Verification method, system and device for ECU calibration file parameters and medium

Abstract

The invention discloses an ECU calibration file parameter verification method, system and device and a medium, and relates to the field of parameter verification. In the method, key features are extracted from to-be-calibrated parameters of an ECU calibration file, and a structured parameter feature vector is constructed according to the key features; candidate parameters with the similarity with the structural parameter feature vectors higher than a threshold value are screened out from a preset calibration library through a mixed matching algorithm, each candidate parameter is endowed with a confidence score, and a candidate parameter set is constructed based on the candidate parameters; recommending an optimal calibration path according to the candidate parameter set, and calibrating the to-be-calibrated parameters according to the optimal calibration path to obtain calibration parameters; and performing basic rule verification on the calibration parameters by using a rule engine, performing anomaly detection on the calibration parameters by using a machine learning model, and determining whether the calibration parameters meet requirements or not according to a basic rule verification result and an anomaly detection result. By implementing the technical scheme provided by the invention, the parameter verification accuracy and efficiency of the ECU calibration file are improved.

Description

Technical Field

[0001] The present application relates to the technical field of parameter verification, and in particular to a method, system, device and medium for verifying parameters of an ECU calibration file. Background Art

[0002] ECU (Electronic Control Unit) calibration technology plays a vital role in modern industrial automation and automotive electronics. It optimizes equipment performance, improves system efficiency and ensures safe operation by accurately setting control parameters. With the rapid development of electronic control technology, ECU calibration has become the core link for efficient debugging and optimization of complex control systems, greatly promoting the development of smart manufacturing and smart transportation.

[0003] In the prior art, in order to solve the parameter matching and optimization problems in the ECU calibration process, a manual calibration method based on human experience is usually used, relying on the professional knowledge of engineers to complete parameter adjustment. However, the above method has the defect of being difficult to fully consider the coupling relationship between parameters and the dynamic change characteristics, resulting in insufficient calibration accuracy or low efficiency, and cannot meet the needs of modern high-precision calibration tasks. Summary of the invention

[0004] The present application provides a method, system, device and medium for verifying ECU calibration file parameters, which improves the accuracy and efficiency of ECU calibration file parameter verification.

[0005] In a first aspect of the present application, a method for verifying parameters of an ECU calibration file is provided, which is applied to a parameter verification platform, and the method comprises: Extract key features from the parameters to be calibrated in the ECU calibration file, and construct a structured parameter feature vector based on the key features, wherein the key features include parameter name, type, value range, physical meaning, maturity index and task relevance; Using a hybrid matching algorithm, candidate parameters with a similarity higher than a threshold to the structural parameter feature vector are screened out from a preset calibration library, a confidence score is assigned to each candidate parameter, and a candidate parameter set is constructed based on the candidate parameters; recommending an optimal calibration path according to the candidate parameter set, and calibrating the parameters to be calibrated according to the optimal calibration path to obtain calibration parameters; A rule engine is used to perform basic rule verification on the calibration parameters, and a machine learning model is used to perform anomaly detection on the calibration parameters. Whether the calibration parameters meet the requirements is determined based on the basic rule verification results and the anomaly detection results.

[0006] Optionally, the step of using a hybrid matching algorithm to select candidate parameters having a similarity with the structural parameter feature vector higher than a threshold from a preset calibration library includes: Generate a semantic vector based on the text field of the candidate parameter in the preset calibration library, and calculate the cosine similarity between the structural parameter feature vector and the semantic vector to obtain the semantic similarity; A global parameter dependency graph is constructed based on the control logic dependency between the parameters, and the topological correlation between the parameters to be calibrated and the parameters to be selected in the global parameter dependency graph is calculated through a graph neural network; Using a bidirectional LSTM to extract a first time series feature of the parameter to be calibrated and a second time series feature of the parameter to be selected, and calculating a time series matching degree between the first time series feature and the second time series feature through a dynamic time warping algorithm; The semantic similarity, the topological association, and the temporal matching are weightedly summed to obtain similarity values ​​of the candidate parameters, and the candidate parameters whose similarity values ​​are higher than the threshold are taken as candidate parameters.

[0007] Optionally, assigning a confidence score to each of the candidate parameters and constructing a candidate parameter set based on the candidate parameters includes: Linearly mapping the similarity value to obtain a basic confidence, obtaining a maturity index of the candidate parameter, and using a logarithmic change value of the maturity index as a correction value; A confidence score is obtained by weighted summing the basic confidence and the correction value, and candidate parameters having confidence scores greater than a confidence threshold are used to construct a candidate parameter set.

[0008] Optionally, the recommending an optimal calibration path according to the candidate parameter set includes: Based on the dependency relationship of each parameter in the candidate parameter set, a calibration dependency graph including node weights and edge weights is constructed, wherein the nodes represent the parameters to be calibrated, and the edge weights include the coupling strength between parameters and the calibration order constraints; determining an initial calibration sequence according to the calibration dependency graph, identifying whether there is a calibration conflict, and when a calibration conflict is detected, adjusting the initial calibration sequence based on a calibration resource occupancy rate; A genetic algorithm is used to optimize and iterate the adjusted initial calibration sequence to obtain an optimal calibration path.

[0009] Optionally, the using a rule engine to perform basic rule verification on the calibration parameters includes: Parsing the metadata structure of the calibration parameters, performing type matching check, value range boundary check and physical unit conversion consistency check, and generating a first-level check flag; Verify the control logic conflict situation based on the global parameter dependency graph and generate a second level check flag; Monitor and calibrate resource occupancy status, implement storage space pre-allocation verification and communication bandwidth peak pressure test, and generate third-level verification flags; A check flag is generated according to the first-level check flag, the second-level check flag and the third-level check flag.

[0010] Optionally, the using of a machine learning model to perform anomaly detection on the calibration parameters, and determining whether the calibration parameters meet the requirements according to the basic rule verification result and the anomaly detection result includes: Extracting the distribution skewness, time series fluctuation entropy value and cross-parameter covariance matrix eigenvalue of the calibration parameters, and constructing a multidimensional calibration vector containing space-time series joint features; Input the multidimensional check vector into a CNN-BiLSTM hybrid neural network, wherein the CNN convolution layer extracts spatial correlation features between parameters, the BiLSTM layer captures long-term and short-term temporal dependencies, and outputs an abnormal probability value; The check flag is encoded as a feature vector and input into the gradient boosting decision tree model together with the abnormal probability value, and the final detection conclusion is generated through a weighted voting mechanism.

[0011] Optionally, the method further includes: Acquire feedback data of the calibration parameters, the feedback data including a maturity increment value, a calibration evaluation report and a task relevance, and dynamically adjust the maturity index, evaluation level and task relevance index of the corresponding parameters in the preset calibration library through a Bayesian update algorithm based on the feedback data; The knowledge graph supplementary data is generated based on the calibration conflict resolution records, and the newly emerged parameter coupling relationships and resource occupancy patterns are updated to the global parameter dependency graph and calibration resource constraint model.

[0012] In a second aspect of the present application, a verification system for ECU calibration file parameters is provided, characterized in that it includes a feature module, a screening module, a calibration module and a verification module, wherein: A feature module is configured to extract key features from the parameters to be calibrated in the ECU calibration file, and construct a structured parameter feature vector according to the key features, wherein the key features include parameter name, type, value range, physical meaning, maturity index and task relevance; A screening module is configured to screen out candidate parameters having a similarity with the structural parameter feature vector higher than a threshold value from a preset calibration library using a hybrid matching algorithm, assign a confidence score to each of the candidate parameters, and construct a candidate parameter set based on the candidate parameters; A calibration module, configured to recommend an optimal calibration path according to the candidate parameter set, and calibrate the parameters to be calibrated according to the optimal calibration path to obtain calibration parameters; The verification module is configured to use a rule engine to perform basic rule verification on the calibration parameters, use a machine learning model to perform anomaly detection on the calibration parameters, and determine whether the calibration parameters meet the requirements based on the basic rule verification results and the anomaly detection results.

[0013] In the third aspect of the present application, an electronic device is provided, including a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes any one of the methods described above.

[0014] In a fourth aspect of the present application, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores instructions, and when the instructions are executed, any of the methods described above is executed.

[0015] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. Standardize the parameters in the ECU calibration file, including unified naming conventions and unit conversion, to ensure the consistency and comparability of the parameters and provide high-quality input for subsequent matching and verification; use a hybrid matching algorithm, combined with multiple matching methods such as semantic similarity matching, topological association analysis, and timing pattern matching, to screen out the candidate parameter set with the highest similarity to the parameters to be verified from the knowledge base, and assign a confidence score to each candidate parameter to improve the accuracy and reliability of matching; 2. Based on the candidate parameter set, use Petri net modeling to mine expert operation sequences, combine parameter dependency graphs and calibration resource occupancy, optimize and iterate through genetic algorithms, recommend the optimal calibration path, reduce blind attempts, improve calibration efficiency and quality, and quickly reach the ideal calibration state; follow the optimal calibration path to adjust and set ECU parameters to ensure the rationality and effectiveness of parameter configuration, improve ECU performance and control accuracy, and reduce performance problems and failure risks caused by improper parameter settings; 3. Use the rule engine to perform basic rule verification and build a multi-dimensional verification rule library, covering rules such as parameter value range constraints, logical dependencies, and physical unit consistency; use machine learning models to detect anomalies in parameters, extract the timing, topology, and calibration process characteristics of parameters, and build multi-dimensional verification vectors. Through model training and optimization, accurately identify abnormal patterns and potential risks in parameters, further improve the comprehensiveness and accuracy of parameter verification, reduce ECU failure rates caused by parameter anomalies, and improve product quality and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1It is a flow chart of a method for verifying ECU calibration file parameters disclosed in an embodiment of the present application; Figure 2 It is another flow chart of the method for verifying the parameters of the ECU calibration file disclosed in the embodiment of the present application; Figure 3 It is a module schematic diagram of the verification system of the ECU calibration file parameters disclosed in the embodiment of the present application; Figure 4 It is a structural schematic diagram of an electronic device disclosed in an embodiment of the present application.

[0017] Explanation of the reference numerals: 301, feature module; 302, screening module; 303, calibration module; 304, verification module; 401, processor; 402, communication bus; 403, user interface; 404, network interface; 405, memory. DETAILED DESCRIPTION

[0018] In order to enable technicians in this field to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of the embodiments.

[0019] In the description of the embodiments of the present application, words such as "for example" or "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "for example" or "for example" is intended to present related concepts in a specific way.

[0020] In the description of the embodiments of the present application, the meaning of the term "multiple" refers to two or more. For example, multiple systems refer to two or more systems, and multiple screen terminals refer to two or more screen terminals. In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "include", "comprise", "have" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0021] This embodiment discloses a method for verifying parameters of an ECU calibration file. Figure 1 It is a flow chart of the verification method of the ECU calibration file parameters disclosed in the embodiment of the present application, such as Figure 1 As shown, the method comprises the following steps: S101, extracting key features from the parameters to be calibrated in the ECU calibration file, and constructing a structured parameter feature vector according to the key features, wherein the key features include parameter name, type, value range, physical meaning, maturity index, and task relevance; S102, using a hybrid matching algorithm to select candidate parameters whose similarity with the structural parameter feature vector is higher than a threshold from a preset calibration library, assigning a confidence score to each candidate parameter, and constructing a candidate parameter set based on the candidate parameters; S103, recommending an optimal calibration path according to the candidate parameter set, and calibrating the parameters to be calibrated according to the optimal calibration path to obtain calibration parameters; S104. Use a rule engine to perform basic rule verification on the calibration parameters, use a machine learning model to perform anomaly detection on the calibration parameters, and determine whether the calibration parameters meet the requirements based on the basic rule verification results and the anomaly detection results.

[0022] Parameter name: uniquely identifies the parameter, such as "engine fuel injection amount" or "brake pedal force feedback coefficient". Type: the data type of the parameter (such as integer, floating point number, Boolean value, etc.), which determines the subsequent processing logic. Value range: the reasonable value range of the parameter (such as 0-100% duty cycle), which is used to constrain the calibration results. Physical meaning: the physical meaning of the parameter in the actual system (such as "fuel injection amount per unit time"), to ensure the physical rationality of the calibration. Maturity index: reflects the historical calibration reliability of the parameter (such as a score of 0-1, 1 indicates high reliability). Task relevance: the degree of relevance between the parameter and the current calibration task (such as a high weight for direct control parameters and a low weight for indirect influence parameters). Convert the above key features into a numerical or encoded vector form (such as [parameter name hash value, type code, upper and lower limits of the value range, physical meaning vector, maturity index, task relevance]). The vector structure facilitates subsequent algorithm processing (such as similarity calculation, machine learning model input). Semantic similarity: Generate semantic vectors based on text fields (such as parameter names and physical meanings), and calculate the semantic matching degree with parameters in the calibration library through cosine similarity. Topological association: Use the global parameter dependency graph (such as the control logic diagram) to calculate the topological association strength between the parameters to be calibrated and the parameters in the library through the graph neural network. Timing matching: Extract the timing characteristics of the parameters (such as dynamic change patterns), and use the dynamic time warping algorithm to calculate the timing similarity. Weighted sum of semantic similarity, topological association, and timing matching is obtained to obtain a comprehensive similarity value. Parameters with similarity higher than the preset threshold are selected as candidate parameters. Each parameter can be assigned a confidence score, and only parameters with confidence scores higher than the threshold are retained to form a candidate set. A directed weighted graph is constructed with candidate parameters as nodes and the dependencies between parameters (such as coupling strength and calibration order constraints) as edge weights. Generate an initial calibration sequence (such as topological sorting) based on the dependency graph. Detect calibration conflicts (such as resource competition and logical contradictions) and adjust the sequence based on the calibration resource occupancy rate. Use genetic algorithms to iteratively optimize the initial sequence, with the goal of minimizing calibration time, resource consumption or error. Output the optimal calibration path (such as parameter calibration sequence and resource allocation plan). Perform actual calibration on the parameters to be calibrated according to the optimal path to generate calibration parameters. Use the rule engine to perform basic rule verification on the calibration parameters to determine whether the calibration parameters meet the basic rules, and use the machine learning model to perform anomaly detection on the calibration parameters to determine whether the calibration parameters are abnormal. Determine whether the calibration parameters meet the requirements based on the basic rule verification results and anomaly detection results.

[0023] Figure 2 is another flow chart of the method for verifying the parameters of the ECU calibration file disclosed in the embodiment of the present application, such as Figure 2As shown in the figure, the calibration work begins, and the calibration parameters are matched for similarity in order to find historical data or cases for reference; based on the similarity matching, accurate matching is performed through parameter names; the maturity of the parameters is evaluated and the parameters of the corresponding maturity level are matched; other parameter tasks related to the current calibration task are determined; the calibration process of the expert is matched to use the expert experience to guide the current calibration; based on the matching results, the specific parameters of this calibration task are determined; the determined calibration parameters are verified to ensure their accuracy and applicability; the calibration parameters are compared with the rule library to verify whether they comply with the established rules; relevant data are collected during the calibration process to provide a basis for subsequent analysis and evaluation; the calibration results are evaluated to determine whether they have achieved the expected results; based on the calibration results, the parameter maturity, evaluation, task relevance and other indicators and files in the calibration library are updated, and the entire calibration process is recorded; the calibration work is completed and the process ends.

[0024] Optionally, the step of using a hybrid matching algorithm to select candidate parameters having a similarity with the structural parameter feature vector higher than a threshold from a preset calibration library includes: Generate a semantic vector based on the text field of the candidate parameter in the preset calibration library, and calculate the cosine similarity between the structural parameter feature vector and the semantic vector to obtain the semantic similarity; A global parameter dependency graph is constructed based on the control logic dependency between the parameters, and the topological correlation between the parameters to be calibrated and the parameters to be selected in the global parameter dependency graph is calculated through a graph neural network; Using a bidirectional LSTM to extract a first time series feature of the parameter to be calibrated and a second time series feature of the parameter to be selected, and calculating a time series matching degree between the first time series feature and the second time series feature through a dynamic time warping algorithm; The semantic similarity, the topological association, and the temporal matching are weightedly summed to obtain similarity values ​​of the candidate parameters, and the candidate parameters whose similarity values ​​are higher than the threshold are taken as candidate parameters.

[0025] Semantic similarity is used to measure the semantic similarity between the parameters to be calibrated and the parameters in the calibration library. A semantic vector can be generated based on the text fields (such as parameter description, name, etc.) of the parameters to be selected in the preset calibration library, and the cosine similarity between the structured parameter feature vector and the semantic vector is calculated to obtain the semantic similarity. Ensure that the selected candidate parameters have a high semantic relevance to the parameters to be calibrated. For example, parameters with similar names or similar descriptions are more likely to be related in function or application. Topological association is used to analyze the association and dependency between the parameters to be calibrated and the parameters in the calibration library in the system. A global parameter dependency graph can be constructed based on the control logic dependency between the parameters, and the topological association between the parameters to be calibrated and the parameters to be selected in the graph can be calculated through a graph neural network. Identify the logical relationship between the parameters, such as mutual exclusion or symbiosis, to ensure that the selected candidate parameters are coordinated and consistent in the system to avoid conflicts between the parameters. Time series matching is used to evaluate the matching degree between the parameters to be calibrated and the parameters in the calibration library in time series. The bidirectional LSTM (Long Short-Term Memory Network) can be used to extract the first time series feature of the parameter to be calibrated and the second time series feature of the candidate parameter, and the time series matching degree of the two time series features is calculated by the dynamic time warping algorithm. Capturing the temporal variation law and trend of the parameter ensures that the selected candidate parameter is similar to the parameter to be calibrated in dynamic behavior, which is particularly important for time-dependent parameters (such as sensor data, control signals, etc.). Taking into account the semantic similarity, topological association and time series matching, the candidate parameter with the highest similarity to the parameter to be calibrated is selected. The above three similarity indicators can be weighted and summed to obtain the comprehensive similarity value of the candidate parameter. The candidate parameter with a similarity value higher than the set threshold is used as the candidate parameter. Through multi-dimensional similarity evaluation, it is ensured that the candidate parameter has a high matching degree with the parameter to be calibrated in terms of semantics, structure and dynamic behavior, improves the accuracy and reliability of screening, and provides a high-quality candidate parameter set for subsequent calibration path recommendation and parameter calibration.

[0026] By combining semantic similarity matching, topological association analysis and temporal pattern matching, the similarity between the parameters to be calibrated and the parameters in the preset calibration library can be evaluated from multiple perspectives. Compared with single-dimensional matching, this comprehensive matching method can more comprehensively and accurately screen out candidate parameters with high similarity to the target parameters, improving the accuracy and reliability of matching. Semantic vectors are generated based on the text fields of the candidate parameters in the preset calibration library, and the cosine similarity with the structured parameter feature vector is calculated, which realizes the in-depth understanding and matching of the parameter semantic information. This method can effectively handle the semantic differences in parameter descriptions and improve the matching accuracy, especially when facing parameters from different projects or different sources, it can better understand their semantic meaning and find more suitable matches. By constructing a global parameter dependency graph and using a graph neural network to calculate the topological association between the parameters to be calibrated and the candidate parameters, this method fully considers the control logic dependency between the parameters. This helps to ensure the consistency and coordination of the selected candidate parameters in the system, avoid problems caused by incompatibility or conflict between parameters, and improve the stability and reliability of the entire ECU system. Using bidirectional LSTM to extract the time series features of the parameters to be calibrated and the parameters to be selected, and calculating the time series matching degree through the dynamic time warping algorithm, can effectively capture the changing rules and trends of the parameters in the time series. This is especially important for ECU parameters with dynamic characteristics, because it can ensure that the matched parameters are not only similar in static features, but also have high consistency in dynamic behavior, which improves the dynamic adaptability of matching.

[0027] Optionally, assigning a confidence score to each of the candidate parameters and constructing a candidate parameter set based on the candidate parameters includes: Linearly mapping the similarity value to obtain a basic confidence, obtaining a maturity index of the candidate parameter, and using a logarithmic change value of the maturity index as a correction value; A confidence score is obtained by weighted summing the basic confidence and the correction value, and candidate parameters having confidence scores greater than a confidence threshold are used to construct a candidate parameter set.

[0028] The similarity values ​​of the candidate parameters are converted into initial confidence scores. The similarity values ​​can be mapped to the interval [0, 1] through a linear function, and the mapped values ​​are used as the basic confidence. The maturity index of the candidate parameters is used to adjust the basic confidence to reflect the reliability of historical calibration. The maturity index of the candidate parameters can be extracted from the preset calibration library. The maturity index is logarithmically transformed (such as the natural logarithm or the logarithm with a base of 10) to generate a correction value. The logarithmic transformation can amplify the difference in maturity index and enhance the discrimination of low-maturity parameters. The basic confidence and the correction value are integrated to generate the final confidence score. The basic confidence and the correction value can be weighted and summed to obtain a comprehensive confidence score. For example, the confidence score = w1×basic confidence + w2×corrected value, where w1 and w2 are the weights corresponding to the basic confidence and the correction value, respectively. This can balance the contribution of the similarity matching result and the reliability of historical calibration. Candidate parameters with confidence scores higher than the threshold are screened out to form the final set. A confidence threshold (such as 0.6) can be set to retain only candidate parameters with confidence scores higher than the threshold. The selected candidate parameters are sorted by confidence score to construct a candidate parameter set.

[0029] The basic confidence is obtained by linearly mapping the similarity value, and the logarithmic change value of the maturity index of the candidate parameter is introduced as the correction value. This quantitative evaluation method comprehensively considers the similarity and maturity of the parameters, and provides a more comprehensive measurement indicator for the reliability of the candidate parameters. The confidence score not only reflects the similarity between the candidate parameter and the target parameter, but also considers the maturity of the parameter, which helps to more accurately screen out high-quality candidate parameters, reduce misjudgments caused by relying solely on similarity, and improve the accuracy of subsequent calibration and verification. The final confidence score is obtained by weighted summation of the basic confidence and the correction value. This method allows the weights of the two to be adjusted according to specific application scenarios and requirements, and flexibly balances the relative importance of similarity and maturity in confidence assessment, so that the confidence score is more in line with the requirements of the actual calibration task. By setting the confidence threshold, the candidate parameters with scores greater than the threshold are constructed as a candidate parameter set, which can quickly and efficiently screen out candidate parameters with high confidence, reduce the amount of data to be processed later, and improve the efficiency of the entire calibration process.

[0030] Optionally, the recommending an optimal calibration path according to the candidate parameter set includes: Based on the dependency relationship of each parameter in the candidate parameter set, a calibration dependency graph including node weights and edge weights is constructed, wherein the nodes represent the parameters to be calibrated, and the edge weights include the coupling strength between parameters and the calibration order constraints; determining an initial calibration sequence according to the calibration dependency graph, identifying whether there is a calibration conflict, and when a calibration conflict is detected, adjusting the initial calibration sequence based on a calibration resource occupancy rate; A genetic algorithm is used to optimize and iterate the adjusted initial calibration sequence to obtain an optimal calibration path.

[0031] The parameters and their dependencies in the candidate parameter set are converted into a graph structure to provide a basis for subsequent path planning. The nodes in the graph represent the parameters to be calibrated (from the candidate parameter set). The edges represent the dependencies between the parameters, and the edge weights include coupling strength and calibration order constraints. Coupling strength: quantifies the degree of mutual influence between parameters (such as a value between 0 and 1, 1 represents strong coupling). Calibration order constraints: indicate the order of parameter calibration (such as parameter A must be calibrated before parameter B). The coupling strength can be determined by analyzing the correlation between parameters (such as control logic dependency, data interaction frequency). The calibration order constraints can be defined by a rule engine or an expert knowledge base. The calibration dependency graph is processed using a topological sorting algorithm to generate an initial calibration sequence. Topological sorting ensures that the calibration order satisfies the dependencies (such as calibrating the pre-conditioning parameters first). Check whether there are calibration conflicts in the initial sequence, for example, resource competition: multiple parameters require the same calibration resource (such as communication channel, storage space) at the same time. Logical contradiction: the calibration order violates the dependency or physical constraint. When a conflict is detected, adjust the sequence based on the calibration resource occupancy: calculate the resource requirements of each parameter (such as bandwidth, memory). Prioritize calibration of parameters with low resource usage to reduce resource competition. Optimize the initial calibration sequence through genetic algorithms to obtain the optimal calibration path. Encode the calibration sequence as a chromosome (such as [parameter 1, parameter 2, parameter 3, ...]). Define a fitness function to evaluate the quality of the calibration path, for example, calibration time: the total time required to calibrate all parameters in the path. Resource consumption: the occupancy of resources (such as CPU, memory) during the calibration process. Error accumulation: the error accumulation caused by the mutual influence between calibration parameters. Select excellent individuals (calibration paths) according to fitness to enter the next generation. Generate new individuals through crossover operations (such as exchanging the order of some parameters in two paths). Randomly change the order of some parameters in the individual to increase the search diversity. Reach the preset number of iterations or fitness value convergence.

[0032] Based on the dependency relationship of each parameter in the candidate parameter set, a calibration dependency graph containing node weights and edge weights is constructed, where the nodes represent the parameters to be calibrated, and the edge weights contain the coupling strength between parameters and the calibration order constraints. This graph model can intuitively display the association and dependency between parameters, providing a basis for subsequent path planning. By constructing the calibration dependency graph, not only the direct dependency between parameters is considered, but also the importance of parameters and the constraints of calibration order are reflected through weights, making path planning more scientific and reasonable. According to the constructed calibration dependency graph, the initial calibration sequence is determined to provide a starting point for subsequent optimization. The calibration conflicts existing in the initial calibration sequence are identified. When conflicts are detected, the initial calibration sequence is adjusted based on the calibration resource occupancy rate. This method can timely discover and solve the resource conflict problems that may occur during the calibration process, ensuring the smooth progress of the calibration process. By reasonably adjusting the initial calibration sequence, unnecessary calibration steps and resource waste are reduced, and the efficiency of the entire calibration process is improved. The adjusted initial calibration sequence is optimized and iterated using a genetic algorithm. The genetic algorithm has global search and optimization capabilities and can effectively find the optimal solution in a large-scale search space. Through continuous iteration and optimization of the genetic algorithm, the optimal calibration path is finally obtained, which enables the calibration process to achieve the best performance indicators, such as minimizing calibration time, resource consumption or maximizing calibration accuracy. The determination of the optimal calibration path ensures that the order and method of parameter calibration are optimal, thereby improving the calibration quality of the entire ECU system and reducing problems caused by improper calibration order or unreasonable resource allocation.

[0033] Optionally, the using a rule engine to perform basic rule verification on the calibration parameters includes: Parsing the metadata structure of the calibration parameters, performing type matching check, value range boundary check and physical unit conversion consistency check, and generating a first-level check flag; Verify the control logic conflict situation based on the global parameter dependency graph and generate a second level check flag; Monitor and calibrate resource occupancy status, implement storage space pre-allocation verification and communication bandwidth peak pressure test, and generate third-level verification flags; A check flag is generated according to the first-level check flag, the second-level check flag and the third-level check flag.

[0034] Parse the metadata structure of the calibration parameters. Extract key information such as parameter type, value range, physical unit, etc., perform type matching check, value range boundary detection and physical unit conversion consistency check. Check whether the data type of the parameter meets the preset requirements, such as integer, floating point number, Boolean value, etc., to ensure the consistency and correctness of the parameter in data type. Verify whether the value of the parameter is within the specified range, including the minimum and maximum limits, to prevent the parameter from exceeding the reasonable range and causing system abnormality or failure. Based on the international system of units conversion matrix, verify the physical unit of the parameter to ensure that the unit conversion between different parameters is correct and consistent, and avoid calculation errors or logical problems caused by unit mismatch. Based on the results of the above three checks, generate the first-level check flag to indicate whether the parameter has passed the check at the basic attribute level. Then, based on the global parameter dependency map, check whether there is a conflict in the control logic between the parameters. The pre-built global parameter dependency map can be used to analyze the mutual dependence and control logic between the parameters to detect whether there are mutually exclusive conditions, circular dependencies or other logical conflicts. Ensure that the collaborative work of the parameters in the system does not cause abnormal behavior or performance problems due to logical conflicts. According to the detection results of control logic conflicts, the second-level check flag is generated to reflect the compliance of the parameters at the logical level. Then pay attention to the resource occupancy during the calibration process. Check whether the storage space required for the calibration parameters has been reasonably pre-allocated to prevent data loss or system crash caused by insufficient storage space. Monitor and test the use of communication bandwidth during the calibration process to ensure that the communication bandwidth can meet the needs of data transmission during the calibration process and avoid communication delays or data packet loss caused by insufficient bandwidth. Ensure the stability and reliability of the calibration process to prevent calibration failure or system performance degradation caused by insufficient resources. According to the monitoring results of the resource occupancy status, generate the third-level check flag to indicate the rationality of the parameters in terms of resource utilization. Combine the results of the first three levels of verification to generate the final check flag. Integrate and evaluate the first, second and third level check flags, and determine whether the calibration parameters pass the basic rule verification according to the preset rules and logic. Provide a clear result identification for the entire basic rule verification process, indicating whether the calibration parameters meet the basic rule requirements, and provide a basis for subsequent anomaly detection and calibration result evaluation.

[0035] The rule engine can automatically parse the metadata structure of the calibration parameters and perform various verification tasks, such as type matching verification, value range boundary detection, and physical unit conversion consistency verification, which greatly improves the efficiency and speed of verification and reduces the time and workload of manual verification. The rule engine can process a large number of calibration parameters at the same time, quickly generate the first, second, and third level verification flags, and realize efficient batch verification of many parameters. By setting the three-level verification flags, comprehensive verification is performed from the basic attributes, control logic, and resource occupancy of the parameters to ensure the compliance, logical correctness, and rationality of resource use of the parameters, avoiding the problems that may be missed by a single verification method. Finally, a comprehensive verification flag is generated based on the three-level verification flags, which fully reflects the verification results of the calibration parameters in all aspects and provides a comprehensive basis for subsequent decision-making. The rule engine can accurately match the predefined business rules and perform strict verification on the calibration parameters. For example, in the type matching verification, ensure that the type of the parameter is consistent with the expectation; in the value range boundary detection, accurately judge whether the parameter value exceeds the allowable range; in the physical unit conversion consistency verification, verify the correctness of the unit conversion, thereby improving the accuracy of the verification. Verifying control logic conflicts based on the global parameter dependency graph can accurately identify possible logical contradictions between parameters and avoid system failures caused by logical errors.

[0036] Optionally, the using of a machine learning model to perform anomaly detection on the calibration parameters, and determining whether the calibration parameters meet the requirements according to the basic rule verification result and the anomaly detection result includes: Extracting the distribution skewness, time series fluctuation entropy value and cross-parameter covariance matrix eigenvalue of the calibration parameters, and constructing a multidimensional calibration vector containing space-time series joint features; Input the multidimensional check vector into a CNN-BiLSTM hybrid neural network, wherein the CNN convolution layer extracts spatial correlation features between parameters, the BiLSTM layer captures long-term and short-term temporal dependencies, and outputs an abnormal probability value; The check flag is encoded as a feature vector and input into the gradient boosting decision tree model together with the abnormal probability value, and the final detection conclusion is generated through a weighted voting mechanism.

[0037] Extract the distribution skewness, time series fluctuation entropy and cross-parameter covariance matrix eigenvalues ​​of the calibration parameters. These features reflect the statistical characteristics and dynamic behaviors of the parameters from different perspectives. Distribution skewness: measures the asymmetry of data distribution and helps identify the skewness of parameter values. Time series fluctuation entropy: reflects the fluctuation complexity of parameters in time series and captures the dynamic change pattern of parameters. Cross-parameter covariance matrix eigenvalues: reveals the linear correlation between different parameters and helps to discover the coordinated change relationship between parameters. Multi-dimensional check vector construction: integrate the above extracted features to construct a multi-dimensional check vector containing space-time series joint features. This vector comprehensively characterizes the characteristics of the calibration parameters in multiple dimensions and provides rich information for subsequent anomaly detection. The constructed multi-dimensional check vector is input into the CNN-BiLSTM hybrid neural network. Among them, the CNN (convolutional neural network) convolution layer is responsible for extracting the spatial correlation features between parameters and can automatically learn the local correlation between features; the BiLSTM (bidirectional long short-term memory network) layer captures the long-term and short-term time series dependencies and effectively captures the dynamic change trend of parameters in the time dimension. Through the processing of the hybrid neural network, the abnormal probability value is finally output. This probability value reflects the possibility of abnormality in the calibration parameters, providing a quantitative basis for subsequent decision-making. The check flag generated by the rule engine is encoded as a feature vector and input into the gradient boosting decision tree (GBDT) model together with the previously obtained abnormal probability value. The GBDT model can comprehensively consider multiple factors through the integrated learning of multiple decision trees to improve the accuracy and robustness of predictions. Within the GBDT model, the input feature vector and abnormal probability value are comprehensively evaluated through a weighted voting mechanism to finally generate a detection conclusion on whether it meets the requirements. This mechanism fully integrates the results of rule verification and anomaly detection, making the final conclusion more reliable and comprehensive.

[0038] The distribution skewness, time series fluctuation entropy and cross-parameter covariance matrix eigenvalues ​​of the calibration parameters are extracted to construct a multidimensional check vector containing space-time joint features. This multidimensional feature extraction method can fully capture the characteristics of parameters in different aspects and provide richer information for subsequent anomaly detection. Combining the spatial correlation features and time series dependencies of the parameters to form a comprehensive feature vector can more accurately reflect the actual situation of the parameters and improve the accuracy of anomaly detection. The use of CNN (Convolutional Neural Network) convolutional layer can effectively extract the spatial correlation features between parameters, automatically learn local patterns and structures in the data, and help discover the complex relationship between parameters. The BiLSTM layer can capture the long-term and short-term time series dependencies of parameters. For ECU parameters with dynamic characteristics, it can better understand their changing laws in time series and improve the sensitivity to abnormal time series patterns. The CNN-BiLSTM hybrid neural network combines the advantages of both, and can process both spatial features and time series features, providing a more comprehensive feature learning capability, thereby improving the accuracy and robustness of anomaly detection. The checksum flag generated by the rule engine is encoded as a feature vector and input into the gradient boosting decision tree model together with the anomaly probability value, and the final detection conclusion is generated through a weighted voting mechanism. This method combines rule-based verification and data-based anomaly detection results, making full use of the advantages of both.

[0039] Optionally, the method further includes: Acquire feedback data of the calibration parameters, the feedback data including a maturity increment value, a calibration evaluation report and a task relevance, and dynamically adjust the maturity index, evaluation level and task relevance index of the corresponding parameters in the preset calibration library through a Bayesian update algorithm based on the feedback data; The knowledge graph supplementary data is generated based on the calibration conflict resolution records, and the newly emerged parameter coupling relationships and resource occupancy patterns are updated to the global parameter dependency graph and calibration resource constraint model.

[0040] Collect feedback data of calibration parameters to evaluate the calibration effect and update parameter indicators. Maturity increment value: the improvement of parameter maturity after calibration (e.g. 0.1 means maturity improvement of 10%). Calibration evaluation report: contains qualitative evaluation of calibration results (e.g. "successful", "partially successful", "failed") and detailed analysis. Task relevance: score of the degree of relevance between calibration parameters and current tasks (e.g. 1-5 points, 5 points is the highest). Use Bayesian update algorithm to dynamically adjust the indicators of corresponding parameters in the preset calibration library. The initial maturity index, evaluation level and task relevance of the parameters in the preset calibration library are used as prior distribution. Construct likelihood function based on feedback data to reflect the relationship between observed data (feedback) and parameter indicators. For example, the maturity increment value can be modeled as a normal distribution, and the evaluation report can be mapped to classification probability. Use Bayesian formula to calculate posterior distribution, and update the maturity index, evaluation level and task relevance of the parameter based on the posterior distribution. For example, the maturity index is updated to the expected value of the posterior distribution. After each calibration is completed, feedback data is obtained immediately and parameter indicators are updated to ensure the timeliness of the calibration library. Historical update records are retained to avoid data overwriting and support long-term trend analysis. Records of conflict resolution during the calibration process are collected to improve the global parameter dependency graph and calibration resource constraint model. Record content: conflict type (such as resource competition, logical contradiction), solution (such as adjusting the calibration order, increasing resource allocation), parameters involved and their coupling relationship. Convert conflict resolution records into supplementary data for the knowledge graph to enhance the model's understanding of parameter dependency and resource constraints. Identify entities such as parameters, resources, and conflict types in the records. Extract the coupling relationship between parameters (such as "parameter A and parameter B have resource competition"). Extract the relationship between resources and parameters (such as "parameter C needs to occupy communication channel X"). Store entities and relationships in a graph structure to form supplementary data for the knowledge graph. Add the newly emerging parameter coupling relationship to the global parameter dependency graph and update the edge weight. Adjust the direction and strength of the dependency relationship based on the conflict resolution record to improve the accuracy of the graph. Analyze the resource allocation plan in the conflict resolution record and learn the resource occupation pattern of the parameter. Add the newly learned resource constraint rules (such as "parameter D must exclusively occupy memory Y during calibration") to the model.

[0041] By obtaining feedback data of calibration parameters, including maturity increment, calibration evaluation report and task relevance, the Bayesian update algorithm is used to dynamically adjust the maturity index, evaluation level and task relevance index of the corresponding parameters in the preset calibration library. This method can reflect the performance and changes of parameters in actual applications in real time, so that the parameter indicators in the calibration library are always kept up to date and accurate. Dynamically updating parameter indicators helps to improve the accuracy and reliability of subsequent calibration work. For example, adjusting the maturity index according to the actual feedback maturity increment can more accurately evaluate the stability and reliability of the parameters, so as to select more appropriate parameters during the calibration process and reduce calibration errors caused by outdated or inaccurate parameters. New parameter coupling relationships and resource occupation patterns may appear during the calibration process. By generating knowledge graphs based on calibration conflict resolution records to supplement data and updating these newly emerged relationships and patterns to the global parameter dependency graph and calibration resource constraint model, the system can flexibly adapt to these changes, continuously optimize its own knowledge structure and model, and improve its adaptability to complex and changing calibration environments. When encountering new calibration conflicts or resource allocation problems, the updated global parameter dependency graph and resource constraint model can provide the system with more comprehensive and accurate information support, helping the system to find solutions more quickly and effectively, and avoiding decision-making errors caused by insufficient or inaccurate information.

[0042] This embodiment also discloses a verification system for ECU calibration file parameters. Figure 3 Schematic diagram of a module of a verification system for ECU calibration file parameters disclosed in an embodiment of the present application. Figure 3 As shown, the system includes a feature module 301, a screening module 302, a calibration module 303 and a verification module 304, wherein: The feature module 301 is configured to extract key features from the parameters to be calibrated in the ECU calibration file, and construct a structured parameter feature vector according to the key features, wherein the key features include parameter name, type, value range, physical meaning, maturity index and task relevance; A screening module 302 is configured to screen out candidate parameters having a similarity with the structural parameter feature vector higher than a threshold value from a preset calibration library using a hybrid matching algorithm, assign a confidence score to each candidate parameter, and construct a candidate parameter set based on the candidate parameters; A calibration module 303 is configured to recommend an optimal calibration path according to the candidate parameter set, and calibrate the parameters to be calibrated according to the optimal calibration path to obtain calibration parameters; The verification module 304 is configured to use a rule engine to perform basic rule verification on the calibration parameters, use a machine learning model to perform anomaly detection on the calibration parameters, and determine whether the calibration parameters meet the requirements based on the basic rule verification results and the anomaly detection results.

[0043] Optionally, the screening module 302 is configured to: Generate a semantic vector based on the text field of the candidate parameter in the preset calibration library, and calculate the cosine similarity between the structural parameter feature vector and the semantic vector to obtain the semantic similarity; A global parameter dependency graph is constructed based on the control logic dependency between the parameters, and the topological correlation between the parameters to be calibrated and the parameters to be selected in the global parameter dependency graph is calculated through a graph neural network; Using a bidirectional LSTM to extract a first time series feature of the parameter to be calibrated and a second time series feature of the parameter to be selected, and calculating a time series matching degree between the first time series feature and the second time series feature through a dynamic time warping algorithm; The semantic similarity, the topological association, and the temporal matching are weightedly summed to obtain similarity values ​​of the candidate parameters, and the candidate parameters whose similarity values ​​are higher than the threshold are taken as candidate parameters.

[0044] Optionally, the screening module 302 is configured to: Linearly mapping the similarity value to obtain a basic confidence, obtaining a maturity index of the candidate parameter, and using a logarithmic change value of the maturity index as a correction value; A confidence score is obtained by weighted summing the basic confidence and the correction value, and candidate parameters having confidence scores greater than a confidence threshold are used to construct a candidate parameter set.

[0045] Optionally, the calibration module 303 is configured to: Based on the dependency relationship of each parameter in the candidate parameter set, a calibration dependency graph including node weights and edge weights is constructed, wherein the nodes represent the parameters to be calibrated, and the edge weights include the coupling strength between parameters and the calibration order constraints; determining an initial calibration sequence according to the calibration dependency graph, identifying whether there is a calibration conflict, and when a calibration conflict is detected, adjusting the initial calibration sequence based on a calibration resource occupancy rate; A genetic algorithm is used to optimize and iterate the adjusted initial calibration sequence to obtain an optimal calibration path.

[0046] Optionally, the verification module 304 is configured to: Parsing the metadata structure of the calibration parameters, performing type matching check, value range boundary check and physical unit conversion consistency check, and generating a first-level check flag; Verify the control logic conflict situation based on the global parameter dependency graph and generate a second level check flag; Monitor and calibrate resource occupancy status, implement storage space pre-allocation verification and communication bandwidth peak pressure test, and generate third-level verification flags; A check flag is generated according to the first-level check flag, the second-level check flag and the third-level check flag.

[0047] Optionally, the verification module 304 is configured to: Extracting the distribution skewness, time series fluctuation entropy value and cross-parameter covariance matrix eigenvalue of the calibration parameters, and constructing a multidimensional calibration vector containing space-time series joint features; Input the multidimensional check vector into a CNN-BiLSTM hybrid neural network, wherein the CNN convolution layer extracts spatial correlation features between parameters, the BiLSTM layer captures long-term and short-term temporal dependencies, and outputs an abnormal probability value; The check flag is encoded as a feature vector and input into the gradient boosting decision tree model together with the abnormal probability value, and the final detection conclusion is generated through a weighted voting mechanism.

[0048] Optionally, the system further includes an update module configured to: Acquire feedback data of the calibration parameters, the feedback data including a maturity increment value, a calibration evaluation report and a task relevance, and dynamically adjust the maturity index, evaluation level and task relevance index of the corresponding parameters in the preset calibration library through a Bayesian update algorithm based on the feedback data; The knowledge graph supplementary data is generated based on the calibration conflict resolution records, and the newly emerged parameter coupling relationships and resource occupancy patterns are updated to the global parameter dependency graph and calibration resource constraint model.

[0049] It should be noted that: when the device provided in the above embodiment realizes its function, only the division of the above functional modules is used as an example. In actual application, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. In addition, the device and method embodiments provided in the above embodiment belong to the same concept, and the specific implementation process is detailed in the method embodiment, which will not be repeated here.

[0050] This embodiment also discloses an electronic device, referring to Figure 4 The electronic device may include: at least one processor 401 , at least one communication bus 402 , a user interface 403 , a network interface 404 , and at least one memory 405 .

[0051] The communication bus 402 is used to realize the connection and communication between these components.

[0052] The user interface 403 may include a display screen (Display) and a camera (Camera). Optionally, the user interface 403 may also include a standard wired interface and a wireless interface.

[0053] The network interface 404 may optionally include a standard wired interface or a wireless interface (such as a WI-FI interface).

[0054] Among them, the processor 401 may include one or more processing cores. The processor 401 uses various interfaces and lines to connect various parts in the entire server, and executes various functions of the server and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 405, and calling data stored in the memory 405. Optionally, the processor 401 can be implemented in at least one hardware form of digital signal processing (Digital Signal Processing, DSP), field programmable gate array (Field-Programmable Gate Array, FPGA), and programmable logic array (Programmable Logic Array, PLA). The processor 401 can integrate one or a combination of a central processing unit (Central Processing Unit, CPU), a graphics processing unit (Graphics Processing Unit, GPU) and a modem. Among them, the CPU mainly processes the operating system, user interface and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 401, and it can be implemented separately through a chip.

[0055] Among them, the memory 405 may include a random access memory (Random Access Memory, RAM) and may also include a read-only memory (Read-Only Memory). Optionally, the memory 405 includes a non-transitory computer-readable storage medium. The memory 405 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 405 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the above-mentioned various method embodiments, etc.; the data storage area may store data involved in the above-mentioned various method embodiments, etc. The memory 405 may also be optionally at least one storage device located away from the aforementioned processor 401. As Figure 4As shown, the memory 405 as a computer storage medium may include an operating system, a network communication module, a user interface module, and an application program of a method for verifying parameters of an ECU calibration file.

[0056] exist Figure 4 In the electronic device shown, the user interface 403 is mainly used to provide an input interface for the user and obtain data input by the user; and the processor 401 can be used to call the application program of the verification method of the ECU calibration file parameters stored in the memory 405. When executed by one or more processors 401, the electronic device executes one or more methods in the above-mentioned embodiments.

[0057] It should be noted that, for the aforementioned method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily required for the present application.

[0058] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0059] In the several embodiments provided in this application, it should be understood that the disclosed devices can be implemented in other ways. For example, the device embodiments described above are only schematic, such as the division of units, which is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some service interfaces, and the indirect coupling or communication connection of devices or units can be electrical or other forms.

[0060] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0061] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0062] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a memory 405, including a number of instructions for a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the various embodiments of the present application. The aforementioned memory 405 includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk.

[0063] The above is only an exemplary embodiment of the present disclosure and cannot be used to limit the scope of the present disclosure. That is, any equivalent changes and modifications made according to the teachings of the present disclosure are still within the scope of the present disclosure. After considering the disclosure of the specification, those skilled in the art will easily think of other embodiments of the present disclosure. This application is intended to cover any modification, use or adaptation of the present disclosure, which follows the general principles of the present disclosure and includes common knowledge or customary technical means in the technical field that are not recorded in the present disclosure. The description and examples are only regarded as exemplary, and the scope and spirit of the present disclosure are defined by the claims.

Claims

1. A method for verifying ECU calibration file parameters, characterized in that: Applied to a parameter verification platform, the method comprises: Extract key features from the parameters to be calibrated in the ECU calibration file, and construct a structured parameter feature vector based on the key features, wherein the key features include parameter name, type, value range, physical meaning, maturity index and task relevance; A hybrid matching algorithm is used to screen out candidate parameters whose similarity with the structural parameter feature vector is higher than a threshold from a preset calibration library, a confidence score is assigned to each candidate parameter, and a candidate parameter set is constructed based on the candidate parameters, wherein the hybrid matching algorithm includes a cosine similarity algorithm, a graph neural network algorithm, and a dynamic time warping algorithm; recommending an optimal calibration path according to the candidate parameter set, and calibrating the parameters to be calibrated according to the optimal calibration path to obtain calibration parameters; A rule engine is used to perform basic rule verification on the calibration parameters, and a machine learning model is used to perform anomaly detection on the calibration parameters. Whether the calibration parameters meet the requirements is determined based on the basic rule verification results and the anomaly detection results.

2. The method for verifying ECU calibration file parameters according to claim 1, characterized in that: The method of using a hybrid matching algorithm to select candidate parameters from a preset calibration library whose similarity with the structural parameter feature vector is higher than a threshold value includes: Generate a semantic vector based on the text field of the candidate parameter in the preset calibration library, and calculate the cosine similarity between the structural parameter feature vector and the semantic vector to obtain the semantic similarity; A global parameter dependency graph is constructed based on the control logic dependency between the parameters, and the topological correlation between the parameters to be calibrated and the parameters to be selected in the global parameter dependency graph is calculated through a graph neural network; Using a bidirectional LSTM to extract a first time series feature of the parameter to be calibrated and a second time series feature of the parameter to be selected, and calculating a time series matching degree between the first time series feature and the second time series feature through a dynamic time warping algorithm; The semantic similarity, the topological association, and the temporal matching are weightedly summed to obtain similarity values ​​of the candidate parameters, and the candidate parameters whose similarity values ​​are higher than the threshold are taken as candidate parameters.

3. The method for verifying ECU calibration file parameters according to claim 2, characterized in that: The step of assigning a confidence score to each of the candidate parameters and constructing a candidate parameter set based on the candidate parameters includes: Linearly mapping the similarity value to obtain a basic confidence, obtaining a maturity index of the candidate parameter, and using a logarithmic change value of the maturity index as a correction value; A confidence score is obtained by weighted summing the basic confidence and the correction value, and candidate parameters having confidence scores greater than a confidence threshold are used to construct a candidate parameter set.

4. The method for verifying ECU calibration file parameters according to claim 1, characterized in that: The recommending the optimal calibration path according to the candidate parameter set comprises: Based on the dependency relationship of each parameter in the candidate parameter set, a calibration dependency graph including node weights and edge weights is constructed, wherein the nodes represent the parameters to be calibrated, and the edge weights include the coupling strength between parameters and the calibration order constraints; determining an initial calibration sequence according to the calibration dependency graph, identifying whether there is a calibration conflict, and when a calibration conflict is detected, adjusting the initial calibration sequence based on a calibration resource occupancy rate; A genetic algorithm is used to optimize and iterate the adjusted initial calibration sequence to obtain an optimal calibration path.

5. The method for verifying ECU calibration file parameters according to claim 2, characterized in that: The using a rule engine to perform basic rule verification on the calibration parameters includes: Parsing the metadata structure of the calibration parameters, performing type matching check, value range boundary check and physical unit conversion consistency check, and generating a first-level check flag; Verify the control logic conflict situation based on the global parameter dependency graph and generate a second level check flag; Monitor and calibrate resource occupancy status, implement storage space pre-allocation verification and communication bandwidth peak pressure test, and generate third-level verification flags; A check flag is generated according to the first-level check flag, the second-level check flag and the third-level check flag.

6. The method for verifying ECU calibration file parameters according to claim 5, characterized in that: The using of the machine learning model to perform anomaly detection on the calibration parameters, and determining whether the calibration parameters meet the requirements according to the basic rule verification result and the anomaly detection result includes: Extracting the distribution skewness, time series fluctuation entropy value and cross-parameter covariance matrix eigenvalue of the calibration parameters, and constructing a multidimensional calibration vector containing space-time series joint features; Input the multidimensional check vector into a CNN-BiLSTM hybrid neural network, wherein the CNN convolution layer extracts spatial correlation features between parameters, the BiLSTM layer captures long-term and short-term temporal dependencies, and outputs an abnormal probability value; The check flag is encoded as a feature vector and input into the gradient boosting decision tree model together with the abnormal probability value, and the final detection conclusion is generated through a weighted voting mechanism.

7. The method for verifying ECU calibration file parameters according to claim 2, characterized in that: The method further comprises: Acquire feedback data of the calibration parameters, the feedback data including a maturity increment value, a calibration evaluation report and a task relevance, and dynamically adjust the maturity index, evaluation level and task relevance index of the corresponding parameters in the preset calibration library through a Bayesian update algorithm based on the feedback data; The knowledge graph supplementary data is generated based on the calibration conflict resolution records, and the newly emerged parameter coupling relationships and resource occupancy patterns are updated to the global parameter dependency graph and calibration resource constraint model.

8. A verification system for ECU calibration file parameters, characterized in that: It includes feature module, screening module, calibration module and verification module, among which: A feature module is configured to extract key features from the parameters to be calibrated in the ECU calibration file, and construct a structured parameter feature vector according to the key features, wherein the key features include parameter name, type, value range, physical meaning, maturity index and task relevance; A screening module is configured to screen out candidate parameters whose similarity with the structured parameter feature vector is higher than a threshold from a preset calibration library using a hybrid matching algorithm, assign a confidence score to each of the candidate parameters, and construct a candidate parameter set based on the candidate parameters, wherein the hybrid matching algorithm includes a cosine similarity algorithm, a graph neural network algorithm, and a dynamic time warping algorithm; A calibration module, configured to recommend an optimal calibration path according to the candidate parameter set, and calibrate the parameters to be calibrated according to the optimal calibration path to obtain calibration parameters; The verification module is configured to use a rule engine to perform basic rule verification on the calibration parameters, use a machine learning model to perform anomaly detection on the calibration parameters, and determine whether the calibration parameters meet the requirements based on the basic rule verification results and the anomaly detection results.

9. An electronic device, characterized in that: It includes a processor, a memory, a user interface and a network interface, the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory so that the electronic device executes the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and when the instructions are executed, the method according to any one of claims 1 to 7 is performed.

Images