Functional navigation method and vehicle diagnosis equipment

Abstract

The invention relates to the technical field of diagnostic equipment, and discloses a functional navigation method and vehicle diagnostic equipment, and the functional navigation method comprises the following steps: obtaining first vehicle data and real-time diagnostic data flow of a target vehicle, and user behavior data; second vehicle data supported by the diagnosis equipment and diagnosis steps associated with the second vehicle data and the fault information are obtained; classifying and mapping the first vehicle data into the second vehicle data through a semantic mapping engine so as to determine target vehicle data of the target vehicle in the diagnosis equipment; generating an optimized diagnosis path of the target vehicle through a dynamic logic generator; and generating a recommended diagnosis step in combination with vehicle type collaborative filtering, user behavior prediction and expert mode recommendation results. The problem that the actual vehicle data and the diagnostic equipment data cannot be aligned is effectively solved, an optimized diagnostic path and personalized recommended diagnostic steps are provided for the user, the operation steps of the vehicle diagnostic equipment are simplified, and the user experience degree is improved.

Description

Technical Field

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

[0002] Currently, there is a common pain point in the industry that after the user connects the target vehicle to the vehicle diagnosis device, they need to click through multiple layers of pages to jump to the page required by the customer, which is extremely unfriendly to novices or beginners. Based on this pain point, an application using an AI large model to improve the function navigation experience on automotive diagnosis devices has been proposed, but there are still many problems. After all, the information scanned by the vehicle diagnosis device, such as the brand, model, and year of manufacture, may not necessarily align with the data on the diagnosis device. Summary of the Invention

[0003] In view of this, embodiments of this application provide a function navigation method and a vehicle diagnosis device, which can effectively solve the problem that the actual vehicle data cannot be aligned with the data of the diagnosis device.

[0004] In a first aspect, embodiments of this application provide a function navigation method, including: Obtaining first vehicle data, real-time diagnostic data stream, and user behavior data of the target vehicle; Obtaining second vehicle data supported by the diagnosis device, and diagnostic steps associated with each of the second vehicle data and fault information; Classifying and mapping the first vehicle data to the second vehicle data through a semantic mapping engine to determine the target vehicle data of the target vehicle in the diagnosis device; Generating an optimized diagnostic path for the target vehicle through a dynamic logic generator based on the target vehicle data, the real-time diagnostic data stream, and the diagnostic steps; Generating recommended diagnostic steps by combining the recommendation results of model-based collaborative filtering, user behavior prediction, and expert mode.

[0005] In a second possible embodiment of the first aspect, the obtaining first vehicle data, real-time diagnostic data stream, and user behavior data of the target vehicle includes: Collecting and parsing the real-time data stream of the target vehicle through an OBD data collector to obtain the first vehicle data and the real-time diagnostic data stream, where the real-time diagnostic data stream includes fault codes and status information of target components; Recording the user behavior data of the user during the use of the diagnosis device through user behavior log buried points.

[0006] In a second possible embodiment of the first aspect, the fault information includes a fault code and a fault phenomenon. The obtaining of the second vehicle data supported by the diagnostic device, and the diagnostic steps associated with each of the second vehicle data and the fault information includes: Performing document recognition on the input manufacturer's technical bulletin document through a multi-document summarization model to obtain the second vehicle data, and a knowledge triple of the second vehicle data, the fault information, and the diagnostic steps; Performing object detection on the maintenance manual illustrations to identify and label the target components in the maintenance manual illustrations; Extracting the diagnostic reference thresholds of the target components in the maintenance manual text through a regular expression template, and generating a knowledge triple of the second vehicle data, the fault information, and the diagnostic steps, wherein the fault phenomenon of the target vehicle is determined according to the status information of the target components and the diagnostic reference thresholds.

[0007] In a third possible embodiment of the first aspect, in the target vehicle data of the target vehicle includes target brand data, target model code, and target model year data. The classifying and mapping the first vehicle data into the second vehicle data through a semantic mapping engine includes: Calculating the brand association degree between the first brand data in the first vehicle data and multiple second brand data in the second vehicle data; Mapping the first brand data to one of the second brand data according to the brand association degree to obtain the target brand data; Calculating the semantic similarity or visual similarity between the first model data in the first vehicle data and multiple second model data in the second vehicle data; Mapping the first model data to one of the second model data according to the semantic similarity or visual similarity, and outputting the model code corresponding to the second model data to obtain the target model code; Calculating the matching weight between the first model year data in the first vehicle data and the second model year data in the second vehicle data according to the exponential decay function of the time decay model, wherein the decay coefficient of the exponential decay function is determined by fitting the historical model year mapping results; Mapping the first model year data to one of the second model year data according to the matching weight to obtain the target model year data.

[0008] In a fourth possible embodiment of the first aspect, the calculating the brand association degree between the first brand data in the first vehicle data and multiple second brand data in the second vehicle data includes: Construct a brand knowledge graph based on the multi-dimensional relationship types and brand embedding vectors of the first brand data and each of the second brand data; Calculate the brand association degrees of the first brand data and each of the second brand data through a multi-head self-attention mechanism based on the brand embedding vectors of each brand node in the brand knowledge graph.

[0009] In a fifth possible embodiment of the first aspect, calculating the semantic similarity or visual similarity between the first vehicle model data in the first vehicle data and multiple second vehicle model data in the second vehicle data includes: Convert the vehicle model text in the input first vehicle model data into a first semantic vector through a pre-trained language model; Calculate the semantic similarity based on the first semantic vector and the second semantic vectors of the second vehicle model data; Perform multi-scale feature extraction on the first vehicle model image in the first vehicle model data to calculate the visual similarity based on the first vehicle model image after feature extraction and the second vehicle model images of the second vehicle model data.

[0010] In a sixth possible embodiment of the first aspect, generating an optimized diagnostic path for the target vehicle based on the target vehicle data, the real-time diagnostic data stream, and the diagnostic steps through a dynamic logic generator includes: Construct a diagnostic decision tree to select the corresponding diagnostic steps according to the target vehicle data and the fault information; Optimize the path of the diagnostic steps through a heuristic search algorithm to obtain the optimized diagnostic path.

[0011] In a seventh possible embodiment of the first aspect, combining the recommendation results of vehicle model collaborative filtering, user behavior prediction, and expert mode to generate recommended diagnostic steps includes: Extract user behavior features from the user behavior data to construct a user behavior model; Recommend the first initial diagnostic steps corresponding to the second vehicle data with the highest vehicle model matching degree with the target vehicle data according to the vehicle model matching degree between the target vehicle data and other second vehicle data; Generate second initial diagnostic steps consistent with the user behavior data by analyzing the user behavior features through a long short-term memory network; Determine third initial diagnostic steps based on the corresponding relationships between the target vehicle data, fault codes, and diagnostic steps in the expert knowledge graph; Determine at least one of the initial diagnostic steps as the recommended diagnostic steps according to the recommendation weights of the vehicle model collaborative filtering, the user behavior prediction, and the expert mode.

[0012] In the eighth possible embodiment of the first aspect, the function navigation method further includes: Predicting each of the diagnostic steps through a pre-trained language model based on the vehicle model feature vector, fault code feature vector of the target vehicle data, and the user behavior data, and outputting the diagnostic steps with corresponding confidence levels and temperature maps.

[0013] 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.

[0014] The embodiments of the present application have the following beneficial effects: A function navigation method according to an embodiment of the present application includes: obtaining first vehicle data and real-time diagnostic data streams of a target vehicle, as well as user behavior data; obtaining second vehicle data supported by a diagnostic device, and diagnostic steps associated with each of the second vehicle data and fault information; classifying and mapping the first vehicle data into the second vehicle data through a semantic mapping engine to determine target vehicle data of the target vehicle in the diagnostic device; generating an optimized diagnostic path of the target vehicle through a dynamic logic generator based on the target vehicle data, real-time diagnostic data stream, and diagnostic steps; generating recommended diagnostic steps by combining the recommendation results of vehicle model collaborative filtering, user behavior prediction, and expert mode. Based on the above solutions, the present application can map the actual vehicle data of the target vehicle to the vehicle data supported by the diagnostic device, and can provide the user with the target vehicle, optimized diagnostic path, and personalized recommended diagnostic steps, greatly simplifying the operation steps of the vehicle diagnostic device and improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for the embodiments will be briefly introduced below. 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 related drawings can be obtained based on these drawings without creative efforts.

[0016] Figure 1 Shows a first flowchart of the function navigation method according to an embodiment of the present application; Figure 2 Shows a second flowchart of the function navigation method according to an embodiment of the present application; Figure 3 Shows a third flowchart of the function navigation method according to an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments.

[0018] Generally, the components of the embodiments of the present application described and illustrated in the accompanying 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 accompanying drawings is not intended to limit the scope of the present application claimed, but merely represents the selected embodiments of the present application. All other embodiments obtained by those skilled in the art based on the embodiments of the present application without creative efforts belong to the scope of protection of the present application.

[0019] In the following, the terms "including", "having" and their cognates that can be used in various embodiments of the present application are only intended to represent 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 increasing the possibility of 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 cannot be understood as indicating or implying relative importance.

[0020] Unless otherwise defined, all terms (including technical terms and scientific terms) used here have the same meaning as commonly understood by those of ordinary skill in the art to which the various embodiments of the present application belong. 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 being overly formal, unless clearly defined in the various embodiments of the present application.

[0021] The following will describe in detail some embodiments of the present application with reference to the accompanying drawings. Without conflict, the following embodiments and the features in the embodiments can be combined with each other.

[0022] Figure 1 A flowchart of a function navigation method according to an embodiment of the present application is shown. Exemplarily, the function navigation method includes the following steps: S110, obtain first vehicle data, real-time diagnostic data stream and user behavior data of the target vehicle.

[0023] In the embodiments of the present application, the present application collects and analyzes the real-time data stream of the target vehicle through an OBD data collector to obtain the first vehicle data and the real-time diagnostic data stream. OBD (On-Board Diagnostics) is a standardized system used in modern vehicles to monitor and diagnose vehicle performance. Through the OBD interface, various operating parameters and fault information of the vehicle can be obtained in real time. In the present application, the OBD data collector is the core part of the target vehicle data collection. The OBD data collector is connected to the OBD-II interface of the vehicle to extract key first vehicle data and real-time diagnostic data stream from the target vehicle.

[0024] Exemplarily, in a vehicle diagnostic device, the OBD data collector needs to process data from different bus protocols, such as the CAN (Controller Area Network) bus protocol, LIN (Local Interconnect Network) bus protocol, etc. To ensure that the data can be correctly parsed and understood, the OBD data collector is configured with a multi-protocol parsing module, which includes an adapter, a listener, and an adapter, respectively responsible for the extraction, listening, and adaptation of different types of real-time diagnostic data streams.

[0025] In one embodiment, the first vehicle data is the vehicle type information of the target vehicle, including the first brand data, the first vehicle type data, and the first model year data. The first vehicle data usually comes from the VIN (Vehicle Identification Number) of the vehicle. The VIN is the unique identifier of each vehicle, containing information such as the manufacturer code, production location, and model year. The OBD data collector can read the VIN data stored in the ECU (Electronic Control Unit) through the OBD protocol. Use a decoder to parse the brand, vehicle type, and model year fields in the VIN.

[0026] Exemplarily, the real-time diagnostic data stream includes the status data and fault codes of the target vehicle. For example, the status data can be the engine speed, vehicle speed, door status, and light status, etc. The fault code (Diagnostic Trouble Code, abbreviated as DTC) is a standardized code used to describe potential problems or faults in the vehicle system. The fault code is generated by the on-board diagnostic system (OBD) and stored in the electronic control unit of the vehicle. In the present application, by reading these fault codes, corresponding diagnostic steps can be executed to locate the problem and perform precise repairs. By reading the status data, the fault phenomenon of the current vehicle can be judged according to the diagnostic reference threshold.

[0027] In one embodiment, the decoder is used to parse the data stream of the CAN / LIN bus protocol, extract the specific data fields and attribute values of the real-time diagnostic data stream, such as engine speed, vehicle speed, etc. For example, the decoder can be a CAN / bus decoder, which parses 11 / 29-bit identifiers according to the ISO 15765-2 standard. The listener is used to monitor and collect the real-time diagnostic data stream on the LIN bus, including periodic status information, such as door status, light status, etc. For example, the listener can be a LIN bus listener, which polls and collects periodic data in the master-slave mode. The adapter (such as a custom protocol adapter) is used to load the vendor-specific DLL dynamic link library to parse the vendor-specific diagnostic protocol and data format, and extract the corresponding data fields and attribute values of the real-time diagnostic data stream, such as fault codes, the inverter temperature threshold of the Toyota hybrid system, etc.

[0028] In one embodiment, the present application records the user behavior data of the user when using the diagnostic device through user behavior log embedding. The user behavior data refers to the operations and interaction behaviors of the user when using the automotive diagnostic device currently or historically. User behavior log embedding is a technical means of recording user behaviors, system states, or business processes in the vehicle diagnostic device application.

[0029] In one implementation, the user behavior log embedding can be triggered when using the diagnosis, including the user's function call sequence and parameter setting values. The user behavior log embedding can be triggered during interface operations to record the user interface operations in real time at a set frequency, record the position where the user clicks on the screen, and record the time the user stays on a certain interface or function module. The user behavior log embedding can be triggered when the user performs voice command operations to record the voice command records. For example, when the user inputs the voice command "Check the high-pressure fuel pump circuit of a certain brand, a certain model, and a certain year of manufacture", record the current user command text, the historical user command text, and the corresponding operations performed.

[0030] S120, obtain the second vehicle data supported by the diagnostic device, and the diagnostic steps associated with each of the second vehicle data and the fault information.

[0031] In one embodiment, the present application performs document recognition on the input manufacturer's technical bulletin documents through a multi-document summarization model to obtain second vehicle data, as well as knowledge triples of vehicle data, fault information, and diagnostic steps. The present application understands the semantic relationships of the documents through a multi-document summarization model based on BERT (Bidirectional Encoder Representations from Transformers). The multi-document summarization model based on BERT is a text summarization generation tool that combines natural language processing (NLP) and deep learning technologies. It understands and extracts key information from multiple documents through a pre-trained language model (such as BERT) and generates concise and accurate summaries. In the present application, the multi-document summarization model based on BERT is used to extract key information from the manufacturer's technical bulletin documents, determine the second vehicle data supported by the diagnostic device, and generate structured knowledge triples.

[0032] Among them, the manufacturer's technical bulletin documents are technical documents issued by automobile manufacturers to provide detailed guidance on vehicle system failures, maintenance, and repairs to maintenance technicians, 4S stores, and third-party diagnostic devices. These documents are usually published in PDF or HTML format, cover various vehicle systems (such as engines, transmissions, chassis, etc.), and provide solutions to specific problems.

[0033] In one implementation manner, the manufacturer's technical bulletin documents in PDF / HTML format are input into the multi-document summarization model based on BERT, and tools (such as LayoutParser) are used to recognize the document structure, extract fault information, diagnostic steps, and applicable vehicle models, and output structured knowledge triples <applicable vehicle model, fault information, solution>.

[0034] In one embodiment, the present application performs object detection on the maintenance manual illustrations to identify and label the target components in the maintenance manual illustrations. For example, by training the YOLOv5 model to identify target components (such as engines), the positions of these target components can be quickly located and labeled.

[0035] In another embodiment, the present application extracts the diagnostic reference thresholds of the target components in the maintenance manual text through a regular expression template, and generates knowledge triples of second vehicle data, fault information, and diagnostic steps. The regular expression (Regex) template is a tool for matching strings and can efficiently extract information of specific patterns from unstructured text. In the present application, the regular expression template is used to extract the diagnostic steps and diagnostic reference thresholds corresponding to the key vehicle models and fault information from the maintenance manual text.

[0036] In this embodiment, the present application determines the fault phenomenon of the target vehicle based on the status information of the target component and the diagnostic reference threshold. For example, the engine power of the target vehicle is matched according to the power diagnostic reference threshold to determine whether the actual engine power is lower than the standard value and the magnitude of the deviation from the standard value; the engine torque of the target vehicle is matched according to the torque diagnostic reference threshold to determine whether the actual engine torque is lower than the standard value and the magnitude of the deviation from the standard value; the engine pressure of the target vehicle is matched according to the pressure threshold diagnostic reference threshold to determine whether the actual engine pressure is lower than the standard value and the magnitude of the deviation from the standard value. Through the above matching, the fault phenomenon of the engine can be determined.

[0037] In one embodiment, when defining the positive standard expression template to extract the diagnostic steps and diagnostic reference thresholds, the specific corresponding indicators may include various technical parameters, fault codes, and key measurement values in the diagnostic steps mentioned in the maintenance manual. The specific content of these indicators depends on the specific content and format of the maintenance manual, as well as the vehicle model and fault type targeted. In the positive standard expression template, the key measurement values need to be defined according to the text characteristics of the maintenance manual and usually correspond to the following indicator types: Power parameters: such as engine power (unit: kW / PS), torque (unit: N·m), displacement (unit: L); Dimension parameters: wheelbase (mm), vehicle length / width / height (mm); Electrical parameters: rated voltage (V), current (A), sensor range (such as temperature range °C); Diagnostic thresholds: fault code trigger conditions (such as speed threshold rpm, pressure threshold bar); Time parameters: maintenance cycle (km / month), test duration (s).

[0038] In another embodiment, the present application builds a dynamic data lake architecture to provide an efficient, flexible, and scalable data storage and processing platform to support the optimization of the functional navigation experience of the AI large model on automotive diagnostic devices. The dynamic data lake can integrate heterogeneous data from multiple sources, including first vehicle data, real-time diagnostic data streams, second vehicle data supported by the diagnostic device, and diagnostic steps. Through steps such as data cleaning, feature extraction, and knowledge graph construction, it provides high-quality data support for subsequent deep semantic mapping, adaptive navigation engines, and continuous learning systems.

[0039] Specifically, the dynamic data lake architecture includes a three-layer storage structure: the raw data layer, Parquet columnar storage, retaining the original byte stream; the feature data layer, Delta Lake format, storing vectorized features; the semantic data layer, Neo4j graph database, storing knowledge graph relationships. The dynamic data lake architecture can be used to implement a data cleaning pipeline, which includes raw data, timestamp alignment, field desensitization, and outlier removal. The raw data refers to unprocessed data, including diagnostic data streams, technical bulletin documents, maintenance manual illustrations and texts, and user behavior logs. These data need to be preprocessed before entering the dynamic data lake, including timestamp alignment to ensure the temporal consistency of the data, field desensitization to protect user privacy and data security, and outlier removal (such as using the Laida principle) to improve data quality and accuracy. Compared with the collected real-time diagnostic data stream, the raw data is more extensive and diverse, including non-real-time and historical data, while the real-time diagnostic data stream focuses on the data stream generated during the real-time diagnostic process.

[0040] S130, mapping the first vehicle data into the second vehicle data by classification through a semantic mapping engine to determine the target vehicle data of the target vehicle in the diagnostic device.

[0041] In the embodiment of the present application, the first vehicle data includes the actual first brand data, first vehicle model data and first model year data of the target vehicle. Since the description or expression of the second vehicle data in the diagnostic device is different from the actual first vehicle data of the target vehicle, it is easy to have the phenomenon that the first vehicle data may not be aligned with the second vehicle data on the diagnostic device. In view of the above problem, the present application maps the first vehicle data into the second vehicle data by category to solve the problem that the vehicle data cannot be aligned.

[0042] In one embodiment, if Figure 2 As shown, the present application maps the first vehicle data classification to the second vehicle data, including the following steps: S131, calculating the brand association between the first brand data in the first vehicle data and the plurality of second brand data in the second vehicle data.

[0043] In one embodiment, the present application constructs a brand knowledge graph based on the multi-dimensional relationship types and brand embedding vectors of the first brand data and each piece of second brand data. The multi-dimensional relationship types include holding, technology alliance, platform sharing, supplier supply, patent cross-licensing, and market area overlap, etc. The definitions of these relationship types enable the brand knowledge graph to more comprehensively capture the complex associations between brands. Each brand can be used as a node in the brand knowledge graph, and different relationship types are used as edges to construct an initial brand knowledge graph. The present application can learn the low-dimensional embedding representation of nodes from the brand knowledge graph through the GraphSAGE (Graph Sample and Aggregate) algorithm. By sampling and aggregating the information of neighbor nodes, the brand embedding vector of the brand node is generated, and the brand embedding vector retains the topological information and semantic features of the brand in the knowledge graph.

[0044] In another embodiment, the present application also calculates the brand association degree between the first brand data and each piece of second brand data based on the brand embedding vectors of each brand node in the brand knowledge graph through a multi-head self-attention mechanism.

[0045] The attention weight calculation formula in the multi-head self-attention mechanism is as follows: ; respectively represent Query, Key, and Value, represents the vector of the target information that needs to be focused on or retrieved currently, is the vector representing the characteristics of each position (or element), used to respond to the "question" of the Query, is the data vector that actually needs to be extracted or used. In the multi-head self-attention mechanism, are respectively generated as the inputs of multiple heads through multiple independent linear transformations. Each head learns different subspace features, and finally the results of all heads are concatenated and linearly transformed to generate the output. This way enhances the expressive ability of the model and can better capture the complex dependencies between the first brand data and the second brand data.

[0046] In matrix form, these three matrices are obtained by multiplying the brand embedding vector e of the first brand data with the linear transformation matrices , , . ; ; . , , is a linear transformation matrix used to map the brand embedding vectors of the first brand data to different spaces in order to better capture the correlation between the first brand data and the second brand data. represents the dimension of the key. When calculating the attention weights, to prevent the inner product from being too large and causing the Softmax function to saturate, the inner product result is divided by . Softmax is a normalization function used to convert each element in the correlation score matrix into a probability value, such that the sum of the correlation strengths between each first brand data and all second brand data is 1. represents the final attention output, which is obtained by multiplying the normalized correlation score matrix with the value matrix. This output matrix contains the correlation information between the first brand data and the second brand data and is used for subsequent brand correlation calculations.

[0047] The final correlation formula is:

[0048] represents the final brand correlation between the first brand data and the second brand data, represents the number of attention heads. The multi-head attention mechanism uses multiple attention heads to capture multiple correlation patterns between the first brand data and the second brand data. represents the th attention head, and the correlation strength between the first brand data i and the second brand data j under this attention head. This value is obtained through the above attention weight calculation formula. represents the holding relationship weight. If there is a holding relationship between the first brand data and the second brand data, then is equal to 1, otherwise is equal to 0. is used to quantify the impact of this holding relationship on the brand correlation. represents the technology alliance weight. If there is a technology alliance relationship between the first brand data and the second brand data, then is equal to 1, otherwise is equal to 0. is used to quantify the impact of this technology alliance relationship on the brand correlation. and are binary identifiers for the graph correspondence relationship. If there is a holding relationship or a technology alliance relationship between the first brand data and the second brand data, the corresponding relationship identifier is 1, otherwise it is 0.

[0049] It can be understood that by substituting the above indicators into the final correlation formula, the final brand correlation between the first brand data and the second brand data can be obtained.

[0050] S132. Map the first brand data to one of the second brand data according to the brand correlation degree to obtain the target brand data.

[0051] Exemplarily, if the correlation degree between the first brand data and one of the second brand data is the largest, the first brand data can be mapped to one of the second brand data, and the target brand data is the second brand data with the largest correlation degree with the first brand data.

[0052] S133. Calculate the semantic similarity or visual similarity between the first vehicle model data in the first vehicle data and multiple second vehicle model data in the second vehicle data.

[0053] In one embodiment, the present application converts the vehicle model text in the input first vehicle model data into a first semantic vector through a pre-trained language model; calculates the semantic similarity according to the first semantic vector and the second semantic vector of the second vehicle model data.

[0054] In one implementation manner, the present application extracts vehicle model text information through regular anchor point detection. Regular anchor point detection is a rule-based method. Through a predefined vehicle model parsing rule set, specific information (such as model year, drive type, power configuration, etc.) is extracted from the vehicle model description text. The pre-trained language model can be the ELECTRA model. ELECTRA is a pre-trained language model based on the Transformer architecture, which is good at capturing semantic information in text. In the present application, the ELECTRA model is used to generate a first semantic vector for the vehicle model description text, so as to support the fuzzy matching task.

[0055] In another embodiment, the present application performs multi-scale feature extraction on the first vehicle model image in the first vehicle model data to calculate the visual similarity according to the first vehicle model image after feature extraction and the second vehicle model image of the second vehicle model data.

[0056] In this embodiment, the present application uses ResNet-50 to extract the global features of the first vehicle model image. ResNet-50 is an architecture based on a deep convolutional neural network, which is good at extracting high-level semantic features from pictures. The first vehicle model image is used as the input and sent into the ResNet-50 model. The ResNet-50 model extracts the features of the first vehicle model image layer by layer through a series of convolutional layers, pooling layers and residual connections, and finally outputs a first vehicle model vector, representing the global features of the picture. The present application determines the visual similarity by calculating the similarity between the first vehicle model vector of the first vehicle model image and the vehicle model vector of the second vehicle model image.

[0057] S134. Map the first vehicle model data to one of the second vehicle model data according to the semantic similarity or visual similarity, and output the vehicle model code corresponding to the second vehicle model data to obtain the target vehicle model code.

[0058] Exemplarily, if the first vehicle model data has the highest degree of association with one of the second vehicle model data, the first vehicle model data can be mapped to one of the second vehicle model data. Among them, different second vehicle model data correspond to different vehicle model codes, and in this application, the vehicle model code corresponding to the second vehicle model data is used as the target vehicle model code.

[0059] S135. Calculate the matching weight between the first-year model data in the first vehicle data and the second-year model data in the second vehicle data according to the exponential decay function of the time decay model, where the decay coefficient of the exponential decay function is determined by fitting the historical model year mapping results.

[0060] In one embodiment, the formula of the exponential decay function is:

[0061] Among them, represents the matching weight between the first vehicle data and the second vehicle data, , represents the current first-year model data and the second-year model data the time difference between them, in years, is the decay coefficient, determined by fitting the historical model year mapping results, and the typical value of the decay coefficient is 0.2 - 0.5.

[0062] For example, if the diagnostic device does not support the 2002 model year, then when mapping the first-year model data, the 2001 model year is selected, then = 1, if = 0.3, then the matching weight is . It can be understood that this time decay model ensures that the closer years have the highest matching priority, while avoiding completely eliminating old data.

[0063] In one embodiment, this application determines the decay coefficient of the exponential decay function by collecting the historical model year mapping results of each first-year model data and combining the maximum likelihood function. When using the maximum likelihood function to determine the decay coefficient of the exponential decay function , our core idea is to find a value such that the probability (i.e., likelihood) of observing the current data is maximized under the given data conditions. For the exponential decay function , set N observation samples, where each sample corresponds to a time difference and a binary label , represents the time difference of the th observation sample, represents the binary label of the th observation sample.

[0064] This application describes the likelihood of observed data through a probability model based on exponentially decaying weights. Under this probability model, the probability of successful matching can be associated with exponentially decaying weights. For example, assume that the probability of successful matching is proportional to (or related to it after a certain transformation) where represents the matching weight corresponding to the time difference between the data of the first model year and the data of the second model year; and the probability of failed matching is .

[0065] In this embodiment, for N independent and identically distributed samples, the likelihood function is the product of the joint probabilities of all samples:

[0066] When assuming that the probability of successful matching , and the probability of failed matching , then the likelihood function can be expressed as:

[0067] For the convenience of calculation, this application takes the logarithm of the likelihood function to obtain the log-likelihood function :

[0068] The goal of this application is to find the value that maximizes the log-likelihood function , which can be achieved by taking the derivative and setting it to zero. Take the derivative of with respect to :

[0069] Among them, the above equation usually has no analytical solution, and numerical optimization methods (such as Newton-Raphson method, gradient ascent method, etc.) need to be used to solve the numerical solution. It can be understood that the maximum likelihood function provides a framework for estimating model parameters (here the decay coefficient ) under given data. By constructing a likelihood function based on the exponential decay function and maximizing it, the decay coefficient that best fits the observed data can be found, enabling the model to better fit the time decay pattern in the data.

[0070] S136. Map the data of the first model year to one of the data of the second model year according to the matching weight to obtain the target model year data.

[0071] Exemplarily, if the weight of matching between the first model year data and one of the second model year data is the largest, the first model year data can be mapped to one of the second model year data, and the present application uses the second model year data as the target vehicle model code.

[0072] S140, generate an optimized diagnostic path for the target vehicle through a dynamic logic generator based on vehicle mapping information, real-time diagnostic data stream, and fault diagnosis data.

[0073] In one embodiment, the present application constructs a diagnostic decision tree to select corresponding diagnostic steps according to target vehicle data and fault information; the diagnostic decision tree is a hierarchical diagnostic model, which finally locates to specific diagnostic steps through step-by-step screening and refinement to form an initial diagnostic path. Each node contains preconditions and execution actions, and the branches point to the next node or the final conclusion according to the diagnostic results. Among them, the preconditions of the diagnostic decision tree include target vehicle data and fault information, and the execution actions are the set of all diagnostic steps corresponding to the target vehicle data and fault information.

[0074] In another embodiment, the present application optimizes the diagnostic path through a heuristic search algorithm (such as the A* search algorithm) to obtain an optimized diagnostic path. This heuristic search algorithm guides the path optimization process through a specific cost function F(n) = G(n) + 0.7H(n), where G(n) represents the time consumed by the executed steps; H(n) represents the estimated remaining step complexity (heuristic function) from the current node to the target node. F(n) represents the total cost, which is used to measure the total cost of passing through the current node n from the starting node to the target node, and the weight 0.7 means that the influence of the heuristic estimate H(n) is weakened, making the algorithm pay more attention to the time consumed by the actually executed steps G(n).

[0075] It can be understood that the present application compares the F(n) values of different nodes, and the A* algorithm selects the node with the smallest F(n) for expansion, gradually approaching the target node. Among multiple diagnostic steps, the execution order of diagnostic steps can be optimized by adjusting the weight to reduce the overall time consumption. For example, according to the collected fault information and the mapped target vehicle data, an optimized diagnostic path for the fuel system of a certain brand, model, and model year of vehicle is dynamically generated. The optimized diagnostic path may include steps such as reading fault codes, performing fuel pressure tests, checking the high-pressure fuel pump circuit, and refreshing the ECU calibration.

[0076] S150, generate recommended diagnostic steps by combining the recommendation results of model-based collaborative filtering, user behavior prediction, and expert mode.

[0077] In one embodiment, the present application extracts user behavior features from user behavior data to construct a user behavior model. Among them, the user behavior features include average step duration, operation correction rate, brand usage preference, etc. The extracted behavior features can comprehensively reflect the user's operation habits and preferences. These features provide important input information for subsequent recommendations, making the recommendation results more personalized.

[0078] In another embodiment, as Figure 3 shown, the present application generates recommended diagnostic steps based on the recommendation results, including the following steps: S151, according to the vehicle model matching degree between the target vehicle data and other second vehicle data, recommend the first initial diagnostic steps corresponding to the second vehicle data with the highest vehicle model matching degree to the target vehicle data.

[0079] Exemplarily, the present application can calculate the vehicle model similarity according to the feature vectors of the target vehicle data and other second vehicle data through similarity calculation methods (such as cosine similarity, Euclidean distance, etc.). For the target vehicle data of new vehicle models or new users, find several second vehicle data that are most similar to it. Analyze the user behavior corresponding to the second vehicle data (such as common diagnostic steps, etc.) to generate the first initial diagnostic steps.

[0080] It can be understood that collaborative filtering based on vehicle models is an effective method to solve the cold start problem. Its core idea is to use the user behavior of similar vehicle models to infer the diagnostic steps of the target vehicle.

[0081] S152, analyze the user behavior features through a long short-term memory network to generate second initial diagnostic steps that conform to the user behavior data.

[0082] In the embodiment of the present application, the long short-term memory network (LSTM) is a special recurrent neural network (RNN) used to process and predict long-term dependencies in time series data. In the present application, LSTM is used to predict user behavior features and predict the diagnostic steps that the user may perform next according to the user's behavior data.

[0083] In one implementation, use historical user behavior data to train the LSTM model so that it learns to extract patterns from past user behavior. When the user is using the diagnostic device, the system will input the current user's operation sequence into the LSTM model to obtain the possible second initial diagnostic steps for the next step.

[0084] S153, based on the correspondence between the target vehicle data, fault codes and diagnostic steps in the expert knowledge graph, determine the third initial diagnostic steps.

[0085] In one embodiment, the data sources of the expert knowledge graph are the second vehicle data, each vehicle data, and the diagnostic steps corresponding to the fault information. When a diagnostic requirement is input (such as selecting target vehicle data and fault codes), the system will search for relevant nodes and relationships in the knowledge graph and output the third initial diagnostic steps.

[0086] S154. Determine at least one initial diagnostic step as the recommended diagnostic step according to the recommendation weights of vehicle model collaborative filtering, user behavior prediction, and expert mode.

[0087] In this embodiment, the behaviors and requirements of users may vary according to scenarios. For example: when a new user starts using the diagnostic device for the first time, the system needs to quickly adapt to their needs (cold start problem); during the diagnostic process, the user's behavior pattern may change, and the system needs to adjust the recommended content in real time (continuous diagnosis); for complex problems, the system needs to rely on professional knowledge for reasoning and judgment (expert mode). To address these diverse needs, this application combines the three modes through a hybrid recommendation strategy and dynamically adjusts the recommendation logic according to specific scenarios.

[0088] In one embodiment, different weights can be configured for the recommendation results of vehicle model collaborative filtering, user behavior prediction, and expert mode. The weight represents the importance of the recommendation result. The greater the weight, the more likely the recommendation result is used as the main recommended diagnostic step. For example, for cold start - vehicle model - based collaborative filtering, the weight is 70%; for continuous diagnosis - LSTM - based behavior prediction, the weight is 85%; for expert mode - knowledge graph reasoning, the weight is 60%. For example, the recommendation results of vehicle model collaborative filtering, user behavior prediction, and expert mode may preferentially display relevant diagnostic steps or solutions such as "rail pressure sensor calibration".

[0089] Exemplarily, this application predicts each diagnostic step through a pre - trained language model based on the vehicle model feature vector, fault code feature vector, and user behavior data of the target vehicle, and outputs the diagnostic steps with corresponding confidence levels and temperature maps.

[0090] In one embodiment, the pre - trained language model is a fine - tuned model based on BERT (Bidirectional Encoder Representations from Transformers). BERT is a pre - trained language model that is good at capturing semantic information in text. The fine - tuned model based on BERT in this application can combine multi - modal inputs (vehicle model feature vector, fault code feature vector, and user behavior data) to achieve more intelligent recommendations and predictions.

[0091] For example, the input of the fine-tuned model based on BERT is vehicle model feature vectors (brand / model / year code), fault code vectors (fault code + snapshot data, such as engine speed, fuel pressure, etc.), and user operation sequences (click coordinates / parameter setting values). The output layer outputs diagnostic steps with corresponding confidence levels and temperature maps. The temperature map is a visualization tool used to intuitively display the reliability of model predictions.

[0092] The temperature map generation logic is to map the confidence level output by the fine-tuned model based on BERT to the HSV color space. The colors or temperature gradients in the temperature map represent the degree of reliability of model predictions. For example, high temperature (or dark color) indicates high reliability, while low temperature (or light color) indicates low reliability. In this way, users can intuitively understand the degree of reliability of model predictions and make decisions accordingly. The correspondence between the temperature map and prediction reliability can be defined and adjusted through a color mapping table or a temperature gradient mapping table. For example, high confidence (>0.9) is mapped to red (H = 0°, S = 100%, V = 100%); medium confidence (0.7 - 0.9) → yellow (H = 60°, S = 80%), low confidence (<0.7) → blue (H = 240°, V = 70%).

[0093] In another embodiment, the present application generates Top3 recommended diagnostic steps according to the confidence level of the diagnostic steps, and recommends the three most likely Top3 recommended diagnostic steps to the user. This recommendation mechanism can automatically generate a sorted recommendation list based on the user's input and context information, as well as the model's understanding and analysis of the target vehicle data and diagnostic steps. Users can select or further operate according to the options in the recommendation list to improve the diagnostic efficiency and accuracy. The generation of Top3 recommendations is an important function in the adaptive navigation engine, which can provide personalized diagnostic suggestions and support for users.

[0094] The present application also provides a vehicle diagnostic device. Exemplarily, the vehicle diagnostic device includes a processor and a memory. The memory stores a computer program, and the processor runs the computer program to enable the vehicle diagnostic device to execute the above-mentioned functional navigation method.

[0095] 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., and can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present application.

[0096] 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.

[0097] 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.

[0098] In several embodiments provided by 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 structural diagrams in the 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 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 structural diagram and / or flowchart, as well as the combination of blocks in the structural 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.

[0099] In addition, each functional module or unit in various embodiments of this application can be integrated together to form an independent part, or each module can exist alone, or two or more modules can be integrated to form an independent part.

[0100] 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 for causing 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 various embodiments of this application.

[0101] 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 by 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: Obtaining first vehicle data, real-time diagnostic data stream of a target vehicle, and user behavior data; Obtaining second vehicle data supported by a diagnostic device, and diagnostic steps associated with each of the second vehicle data and fault information; Classifying and mapping the first vehicle data to the second vehicle data through a semantic mapping engine to determine target vehicle data of the target vehicle in the diagnostic device; Generating an optimized diagnostic path of the target vehicle through a dynamic logic generator based on the target vehicle data, the real-time diagnostic data stream, and the diagnostic steps; Generating recommended diagnostic steps by combining the recommendation results of vehicle model collaborative filtering, user behavior prediction, and expert mode.

2. The functional navigation method according to claim 1, wherein The obtaining first vehicle data, real-time diagnostic data stream of a target vehicle, and user behavior data includes: Collecting and parsing the real-time data stream of the target vehicle through an OBD data collector to obtain the first vehicle data and the real-time diagnostic data stream, where the real-time diagnostic data stream includes fault codes and status information of target components; Recording the user behavior data of the user during using the diagnostic device through user behavior log embedding.

3. The functional navigation method according to claim 2, wherein The fault information includes fault codes and fault phenomena. The obtaining second vehicle data supported by a diagnostic device, and diagnostic steps associated with each of the second vehicle data and fault information includes: Performing document recognition on the input manufacturer's technical bulletin document through a multi-document summarization model to obtain the second vehicle data, and a knowledge triple of the second vehicle data, the fault information, and the diagnostic steps; Performing object detection on the repair manual illustrations to identify and label the target components in the repair manual illustrations; Extracting diagnostic reference thresholds of the target components in the repair manual text through a regular expression template, and generating a knowledge triple of the second vehicle data, the fault information, and the diagnostic steps, where the fault phenomena of the target vehicle are determined according to the status information of the target components and the diagnostic reference thresholds.

4. The functional navigation method according to claim 1, wherein The target vehicle data of the target vehicle includes target brand data, target vehicle model code, and target model year data. The classifying and mapping the first vehicle data to the second vehicle data through a semantic mapping engine includes: Calculating the brand association degree between the first brand data in the first vehicle data and multiple second brand data in the second vehicle data; Mapping the first brand data to one of the second brand data according to the brand association degree to obtain the target brand data; Calculating the semantic similarity or visual similarity between the first vehicle model data in the first vehicle data and multiple second vehicle model data in the second vehicle data; Mapping the first vehicle model data to one of the second vehicle model data according to the semantic similarity or visual similarity, and outputting the vehicle model code corresponding to the second vehicle model data to obtain the target vehicle model code; Calculate the matching weights of the first model-year data in the first vehicle data and the second model-year data in the second vehicle data according to the exponential decay function of the time decay model, where the decay coefficient of the exponential decay function is determined by fitting the historical model-year mapping results; Map the first model-year data to one of the second model-year data according to the matching weights to obtain the target model-year data.

5. The functional navigation method according to claim 4, wherein The calculating the brand association degrees of the first brand data in the first vehicle data and multiple second brand data in the second vehicle data includes: Construct a brand knowledge graph according to the multi-dimensional relationship types and brand embedding vectors of the first brand data and each of the second brand data; Calculate the brand association degrees of the first brand data and each of the second brand data based on the brand embedding vectors of each brand node in the brand knowledge graph through the multi-head self-attention mechanism.

6. The functional navigation method according to claim 4, wherein The calculating the semantic similarity or visual similarity of the first vehicle model data in the first vehicle data and multiple second vehicle model data in the second vehicle data includes: Convert the vehicle model text in the input first vehicle model data into a first semantic vector through a pre-trained language model; Calculate the semantic similarity according to the first semantic vector and the second semantic vectors of the second vehicle model data; Perform multi-scale feature extraction on the first vehicle model image in the first vehicle model data to calculate the visual similarity according to the first vehicle model image after feature extraction and the second vehicle model image of the second vehicle model data.

7. The functional navigation method according to claim 1, wherein The generating the optimized diagnostic path of the target vehicle through a dynamic logic generator based on the target vehicle data, the real-time diagnostic data stream, and the diagnostic steps includes: Construct a diagnostic decision tree to select the corresponding diagnostic steps according to the target vehicle data and the fault information; Optimize the path of the diagnostic steps through a heuristic search algorithm to obtain the optimized diagnostic path.

8. The functional navigation method according to claim 1, wherein The generating the recommended diagnostic steps by combining the recommendation results of vehicle model collaborative filtering, user behavior prediction, and expert mode includes: Extract user behavior features from the user behavior data to construct a user behavior model; Recommend the first initial diagnostic steps corresponding to the second vehicle data with the highest vehicle model matching degree with the target vehicle data according to the vehicle model matching degree between the target vehicle data and other second vehicle data; Analyze the user behavior features through a long short-term memory network to generate second initial diagnostic steps consistent with the user behavior data; Determine the third initial diagnostic steps based on the corresponding relationship between the target vehicle data, fault codes, and the diagnostic steps in the expert knowledge graph; Determine at least one of the initial diagnostic steps as the recommended diagnostic steps according to the recommendation weights of the vehicle model collaborative filtering, the user behavior prediction, and the expert mode.

9. The functional navigation method according to claim 1, wherein Further includes: Predict each of the diagnostic steps through a pre-trained language model based on the vehicle model feature vector, fault code feature vector, and the user behavior data of the target vehicle data, and output the diagnostic steps with corresponding confidence levels and temperature maps.

10. 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 1-9.

Images