Functional navigation method, vehicle diagnosis equipment and storage medium

Abstract

The invention relates to the technical field of vehicle diagnosis, and discloses a functional navigation method, vehicle diagnosis equipment and a storage medium, the functional navigation method comprises the following steps: constructing a training data set of a functional navigation model of a vehicle-mounted terminal; performing feature extraction based on the training data set, and determining feature information of the candidate function instruction; constructing a function navigation model based on the training data set and the feature information; and matching an input preset query instruction through the function navigation model, and generating a recommendation list for recommending the function instruction. The functional navigation experience on the vehicle diagnosis equipment is improved, the accuracy of functional navigation is improved, and the working efficiency of a user is further improved.

Description

Technical Field

[0001] This application relates to the technical field of vehicle diagnosis, and particularly relates to a function navigation method, a vehicle diagnosis device, and a storage medium. Background Art

[0002] The functions of a vehicle diagnosis device cover multiple aspects. After the vehicle diagnosis device is connected to related devices or systems, if a user wants to reach a specific diagnosis function required, they need to click through multiple layers of pages, experiencing a cumbersome operation process. This not only wastes time but also is very likely to cause operation errors due to the complexity of the operation, thereby affecting the normal progress of the entire diagnosis process or related services.

[0003] The actual function navigation method of existing vehicle diagnosis devices is to identify the corresponding function through human expression. However, human expression may be inaccurate, resulting in the inability to accurately match the corresponding function. Moreover, the descriptions or names of the same diagnosis function for different vehicle models may be inconsistent, leading to the inability to accurately identify the corresponding function. Summary of the Invention

[0004] In view of this, embodiments of this application provide a function navigation method, a vehicle diagnosis device, and a storage medium, which can effectively solve the problem of inability to accurately identify the corresponding function.

[0005] In a first aspect, embodiments of this application provide a function navigation method, including:

[0006] Construct a training data set for the function navigation model of the in-vehicle terminal;

[0007] Extract features based on the training data set to determine the feature information of candidate function instructions;

[0008] Construct the function navigation model based on the training data set and the feature information;

[0009] Match the input preset query instruction through the function navigation model to generate a recommended list of recommended function instructions.

[0010] In a first possible embodiment of the first aspect, the training data set includes a vehicle model feature fingerprint library and a diagnosis instruction library. The construction of the training data set for the function navigation model of the in-vehicle terminal includes:

[0011] Generate a vehicle model feature vector according to the extracted ECU feature parameters to establish the vehicle model feature fingerprint library;

[0012] Collect the original diagnostic instructions of multiple vehicle models, determine the call frequency of each candidate function instruction based on the call frequency of the original diagnostic instructions, and determine the usage heat index of each candidate function instruction to establish the diagnostic instruction library.

[0013] In the second possible embodiment of the first aspect, the training data set further includes a semantic association graph, the candidate function instructions include a plurality of identification diagnostic instructions, and each identification diagnostic instruction includes function parameters. The steps of constructing the training data set of the function navigation model of the vehicle-mounted terminal further include:

[0014] Encode the name, function parameter list, and operation scenario of each candidate function instruction to obtain the semantic vector of each candidate function instruction;

[0015] Construct the semantic association graph of the candidate function instructions based on the semantic vectors of the candidate function instructions.

[0016] In the third possible embodiment of the first aspect, constructing the semantic association graph of the candidate function instructions based on the semantic vectors of the candidate function instructions includes:

[0017] Take each candidate function instruction as a node of the semantic association graph;

[0018] Take the semantic vector and call frequency of each candidate function instruction, and the vehicle model feature vector to which the candidate function instruction belongs as the node attributes of the semantic association graph;

[0019] Determine the edge weights of the semantic association graph according to the semantic similarity and co-occurrence frequency between the candidate function instructions.

[0020] In the fourth possible embodiment of the first aspect, constructing the function navigation model based on the training data set and the feature information includes:

[0021] Construct an initial classifier and train the initial classifier based on the training data set and the feature information;

[0022] Stop the training of the initial classifier under preset conditions through an early stopping mechanism to obtain the optimal function navigation model parameters and obtain an initial function navigation model;

[0023] Perform online learning optimization on the initial function navigation model through an incremental learning pipeline to obtain the final function navigation model.

[0024] In the fifth possible embodiment of the first aspect, matching the input preset query instruction through the function navigation model to generate a recommended list of recommended function instructions includes:

[0025] Semantically parse the preset query instruction to obtain the semantic feature vector of the preset query instruction;

[0026] Through nearest neighbor search, find multiple nearest neighbors of the semantic feature vector of the preset query instruction in the feature space of the candidate function instructions, and determine the initial recommended function instructions;

[0027] After filtering the initial recommended function instructions by vehicle model and sorting them by popularity, determine the final recommended function instructions.

[0028] In the sixth possible embodiment of the first aspect, the feature information includes the vehicle model matching degree of the candidate function instructions, and the filtering the initial recommended function instructions by vehicle model and sorting them by popularity includes:

[0029] Based on the vehicle model matching degree, filter the recommended function instructions compatible with the target vehicle model;

[0030] Based on the popularity index of the filtered recommended function instructions, sort the recommended function instructions in descending order to obtain the recommended list.

[0031] In the seventh possible embodiment of the first aspect, the function navigation method further includes:

[0032] [[ID=2,2]]Determine a reward signal according to the user's selection result, so as to update the parameters of the function navigation model according to the reward signal.

[0033] In a second aspect, an embodiment of the present application provides a vehicle diagnostic device, which includes a processor and a memory. The memory stores a computer program, and the processor is configured to execute the computer program to implement the above function navigation method.

[0034] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed on a processor, the above function navigation method is implemented.

[0035] The embodiments of the present application have the following beneficial effects:

[0036] A function navigation method according to this embodiment includes: constructing a training data set for the function navigation model of the vehicle-mounted terminal; extracting features based on the training data set to determine the feature information of candidate function instructions; constructing the function navigation model based on the training data set and the feature information; and matching the input preset query instruction through the function navigation model to generate a recommended list of recommended function instructions. Based on the above solution, the present application realizes a more intelligent and flexible user interaction experience. Without precisely describing the complete function names under this vehicle model, only by inputting common preset query instructions can the corresponding specific function instructions of the vehicle model be matched, reducing the usage threshold and enhancing the user experience at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] To more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments. It should be understood that the following drawings only show some embodiments of the present application and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without creative efforts.

[0038] Figure 1 FIG. 1 shows the first flow diagram of the function navigation method according to the embodiment of the present application;

[0039] Figure 2 FIG. 2 shows the second flow diagram of the function navigation method according to the embodiment of the present application;

[0040] Figure 3 FIG. 3 shows the third flow diagram of the function navigation method according to the embodiment of the present application;

[0041] Figure 4 FIG. 4 shows the fourth flow diagram of the function navigation method according to the embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all of them.

[0043] The components of the embodiments of the present application generally described and illustrated in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the present application to be protected, but only represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0044] In the following, the terms "comprising", "having" and their cognates that may be used in various embodiments of the present application are only intended to indicate specific features, numbers, steps, operations, elements, components or combinations of the foregoing items, and should not be construed as first excluding the existence of one or more other features, numbers, steps, operations, elements, components or combinations of the foregoing items or precluding the possibility of adding one or more features, numbers, steps, operations, elements, components or combinations of the foregoing items. In addition, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0045] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the present application belong. The terms (such as those defined in a general-use dictionary) will be interpreted as having the same meaning as the contextual meaning in the relevant technical field and will not be interpreted as having an idealized meaning or an overly formal meaning unless clearly defined in the various embodiments of the present application.

[0046] The following will, with reference to the accompanying drawings, elaborate on some embodiments of the present application. Without conflict, the following embodiments and the features in the embodiments may be combined with each other.

[0047] The functions of a vehicle diagnostic device include: reading and clearing vehicle fault codes, real-time monitoring of the working state and data changes, recording working data for analyzing the cause of faults or optimizing performance. Testing each sensor of the vehicle to determine whether it is working properly. Simulating various driving conditions to check whether the actuators and control systems of the vehicle are normal. Providing corresponding repair suggestions and guidance based on the fault codes and test results. The functions of the vehicle diagnostic device also include other functions such as system upgrade and viewing vehicle information.

[0048] In the process of vehicle repair and diagnosis, using a diagnostic device generally faces a significant efficiency bottleneck: it requires multiple menu jumps from connecting to the diagnostic device to the target function page. In this situation, novices need to memorize complex paths and are prone to getting lost in the menu. Skilled technicians may also waste valuable time due to too many levels (such as in emergency fault troubleshooting), need to retrace the process after misclicking, and the work efficiency is reduced.

[0049] Although large AI models can simplify the operation process of automotive diagnostic devices through natural language interaction, there are still problems such as the ambiguity of user expressions and the differences in terms between vehicle models in practical applications. Furthermore, in order to improve the various defects in the above-mentioned prior art, the embodiments of the present application propose a function navigation method, a vehicle diagnostic device, and a storage medium, which can construct a function navigation model based on the training data set and the feature information of candidate function instructions, match the input preset query instruction through the function navigation model, and obtain the corresponding recommended function instruction. The user can select one or more recommended function instructions to detect the target vehicle. The present application improves the function navigation experience on the vehicle diagnostic device, improves the accuracy of answers, and further improves the work efficiency of users.

[0050] The following will illustrate the function navigation method with some specific embodiments.

[0051] Figure 1 A flowchart of the function navigation method according to an embodiment of the present application is shown. Exemplarily, the function navigation method includes the following steps:

[0052] S110, construct a training data set for the function navigation model of the in-vehicle terminal.

[0053] In one embodiment, the training data set includes a vehicle model feature fingerprint library. The present application generates a vehicle model feature vector based on the extracted ECU (Electronic Control Unit) feature parameters to establish a vehicle model feature fingerprint library.

[0054] In the embodiments of the present application, the ECU feature parameters include power type, ECU version number, bus architecture, and production batch. The vehicle model feature vector can be expressed as: V_car = [power type, ECU version number, bus architecture, production batch]. Among them, the power type determines the functional modules and diagnostic logic differences of the ECU, including fuel vehicles, hybrid vehicles, pure electric vehicles, and fuel cell vehicles, etc. The ECU version number is the unique identifier of the software, hardware, and calibration data of the electronic control unit, and is used to distinguish the ECU designs in different iteration stages. The bus architecture defines the communication method between ECUs, affects the protocol selection and connection logic of diagnostic instruments, including CAN bus (Controller Area Network), LIN bus (Local Interconnect Network), and Ethernet, etc. The production batch is the production group identifier during the manufacturing process of the ECU, and is used to trace raw materials, processes, and quality inspection information.

[0055] In one embodiment, the present application can access the OEM (Original Equipment Manufacturer) technical document library to obtain the ECU characteristic parameters of each vehicle model. Among them, the OEM technical document library is a resource library for storing technical documents related to the original equipment manufacturer.

[0056] In another embodiment, the training data set further includes a diagnostic instruction library. The present application collects the original diagnostic instructions of multiple vehicle models, determines the call frequency of each candidate function instruction based on the call frequency of the diagnostic instructions, and determines the usage heat index of each diagnostic function to establish a diagnostic instruction library.

[0057] In one embodiment, the present application can collect the original diagnostic instructions of multiple vehicle models based on the dynamic acquisition protocol stack. Each original diagnostic instruction includes the ECU characteristic parameters to which it belongs. The dynamic acquisition protocol stack is a software protocol stack for realizing the dynamic acquisition of vehicle diagnostic data. The dynamic acquisition protocol stack mainly includes [SAE J2534-1 standard instruction set, OEM-specific instruction decoder, user operation behavior recorder]. The SAE J2534-1 standard instruction set is used to realize the standardized communication between the diagnostic device and the vehicle ECU and covers the basic diagnostic functions of mainstream vehicle models. The OEM-specific instruction decoder is used to decode the diagnostic instruction set and convert the instruction into an operation that the device can execute. The user operation behavior recorder is used to record and analyze the user operation behavior data.

[0058] Exemplarily, the candidate function instructions of the present application include multiple identification diagnostic instructions. The identification diagnostic instruction is the original diagnostic instruction of the standardized identification of the additional function. When the identification diagnostic instruction is called, different function parameters can be attached for the call. For example, for the steering angle zeroing learning function, the corresponding parameters can be the steering wheel angle, the steering angle zero point, etc., which are not limited here.

[0059] In another embodiment, the present application records the call frequency of each identification diagnostic instruction through the feature data layer, and the present application records the call frequency of the candidate function instruction according to the call frequency of the identification diagnostic instruction.

[0060] Among them, the feature data layer refers to a structured data storage and analysis layer, which is mainly used to record, count and analyze the function call characteristics of the identification diagnostic instruction. The core function of the feature data layer is to convert the identification diagnostic instruction into quantifiable feature data to facilitate subsequent analysis and model training, and mainly includes function call frequency statistics, multi-dimensional feature labels and dynamic behavior modeling.

[0061] Function call frequency statistics refers to the number of calls of each identification diagnostic instruction in different vehicle models and different sessions. Multidimensional feature tags refer to functions (standardized identifiers of identification diagnostic instructions), associated parameters (functional parameters attached when the identification diagnostic instruction is called), timing features (timing relationships such as the call time, interval, and order of the identification diagnostic instruction), and vehicle model association (associated with the vehicle model feature vector, identifying the ECU feature parameters to which the identification diagnostic instruction belongs, including power type, ECU version number, bus architecture, and production batch). This application can determine the ECU feature parameters and vehicle model feature vector to which the candidate function instruction belongs based on the ECU feature parameters applicable to the identification diagnostic instruction. Dynamic behavior modeling is to discover typical diagnostic patterns and identify abnormal calls through the frequency statistics of identification diagnostic instructions.

[0062] Exemplarily, the calculation formula for the usage heat index of the candidate function instruction is: Heat index = ln(average weekly call times) × 10 + emergency degree coefficient. The emergency degree coefficient refers to the emergency degree or impact degree of the function, and the range of the emergency degree coefficient can be [1, 10]. This application sets the corresponding emergency degree coefficient for the candidate function instruction according to the data items involved in each candidate function instruction, Special function > Action test > Data stream.

[0063] Special function refers to advanced functions that require in-depth operations or manufacturer authorization in vehicle diagnosis and are used to perform non-standard tasks, such as coding (modifying ECU parameters, such as lighting coding, vehicle model configuration), matching / adapting (calibrating sensors or actuators), programming (flashing ECU firmware, such as software upgrade, initialization after replacing the control unit), and security access (lifting manufacturer protection to perform sensitive operations). Action test refers to actively controlling vehicle components through function instructions to verify whether their functions are normal, such as activating actuators such as fuel injectors, relays, and solenoid valves, mainly to locate hardware faults and verify the function recovery after maintenance. Data stream refers to the operating parameters read from the ECU, such as engine speed, water temperature, and vehicle speed, etc., mainly for auxiliary diagnosis.

[0064] In one embodiment, the training data set further includes a semantic association graph. This application encodes the name of each candidate function instruction, the list of functional parameters of the identification diagnostic instruction of the candidate function instruction, and the operation scenario to obtain the semantic vector of each candidate function instruction.

[0065] In one implementation manner, this application performs context analysis on the original DTC code (Diagnostic Trouble Code) and execution logs through a Bi-LSTM model to generate a distributed representation of the function name: E_func = Encoder(function name + list of functional parameters + operation scenario), where E_func represents the semantic vector of the candidate function instruction, and Encoder specifically refers to the Bi-LSTM model here.

[0066] In this application, the Bi-LSTM model is used to encode the function name, function parameter list, and operation scenario, so as to obtain a semantic vector. For example, for the diagnostic function "ABS pump reset", its function name is "ABS pump reset", the function parameter list may be empty (because this function does not require additional parameters), and the operation scenario may be "perform a reset operation after a fault occurs in the ABS system".

[0067] Function name encoding: Input the string "ABS pump reset" into the Bi-LSTM model. The Bi-LSTM model processes the string character by character and outputs a vector representation of a fixed dimension. This vector captures the semantic information of the function "ABS pump reset".

[0068] Function parameter list encoding: If the parameter list is empty, it can be directly represented by a zero vector or a specific placeholder vector. If the parameter list is not empty, each parameter can be string-encoded, and these encoded vectors can be aggregated (such as summing, averaging, etc.) to obtain a unified parameter list vector.

[0069] Operation scenario encoding: Similarly, input the string "perform a reset operation after a fault occurs in the ABS (Antilock Brake System) system" into the Bi-LSTM model to obtain a vector representing the operation scenario.

[0070] Merge vectors: Merge the function name vector, function parameter list vector, and operation scenario vector (such as concatenation, weighted summation, etc.) to obtain a final semantic vector E_func.

[0071] For example, after Bi-LSTM encoding, the function name vector of "ABS pump reset" may be [0.1, -0.2, 0.3,...], the operation scenario vector may be [0.2, 0.1, -0.3,...], and if the parameter list is empty, it is replaced by a zero vector [0, 0, 0,...]. Concatenating these vectors gives the final semantic vector E_func = [0.1, -0.2, 0.3, 0.2, 0.1, -0.3, 0, 0, 0,...].

[0072] In another embodiment, this application constructs a semantic association graph of candidate function instructions based on the semantic vectors of each candidate function instruction. As Figure 2 shown, constructing the semantic association graph includes the following steps:

[0073] S111, Take each candidate function instruction as a node of the semantic association graph.

[0074] Exemplarily, the semantic association graph is a relational network graph constructed based on each candidate function instruction, revealing the logical relationship between candidate function instructions with nodes and edges as basic elements, mapping the complex relationships among candidate function instructions in terms of semantics, the vehicle models they belong to, and the call frequencies, and improving the accuracy of function navigation.

[0075] S112, taking the semantic vectors and call frequencies of each candidate function instruction, as well as the vehicle model feature vectors to which the candidate function instructions belong, as the node attributes of the semantic association graph.

[0076] S113, determining the edge weights of the semantic association graph according to the semantic similarity and co-occurrence frequency between each candidate function instruction.

[0077] In one embodiment, the directed edge eij of the semantic association graph represents the association relationship from diagnostic function i to diagnostic function j. The edge weight wij combines the semantic similarity and co-occurrence frequency. The edge weight calculation formula is: w_ij = sim(E_i, E_j)^0.8 + co-occurrence^0.2.

[0078] co-occurrence represents the co-occurrence frequency, that is, the frequency of the joint occurrence of candidate function instruction i and candidate function instruction j. In this application, the co-occurrence frequency of candidate function instruction i and candidate function instruction j within the same time period can be statistically calculated according to the time and frequency of identifying diagnostic instructions. sim(E_i, E_j) represents the semantic similarity between candidate function instruction i and candidate function instruction j. E_i represents the semantic vector of candidate function instruction i, and E_j represents the semantic vector of candidate function instruction j. In this application, the cosine similarity calculation is performed based on the semantic vector of candidate function instruction i and the semantic vector of candidate function instruction j to determine the semantic similarity between candidate function instruction i and candidate function instruction j.

[0079] It can be understood that during the process of training the function navigation model, the information in the semantic association graph is introduced as input into the model training, thereby enhancing the ability of the function navigation model in knowledge representation and reasoning. Knowledge enhancement can help the large model better utilize domain knowledge for learning and improve the generalization ability of the model.

[0080] In one embodiment, the training data set further includes historical diagnostic logs and a semantic generalization result set. The historical diagnostic logs are a set of structured operation data recorded during the vehicle diagnosis process, used to track the diagnosis process, analyze failure modes, and optimize maintenance strategies. The semantic generalization result set is the semantic generalization result of the function names of each candidate function instruction. By specific methods, the function names of candidate function instructions are extended or enhanced, enabling the model to understand a wider range of semantic concepts. For example, through synonym replacement, the original word can be replaced with a semantically similar word to obtain the semantic generalization result.

[0081] In one embodiment, the training data set includes labeled data. This application labels each candidate functional instruction to construct the labeled data. According to the corresponding relationship between the preset query instruction and each candidate functional instruction, this application obtains the candidate functional instruction for labeling the preset query instruction, and then learns the corresponding relationship between the input (preset query instruction) and the output (candidate functional instruction) through a large amount of labeled data for supervised learning.

[0082] S120. Perform feature extraction based on the training data set to determine the feature information of the candidate functional instruction.

[0083] Exemplarily, the feature information of the candidate functional instruction includes the vehicle model matching degree between the candidate functional instruction and the target vehicle model, the historical success rate of the candidate functional instruction, and the semantic similarity between the candidate functional instruction and the preset query instruction.

[0084] This application can match the ECU feature parameters to which the candidate functional instruction belongs with the target vehicle model to obtain the vehicle model matching degree. For example, if the power type, ECU version number, and bus architecture in the ECU feature parameters to which the candidate functional instruction belongs are all the same as those of the target vehicle model, the vehicle model matching rate is 75%.

[0085] The historical success rate is the frequency at which the candidate functional instruction successfully completes the expected operation in historical executions, which can be determined from the number of successful times and the number of invocations in the historical diagnostic log. The semantic similarity between the candidate functional instruction and the preset query instruction can be obtained by calculating the cosine similarity of their semantic vectors.

[0086] S130. Construct a function navigation model based on the training data set and the feature information.

[0087] In one embodiment, as Figure 3 shown, the steps for this application to construct the function navigation model are as follows:

[0088] S131. Construct an initial classifier and train the initial classifier based on the training data set and the feature information.

[0089] Exemplarily, this application combines multiple weak classifiers (decision trees) through the XGBoost algorithm (Extreme Gradient Boosting) to obtain the initial classifier. Each decision tree is trained based on the residuals of the previous tree, and the loss function is continuously iteratively optimized to gradually reduce the residuals. The XGBoost algorithm constructs a strong learner by iteratively training weak classifiers on the training data set to obtain the initial classifier. In each iteration, the XGBoost algorithm adjusts the model parameters according to the performance of the current model on the training data set and the feature information to improve the accuracy of the model. The model parameters include but are not limited to the learning rate, regularization parameters, tree-related parameters, and data sampling parameters, etc.

[0090] S132, stop the training of the initial classifier under preset conditions through an early stopping mechanism to obtain the optimal functional navigation model parameters, and obtain the initial functional navigation model.

[0091] In one embodiment, the early stopping mechanism of the present application uses a dynamic monitoring strategy to randomly select 20% of the training data as the validation set. During the model training process, the training process of passing through all samples in the training data set once is called an epoch. Each epoch will evaluate the validation set and calculate the accuracy rate. The basic condition for the early stopping condition is that the accuracy rate of the validation set increases by less than 0.2% for five consecutive epochs. The enhanced condition is that if the AUC index drops by more than 0.5% at the same time, the training will be terminated immediately. A backtracking mechanism is added to intelligently roll back to the historical optimal model state after the early stopping condition or the enhanced condition is triggered, that is, save the model parameter snapshot when the validation set performs best.

[0092] AUC (Area Under the Curve) is an index used to evaluate the performance of the model, which measures the area under the ROC curve (Receiver Operating Characteristic curve). The ROC curve is a curve plotted with the false positive rate as the horizontal axis and the true positive rate as the vertical axis. The higher the AUC value, the better the classification performance of the model. If while meeting the basic condition, the AUC index drops by more than 0.5%, this indicates that the model not only has a slow improvement in accuracy rate, but also shows an obvious decline in the overall classification performance, further indicating that the model may have been overfitted or fallen into a local optimal solution, and the training should be stopped at this time.

[0093] In an optional embodiment, the present application can verify the trained functional navigation model. For example, when the user inputs a preset query instruction "Execute Steering Angle Learning" for the target vehicle model, the functional navigation model automatically matches to: Steering Angle Sensor Reset corresponding to the target vehicle model, X208A calibration, and steering angle zeroing learning. The present application verifies the accuracy of the functional navigation model according to the matching degree between the input preset query instruction and the matched functions. If the matching degrees all reach the preset matching degree, the initial functional navigation model that meets the target can be verified.

[0094] The matching degree calculation formula is: final_score = α×a + β×b + γ×c + δ×d, where α + β + γ + δ = 1. Here, final_score is the matching degree between the preset query instruction and the matched function, a is the semantic similarity between the input preset query instruction and the matched candidate function instruction, b is the co-occurrence frequency between the input preset query instruction and the matched candidate function instruction, c is the vehicle model matching degree between the input preset query instruction and the matched candidate function instruction, d is the user preference degree for the matched candidate function instruction, α is the semantic similarity weight, β is the co-occurrence frequency weight, γ is the vehicle model matching degree weight, and δ is the user preference degree weight.

[0095] S133, online learning optimization is performed on the initial function navigation model through an incremental learning pipeline to obtain the final function navigation model.

[0096] Exemplarily, the incremental learning pipeline is an implementation method of online learning. Based on the original initial function navigation model, new data is used to fine-tune the initial function navigation model, thereby continuously optimizing the model performance and avoiding the forgetting of old knowledge caused by the introduction of new data.

[0097] In one embodiment, the incremental learning pipeline includes: new data input, feature extraction, similar sample retrieval, and local parameter update. The new data arrives in an incremental form. For new data samples, a feature extraction process is performed to obtain a semantic vector and the vehicle model feature vector to which it belongs. Similar sample retrieval is used to quickly find the set of samples most similar to the new data from the historical dataset. The new samples have similar vehicle model features to the historical database.

[0098] This application adopts an efficient nearest neighbor search algorithm, including index structure optimization, vector database, and rule filtering. Among them, index structure optimization mainly uses k-d tree (K-Dimension Tree, a binary tree for searching in the dimensional space, suitable for low-dimensional feature spaces, quickly splitting the vehicle model feature space) and locality-sensitive hashing (suitable for high-dimensional semantic vectors, performing hash bucketing for approximate nearest neighbor search). The vector space library uses special tools (FAISS, Milvus) to manage historical feature vectors and supports millisecond-level retrieval. Rule filtering includes vehicle model hard matching (if the data source is a pure electric vehicle, directly filter fuel vehicle samples) and protocol version constraint (only retrieve historical data that supports the current bus architecture).

[0099] Finally, based on finding similar samples, this application performs an operation of local parameter update, by fine-tuning or updating some parameters on the existing model parameters to adapt to the characteristics of new data. This process can help improve the adaptability and generalization ability of the model to new data.

[0100] In an optional embodiment, the present application restricts the adjustment of weights important for previous tasks through the Elastic Weight Consolidation (EWC) algorithm, thereby protecting these weights from being overly changed to maintain the model's memory of old tasks and prevent catastrophic forgetting. The Elastic Weight Consolidation algorithm is based on the Fisher information matrix, which measures the importance of each weight in previous tasks. Specifically, EWC slows down the rate of change of each weight by calculating the importance of each weight and imposing a regularization constraint on the weights during the training of new tasks.

[0101] S140. Match the input preset query instruction through the function navigation model to generate a recommended list of recommended function instructions.

[0102] In one embodiment, the preset query instruction is derived from a predefined diagnostic instruction library. As Figure 4 shown, the present application obtains the recommended function instruction based on the function navigation model, including the following steps:

[0103] S141. Semantically parse the preset query instruction to obtain the semantic feature vector of the preset query instruction.

[0104] Exemplarily, the preset query instruction includes the target function name corresponding to the preset query instruction, the function parameter list identifying the diagnostic instruction, and the target operation scenario. Similar to determining the semantic vector of the candidate function instruction, the present application inputs the string of the preset query instruction into the Bi-LSTM model. The Bi-LSTM model processes this string character by character and outputs a vector representation of a fixed dimension to obtain the semantic feature vector of the preset query instruction.

[0105] S142. Find multiple neighbors of the semantic feature vector of the preset query instruction in the feature space of the candidate function instructions through nearest neighbor retrieval to determine the initial recommended function instructions.

[0106] Exemplarily, the present application can perform intelligent matching through the function navigation model, calculate the semantic similarity based on the semantic feature vector of the preset query instruction and the semantic vector of the candidate function instruction, and find the top n candidate diagnostic instructions with the largest semantic similarity as the initial recommended function instructions.

[0107] S143. After filtering the models and sorting the popularity of the initial recommended function instructions, determine the final recommended function instructions and generate a recommended list.

[0108] In one implementation manner, the feature information includes the model matching degree of the candidate function instruction. The present application filters the recommended function instructions compatible with the target model based on the model matching degree, and sorts the recommended function instructions in descending order based on the popularity index of the filtered recommended function instructions to obtain a recommended list.

[0109] Exemplarily, the present application determines the vehicle model matching degree between the initial recommended function instruction and the preset query instruction according to the target ECU characteristic parameters of the preset query instruction and the ECU characteristic parameters of the initial recommended function instruction, and filters and deletes the initial recommended function instructions that do not match the vehicle model of the preset query instruction. For example, a preset vehicle model matching degree can be set, the initial recommended function instructions exceeding the preset vehicle model matching degree are retained, and the initial recommended function instructions not exceeding the preset vehicle model matching degree are filtered and deleted.

[0110] In one embodiment, the present application also determines a reward signal according to the user's selection result to update the parameters of the function navigation model according to the reward signal.

[0111] The present application uses a feedback learning mechanism to collect the user's selection results in the recommendation list to calculate the reward signal: r = 1 - (selection position / list length), and updates the parameters of the function navigation model based on the expression for updating the model parameters in reinforcement learning.

[0112] The expression for updating the model parameters is: where θ , represents the updated model parameters, θ represents the model parameters before update, ε represents the learning rate, the learning rate is used to control the influence degree of new data on the old parameters during the update process, the larger the learning rate, the greater the update amplitude of the model parameters. r represents the reward signal, and the reward signal represents the reward obtained after executing a certain action. represents the gradient of the Q value obtained by executing action a in state s, the Q value represents the long-term reward expected to be obtained by executing a certain action in a given state. The gradient represents the change rate of the Q value with respect to the model parameters, so it indicates how to adjust the model parameters to maximize the Q value. s is the state, representing the current environment or situation, in the present application, the state can be information such as the semantic vector of the preset query instruction, the vehicle model feature vector, etc. A is the action, which represents the execution instruction selected by the model in a given state, in the present application, the action can be the recommended function instruction recommended by the model for the user from the candidate function instructions.

[0113] The present application also provides a vehicle diagnostic device, which can be a vehicle diagnostic device such as an intelligent vehicle diagnostic device, a remote vehicle diagnostic device, and a remote video vehicle diagnostic device, and is not limited here.

[0114] Exemplarily, the vehicle diagnostic device includes a processor and a memory, wherein the memory stores a computer program, and the processor runs the computer program to enable the vehicle diagnostic device to execute the above-mentioned function navigation method.

[0115] Among them, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, including at least one of a central processing unit (CPU), a graphics processing unit (GPU), a network processor (NP), a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, and discrete hardware components. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc., which can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present application.

[0116] The memory can be, but is not limited to, a random access memory (RAM), a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), etc. Among them, the memory is used to store a computer program, and after receiving an execution instruction, the processor can execute the computer program accordingly.

[0117] The present application also provides a computer-readable storage medium for storing the computer program used in the above vehicle diagnostic device. For example, the computer-readable storage medium can include, but is not limited to: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disc that can store program codes.

[0118] In several embodiments provided in this application, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architectures, functions, and operations of devices, methods, and computer program products according to multiple embodiments of this application. In this regard, each block in the flowchart or block diagram can represent a module, a program segment, or a part of code, and the module, program segment, or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in an alternative implementation, the functions marked in the blocks can occur in a different order from that marked in the accompanying drawings. For example, two consecutive blocks can actually be executed substantially in parallel, and they can sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the block diagram and / or flowchart, as well as the combination of blocks in the block diagram and / or flowchart, can be implemented by a dedicated hardware-based system that performs the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0119] In addition, in each embodiment of this application, each functional module or unit can be integrated together to form an independent part, or each module can exist separately, or two or more modules can be integrated to form an independent part.

[0120] If the above functions are implemented in the form of software functional modules and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a smart phone, a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of this application.

[0121] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed in this application can easily think of changes or substitutions, which should all be covered by the protection scope of this application.

Claims

1. A functional navigation method, characterized in that, Including: Constructing a training dataset for the functional navigation model of the vehicle terminal; Performing feature extraction based on the training dataset to determine the feature information of candidate functional instructions; Constructing the functional navigation model based on the training dataset and the feature information; Matching the input preset query instruction through the functional navigation model to generate a recommendation list of recommended functional instructions.

2. The functional navigation method according to claim 1, wherein The training dataset includes a vehicle model feature fingerprint library and a diagnostic instruction library. The constructing of the training dataset for the functional navigation model of the vehicle terminal includes: Generating a vehicle model feature vector according to the extracted ECU feature parameters to establish the vehicle model feature fingerprint library; Collecting the original diagnostic instructions of multiple vehicle models, determining the call frequency of each candidate functional instruction based on the call frequency of the original diagnostic instructions, and determining the usage heat index of each candidate functional instruction to establish the diagnostic instruction library.

3. The functional navigation method according to claim 2, wherein The training dataset further includes a semantic association graph. The candidate functional instructions include multiple identification diagnostic instructions, and each identification diagnostic instruction includes functional parameters. The constructing of the training dataset for the functional navigation model of the vehicle terminal further includes: Encoding the name, functional parameter list, and operation scenario of each candidate functional instruction to obtain the semantic vector of each candidate functional instruction; Constructing the semantic association graph of the candidate functional instructions based on the semantic vectors of the candidate functional instructions.

4. The functional navigation method according to claim 3, wherein The constructing of the semantic association graph of the candidate functional instructions based on the semantic vectors of the candidate functional instructions includes: Taking each candidate functional instruction as a node of the semantic association graph; Taking the semantic vector and call frequency of each candidate functional instruction, and the vehicle model feature vector to which the candidate functional instruction belongs as the node attributes of the semantic association graph; Determining the edge weights of the semantic association graph according to the semantic similarity and co-occurrence frequency between the candidate functional instructions.

5. The functional navigation method according to claim 1, wherein The constructing of the functional navigation model based on the training dataset and the feature information includes: Constructing an initial classifier and training the initial classifier based on the training dataset and the feature information; Stopping the training of the initial classifier under preset conditions through an early stopping mechanism to obtain the optimal functional navigation model parameters and obtain an initial functional navigation model; Performing online learning optimization on the initial functional navigation model through an incremental learning pipeline to obtain the final functional navigation model.

6. The functional navigation method according to claim 1, characterized in that The matching of the input preset query instruction through the functional navigation model to generate a recommendation list of recommended functional instructions includes: Performing semantic parsing on the preset query instruction to obtain the semantic feature vector of the preset query instruction; Finding multiple neighbors of the semantic feature vector of the preset query instruction in the feature space of the candidate functional instructions through nearest neighbor retrieval to determine the initial recommended functional instructions; Determining the final recommended functional instructions after filtering by vehicle model and sorting by heat for the initial recommended functional instructions.

7. The functional navigation method according to claim 6, characterized in that The feature information includes the vehicle model matching degree of the candidate functional instruction. The filtering by vehicle model and sorting by heat for the initial recommended functional instructions includes: Screen the recommended function instructions compatible with the target vehicle model based on the vehicle model matching degree; Based on the popularity index of the screened recommended function instructions, sort the recommended function instructions in descending order to obtain the recommended list.

8. The functional navigation method according to claim 1, characterized in that The method further includes: Determine a reward signal according to the user's selection result to update the parameters of the function navigation model according to the reward signal.

9. A vehicle diagnostic device, characterized in that, The vehicle diagnostic device includes a processor and a memory, the memory stores a computer program, and the processor is configured to execute the computer program to implement the function navigation method according to any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, It stores a computer program, and when the computer program is executed on a processor, it implements the function navigation method according to any one of claims 1-8.

Images