adjust the threshold to 40 milliseconds.

[0053] Embed the dynamic adjustment parameters into the decision-making process of the test diagnosis module to achieve closed-loop policy iteration. When the dynamic adjustment parameters are generated, the system will pass these parameters to the test diagnosis module. After receiving the parameters, the test diagnosis module will process the real-time test data according to the new fault determination threshold. For example, when performing ECU function tests, the test diagnosis module will analyze and judge the real-time collected test data based on the newly adjusted communication response delay threshold, signal amplitude error threshold, power consumption threshold, etc. If it is found that the real-time test data exceeds the new threshold range, it will more accurately determine the possible fault types and degrees of the ECU. This closed-loop policy iteration enables the system to continuously optimize its test diagnosis capabilities according to the actual test results and abnormal situations, improving the accuracy and timeliness of ECU fault judgment, and thus providing more reliable support for the development, production, and maintenance of the ECU.

[0054] Embodiment 6: As an important interaction interface between the system and the operator, the visualization module presents the key information during the system operation in an intuitive and understandable way through data mapping algorithms and graphics rendering technologies, facilitating the operator to comprehensively understand the test and monitoring situation of the ECU.

[0055] The visualization module first converts the dynamic power supply parameters of the power intelligent regulation module, the burn-in integrity report of the burn-in verification module, the diagnostic result set of the test diagnosis module, and the real-time abnormal alarm of the abnormal monitoring module into visual elements through data mapping algorithms. For the dynamic power supply parameters of the power intelligent regulation module, such as voltage, current, power, etc., the data mapping algorithm will map them into elements such as axis scales, graphic sizes, and color shades in the visual chart according to the value ranges and change trends of these parameters. For example, map the voltage value to the height of the bar in the bar chart, the higher the voltage, the higher the bar; map the change trend of the current to the line trend in the line chart to intuitively show the fluctuation of the current over time. For the burn-in integrity report of the burn-in verification module, if the burn-in data is all correct, it may be mapped to a green tick icon; if there are some data verification failures, it will be marked with a red warning icon, and the positions and quantities of the failed data blocks will be detailedly shown through a chart. The diagnostic result set of the test diagnosis module will be mapped to icons of different colors and shapes according to different fault types and confidence levels, and at the same time, the specific fault information, such as fault type, fault occurrence time, etc., will be shown in combination with a table. The real-time abnormal alarm of the abnormal monitoring module will be mapped to flashing hint boxes of different colors according to the alarm level, with a red flashing box for high-level alarms, a yellow flashing box for medium-level alarms, and a blue flashing box for low-level alarms, so that the operator can quickly identify the severity of the abnormality.

[0056] The visualization elements are laid out and rendered by using a graphics rendering engine in combination with a preset visualization template to generate a multi-dimensional visualization interface. The graphics rendering engine is the core technology for realizing visualization, which can draw and display the visualization elements generated by data mapping according to certain rules. The preset visualization template stipulates the overall layout and style of the visualization interface. For example, in the overall layout, the dynamic power parameters may be displayed in the upper left corner of the interface and presented in the form of bar charts and line charts; the burn-in integrity report is displayed in the upper right corner, and the burn-in situation is explained with simple icons and text; the test diagnosis result set and the anomaly monitoring alerts are displayed at the bottom of the interface and presented in the form of a table and a warning box respectively. In terms of style, the visualization template may adopt a simple and clear design style with coordinated color matching, which is convenient for operators to observe and analyze. During the rendering process, the graphics rendering engine will finely draw each element according to the attributes of the visualization elements and the requirements of the visualization template. For example, for a bar chart, clear column boundaries and scale labels will be drawn; for a line chart, the line will be smoothly drawn and data point markers will be added. Through this layout and rendering process, a multi-dimensional visualization interface is finally generated. Operators can use this interface to understand the operating status of the ECU power-on burn-in test monitoring system in real time and comprehensively, discover problems in a timely manner and take corresponding measures, which greatly improves work efficiency and the operability of the system.

[0057] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device.

[0058] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An integrated ECU power-on burning test monitoring system, characterized in that: include: The power intelligent control module is used to track and process the voltage and current data of the ECU during power-on in real time through a dynamic voltage and frequency adjustment algorithm to generate dynamic power parameters; Use temperature adaptive algorithms combined with real-time temperature monitoring data to optimize the cooling strategy of dynamic power supply parameters and generate stable power output; The burning verification module is used to perform segmented hash calculation processing on the burning data stream through a block redundancy verification algorithm to obtain a data verification sequence; And use the incremental verification algorithm combined with the preset benchmark hash value to perform integrity matching on the data verification sequence and generate a burning integrity report; The test diagnosis module is used to perform feature extraction processing on ECU function parameters based on a multi-dimensional test data set to obtain a test feature vector; a decision tree classification model is used to perform fault pattern matching processing on the test feature vector to generate a diagnosis result set; The abnormality monitoring module is used to perform sliding window sampling processing on the ECU operation status data through real-time stream data processing technology to obtain status time series data; and use the isolation forest algorithm to calculate the abnormal probability of the status time series data to generate real-time abnormality alarms; The feedback optimization module is used to perform correlation analysis on the diagnosis result set and the real-time abnormal alarm through the incremental learning algorithm to obtain the optimized feature parameters; and use the Bayesian optimization algorithm combined with the optimized feature parameters to dynamically adjust the threshold of the test diagnosis module to generate a closed-loop test strategy.

2. The integrated ECU power-on burning test monitoring system according to claim 1 is characterized in that: The power intelligent control module includes: The voltage rise slope and current peak value during the ECU power-on phase are dynamically modeled through a dynamic voltage-frequency adjustment algorithm to generate a power control model. Based on the real-time temperature sensor data, the cooling fan speed in the power control model is optimized by gradient descent to obtain the temperature adaptive parameters; The temperature adaptation parameters are compared with the preset power stability threshold to generate cooling strategy instructions.

3. The integrated ECU power-on burning test monitoring system according to claim 2 is characterized in that: The function formula of the dynamic voltage frequency adjustment algorithm is as follows: ， In the formula, is the optimized output voltage, is the voltage regulation weight coefficient, is the power change rate, is the current compensation factor, is the real-time current peak value, is the reference current threshold.

4. The integrated ECU power-on burning test monitoring system according to claim 1 is characterized in that: The burning verification module comprises: The burning data stream is divided into data blocks according to the fixed byte length through the block redundancy check algorithm, and each data block is double hashed to generate a local check code; Use the incremental verification algorithm to compare the local checksum with the cloud benchmark hash library block by block to filter out abnormal data block indexes; The data retransmission protocol is triggered based on the abnormal data block index until all data blocks pass the integrity verification.

5. The integrated ECU power-on burning test monitoring system according to claim 1 is characterized in that: The test diagnosis module comprises: The communication response delay, signal amplitude error and power consumption data of the ECU are standardized through multi-dimensional test data sets to generate a normalized test matrix; Use the decision tree classification model to perform feature weight distribution processing on the normalized test matrix and build a fault decision tree; Perform path matching processing on real-time test data based on the fault decision tree and output the fault type and confidence level.

6. The integrated ECU power-on burning test monitoring system according to claim 1 is characterized in that: The abnormality monitoring module includes: The ECU operation status data is processed in time series segments through the sliding window technology to generate window data segments; The isolation forest algorithm is used to calculate the abnormal score of the window data segment to obtain the abnormal probability of each data segment; The abnormal probability is classified based on the dynamic threshold segmentation method to trigger the graded alarm instructions.

7. The integrated ECU power-on burning test monitoring system according to claim 6 is characterized in that: The anomaly score calculation formula of the isolation forest algorithm is as follows: ， In the formula, For data points The anomaly score, For data points The expected path length in an isolation tree, is the standardization factor, is the sample size.

8. The integrated ECU power-on burning test monitoring system according to claim 1 is characterized in that: The feedback optimization module comprises: The incremental learning algorithm is used to perform feature correlation processing on historical diagnosis results and real-time abnormal alarms to generate an optimized weight matrix; The Bayesian optimization algorithm is used to perform a posteriori probability update on the fault judgment threshold of the test diagnosis module to generate dynamic adjustment parameters; The dynamic adjustment parameters are embedded into the decision-making process of the test diagnosis module to achieve closed-loop strategy iteration.

9. The integrated ECU power-on burning test monitoring system according to claim 1 is characterized in that: Also includes: The visualization module is used to convert the dynamic power parameters of the power intelligent control module, the burning integrity report of the burning verification module, the diagnosis result set of the test diagnosis module, and the real-time abnormal alarm of the abnormal monitoring module into visualization elements through the data mapping algorithm; the visualization elements are laid out and rendered using the graphics rendering engine combined with the preset visualization template to generate a multi-dimensional visualization interface.

10. An integrated ECU power-on burn-in test monitoring method, characterized in that: The following steps are involved: Step 1: Use the dynamic voltage and frequency adjustment algorithm to track and process the voltage and current data during the ECU power-on process in real time to generate dynamic power parameters. Then use the temperature adaptive algorithm combined with real-time temperature monitoring data to optimize the heat dissipation strategy of the dynamic power parameters to generate stable power output. Step 2: Perform segmented hash calculation on the burning data stream through the block redundancy check algorithm to obtain a data check sequence, and then use the incremental verification algorithm combined with the preset reference hash value to perform integrity matching on the data check sequence to generate a burning integrity report; Step 3: Extract features of ECU function parameters based on multi-dimensional test data sets to obtain test feature vectors, and then use the decision tree classification model to perform fault pattern matching on the test feature vectors to generate a diagnosis result set; Step 4: Use real-time stream data processing technology to perform sliding window sampling on the ECU operating status data to obtain status time series data, and then use the isolation forest algorithm to calculate the abnormal probability of the status time series data to generate a real-time abnormal alarm; Step 5: Use the incremental learning algorithm to correlate and analyze the diagnostic result set with the real-time abnormal alarm to obtain the optimized feature parameters. Finally, use the Bayesian optimization algorithm combined with the optimized feature parameters to dynamically adjust the threshold of the test diagnosis module to generate a closed-loop test strategy.

Images