Vehicle diagnostic software starting method and system, vehicle diagnostic equipment and medium

Abstract

The invention relates to the technical field of vehicle diagnosis, and discloses a vehicle diagnosis software starting method and system, vehicle diagnosis equipment and a medium, and the vehicle diagnosis equipment comprises the following steps: obtaining a voice instruction for starting vehicle diagnosis software input by a user, and converting the voice instruction into a voice text; performing entity recognition on the voice text to obtain a target entity in the voice text; the association relationship between the target entities is analyzed based on a knowledge graph, target vehicle diagnosis software corresponding to the target entities in the knowledge graph is determined, and the knowledge graph comprises the association relationship between brands and vehicle types of different languages and the vehicle diagnosis software; and generating a structured instruction corresponding to the target vehicle diagnosis software so as to start the target vehicle diagnosis software through the structured instruction. According to the method and the device, the conversion from the voice instruction to the executable diagnosis command is realized, the step of manually searching diagnosis software is reduced, the interaction efficiency is improved, and the problems of synonyms, multiple languages and multiple brand terms are solved.

Description

Technical Field

[0001] The present application relates to the field of vehicle diagnostic technology, and in particular to a vehicle diagnostic software startup method, system, vehicle diagnostic equipment, and medium. Background Art

[0002] In the field of smart cars and smart devices, users often need to control vehicle diagnostic software for specific brands through voice commands. Existing technologies have the following drawbacks: existing voice systems are mostly monolingual and cannot recognize or adapt to brand and model names in other languages. They cannot accurately identify brands and models, often causing confusion, leading to incorrect calls or failed responses to vehicle diagnostic software. The unclear mapping between software, brands, and models leads to delayed voice command responses and a poor user experience. Summary of the Invention In view of this, the embodiments of the present application provide a vehicle diagnostic software startup method, system, vehicle diagnostic device and medium, which can effectively solve the problem of easy errors and slow response in the vehicle diagnostic software startup response.

[0003] In a first aspect, an embodiment of the present application provides a method for starting vehicle diagnostic software, comprising: Obtaining a voice command input by a user to start vehicle diagnostic software, and converting the voice command into voice text; Performing entity recognition on the speech text to obtain a target entity in the speech text; Analyzing associations between the target entities based on a knowledge graph, and determining target vehicle diagnostic software corresponding to the target entities in the knowledge graph, wherein the knowledge graph includes associations between brands, models, and vehicle diagnostic software in different languages; Generate structured instructions corresponding to the target vehicle diagnostic software to start the target vehicle diagnostic software through the structured instructions.

[0004] In a first possible embodiment of the first aspect, the target entity includes a target brand and a target vehicle model, and performing entity recognition on the speech text to obtain the target entity in the speech text includes: Performing semantic analysis on the speech text using the shared bottom layer of a pre-trained multilingual understanding model to obtain a semantic vector of the speech text; Entity recognition is performed on the semantic vector using the entity recognition head of the multilingual understanding model to obtain the target brand and the target vehicle model in the speech text.

[0005] In a second possible embodiment of the first aspect, analyzing the association relationship between the target entities based on the knowledge graph and determining the target vehicle diagnostic software corresponding to the target entity in the knowledge graph includes: Analyzing the association relationship between the target entities in the knowledge graph through the graph self-attention network of the multilingual understanding model to generate a relationship label between the target model and the target brand; the relationship label between the target entities is the attribution relationship between the target model and the target brand; Under the condition that the target vehicle model belongs to the target brand, the target vehicle diagnostic software corresponding to the target vehicle model and the target brand in the knowledge graph is determined through the graph self-attention network.

[0006] In a third possible embodiment of the first aspect, generating structured instructions corresponding to the target vehicle diagnostic software includes: The semantic vector of the speech text and the target vehicle diagnostic software are combined by the decoder of the multilingual understanding model to generate structured instructions corresponding to the target vehicle diagnostic software.

[0007] In a fourth possible embodiment of the first aspect, the pre-training process of the multilingual understanding model includes: Perform data enhancement processing on the collected multilingual data; Performing a model training task on a neural network model based on the multilingual data after data augmentation processing to obtain a multilingual model after the training task; the model training task includes a masked language training task and a language prediction task; Constructing multiple independent task heads of the multilingual model after performing the training task, and jointly optimizing each of the independent task heads based on a multi-task loss function.

[0008] In a fifth possible embodiment of the first aspect, the process of constructing the knowledge graph includes: Collecting multilingual brands, models, and vehicle diagnostic software, and performing data cleaning on the collected multilingual brands, models, and vehicle diagnostic software; Construct a multi-language mapping table for each of the brands, the models, and the vehicle diagnostic software, and construct the knowledge graph based on the multi-language mapping table.

[0009] In a sixth possible embodiment of the first aspect, the present invention further includes: Identify whether the voice text input by the user is the preset wake-up word through the acoustic model; Activate voiceprint recognition when the preset wake-up word is recognized, extract the user's voiceprint features, and verify the user's identity based on the comparison results of the voiceprint features with the pre-stored voiceprint library; On the condition that the user identity authentication is successful, the voice instruction for starting the vehicle diagnostic software is converted into the voice text.

[0010] In a second aspect, an embodiment of the present application provides a vehicle diagnostic software startup system, comprising: a speech recognition module, configured to obtain a voice command input by a user for starting the vehicle diagnostic software, and convert the voice command into speech text; An entity recognition module is used to perform entity recognition on the speech text to obtain a target entity in the speech text; a knowledge graph analysis module for analyzing associations between the target entities based on a knowledge graph, and determining target vehicle diagnostic software corresponding to the target entities in the knowledge graph, wherein the knowledge graph includes associations between brands, models, and vehicle diagnostic software in different languages; The software startup module is used to generate structured instructions corresponding to the target vehicle diagnostic software, so as to start the target vehicle diagnostic software through the structured instructions.

[0011] In a third aspect, an embodiment of the present application provides a vehicle diagnostic device, which includes a processor and a memory, wherein the memory stores a computer program, and the processor is used to execute the computer program to implement the above-mentioned vehicle diagnostic software startup method.

[0012] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium storing a computer program, which, when executed on a processor, implements the above-mentioned vehicle diagnostic software startup method.

[0013] The embodiments of the present application have the following beneficial effects: A method for starting vehicle diagnostic software in this embodiment includes: obtaining a user-input voice command for starting vehicle diagnostic software, and converting the voice command into voice text; performing entity recognition on the voice text to obtain a target entity in the voice text; analyzing the association relationship between target entities based on a knowledge graph, and determining the target vehicle diagnostic software corresponding to the target entity in the knowledge graph, wherein the knowledge graph includes the association relationship between brands, models, and vehicle diagnostic software in different languages; generating a structured instruction corresponding to the target vehicle diagnostic software to start the target vehicle diagnostic software through the structured instruction. This application implements a unified semantic representation in multiple languages, and regardless of whether the user uses Chinese, English, Spanish, etc., it can be recognized and unified into standard brands and models. This application achieves end-to-end automation from voice commands to executable diagnostic commands, reduces the steps of manually searching for diagnostic software, improves interaction efficiency, supports multilingual input and entity recognition, and solves the problem of multilingual and multi-brand terminology. It significantly improves the accuracy of voice recognition in multilingual scenarios and reduces the command failure rate due to language mismatch. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be obtained based on these drawings without creative work.

[0015] Figure 1 A first flow chart of a method for starting vehicle diagnostic software according to an embodiment of the present application is shown; Figure 2 A second flow chart of the method for starting vehicle diagnostic software according to an embodiment of the present application is shown; Figure 3 A third flow chart of the method for starting vehicle diagnostic software according to an embodiment of the present application is shown; Figure 4 A structural diagram of a vehicle diagnostic software startup system according to an embodiment of the present application is shown.

[0016] Description of main component symbols: 200-Vehicle diagnostic software startup system; 210-Speech recognition module; 220-Entity recognition module; 230-Knowledge graph analysis module; 240-Software startup module. DETAILED DESCRIPTION

[0017] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments.

[0018] The components of the embodiments of the present application generally described and illustrated in the drawings herein may be arranged and designed in a variety of 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 claimed application, but rather merely represents 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 effort are within the scope of protection of the present application.

[0019] Hereinafter, the terms "including", "having" and their cognates used in various embodiments of the present application are intended only to indicate specific features, numbers, steps, operations, elements, components or combinations of the aforementioned items, and should not be understood as excluding the existence of one or more other features, numbers, steps, operations, elements, components or combinations of the aforementioned items or adding the possibility of one or more features, numbers, steps, operations, elements, components or combinations of the aforementioned items. In addition, the terms "first", "second", "third" and the like are only used to distinguish descriptions and should not be understood as indicating or implying relative importance.

[0020] Unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by those skilled in the art to which the various embodiments of the present application belong. The terms (such as those defined in generally used dictionaries) will be interpreted as having the same meaning as in the context of 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.

[0021] The following describes some embodiments of the present application in detail with reference to the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments may be combined with each other.

[0022] The method for starting the vehicle diagnostic device is described below with reference to some specific embodiments.

[0023] Figure 1 A flow chart of a method for starting a vehicle diagnostic device according to an embodiment of the present application is shown. Exemplarily, the method for starting a vehicle diagnostic device includes the following steps: S110 , obtaining a voice instruction input by a user to start vehicle diagnostic software, and converting the voice instruction into voice text.

[0024] In one embodiment, the present application uses an acoustic model to identify whether the user's input voice text is a preset wake-up word. Upon recognizing the preset wake-up word, voiceprint recognition is activated, the user's voiceprint features are extracted, and the user's identity is verified based on the comparison of the voiceprint features with a pre-stored voiceprint library. In this embodiment, the present application uses a CNN (Convolutional Neural Network) + Gated Recurrent Unit (GRU) acoustic model to identify whether the user's input wake-up word is a preset wake-up word, and supports custom preset wake-up words. The CNN is used to extract local speech features of the user's input speech using multiple convolution kernels, and the GRU is used to receive the local speech features output by the CNN, model the temporal dependencies of the speech signal, and output whether it is a wake-up word (binary classification) or which wake-up word it belongs to (multi-classification).

[0025] This application uses traditional acoustic features or deep learning features to extract the voiceprint vector of the voice input by the speaking user. For example, i-vector (Identity Vector), i-vector is a low-dimensional speaker feature representation method that maps a speech segment into a fixed-length voiceprint vector to represent the speaker's identity information. When a user uses it for the first time, he needs to record a voice segment to build his pre-stored voiceprint template. The pre-stored voiceprint template is stored in the pre-stored voiceprint library in the form of an embedded vector. It supports multi-user registration, and each user can set a personalized wake-up word. The voiceprint vector extracted from the real-time voice stream is compared with the pre-stored voiceprint template for cosine similarity; if the match is successful and the wake-up word is correct, personalized wake-up is triggered; otherwise, it is ignored or prompted that the authority is insufficient. Voiceprint recognition is introduced to identify user identity to improve system intelligence, security and user experience.

[0026] In another embodiment, upon successful user authentication, the voice command to activate the vehicle diagnostic software is converted to speech text. This application uses ASR (Automatic Speech Recognition) to convert the user's voice command to activate the vehicle diagnostic software into speech text, supporting recognition in multiple languages ​​such as Chinese, English, and Spanish.

[0027] S120: Perform entity recognition on the speech text to obtain a target entity in the speech text.

[0028] In one embodiment, the present application performs entity recognition tasks, knowledge graph analysis tasks, and structured instruction generation tasks through a pre-trained multilingual understanding model. Figure 2 As shown in Figure 2, the pre-training process of the multilingual understanding model includes: S121: Perform data enhancement processing on the collected multilingual data.

[0029] Exemplarily, this application uses multilingual Wikipedia data as multilingual data for model training. Multilingual Wikipedia data is language data from multiple language versions on the Wikipedia platform and is the data source for building a multilingual understanding model.

[0030] In one embodiment, the data enhancement processing of the present application includes: synonym replacement (SynonymReplacement), building a cross-language synonym dictionary that includes expression variants in different languages, and randomly replacing keywords in multilingual data during training; for example, the original sentence: open a diagnostic system for a certain brand and a certain model, after replacement: start a diagnostic tool for a certain brand and a certain model. The data enhancement processing of the present application also includes: random replacement of model entities in multilingual data, with the replacement range limited to the same brand, to enhance the model's ability to understand the brand-model relationship, avoid the model's dependence on a single model, and improve robustness. Replace vehicle diagnostic software within semantic categories; the replacement rules are based on the functional similarity of vehicle diagnostic software, which improves the model's ability to recognize software intent categories, promotes intent generalization learning, and improves intent recognition accuracy.

[0031] S122, performing a model training task on the neural network model based on the multilingual data after data augmentation processing to obtain a multilingual model after performing the training task; the model training task includes a masked language training task and a language prediction task.

[0032] In this embodiment, the neural network model can be a Transformer architecture. This application is based on the Transformer architecture to build a unified multilingual semantic space to achieve cross-language knowledge transfer and generalized understanding.

[0033] In one embodiment, the present application uses a masked language training task to randomly mask approximately 15% of the tokens in the input text, allowing the model to predict the masked words. The training goal of the masked language training task is to learn contextual semantic information, allowing unsupervised training to improve the model's understanding of multilingual text. The present application also uses a language prediction task (Next Sentence Prediction) to train the model to understand the logical relationship between contexts. The language prediction task is to continuously select two sentences A and B from multilingual data, and randomly select another sentence C from multilingual data. The model uses a classifier to determine the relationship between the sentence pairs, allowing the model to learn the contextual logical relationship between the two consecutive sentences. The model splices A and B, and B and C into input sequences, respectively, and adds special tags before and after. For example, the concatenation of A and B is: [CLS] + A + [SEP] + B + [SEP], where [CLS] is the classification marker, used by the classifier to extract the overall semantic representation of A and B; [SEP] is the separation marker, distinguishing the two sentences. The two concatenated sequences are passed through a fully connected layer (typically a linear classifier) ​​to output the probabilities of two categories: continuous sentences and random sentences. This application uses the cross-entropy loss function for training, and the optimization goal is to enable the model to correctly distinguish continuous sentences from random sentences.

[0034] S123: construct multiple independent task heads of the multilingual model after executing the training task, and jointly optimize each independent task head based on the multi-task loss function.

[0035] In this embodiment, multiple independent task heads refer to the simultaneous execution of multiple independent tasks in a multilingual model, improving the overall generalization ability and learning efficiency of the model by sharing the underlying representation. Multiple independent task heads include entity recognition tasks, knowledge graph analysis tasks, and structured instruction generation tasks. The entity recognition task is used to identify target entities in the speech text; the knowledge graph analysis task is used to determine the association relationship between target entities and determine the target vehicle diagnostic software corresponding to the target entity in the knowledge graph; the structured instruction generation task is used to generate structured instructions corresponding to the target vehicle diagnostic software.

[0036] In one embodiment, the present application uses a multi-task learning (MTL) architecture to simultaneously optimize multiple independent tasks and jointly optimize multiple independent tasks through a weighted loss function, wherein the expression of the weighted loss function is:

[0037] in, Represents the joint loss, which is the total loss to be minimized during training of the multilingual understanding model. The joint loss takes into account the losses of the three independent tasks. Represents the entity recognition loss, which is used to measure whether the entity recognition task is accurate in identifying the target entity; represents the knowledge graph analysis loss, which is used to measure the prediction error of the knowledge graph analysis task in judging the relationship between two target entities and determining the target vehicle diagnostic software corresponding to the target entity; Represents the structured instruction generation loss, which is used to measure the error in generating the structured instructions corresponding to the target vehicle diagnostic software. The entity recognition loss, knowledge graph analysis loss, and structured instruction generation loss can be calculated using the cross entropy loss, which is not limited here. Indicates the importance of the entity recognition task in the total loss function, Indicates the importance of knowledge graph analysis loss in the total loss function, Indicates the importance of the structured instruction generation loss in the total loss function. For example, Can be set to 0.4, Can be set to 0.3, Can be set to 0.3.

[0038] In one embodiment, the target entity includes a target brand and a target model. The present application performs semantic analysis on the speech text through the shared bottom layer of a pre-trained multilingual understanding model to obtain the semantic vector of the speech text; and performs entity recognition on the semantic vector through the entity recognition head of the multilingual understanding model to obtain the target brand and target model in the speech text.

[0039] In one embodiment, the shared bottom layer is an encoding layer, which acts as a feature extractor and provides shared semantic vectors for entity recognition and relationship classification. This encoding layer uses a shared vocabulary to map similar semantics of spoken text in different languages ​​to similar vector spaces. A shared vocabulary refers to sharing a unified vocabulary representation space across multiple languages. Unlike constructing a separate vocabulary for each language, a shared vocabulary encodes word units from all languages ​​into a single vocabulary. This allows words with the same or similar semantics in different languages ​​to be encoded using the same word unit vector, facilitating transfer learning for models in unseen language tasks.

[0040] In one embodiment, the entity recognition head of the multilingual understanding model classifies each word to determine whether it belongs to a specific entity category (such as brand, model, etc.). The entity recognition head maps semantic vectors to entity category labels through a fully connected layer. A fully connected layer is one of the fundamental layers in a neural network. It linearly transforms the input vector using a weight matrix and a bias vector. The output label space represents the probability distribution of the word belonging to different entity category labels. The category with the highest probability is selected as the word's category to identify the target brand and model entity. Entity category labels include brand and model.

[0041] S130, analyzing the association relationship between target entities based on the knowledge graph, and determining the target vehicle diagnostic software corresponding to the target entity in the knowledge graph, the knowledge graph including the association relationship between brands, models and vehicle diagnostic software in different languages.

[0042] In one embodiment, if Figure 3 As shown in Figure 2, the knowledge graph construction process includes: S131, collects multi-language brands, models and vehicle diagnostic software, and performs data cleaning processing on the multi-language brands, models and vehicle diagnostic software.

[0043] In one embodiment, the present application can use a crawler framework (such as Scrapy or Selenium) to crawl multilingual web pages and obtain multilingual brands, models, and vehicle diagnostic software from the multilingual web pages. It can also obtain vehicle diagnostic software usage records from automotive forums / communities and obtain multilingual brands, models, and vehicle diagnostic software from the vehicle diagnostic software usage records. The present application can also collect software call logs output by the vehicle diagnostic interface and obtain multilingual brands, models, and vehicle diagnostic software from the software call logs.

[0044] In one embodiment, the present application uses the SimHash algorithm (Similarity Hashing Algorithm) to detect duplicate text. This algorithm can be used to quickly identify duplicate or similar content and detect duplicate brands, models, and vehicle diagnostic software names across multiple forums, logs, and official websites. The present application removes all detected duplicate brands, models, and vehicle diagnostic software names, retaining only one. The present application also unifies the cleaned brand, model, and software information into a structured format to facilitate subsequent operations such as multilingual alignment and knowledge graph construction.

[0045] S132, constructing a multilingual mapping table for each brand, model, and vehicle diagnostic software, and constructing a knowledge graph based on the multilingual mapping table.

[0046] In one embodiment, the multi-language mapping table utilizes a nested JSON (JavaScript Object Notation) structure, constructing entity mappings centered around the language dimension. Its core structure consists of three nested layers: brand, which stores the multi-language mappings for vehicle brands; vehicle model, which records the multi-language mappings for specific vehicle models; and vehicle diagnostic software, which contains the multi-language mappings for the vehicle's onboard software systems. Each entity entry includes a multi-language dictionary, supporting unlimited language expansion. JSON Schema validation enables type-safe expansion, allowing new languages ​​to be added by simply appending key-value pairs to the corresponding entity dictionary without modifying the overall structure.

[0047] In this embodiment, the knowledge graph is a structured semantic network used to represent entities (brands, models, and vehicle diagnostic software) and their relationships. The nodes of the knowledge graph represent entities and the various languages ​​used to record each entity. The edges of the knowledge graph represent relationships between entities, such as "produced by," "adapted to," and "belongs to." The attributes of the knowledge graph describe the metadata of nodes or edges, such as the brand's founding year, the vehicle's powertrain, and the software version.

[0048] In one embodiment, the present application uses a Graph Attention Network (GAT) of a multilingual understanding model to analyze the association relationship between target entities in a knowledge graph to generate relationship labels between target models and target brands; the relationship labels between target entities are the attribution relationship between the target models and the target brands; under the condition that the target models belong to the target brand, the target vehicle diagnostic software corresponding to the target models and target brands in the knowledge graph is determined through the Graph Attention Network.

[0049] In this embodiment, the graph self-attention network is used to dynamically aggregate node information in the knowledge graph through the attention mechanism. It can process different types of nodes such as brands, models, and software. It can be used to determine whether the target model belongs to the target brand and determine the target vehicle diagnostic software corresponding to the target model and target brand.

[0050] S140 , generating a structured instruction corresponding to the target vehicle diagnostic software, so as to start the target vehicle diagnostic software through the structured instruction.

[0051] In one embodiment, the present application uses a decoder of a multilingual understanding model to generate structured instructions corresponding to the target vehicle diagnostic software from the semantic vectors of the speech text and the target vehicle diagnostic software. In this embodiment, the structured instruction generation task of the multilingual understanding model of the present application generates structured instructions for the target vehicle diagnostic software through a decoder. The decoder can be a Transformer decoder. The Transformer decoder is used to fuse the semantic vectors of the speech instruction with the target vehicle diagnostic software, and based on the fused semantic representation, generates structured instructions corresponding to the target vehicle diagnostic software. The structured instructions are used to call the vehicle system interface, switch the software interface language, and perform permission control, so that the vehicle system can call and start the target vehicle diagnostic software.

[0052] In one embodiment, the semantic vector is used as the encoder output, and the target vehicle diagnostic software is used as the decoder input context information. The Transformer decoder uses the semantic vector as a global guide through the self-attention mechanism to ensure that the generated content is consistent with the user's intention. Figure 1For example, if the semantic vector includes launching the vehicle diagnostic software, {intent: activate_software} is generated. The semantic vector is then fused with contextual information through a self-attention mechanism to generate structured fields. This supports outputting structured instructions in JSON format of any length. The structured instructions represent the semantic intent information parsed from the voice command that can be used to execute the operation. The structured instructions include an intent identifier, a target software identifier, and a language identifier. For example, the output format of the structured instruction can be: {intent: activate_software, software_id: UUID, lang: zh}, where intent represents the user's operation intention, activate_software represents launching the vehicle diagnostic software, software_id represents the software unique identifier, UUID represents the universally unique identifier of a vehicle diagnostic software, lang represents the language identifier, which is used to identify the language version of the current voice command, and zh represents the Chinese identifier.

[0053] Figure 4 A schematic diagram of the structure of a vehicle diagnostic software startup system 200 according to an embodiment of the present application is shown. Exemplarily, the vehicle diagnostic software startup system 200 includes: The speech recognition module 210 is configured to obtain a voice command input by a user to start the vehicle diagnostic software and convert the voice command into speech text.

[0054] The entity recognition module 220 is used to perform entity recognition on the speech text to obtain the target entity in the speech text.

[0055] The knowledge graph analysis module 230 is used to analyze the association relationship between target entities based on the knowledge graph, and to determine the target vehicle diagnostic software corresponding to the target entity in the knowledge graph. The knowledge graph includes the association relationship between brands, models and vehicle diagnostic software in different languages.

[0056] The software startup module 240 is used to generate structured instructions corresponding to the target vehicle diagnostic software, so as to start the target vehicle diagnostic software through the structured instructions.

[0057] It can be understood that the system of this embodiment corresponds to the vehicle diagnostic software startup method of the above embodiment, and the optional items in the above embodiment are also applicable to this embodiment, so they will not be repeated here.

[0058] The present application also provides a vehicle diagnostic device. 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 vehicle diagnostic software startup method or the functions of each module in the above-mentioned vehicle diagnostic software startup system.

[0059] 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, or discrete hardware components. The general-purpose processor can be a microprocessor or any conventional processor, etc., and can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of this application.

[0060] The memory may be, but is not limited to, random access memory (RAM), read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), and electrically erasable programmable read-only memory (EEPROM). The memory is used to store computer programs, and the processor can execute the computer programs accordingly after receiving execution instructions.

[0061] This application also provides a computer-readable storage medium for storing a computer program used in the vehicle diagnostic device. For example, the computer-readable storage medium may include, but is not limited to, a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, among other media capable of storing program code.

[0062] In the 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 schematic. For example, the flowcharts and structure diagrams in the accompanying drawings show the possible architectures, functions and operations of the devices, methods and computer program products according to the multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and the module, program segment or a part of the code contains one or more executable instructions for implementing the specified logical functions. It should also be noted that in an alternative implementation, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the structure diagram and / or flowchart, and the combination of boxes in the structure diagram and / or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0063] In addition, the functional modules or units in the various embodiments of the present application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0064] If a function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, or the part that contributes to the existing technology, or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes a number of instructions for enabling a computer device (which can be a smart phone, personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the method of this application.

[0065] The above is only a specific implementation method of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with this technical field can easily think of changes or replacements within the technical scope disclosed in this application, which should be covered by the protection scope of the present application.

Claims

1. A method for starting vehicle diagnostic software, characterized in that: include: Obtaining a voice command input by a user to start vehicle diagnostic software, and converting the voice command into voice text; Performing entity recognition on the speech text to obtain a target entity in the speech text; Analyzing associations between the target entities based on a knowledge graph, and determining target vehicle diagnostic software corresponding to the target entities in the knowledge graph, wherein the knowledge graph includes associations between brands, models, and vehicle diagnostic software in different languages; Generate structured instructions corresponding to the target vehicle diagnostic software to start the target vehicle diagnostic software through the structured instructions.

2. The vehicle diagnostic software startup method according to claim 1, characterized in that: The target entity includes a target brand and a target vehicle model, and performing entity recognition on the speech text to obtain the target entity in the speech text includes: Performing semantic analysis on the speech text using the shared bottom layer of a pre-trained multilingual understanding model to obtain a semantic vector of the speech text; Entity recognition is performed on the semantic vector using the entity recognition head of the multilingual understanding model to obtain the target brand and the target vehicle model in the speech text.

3. The vehicle diagnostic software startup method according to claim 2, characterized in that: The analyzing the association relationship between the target entities based on the knowledge graph and determining the target vehicle diagnostic software corresponding to the target entity in the knowledge graph includes: Analyzing the association relationship between the target entities in the knowledge graph through the graph self-attention network of the multilingual understanding model to generate a relationship label between the target model and the target brand; the relationship label between the target entities is the attribution relationship between the target model and the target brand; Under the condition that the target vehicle model belongs to the target brand, the target vehicle diagnostic software corresponding to the target vehicle model and the target brand in the knowledge graph is determined through the graph self-attention network.

4. The vehicle diagnostic software startup method according to claim 2, characterized in that: The generating of structured instructions corresponding to the target vehicle diagnostic software includes: The semantic vector of the speech text and the target vehicle diagnostic software are combined by the decoder of the multilingual understanding model to generate structured instructions corresponding to the target vehicle diagnostic software.

5. The vehicle diagnostic software startup method according to claim 2, characterized in that: The pre-training process of the multilingual understanding model includes: Perform data enhancement processing on the collected multilingual data; Performing a model training task on a neural network model based on the multilingual data after data augmentation processing to obtain a multilingual model after the training task; the model training task includes a masked language training task and a language prediction task; Constructing multiple independent task heads of the multilingual model after performing the training task, and jointly optimizing each of the independent task heads based on a multi-task loss function.

6. The vehicle diagnostic software startup method according to claim 1, characterized in that: The construction process of the knowledge graph includes: Collecting multilingual brands, models, and vehicle diagnostic software, and performing data cleaning on the collected multilingual brands, models, and vehicle diagnostic software; Construct a multi-language mapping table for each of the brands, the models, and the vehicle diagnostic software, and construct the knowledge graph based on the multi-language mapping table.

7. The vehicle diagnostic software startup method according to claim 1, characterized in that: Also includes: Identify whether the voice text input by the user is the preset wake-up word through the acoustic model; Activate voiceprint recognition when the preset wake-up word is recognized, extract the user's voiceprint features, and verify the user's identity based on the comparison results of the voiceprint features with the pre-stored voiceprint library; On the condition that the user identity authentication is successful, the voice instruction for starting the vehicle diagnostic software is converted into the voice text.

8. A vehicle diagnostic software startup system, characterized in that: include: a speech recognition module, configured to obtain a voice command input by a user for starting the vehicle diagnostic software, and convert the voice command into speech text; An entity recognition module is used to perform entity recognition on the speech text to obtain a target entity in the speech text; a knowledge graph analysis module for analyzing associations between the target entities based on a knowledge graph, and determining target vehicle diagnostic software corresponding to the target entities in the knowledge graph, wherein the knowledge graph includes associations between brands, models, and vehicle diagnostic software in different languages; The software startup module is used to generate structured instructions corresponding to the target vehicle diagnostic software, so as to start the target vehicle diagnostic software through the structured instructions.

9. A vehicle diagnostic device, characterized in that: The vehicle diagnostic device includes a processor and a memory, wherein the memory stores a computer program, and the processor is configured to execute the computer program to implement the vehicle diagnostic software startup method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that The device stores a computer program, which, when executed on a processor, implements the vehicle diagnostic software startup method according to any one of claims 1 to 7.

Images