Vehicle NVH anomaly diagnosis method, chat robot, system and electronic device

Abstract

The present application relates to the technical field of artificial intelligence, and in particular to a vehicle NVH abnormality diagnosis method, a chat robot, a system and an electronic device, the method comprising: receiving to-be-diagnosed data; calling a preset large language model to enable the large language model to perform vehicle NVH abnormality diagnosis on the to-be-diagnosed data, wherein performing vehicle NVH abnormality diagnosis on the to-be-diagnosed data comprises: if the to-be-diagnosed data is natural language dialogue data, performing vehicle NVH abnormality diagnosis based on the natural language dialogue data and a preset knowledge graph, obtaining a first abnormality diagnosis result and first response information and feeding back the first abnormality diagnosis result and the first response information, and the first response information being used to guide a user to supplement target data associated with a current NVH analysis stage, the target data being machine data or subjective perception feature data of the user; the method can effectively improve the efficiency and accuracy of vehicle NVH abnormality diagnosis, has a relatively low cost, and has relatively strong implementability.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a method for diagnosing vehicle NVH anomalies, a chatbot, a system, and electronic equipment. Background Technology

[0002] NVH (Noise, Vibration, and Harshness) is a key indicator for measuring the comfort and performance of mechanical systems (such as vehicles, aircraft, and industrial equipment). Taking vehicles as an example, superior NVH performance means a quieter, more comfortable, and smoother ride, while poor NVH performance, such as abnormal vibrations and noise, reflects potential safety and reliability issues. Therefore, efficiently troubleshooting abnormal NVH problems in vehicles is crucial.

[0003] Current NVH anomaly diagnosis typically relies on manual analysis; however, this approach has several drawbacks. For example, because human perception of noise and vibration is somewhat subjective, it's difficult to achieve standardized quantification of vehicle noise issues, leading to poor accuracy and low reliability in NVH anomaly diagnosis results. Furthermore, manual analysis and diagnosis are inefficient and costly. Summary of the Invention

[0004] This invention provides a method, system, electronic device, and storage medium for diagnosing vehicle NVH anomalies, in order to solve the problems of low efficiency and poor accuracy of diagnosis results in the prior art.

[0005] This invention provides a method for diagnosing vehicle NVH anomalies, comprising:

[0006] Receive data to be diagnosed;

[0007] A preset large language model is invoked to perform vehicle NVH anomaly diagnosis on the data to be diagnosed. The vehicle NVH anomaly diagnosis on the data to be diagnosed includes:

[0008] If the data to be diagnosed is natural language dialogue data, then based on the natural language dialogue data and the preset knowledge graph, vehicle NVH anomaly diagnosis is performed to obtain a first anomaly diagnosis result and a first response information and feedback is provided. The first response information is used to guide the user to supplement target data associated with the current NVH analysis stage. The target data is machine data or user subjective perception feature data. The machine data includes at least one of the following: vehicle vibration data, vehicle sound data, image data, video data, vehicle operating condition data, and vehicle environmental data.

[0009] In one embodiment of the present invention, it further includes:

[0010] If the target data is received, the large language model is invoked so that the large language model can re-diagnose vehicle NVH anomalies based on the data to be diagnosed and the target data.

[0011] In one embodiment of the present invention, based on the natural language dialogue data and a preset knowledge graph, vehicle NVH anomaly diagnosis is performed to obtain a first anomaly diagnosis result and a first response information, and the feedback includes:

[0012] Semantic recognition is performed on the natural language dialogue data to obtain semantic information;

[0013] The semantic information is processed in a structured manner to obtain a structured query statement;

[0014] Based on the structured query statement and the knowledge graph, vehicle NVH anomaly diagnosis is performed, and the first anomaly diagnosis result and the first response information are obtained and fed back.

[0015] In one embodiment of the present invention, the process of diagnosing vehicle NVH anomalies from the data to be diagnosed further includes:

[0016] If the data packet to be diagnosed contains vehicle vibration data and / or vehicle sound data, a preset filtering playback interface is invoked to access the preset filtering playback model. This allows the filtering playback model to selectively filter and enhance the signal playback of the vehicle vibration data and / or vehicle sound data, thereby obtaining an NVH abnormality feature vector. The selective filtering refers to filtering out noise unrelated to NVH abnormalities, and the signal enhancement playback refers to amplifying the selectively filtered signal.

[0017] Call the preset feature extraction interface to access the preset feature extraction model, so that the feature extraction model can extract features from the data to be diagnosed and obtain the feature vector to be diagnosed.

[0018] The system calls a preset diagnostic analysis interface to access a preset diagnostic analysis model, thereby enabling the diagnostic analysis model to perform vehicle NVH anomaly diagnosis based on the NVH anomaly feature vector and the feature vector to be diagnosed, obtain a second anomaly diagnosis result and a second response information, and provide feedback.

[0019] In one embodiment of the present invention, before calling the diagnostic analysis interface, the method further includes:

[0020] Call the preset data retrieval interface to access the preset data retrieval model, so that the data retrieval model can perform target retrieval. The target retrieval includes: based on the NVH abnormal feature vector and the feature vector to be diagnosed, performing case matching in the preset database, and determining historical cases with a matching degree exceeding the preset matching degree threshold as target cases. The database stores multiple historical cases, and the historical cases include: historical feature vectors and corresponding historical abnormal diagnosis results.

[0021] Calling the preset diagnostic analysis interface to access the preset diagnostic analysis model includes: enabling the diagnostic analysis model to perform vehicle NVH abnormality diagnosis based on the NVH abnormality feature vector, the feature vector to be diagnosed, and the target case, obtaining the second abnormality diagnosis result and the second response information, and feeding it back.

[0022] In one embodiment of the present invention, the data in the database is further used for model training, the model training including:

[0023] According to the preset model training cycle, the filtering playback model, the feature extraction model, the diagnostic analysis model, and the data retrieval model are periodically trained and iterated based on the data in the database to complete the model update.

[0024] In one embodiment of the present invention, the filtering playback model, the feature extraction model, the diagnostic analysis model, and the data retrieval model are periodically trained and iterated based on data in the database, including:

[0025] According to the preset data partitioning rules and the data type of the data in the database, the data in the database is partitioned to obtain training sample data for multiple models to be trained. The models to be trained are the filtering playback model, the feature extraction model, the diagnostic analysis model, or the data retrieval model.

[0026] The training sample data is preprocessed to obtain preprocessed sample data. The preprocessing includes invalid data removal, outlier data correction, and normalization.

[0027] Based on the preprocessed sample data, the corresponding model to be trained is trained and iterated to obtain the updated model;

[0028] The updated model is evaluated according to the preset model evaluation rules. If the evaluation is successful, the deployment of the updated model is completed.

[0029] This invention also provides a chatbot for use in a vehicle NVH anomaly diagnostic system, the chatbot comprising:

[0030] The data receiving module is used to receive data to be diagnosed.

[0031] The NVH anomaly diagnosis module is used to call a preset large language model, so that the large language model can perform vehicle NVH anomaly diagnosis on the data to be diagnosed. The vehicle NVH anomaly diagnosis on the data to be diagnosed includes:

[0032] If the data to be diagnosed is natural language dialogue data, then based on the natural language dialogue data and the preset knowledge graph, vehicle NVH anomaly diagnosis is performed to obtain a first anomaly diagnosis result and a first response information and feedback is provided. The first response information is used to guide the user to supplement target data associated with the current NVH analysis stage. The target data is machine data or user subjective perception feature data. The machine data includes at least one of the following: vehicle vibration data, vehicle sound data, image data, video data, vehicle operating condition data, and vehicle environmental data.

[0033] The present invention also provides a vehicle NVH anomaly diagnostic system, comprising:

[0034] The chatbot described above;

[0035] The large language model is used by the chatbot for invocation;

[0036] The knowledge graph module provides a knowledge graph for querying by the large language model.

[0037] The present invention also provides an electronic device, including a processor, a memory, and a communication bus; the communication bus is used to connect the processor and the memory; the processor is used to execute a computer program stored in the memory to implement the method provided in any of the above embodiments.

[0038] The beneficial effects of this invention are as follows: The vehicle NVH anomaly diagnosis method, chatbot, system, and electronic device provided by this invention receive data to be diagnosed; invoke a preset large language model to enable the large language model to perform vehicle NVH anomaly diagnosis on the data to be diagnosed. The vehicle NVH anomaly diagnosis on the data to be diagnosed includes: if the data to be diagnosed is natural language dialogue data, then based on the natural language dialogue data and a preset knowledge graph, performing vehicle NVH anomaly diagnosis, obtaining a first anomaly diagnosis result and a first response information, and providing feedback. The first response information is used to guide the user to supplement target data related to the current NVH analysis stage. The target data is machine data or user subjective perception feature data. Machine data includes at least one of the following: vehicle vibration data, vehicle sound data, image data, video data, vehicle operating condition data, and vehicle environmental data. This method can effectively improve the efficiency and accuracy of vehicle NVH anomaly diagnosis, has low cost, and is highly feasible. Attached Figure Description

[0039] Figure 1 This is a flowchart illustrating a vehicle NVH anomaly diagnosis method according to an embodiment of the present invention.

[0040] Figure 2 This is a schematic diagram of the structure of a chatbot provided in an embodiment of the present invention;

[0041] Figure 3 This is an example diagram of a dialogue between a driver and a chatbot provided in one embodiment of the present invention;

[0042] Figure 4 This is an example diagram of a dialogue between a maintenance worker and a chatbot, provided according to an embodiment of the present invention.

[0043] Figure 5 This is a schematic diagram of the architecture of a vehicle NVH anomaly diagnostic system provided in an embodiment of the present invention;

[0044] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0045] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, unless otherwise specified, the following embodiments and features described therein can be combined with each other.

[0046] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0047] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0048] The following is combined Figures 1 to 6This paper explains and describes the vehicle NVH abnormality diagnosis method, chatbot, system, and electronic equipment provided by the present invention.

[0049] Please see Figure 1 , Figure 1 This is a flowchart illustrating a vehicle NVH anomaly diagnosis method according to an embodiment of the present invention. This method is applied to a chatbot, such as... Figure 1 As shown, the method includes:

[0050] S110: Receive data to be diagnosed.

[0051] It should be noted that the data to be diagnosed is data input or uploaded by the user through a preset human-machine interface. This data can be text data, vehicle vibration data, vehicle sound data, image data, video data, vehicle operating condition data, and vehicle environmental data, etc.

[0052] It should also be noted that by setting up a human-machine interface, users can easily input or upload data to be diagnosed, thus lowering the threshold for diagnosing vehicle NVH abnormalities.

[0053] S120: Invoke a preset Large Language Model (LLM) to enable the LLM to perform vehicle NVH anomaly diagnosis on the data to be diagnosed. The vehicle NVH anomaly diagnosis on the data to be diagnosed includes:

[0054] If the data to be diagnosed is natural language dialogue data, then based on the natural language dialogue data and the preset knowledge graph, vehicle NVH anomaly diagnosis is performed to obtain a first anomaly diagnosis result and a first response information and feedback is provided. The first response information is used to guide the user to supplement target data associated with the current NVH analysis stage. The target data is machine data or user subjective perception feature data. The machine data includes at least one of the following: vehicle vibration data, vehicle sound data, image data, video data, vehicle operating condition data, and vehicle environmental data.

[0055] It should be noted that natural language dialogue data refers to text data input by users through the human-computer interaction interface for conversation with the chatbot, such as "My car makes a buzzing sound while driving, can you help me figure out what the problem is?" or "I have a customer's car, and the customer reports obvious vibration and noise while driving. Can you guide me on how to collect data?" etc. User subjective perception feature data refers to information obtained by users based on their own perception and judgment, such as "The sound seems to be coming from under the car," and "I feel the steering wheel vibrating a bit at high speeds," etc.

[0056] It should also be noted that by setting up a large language model and using a pre-defined knowledge graph to diagnose vehicle NVH anomalies when the data to be diagnosed is natural language dialogue data, the large language model can obtain highly accurate first anomaly diagnosis results and first response information. Both the first anomaly diagnosis results and the first response information are displayed on the human-machine interface for user viewing.

[0057] It's worth mentioning that this knowledge graph stores structured knowledge in the field of vehicle NVH (Noise, Vibration, and Harshness), such as entities (e.g., failure modes, components) and relationships between entities (e.g., "cause," "association," etc.). It supports complex reasoning and queries, enabling effective systematic utilization of knowledge in the vehicle NVH field. Furthermore, the structured knowledge representation in the knowledge graph ensures the reliability of information sources. Additionally, the knowledge graph supports dynamic updates to incorporate new knowledge and user feedback.

[0058] It should also be mentioned that, since the content described by users is limited and usually incomplete, the above embodiments, by obtaining the first response information at the same time as obtaining the first abnormality diagnosis result and displaying the first response information on the human-machine interface, can facilitate guiding users to supplement the above target data, so that the vehicle NVH abnormality diagnosis can be carried out again based on the supplemented target data and the above-mentioned data to be diagnosed, thereby improving the accuracy of diagnosis.

[0059] Understandably, vehicle vibration data can be obtained through onboard sensors installed in vehicle bearings, suspension, and engine locations; vehicle sound data can be obtained through onboard microphones or external recording equipment; image data can be obtained through onboard cameras; video data can be obtained through dashcams and high-speed cameras; vehicle operating condition data can be obtained through the onboard CAN (Controller Area Network Bus); and vehicle environmental data can be obtained through weather stations and high-precision maps.

[0060] In some embodiments, the large language model can be used not only for vehicle NVH anomaly diagnosis, but also for answering everyday questions, such as: User: "Please help me find a nearby repair shop", chatbot: "Okay, based on your location, I found the following highly rated repair shops..." etc.

[0061] In some embodiments, the first response information can be used not only to guide the user to supplement the aforementioned target data, but also to provide the user with suggestions such as checking specific components.

[0062] The large language model in the above embodiments, through training on massive amounts of data, possesses the ability to understand natural language and can accurately diagnose vehicle NVH anomalies while enabling interaction with users.

[0063] In some embodiments, the method further includes:

[0064] If the target data is received, the large language model is invoked so that the large language model can re-diagnose vehicle NVH anomalies based on the data to be diagnosed and the target data.

[0065] In some examples of this embodiment, by fusing the data to be diagnosed and the target data, and then re-diagnosing the vehicle's NVH anomaly based on the fused data, the accuracy of the vehicle's NVH anomaly diagnosis can be improved. This fusion can be achieved through methods such as splicing.

[0066] In some embodiments, based on the natural language dialogue data and a preset knowledge graph, vehicle NVH anomaly diagnosis is performed to obtain a first anomaly diagnosis result and a first response information, and the feedback includes:

[0067] First, perform semantic recognition on the natural language dialogue data to obtain semantic information.

[0068] Understandably, natural language dialogue data may contain ambiguities. Therefore, by performing semantic recognition on this natural language dialogue data, we can mine or identify the key information implicit in the dialogue statements, i.e., its semantic information. This semantic information includes things like fault type and fault occurrence conditions.

[0069] Second, the semantic information is processed in a structured manner to obtain a structured query statement.

[0070] It should be noted that structuring semantic information facilitates subsequent knowledge graph queries.

[0071] Third, based on the structured query statement and the knowledge graph, perform vehicle NVH anomaly diagnosis, obtain the first anomaly diagnosis result and the first response information, and provide feedback.

[0072] In some embodiments, performing vehicle NVH anomaly diagnosis on the data to be diagnosed further includes:

[0073] 1. If the data packet to be diagnosed contains vehicle vibration data and / or vehicle sound data, a preset filtering playback interface is invoked to access the preset filtering playback model. This allows the filtering playback model to selectively filter and enhance the signal playback of the vehicle vibration data and / or vehicle sound data, thereby obtaining an NVH abnormality feature vector. The selective filtering refers to filtering out noise unrelated to NVH abnormalities, and the signal enhancement playback refers to amplifying the selectively filtered signal.

[0074] In some examples of this embodiment, the methods for determining noise unrelated to NVH anomalies include:

[0075] Based on a pre-defined feature library, the noise unrelated to NVH anomalies is identified. This feature library stores various common irrelevant noise information, such as random road vibrations during vehicle vibration or environmental noise in vehicle sound data.

[0076] Understandably, by performing the aforementioned selective filtering, the noise frequencies (such as 200 to 300 Hz) or vibration bands causing NVH abnormalities in vehicle vibration data can be identified, which helps in subsequent vehicle NVH abnormality diagnosis. Furthermore, this NVH abnormality feature vector can be displayed in the subsequent second response information, thereby confirming the accuracy of the NVH abnormality feature vector through interaction with the user. For example, displaying "the noise frequency causing NVH abnormality is between 200 and 300 Hz" in the second response information allows the accuracy of the selective filtering process to be determined based on the user's response. Through multiple interactions, the noise frequencies or vibration bands causing NVH abnormalities can be accurately identified.

[0077] Second, call the preset feature extraction interface to access the preset feature extraction model, so that the feature extraction model can extract features from the data to be diagnosed and obtain the feature vector to be diagnosed.

[0078] Third, call the preset diagnostic analysis interface to access the preset diagnostic analysis model, so that the diagnostic analysis model performs vehicle NVH abnormality diagnosis based on the NVH abnormality feature vector and the feature vector to be diagnosed, obtains the second abnormality diagnosis result and the second response information and feeds it back.

[0079] Understandably, inputting both the NVH abnormality feature vector and the feature vector to be diagnosed into the diagnostic analysis model to perform vehicle NVH abnormality diagnosis can help improve the accuracy of the obtained second abnormality diagnosis results.

[0080] In some embodiments, prior to invoking the diagnostic analysis interface, the method further includes:

[0081] Call the preset data retrieval interface to access the preset data retrieval model, thereby enabling the data retrieval model to perform target retrieval. The target retrieval includes: based on the NVH abnormal feature vector and the feature vector to be diagnosed, performing case matching in a preset database (DB), and determining historical cases with a matching degree exceeding a preset matching degree threshold as target cases. The database stores multiple historical cases, and the historical cases include: historical feature vectors and corresponding historical abnormality diagnosis results.

[0082] It should be noted that this historical case also includes user feedback information, such as whether the historical diagnostic results in this case were correct. By calling the data retrieval interface, multiple feature vectors similar to the current NVH anomaly feature vector and the feature vector to be diagnosed can be obtained. This facilitates the subsequent provision of auxiliary information for the vehicle NVH anomaly diagnosis process based on these similar feature vectors, the historical diagnostic results in the corresponding case, and the user feedback information.

[0083] Understandably, by providing the above-mentioned multiple interfaces, namely the filtering and playback interface, feature extraction interface, diagnostic analysis interface, and data retrieval interface, the large language model can easily call these multiple interfaces, thereby improving the large language model's ability to integrate and process multi-source data input by users or uploaded.

[0084] It should be mentioned that all the data involved in the above processing is stored in a pre-set database.

[0085] In some embodiments, calling a preset diagnostic analysis interface to access a preset diagnostic analysis model includes: enabling the diagnostic analysis model to perform vehicle NVH anomaly diagnosis based on the NVH anomaly feature vector, the feature vector to be diagnosed, and the target case, obtaining the second anomaly diagnosis result and the second response information, and feeding it back.

[0086] Understandably, fusing NVH anomaly feature vectors, the feature vector to be diagnosed, and the target case for vehicle NVH anomaly diagnosis can help improve diagnostic accuracy. The fusion method can also include stitching, etc.

[0087] In some embodiments, the data in the database is also used for model training, the model training including:

[0088] According to the preset model training cycle, the filtering playback model, the feature extraction model, the diagnostic analysis model, and the data retrieval model are periodically trained and iterated based on the data in the database to complete the model update.

[0089] Understandably, by periodically training and iterating each model as described above, the accuracy and reliability of each model to be trained can be guaranteed. Compared to manual model training, the method of periodically and automatically completing model training in this embodiment can effectively improve model training efficiency and model performance, and has a high degree of automation.

[0090] In some embodiments, the filtering playback model, the feature extraction model, the diagnostic analysis model, and the data retrieval model are periodically trained and iterated based on data in the database, including:

[0091] 1. According to the preset data partitioning rules and the data type of the data in the database, the data in the database is partitioned to obtain training sample data for multiple models to be trained. The models to be trained are the filtering playback model, the feature extraction model, the diagnostic analysis model, or the data retrieval model.

[0092] 2. The training sample data is preprocessed to obtain preprocessed sample data. The preprocessing includes invalid data removal, abnormal data correction, and normalization.

[0093] Third, based on the preprocessed sample data, the corresponding model to be trained is trained and iterated to obtain the updated model.

[0094] Fourth, evaluate the updated model according to the preset model evaluation rules. If the evaluation is successful, the deployment of the updated model is completed.

[0095] Understandably, the above steps can effectively complete the regular training and deployment of each model to be trained, with a high degree of automation and feasibility.

[0096] The following specific example illustrates the steps involved in periodically completing model training as described above:

[0097] First, the model training task is triggered periodically.

[0098] Secondly, when the model training task is triggered, the data in the database is loaded, preprocessed, and partitioned.

[0099] Then, select the corresponding model to be trained.

[0100] Next, the model training process is executed.

[0101] Next, the trained model is evaluated to determine whether it should be deployed.

[0102] Finally, if the evaluation is successful, the trained model, i.e. the updated model, will be deployed to the production environment.

[0103] In some embodiments, the aforementioned database can also meet data query and retrieval needs, facilitating data query and retrieval by users through the human-computer interaction interface of the chatbot. Furthermore, the aforementioned database can also meet the data update needs of knowledge graphs.

[0104] The chatbot provided by this invention is described below. The chatbot described below can be referred to in correspondence with the vehicle NVH abnormality diagnosis method described above.

[0105] Please refer to Figure 2 The chatbot provided in this embodiment is applied to a vehicle NVH anomaly diagnostic system. The chatbot includes:

[0106] Data receiving module 210 is used to receive data to be diagnosed;

[0107] The NVH anomaly diagnosis module 220 is used to call a preset large language model, so that the large language model performs vehicle NVH anomaly diagnosis on the data to be diagnosed, wherein performing vehicle NVH anomaly diagnosis on the data to be diagnosed includes:

[0108] If the data to be diagnosed is natural language dialogue data, then based on the natural language dialogue data and a preset knowledge graph, vehicle NVH anomaly diagnosis is performed, obtaining a first anomaly diagnosis result and a first response information, which is then fed back. The first response information is used to guide the user to supplement target data related to the current NVH analysis stage. The target data is machine data or user subjective perception feature data. The machine data includes at least one of the following: vehicle vibration data, vehicle sound data, image data, video data, vehicle operating condition data, and vehicle environmental data. The chatbot in this embodiment can achieve the technical effects achieved by the vehicle NVH anomaly diagnosis method in the above embodiments, which will not be elaborated here.

[0109] It should be noted that the vehicle NVH anomaly diagnosis method provided in the above embodiments is based on the same concept as the chatbot. The specific operation methods of each module have been described in detail in the method embodiments and will not be repeated here. In practical applications, the chatbot provided in the above embodiments can assign the above functions to different functional modules as needed, that is, divide the internal structure of the chatbot into different functional modules to complete all or part of the functions described above. This is not a limitation here.

[0110] Figure 3 This is an example diagram of a dialogue between a driver and a chatbot provided in one embodiment of the present invention. Please refer to it. Figure 3 , Figure 3The example demonstrates the human-computer interaction interface between the driver and the chatbot. The chatbot performs vehicle NVH anomaly diagnosis based on data input or uploaded by the driver and provides corresponding diagnostic results and responses. Through multiple interactions, the chatbot gradually guides the driver to provide more detailed information about the NVH anomalies, thereby obtaining accurate diagnostic results. The entire interaction process is smooth and comfortable, contributing to an improved driver experience. Figure 3 The specific dialogue content is for illustrative purposes only and has no actual referential meaning, so it will not be elaborated further here.

[0111] Figure 4 This is an example diagram of a dialogue between a maintenance worker and a chatbot, provided in an embodiment of the present invention. Please refer to it. Figure 4 In this human-computer interaction interface, the chatbot provides repair personnel with guidance on data collection and performs corresponding vehicle NVH anomaly diagnosis, helping them with data collection and vehicle NVH anomaly diagnosis. Furthermore, the chatbot supports the generation and download of diagnostic reports, meeting the technical guidance and vehicle NVH anomaly diagnosis needs of repair personnel. Figure 4 The specific dialogue content is for illustrative purposes only and has no actual referential meaning, so it will not be elaborated further here.

[0112] Understandably, traditional vehicle NVH anomaly diagnosis often relies on expert experience, such as using verbal descriptions from experts as diagnostic results. However, this approach is inefficient and prone to missing crucial information. The chatbot in the above embodiment, however, interacts with users through natural language, quickly collecting user descriptions and guiding them to provide key information, thereby offering accurate vehicle NVH anomaly diagnostic results.

[0113] Figure 5 This is a schematic diagram of the architecture of a vehicle NVH anomaly diagnostic system provided in an embodiment of the present invention. Please refer to it. Figure 5 The vehicle NVH anomaly diagnostic system provided in this embodiment includes:

[0114] The chatbot described above;

[0115] The large language model is used by the chatbot for invocation;

[0116] The knowledge graph module provides a knowledge graph for querying the large language model. The vehicle NVH anomaly diagnosis system in this embodiment effectively improves the efficiency and accuracy of vehicle NVH anomaly diagnosis at a low cost.

[0117] In some embodiments, the vehicle NVH anomaly diagnostic system further includes:

[0118] The Tool API module includes: a filter playback interface (Filter Playback API), a feature extraction interface (Feature Extraction API), a diagnostic analysis interface (Diagnostic Analysis API), and a data retrieval interface (Data Retrieval API). All of these interfaces are used by large language models to perform vehicle NVH anomaly diagnosis.

[0119] In some embodiments, the vehicle NVH anomaly diagnostic system further includes a database for storing structured data, unstructured data, and image data. By setting up this database, the large language model can easily retrieve similar features during vehicle NVH anomaly diagnosis, thereby improving the accuracy of the diagnosis. Furthermore, the database facilitates the storage of various types of information, enabling subsequent data retrieval. Additionally, the database allows for the storage of user feedback, facilitating the periodic training and iteration of the filtering playback model, feature extraction model, diagnostic analysis model, and data retrieval model based on data from the database.

[0120] In some embodiments, the vehicle NVH anomaly diagnosis system further includes a model training module, used to periodically train and iterate the filter playback model, feature extraction model, diagnostic analysis model, or data retrieval model based on data in the database.

[0121] In summary, the vehicle NVH anomaly diagnosis method, chatbot, system, and electronic equipment in the above embodiments have the following technical effects:

[0122] 1. Significantly improved efficiency in vehicle NVH anomaly diagnosis. Understandably, by setting up a chatbot and utilizing it to invoke a large language model, natural language interaction between the chatbot and the user, along with an automated NVH anomaly diagnosis process, is effectively achieved, significantly shortening the time required from problem description to diagnostic result generation. Compared to on-site inspections by technicians, the vehicle NVH anomaly diagnosis method, chatbot, system, and electronic equipment described in the above embodiments greatly reduce diagnosis time and effectively improve diagnostic efficiency.

[0123] 2. Significantly Improved Diagnostic Accuracy. Understandably, combining knowledge graphs and application programming interface (API) modules allows for more complex reasoning and multi-source data integration and analysis, leveraging knowledge in the field of NVH anomaly diagnosis to provide more accurate diagnostic results. In contrast, the diagnostic accuracy of traditional methods is limited by the experience of technical personnel. The vehicle NVH anomaly diagnosis method, chatbot, system, and electronic equipment described in the above embodiments can significantly improve the accuracy of vehicle NVH anomaly diagnosis.

[0124] 3. Significantly reduces the cost of NVH anomaly diagnosis. Understandably, the model training module reduces manual intervention, lowering the cost of model training and deployment. Simultaneously, through multi-source data integration and efficient processing, it reduces reliance on specialized equipment and manpower.

[0125] 4. Enhanced interpretability of diagnostic results. Understandably, knowledge graphs and large language models can provide detailed explanations of the diagnostic process, including the reasoning paths of failure modes and relevant knowledge sources. Users can clearly understand the basis for the diagnostic results. Traditional methods, on the other hand, lack interpretability, making it difficult for technicians to explain the basis of their diagnostic results to users.

[0126] 5. Lowering the barrier to vehicle NVH anomaly diagnosis and effectively improving user experience. As is understandable, chatbots support natural language interaction, allowing users to describe problems through simple dialogue and obtain diagnostic results without requiring professional skills. Furthermore, chatbots support multi-source data upload and download, facilitating user operation. In contrast, existing vehicle NVH anomaly diagnosis methods typically require users to possess certain professional knowledge, and the interaction process is relatively complex. In comparison, the vehicle NVH anomaly diagnosis method, chatbot, system, and electronic equipment in the above embodiments significantly improve the user experience through a convenient human-computer interaction interface and natural language interaction.

[0127] 6. Maximizes knowledge utilization. Understandably, knowledge graphs can dynamically update and absorb new fault modes and user feedback, ensuring that the vehicle NVH anomaly diagnostic system always performs diagnoses based on the latest domain knowledge.

[0128] 7. Strong multi-source data integration capability. Understandably, the above embodiments, by setting up a tool application programming interface module and a database, can achieve integrated processing of multi-source data (vehicle vibration data, vehicle sound data, image data, video data, vehicle operating condition data, vehicle environmental data, historical cases, etc.), providing more comprehensive information support for vehicle NVH anomaly diagnosis. Compared to a single data source approach, the above embodiments, by setting up the aforementioned standardized interfaces (filter playback interface, feature extraction interface, diagnostic analysis interface, and data retrieval interface) and database, better achieve unified management and efficient utilization of multi-source data.

[0129] 8. The system has strong scalability. It is understood that the vehicle NVH anomaly diagnosis system in the above embodiments adopts a modular design. This modular design allows each part of the system (such as the knowledge graph module, model training module, etc.) to support independent upgrades and expansions, thereby achieving applicability to different application scenarios, meeting different technical requirements, and exhibiting high flexibility.

[0130] The following explains some of the extended embodiments of this application.

[0131] In some embodiments, edge computing technology can be combined to deploy some diagnostic functions (such as data preprocessing and preliminary diagnosis) locally in the vehicle (e.g., in-vehicle infotainment system or in-vehicle edge device). The local edge device performs preliminary processing and feature extraction on the collected NVH data (such as vehicle vibration data and vehicle sound data), and then sends key information to the cloud for further analysis. The cloud is responsible for model training, knowledge graph management, and complex inference, while the edge device is responsible for real-time response and preliminary diagnosis. Advantages include reduced data transmission volume, lower dependence on network bandwidth, and improved system real-time performance and response speed. It also enhances system privacy protection, allowing preliminary diagnosis to be completed without uploading some data to the cloud. It is suitable for scenarios with unstable network conditions or high real-time requirements, such as real-time fault monitoring of autonomous vehicles.

[0132] In some embodiments, the model training module can be combined with reinforcement learning, enabling each model to be trained to improve its diagnostic capabilities through reinforcement learning and adaptive learning. This is applicable to scenarios where vehicle NVH problems are constantly evolving, such as new failure modes in new energy vehicles.

[0133] In some embodiments, in addition to storing knowledge from the vehicle NVH domain, the knowledge graph can also include knowledge from related domains such as vehicle dynamics and electronic / electrical systems. By associating knowledge entities and relationships across different domains, more comprehensive fault diagnosis can be achieved. During diagnosis, the large language model can invoke cross-domain knowledge for comprehensive reasoning, thereby improving the comprehensiveness and accuracy of the diagnosis, especially for complex faults involving multiple systems. Through cross-domain knowledge fusion, the system can provide more in-depth fault cause analysis. This is suitable for complex fault scenarios, such as when vehicle NVH problems are interrelated with powertrain or electronic / electrical system faults.

[0134] In some embodiments, a user feedback module can be added, allowing users to evaluate diagnostic results (e.g., accuracy, satisfaction). The system dynamically adjusts the knowledge graph and diagnostic analysis model based on user feedback. The knowledge graph is updated in real time through user feedback, optimizing diagnostic logic and model parameters. Reward mechanisms (e.g., points, coupons) can also be provided to encourage active feedback, enhancing user engagement and dynamic optimization capabilities. Through user feedback, the system can quickly correct erroneous diagnoses, improving user experience. This is suitable for large-scale user scenarios, such as after-sales service networks for automotive manufacturers.

[0135] In some embodiments, the vehicle NVH anomaly diagnostic system can also be deeply integrated with the vehicle health management system (VHM) to achieve fault early warning and preventative maintenance. By combining vehicle sensor data, potential NVH problems can be predicted in advance. Based on the diagnostic results, preventative maintenance recommendations are provided to vehicle manufacturers or maintenance personnel. This improves vehicle reliability and safety, reduces the risk of sudden failures, lowers vehicle maintenance costs, and extends vehicle lifespan. It is suitable for the entire lifecycle management of vehicle manufacturers and the operational management of large fleets.

[0136] In some embodiments, an electronic device is also provided, which may be a server, and its internal structure diagram is shown below. Figure 6 As shown, the electronic device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used to communicate with external clients via a network connection. When the computer program is executed by the processor, it implements the functions or steps of the server-side method described above.

[0137] In some embodiments, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it performs the following steps: receiving data to be diagnosed; invoking a preset large language model to enable the large language model to perform vehicle NVH anomaly diagnosis on the data to be diagnosed. The vehicle NVH anomaly diagnosis on the data to be diagnosed includes: if the data to be diagnosed is natural language dialogue data, then performing vehicle NVH anomaly diagnosis based on the natural language dialogue data and a preset knowledge graph, obtaining a first anomaly diagnosis result and a first response information and feeding it back. The first response information is used to guide the user to supplement target data associated with the current NVH analysis stage. The target data is machine data or user subjective perception feature data. The machine data includes at least one of the following: vehicle vibration data, vehicle sound data, image data, video data, vehicle operating condition data, and vehicle environmental data.

[0138] In some embodiments, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, it performs the following steps: receiving data to be diagnosed; invoking a preset large language model to enable the large language model to perform vehicle NVH anomaly diagnosis on the data to be diagnosed, wherein performing vehicle NVH anomaly diagnosis on the data to be diagnosed includes: if the data to be diagnosed is natural language dialogue data, then performing vehicle NVH anomaly diagnosis based on the natural language dialogue data and a preset knowledge graph, obtaining a first anomaly diagnosis result and a first response information and feeding it back, wherein the first response information is used to guide the user to supplement target data associated with the current NVH analysis stage, wherein the target data is machine data or user subjective perception feature data, and the machine data includes at least one of the following: vehicle vibration data, vehicle sound data, image data, video data, vehicle operating condition data, and vehicle environmental data.

[0139] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or electronic device described above can be referred to the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0140] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0141] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A method for diagnosing vehicle NVH anomalies, characterized in that, include: Receive data to be diagnosed; A preset large language model is invoked to perform vehicle NVH anomaly diagnosis on the data to be diagnosed. The vehicle NVH anomaly diagnosis on the data to be diagnosed includes: If the data to be diagnosed is natural language dialogue data, then based on the natural language dialogue data and the preset knowledge graph, vehicle NVH anomaly diagnosis is performed to obtain a first anomaly diagnosis result and a first response information and feedback is provided. The first response information is used to guide the user to supplement target data associated with the current NVH analysis stage. The target data is machine data or user subjective perception feature data. The machine data includes at least one of the following: vehicle vibration data, vehicle sound data, image data, video data, vehicle operating condition data, and vehicle environmental data.

2. The vehicle NVH abnormality diagnosis method according to claim 1, characterized in that, Also includes: If the target data is received, the large language model is invoked so that the large language model can re-diagnose vehicle NVH anomalies based on the data to be diagnosed and the target data.

3. The vehicle NVH abnormality diagnosis method according to claim 1, characterized in that, Based on the natural language dialogue data and the preset knowledge graph, vehicle NVH anomaly diagnosis is performed, and a first anomaly diagnosis result and a first response information are obtained and fed back, including: Semantic recognition is performed on the natural language dialogue data to obtain semantic information; The semantic information is processed in a structured manner to obtain a structured query statement; Based on the structured query statement and the knowledge graph, vehicle NVH anomaly diagnosis is performed, and the first anomaly diagnosis result and the first response information are obtained and fed back.

4. The vehicle NVH abnormality diagnosis method according to claim 1, characterized in that, The vehicle NVH anomaly diagnosis based on the data to be diagnosed also includes: If the data packet to be diagnosed contains vehicle vibration data and / or vehicle sound data, a preset filtering playback interface is invoked to access the preset filtering playback model. This allows the filtering playback model to selectively filter and enhance the signal playback of the vehicle vibration data and / or vehicle sound data, thereby obtaining an NVH abnormality feature vector. The selective filtering refers to filtering out noise unrelated to NVH abnormalities, and the signal enhancement playback refers to amplifying the selectively filtered signal. Call the preset feature extraction interface to access the preset feature extraction model, so that the feature extraction model can extract features from the data to be diagnosed and obtain the feature vector to be diagnosed. The system calls a preset diagnostic analysis interface to access a preset diagnostic analysis model, thereby enabling the diagnostic analysis model to perform vehicle NVH anomaly diagnosis based on the NVH anomaly feature vector and the feature vector to be diagnosed, obtain a second anomaly diagnosis result and a second response information, and provide feedback.

5. The vehicle NVH abnormality diagnosis method according to claim 4, characterized in that, Before calling the diagnostic analysis interface, the following is also included: Call the preset data retrieval interface to access the preset data retrieval model, so that the data retrieval model can perform target retrieval. The target retrieval includes: based on the NVH abnormal feature vector and the feature vector to be diagnosed, performing case matching in the preset database, and determining historical cases with a matching degree exceeding the preset matching degree threshold as target cases. The database stores multiple historical cases, and the historical cases include: historical feature vectors and corresponding historical abnormal diagnosis results. Calling the preset diagnostic analysis interface to access the preset diagnostic analysis model includes: enabling the diagnostic analysis model to perform vehicle NVH abnormality diagnosis based on the NVH abnormality feature vector, the feature vector to be diagnosed, and the target case, obtaining the second abnormality diagnosis result and the second response information, and feeding it back.

6. The vehicle NVH abnormality diagnosis method according to claim 5, characterized in that, The data in the database is also used for model training, which includes: According to the preset model training cycle, the filtering playback model, the feature extraction model, the diagnostic analysis model, and the data retrieval model are periodically trained and iterated based on the data in the database to complete the model update.

7. The vehicle NVH abnormality diagnosis method according to claim 6, characterized in that, Periodically train and iterate the filtering playback model, the feature extraction model, the diagnostic analysis model, and the data retrieval model based on data in the database, including: According to the preset data partitioning rules and the data type of the data in the database, the data in the database is partitioned to obtain training sample data for multiple models to be trained. The models to be trained are the filtering playback model, the feature extraction model, the diagnostic analysis model, or the data retrieval model. The training sample data is preprocessed to obtain preprocessed sample data. The preprocessing includes invalid data removal, outlier data correction, and normalization. Based on the preprocessed sample data, the corresponding model to be trained is trained and iterated to obtain the updated model; The updated model is evaluated according to the preset model evaluation rules. If the evaluation is successful, the deployment of the updated model is completed.

8. A chatbot, characterized in that, The chatbot, used in vehicle NVH anomaly diagnostic systems, includes: The data receiving module is used to receive data to be diagnosed. The NVH anomaly diagnosis module is used to call a preset large language model, so that the large language model can perform vehicle NVH anomaly diagnosis on the data to be diagnosed. The vehicle NVH anomaly diagnosis on the data to be diagnosed includes: If the data to be diagnosed is natural language dialogue data, then based on the natural language dialogue data and the preset knowledge graph, vehicle NVH anomaly diagnosis is performed to obtain a first anomaly diagnosis result and a first response information and feedback is provided. The first response information is used to guide the user to supplement target data associated with the current NVH analysis stage. The target data is machine data or user subjective perception feature data. The machine data includes at least one of the following: vehicle vibration data, vehicle sound data, image data, video data, vehicle operating condition data, and vehicle environmental data.

9. A vehicle NVH anomaly diagnostic system, characterized in that, include: The chatbot as described in claim 8; The large language model is used by the chatbot for invocation; The knowledge graph module provides a knowledge graph for querying by the large language model.

10. An electronic device, characterized in that, It includes a processor, a memory, and a communication bus; the communication bus is used to connect the processor and the memory; the processor is used to execute a computer program stored in the memory to implement the vehicle NVH abnormality diagnosis method as described in any one of claims 1 to 7.

Images