Intelligent cabin evaluation method and system based on multi-source data fusion

Abstract

The invention relates to the technical field of intelligent automobiles, and discloses an intelligent cabin evaluation method and system based on multi-source data fusion, and the method comprises the following steps: S1, defining a customized evaluation task for intelligent cabin evaluation based on a selected automobile model, an evaluation scene and a user portrait; and S2, in response to execution of the customized evaluation task, multi-source data is collected, the multi-source data comprises structured scoring data and unstructured user sound data, and each piece of unstructured user sound data is bound with the corresponding structured scoring data. According to the method, the customized evaluation task is defined based on the selected vehicle type, the evaluation scene and the user portrait, and the dynamic weight is applied, so that the evaluation accuracy and pertinence are remarkably improved, and the exclusive evaluation content and evaluation standard can be automatically generated according to the evaluation context, and a scientific basis is provided for differential positioning and accurate optimization of products.

Description

Technical Field

[0001] This invention relates to the field of intelligent vehicle technology, specifically to an intelligent cockpit evaluation method and system based on multi-source data fusion. Background Technology

[0002] With the rapid development of automotive intelligence and connectivity technologies, the intelligent cockpit has evolved from a traditional driver information display and entertainment center into a "third living space" that integrates multimodal interaction, personalized services, and a rich in-vehicle ecosystem. It has become a key factor in measuring the core competitiveness of vehicle products. Therefore, how to scientifically, comprehensively, and accurately evaluate user satisfaction with intelligent cockpits has become an important issue that the automotive industry urgently needs to address.

[0003] However, existing intelligent cockpit evaluation technologies generally exhibit a serious disconnect between system and execution when facing this trend. Evaluation methods often remain at the level of fragmented and localized functional verification, lacking an integrated software platform capable of supporting a complete evaluation system and achieving full-process online management. This results in a broken evaluation process, fragmented data distribution, and difficulty in forming an effective management and traceability loop. Existing technologies generally adopt a one-size-fits-all general evaluation template, which ignores the huge differences between different user groups (such as professional evaluators and ordinary car owners) and different driving scenarios (such as urban congestion and highway cruising). This leads to a mismatch between evaluation content and actual needs, and the conclusions drawn often lack specificity, making it difficult to provide effective guidance for differentiated product improvements.

[0004] Meanwhile, at the data processing level, existing technologies are clearly insufficient in their ability to integrate and mine multi-source data. Although the evaluation process can collect quantitative scores from users, it is insufficient to understand the qualitative reasons behind the scores, i.e., the users' true voices. Often, evaluation methods rely on simple recording or extensive manual analysis, which is not only inefficient but, more importantly, fails to establish a deep connection and linkage analysis mechanism between structured scoring and unstructured feedback. This fails to truly reveal the fundamental driving factors behind user satisfaction. Furthermore, most current evaluation systems are essentially static, lacking the ability to evolve dynamically with technological advancements and changing user needs. With the continuous emergence of new elements such as in-vehicle artificial intelligence and data security regulations, the fixed evaluation dimensions and weighting systems will soon become outdated, unable to continuously and accurately measure the comprehensive competitiveness of intelligent cockpits, and their long-term applicability is greatly limited. Therefore, this invention designs an intelligent cockpit evaluation method and system based on multi-source data fusion to address the aforementioned problems. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a smart cockpit evaluation method and system based on multi-source data fusion. This solves the problems in existing smart cockpit evaluation methods, such as rigid evaluation systems, inability to adapt to the needs of multiple scenarios and user groups, and insufficient fusion analysis of multi-source data, which leads to evaluation conclusions lacking specificity, in-depth insight, and interpretability.

[0006] To achieve the above objectives, the present invention provides a method for evaluating intelligent cockpits based on multi-source data fusion, comprising the following steps: S1. Based on the selected vehicle model, evaluation scenario, and user profile, define customized evaluation tasks for intelligent cockpit evaluation; S2. In response to the execution of the customized evaluation task, multi-source data is then collected, including structured scoring data and unstructured user feedback. The data, wherein each of the unstructured user voice data is bound to its corresponding structured scoring data; S3. The collected multi-source data is fused, and the fusion process is performed by automatically analyzing the unstructured user voice. The data is used to interpret the corresponding bound structured scoring data, thereby generating comprehensive analysis results; S4. Finally, based on the comprehensive analysis results, an intelligent cockpit evaluation report can be generated.

[0007] Preferably, the step of defining the customized evaluation task in step S1 further includes: Based on the vehicle model, evaluation scenario, and user profile, dynamic weights are loaded from a preset weighting system and applied to the customized evaluation task to achieve adaptive adjustment of the evaluation focus.

[0008] Preferably, in step S2, the step of collecting multi-source data is as follows: The structured scoring data consists of evaluation scores based on a Likert scale; the unstructured user voice data... The data consists of user reviews in voice or text format.

[0009] Preferably, the binding step specifically includes: After collecting the structured scoring data for any evaluation item, the unstructured user voice for that evaluation item is immediately triggered. The data collection entry point is used to establish the structured scoring data and the unstructured user voice. One-to-one correspondence between data.

[0010] Preferably, in the fusion processing step described in step S3, the unstructured user voice is automatically analyzed. The data includes: Natural Language Processing (NLP) technology is used to process the unstructured user voice. The data is cleaned and automatically classified and clustered according to preset dimensions, scenarios, or emotional polarities.

[0011] Preferably, the step of generating the comprehensive analysis results in step S3 further includes: By combining the dynamic weights, the structured scoring data is weighted and calculated to obtain a quantitative comprehensive evaluation score.

[0012] Preferably, the step of generating the intelligent cockpit evaluation report in step S4 further includes: The smart cockpit evaluation report provides a range of data from the overall evaluation score to the corresponding bound unstructured user voice data. Drill-down tracing of data paths.

[0013] Preferably, the method further includes the step of forming a feedback closed loop: Based on the issues revealed in the intelligent cockpit evaluation report, the intelligent cockpit was optimized, and after a preset period, the same evaluation group was invited to perform a retest to verify the optimization effect.

[0014] Preferably, the method further includes a system evolution step: In response to external triggering conditions such as changes in user needs, technological iterations, or regulatory updates, the evaluation dimensions or weighting system used in the customized evaluation task steps are dynamically adjusted. The evaluation dimensions or weighting system includes a matrix of 10 primary dimensions and 54 secondary dimensions.

[0015] Preferably, an intelligent cockpit evaluation system based on multi-source data fusion includes: Task definition module: Used to define customized evaluation tasks for intelligent cockpit evaluation based on selected vehicle models, evaluation scenarios, and user profiles; Data acquisition module: Used to collect multi-source data in response to the execution of the customized evaluation task, including structured scoring data and unstructured user feedback. Data, and implement each of the aforementioned unstructured user voices. The data is bound one-to-one with its corresponding structured scoring data; Fusion Processing Module: Used to fuse the collected multi-source data, and automatically analyze the unstructured user voice. The data is used to interpret the corresponding bound structured scoring data, thereby generating comprehensive analysis results; And a report generation module: used to generate a smart cockpit evaluation report based on the comprehensive analysis results.

[0016] This invention provides a method and system for evaluating intelligent cockpits based on multi-source data fusion. It has the following beneficial effects: 1. This invention defines customized evaluation tasks based on selected vehicle models, evaluation scenarios, and user profiles, and applies dynamic weights, significantly improving the accuracy and relevance of the evaluation. Furthermore, it can automatically generate exclusive evaluation content and evaluation criteria based on the evaluation context, enabling the evaluation results to more realistically reflect the product's performance in specific scenarios. This avoids conclusion bias caused by the mismatch between evaluation content and actual needs, providing a scientific basis for product differentiation and precise optimization.

[0017] 2. This invention triggers unstructured scoring immediately after collecting structured scoring data. By collecting data and binding it one-to-one with qualitative factors, a strong correlation was established between quantitative evaluation and qualitative causes, achieving deep data fusion and ensuring the accuracy of every result. The evaluations can be precisely traced back to their corresponding specific evaluation items and scores, forming the smallest data unit of "score + cause". This design lays a solid foundation for subsequent root cause analysis (RCA).

[0018] 3. This invention employs Natural Language Processing (NLP) technology to bind unstructured data. Automated data classification and clustering greatly improve the efficiency and depth of user feedback analysis. It can automatically and quickly extract high-frequency issues, sentiment tendencies and scenario features from tens of thousands of VoCs and transform this unstructured information into quantifiable and statistically significant structured labels. This not only reduces the workload of evaluators, but also reveals deep-seated relationships and trends that are difficult for humans to detect.

[0019] 4. This invention provides a range of evaluation reports, from overall scores to specific details. The drill-down approach to data significantly enhances the interpretability and operability of the evaluation results. The interactive reports allow users to drill down from the overall score to the specific user complaints or recordings that led to a low score in a particular dimension. This transparent and traceable analysis path ensures that every evaluation conclusion is based on evidence, greatly enhancing the credibility of the reports.

[0020] 5. By constructing a feedback loop and system evolution mechanism, this invention enables the entire evaluation method to have the ability to self-optimize and adapt over a long period of time, solving the rigidity problem of static evaluation systems. It not only designs a product iteration closed loop of "evaluation-optimization-retesting", but also introduces a system evolution trigger that responds to external changes. When significant changes in technology, requirements or regulations are detected, the evaluation system itself can dynamically adjust its dimensions and weights, thereby ensuring the long-term effectiveness of the evaluation method. Attached Figure Description

[0021] Figure 1 This is one of the schematic diagrams of the method flow of the present invention; Figure 2 This is a second schematic diagram of the method flow of the present invention; Figure 3 This is the third schematic diagram of the method flow of the present invention; Figure 4 This is the fourth schematic diagram of the method flow of the present invention; Figure 5 This is the fifth schematic diagram of the method flow of the present invention; Figure 6 This is one of the system flow diagrams of the present invention; Figure 7 This is the second system flow diagram of the present invention; Figure 8 This is the third system flow diagram of the present invention; Figure 9 This is the fourth system flow diagram of the present invention; Figure 10 This is the fifth system flow diagram of the present invention; Figure 11 This is a schematic diagram of the system framework of the present invention; Figure 12 This is a flowchart of the management backend system of the present invention. Detailed Implementation

[0022] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Please see the appendix Figure 1 - Appendix Figure 5 This invention provides a smart cockpit evaluation method based on multi-source data fusion, comprising the following steps: S1. Based on the selected vehicle model, evaluation scenario, and user profile, define a customized evaluation task for intelligent cockpit evaluation. The steps for defining the customized evaluation task also include: loading and applying dynamic weights from the preset weight system to the customized evaluation task according to the vehicle model, evaluation scenario, and user profile, so as to achieve adaptive adjustment of the evaluation focus. Specifically, the system's backend pre-builds the basic database required for the evaluation, which includes: Vehicle version library: Stores information such as the brand, model, hardware configuration, and software version number of the vehicles to be evaluated; Evaluation Scenario Library: Presets multiple typical vehicle usage scenarios, with each scenario associated with a specific subset of evaluation dimensions; User profile database: Stores the identity tags of the evaluators; Evaluation Dimensions and Weighting System Library: A comprehensive evaluation indicator system is designed, including a matrix of 10 primary dimensions and 54 secondary dimensions. The specific matrix of the 10 primary dimensions and 54 secondary dimensions is as follows: Configuration satisfaction: Chip configuration, screen configuration, HUD, speaker (audio) configuration, DMS & OMS, microphone; Interaction satisfaction: Satisfaction with ease of interaction, satisfaction with interactive feedback, and satisfaction with voice interaction; Visual satisfaction: Layout rationality, 2D visual aesthetics, 3D visual aesthetics, and visual interference resistance; Comfort and satisfaction: Seat satisfaction, air environment satisfaction, lighting satisfaction, tactile satisfaction; Functional satisfaction: Overall performance, imaging, audio-visual entertainment, navigation, system, vehicle control and design, task scenarios, social, life applications, hardware expansion; Smart satisfaction: Intelligent perception, intelligent interaction, personalized services, proactive scenario services, AIGC, and intelligent driving; Service satisfaction expectations: OTA expectations, holiday operation service expectations, user feedback response expectations, easy-to-use vehicle guidance service, maintenance and repair reminder service, and emergency rescue service expectations; Vehicle safety satisfaction: Intelligent driving safety, road safety, parking safety, emergency safety; Privacy protection satisfaction: Clear user authorization, clear notification of data collection and use, temporary customer mode, data deletion, and data exportability; Simultaneously, based on the principles of "user perception + core technology + bottom-line guarantee," basic weights are preset for each dimension (such as "interaction satisfaction" weight of 14%-16%) and scenario weight adjustment coefficients. Before being put into use, this system needs to be tested using statistical methods (such as a KMO value greater than 0.7). To ensure its scientific validity, reliability and validity are tested. When an administrator creates a task, they select the vehicle model, evaluation scenario, and user profile through the system interface. After receiving these three elements, the system automatically performs the following operations: Content customization: Based on the selected evaluation scenario and user profile, relevant evaluation items are selected and combined from the complete evaluation dimension matrix to form the evaluation questionnaire for this task; Dynamic weight loading: Based on the three selected elements, load the basic weights from the weight system library and apply the weight adjustment coefficients that match the scenario to generate the dynamic weights for this task.

[0024] S2. Responding to the execution of customized evaluation tasks, it then collects multi-source data, including structured scoring data and unstructured user feedback. The data, in which each unstructured user voice data point is bound to its corresponding structured scoring data, involves collecting multi-source data: the structured scoring data consists of evaluation scores based on a Likert scale; the unstructured user voice data... For user reviews in voice or text format, the binding process is as follows: after collecting structured scoring data for any evaluation item, immediately trigger unstructured user voice data for that evaluation item. The data collection entry point is used to establish structured scoring data and unstructured user feedback. One-to-one correspondence between data; Specifically, after logging into the evaluation client (such as a PAD or web browser), the evaluator receives and opens the customized evaluation task defined by S1. The evaluation interface will then display the following evaluation content item by item: Collect structured scoring data: For each evaluation item (such as "voice assistant wake-up success rate"), the evaluators click to score on the Likert 5-level scale provided on the interface based on their actual experience (1 = very dissatisfied, 5 = very satisfied). This part of the data is the structured scoring data. Collecting and binding unstructured VoC data: After the evaluator completes the scoring of a certain item, the system will immediately pop up an input box or activate a recording button next to the scoring item. The evaluator can enter text on the keyboard or record voice through the microphone to describe in detail the reasons for the score. For example, after scoring 2 points, the evaluator can enter: "When driving at high speed with the window open, I said 'Hello, Xiaodi' five times, and only once was it successfully woke up." This text or voice is the unstructured user voice VoC data. In the background, the system will strongly bind this VoC data with the score of 2 and the evaluation item "voice assistant wake-up success rate" to form an indivisible data unit. This interactive process of "immediately triggering VoC input after scoring" ensures the real-time and accuracy of data binding. After the entire evaluation process is completed, the system will package and encrypt all collected data units (including scores, VoC, evaluation item ID, user ID, timestamp, etc.) and upload them to the server.

[0025] S3. Perform fusion processing on the collected multi-source data. The fusion processing automatically analyzes unstructured user voice. The data is used to interpret the corresponding structured scoring data, thereby generating a comprehensive analysis result. In the fusion processing step, unstructured user voice is automatically analyzed. The data includes: unstructured user voice processed using Natural Language Processing (NLP) technology. The steps of cleaning the data and automatically classifying and clustering it according to preset dimensions, scenarios or sentiment polarities to generate comprehensive analysis results also include: combining dynamic weights to perform weighted calculations on the structured scoring data to obtain a quantitative comprehensive evaluation score. Specifically, by utilizing computer technology, especially Natural Language Processing (NLP), unstructured data is deeply processed, enabling it to be semantically integrated with structured data. This allows valuable insights to be extracted from massive amounts of raw data. After receiving the data packet uploaded by S2, the server initiates an automated processing pipeline: Data cleaning: The system first performs preliminary verification of the data, such as automatically removing invalid data where the vast majority (e.g., more than 95%) of the data has the same score; VoC Deep Analysis: The system inputs the bound VoC data into a pre-trained NLP model and performs the following: Cleaning: removing invalid characters and stop words, and standardizing the format; Classification and Clustering: the model automatically assigns multi-dimensional labels to each VoC. For example, for the VoC "Cannot wake up when opening a window at high speed", the model will assign labels such as [Dimension: Voice Interaction], [Scenario: High Speed / Noisy], [Problem: Wake-up Failure], [Emotion: Negative]. Data Fusion and Comprehensive Analysis: At this point, the original "score + raw VoC" data unit has been upgraded to a rich information data unit of "score + structured VoC tag + raw VoC". This is the manifestation of fusion processing: structured tags are used to interpret the quantified score. At the same time, the system calls the dynamic weights loaded for this task in S1 to perform a weighted average calculation on all valid structured scoring data to obtain the scores of each dimension and the final quantified comprehensive evaluation score. The final output of this step is the comprehensive analysis result, which includes both the macro-level quantified score and the micro-level, deeply analyzed problem / highlight attribution.

[0026] S4. Finally, based on the comprehensive analysis results, a smart cockpit evaluation report can be generated. The steps for generating the smart cockpit evaluation report also include: providing the smart cockpit evaluation report with the corresponding bound specific unstructured user voice data, from the comprehensive evaluation score. The data drilling-through tracing path method also includes the step of forming a feedback loop: based on the problems revealed in the intelligent cockpit evaluation report, the intelligent cockpit is optimized, and after a preset period, the same evaluation group is invited to perform a retest task to verify the optimization effect. The method also includes the step of system evolution: in response to external triggering conditions such as changes in user needs, technology iteration or regulatory updates, the evaluation dimensions or weight system based on the customized evaluation task steps is dynamically adjusted. The evaluation dimensions or weight system includes a matrix of 10 primary dimensions and 54 secondary dimensions. Specifically, the system automatically generates an interactive evaluation report on the evaluation results display platform.

[0027] Multi-dimensional visualization: The report homepage displays the scores of each primary dimension using a radar chart, giving users a clear understanding of the balance of the cockpit's capabilities. It also uses a word cloud to show the most frequently occurring keywords in VoC and a bar chart to compare the satisfaction differences among different user profile groups. Drill-down traceability path: The interactivity of the report is reflected in its powerful traceability capabilities. For example, if a user sees a low score for "Interaction Satisfaction" on the radar chart, they can click on that dimension, and the interface will drill down to all the scores of the second-level dimensions under that first-level dimension. At this point, the user finds that the "Speech Recognition Accuracy" score is the lowest. Clicking again will further display a list of all related VoC entries processed by NLP, clearly listing the specific problems complained about by the user (such as "wake-up failure", "command misrecognition", etc.). The user can even click to play the original recording. This path, which drills down from the macro score layer by layer, and finally locates the original text of the specific user complaint, is the protected drill-down traceability path.

[0028] Furthermore, the method of the present invention may also include subsequent feedback loops and system evolution steps to ensure the long-term value of the evaluation and the continued viability of the method; Regarding feedback loop: The system can record the results of each evaluation. When the car manufacturer performs an OTA upgrade based on a certain evaluation report (for example, the report points out the voice wake-up problem), the system can invite the same batch of evaluation users to retest using the same evaluation task. By comparing the scores of "voice recognition accuracy" and changes in VoC content in the two evaluation reports, the real effect of this OTA upgrade can be quantitatively verified. Regarding system evolution: The system can integrate external triggers. For example, when it detects that the discussion of new technologies in the industry (such as in-vehicle AIGC) or user demand (analyzed by the KANO model) has increased significantly for several consecutive quarters, or when the country introduces new automotive data security regulations, the system will prompt the administrator to return to step S1 to review and adjust the evaluation dimensions (such as adding the "AIGC experience" dimension) and weighting system, thereby ensuring that the entire evaluation method can keep pace with the times.

[0029] Please see the appendix Figure 6 - Appendix Figure 12 A smart cockpit evaluation system based on multi-source data fusion includes: Task definition module: Used to define customized evaluation tasks for intelligent cockpit evaluation based on selected vehicle models, evaluation scenarios, and user profiles; Specifically, the task definition module is usually deployed on the backend server, providing an operation interface for administrators, and integrating the basic database and weight algorithm to execute all the functions of S1.

[0030] Data Acquisition Module: Used to respond to customized evaluation tasks by collecting multi-source data, including structured scoring data and unstructured user feedback. Data, and realize each unstructured user voice The data is bound one-to-one with its corresponding structured scoring data; Specifically, the data acquisition module will exist in the form of a client application (App or Web page), deployed on the device used by the testers, and will be responsible for executing all the functions of S2.

[0031] Fusion Processing Module: Used to fuse collected multi-source data, automatically analyzing unstructured user voice. The data is used to interpret the corresponding structured scoring data that has been bound to it, thereby generating comprehensive analysis results; Specifically, the fusion processing module is deployed on the backend server, integrating NLP models and data computing engines, and is responsible for executing all the functions of S3.

[0032] And a report generation module: used to generate smart cockpit evaluation reports based on comprehensive analysis results; Specifically, the report generation module is deployed on the backend server and provides a frontend display interface. It is responsible for executing all the functions of S4. These modules communicate and exchange data through network APIs, work together to complete all the technical solutions of this invention.

[0033] In summary, this invention provides a smart cockpit evaluation method and system based on multi-source data fusion. By defining customized evaluation tasks based on selected vehicle models, evaluation scenarios, and user profiles, and applying dynamic weights, it significantly improves the accuracy and relevance of the evaluation. Furthermore, it can automatically generate exclusive evaluation content and evaluation criteria based on the evaluation context, enabling the evaluation results to more realistically reflect the product's performance in specific scenarios. This avoids conclusion bias caused by a mismatch between evaluation content and actual needs, providing a scientific basis for product differentiation and precise optimization. Moreover, by providing the evaluation report with details ranging from overall evaluation scores to specific... The drill-down approach to data significantly enhances the interpretability and operability of the evaluation results. The interactive reports provided allow users to view them. This transparent and traceable analytical path ensures that every evaluation conclusion is verifiable, greatly enhancing the credibility of the reports.

[0034] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A smart cockpit evaluation method based on multi-source data fusion, characterized in that, Includes the following steps: S1. Based on the selected vehicle model, evaluation scenario, and user profile, define customized evaluation tasks for intelligent cockpit evaluation; S2. In response to the execution of the customized evaluation task, multi-source data is then collected, including structured scoring data and unstructured user feedback. The data, wherein each of the unstructured user voice data is bound to its corresponding structured scoring data; S3. The collected multi-source data is fused, and the fusion process is performed by automatically analyzing the unstructured user voice. The data is used to interpret the corresponding bound structured scoring data, thereby generating comprehensive analysis results; S4. Finally, based on the comprehensive analysis results, an intelligent cockpit evaluation report can be generated.

2. The intelligent cockpit evaluation method based on multi-source data fusion according to claim 1, characterized in that, The step of defining the customized evaluation task described in step S1 also includes: Based on the vehicle model, evaluation scenario, and user profile, dynamic weights are loaded from a preset weighting system and applied to the customized evaluation task to achieve adaptive adjustment of the evaluation focus.

3. The intelligent cockpit evaluation method based on multi-source data fusion according to claim 1, characterized in that, In step S2, the step of collecting multi-source data is described as follows: The structured scoring data consists of evaluation scores based on a Likert scale; the unstructured user voice data... The data consists of user reviews in voice or text format.

4. The intelligent cockpit evaluation method based on multi-source data fusion according to claim 3, characterized in that, The specific steps for binding are as follows: After collecting the structured scoring data for any evaluation item, the unstructured user voice for that evaluation item is immediately triggered. The data collection entry point is used to establish the structured scoring data and the unstructured user voice. One-to-one correspondence between data.

5. The intelligent cockpit evaluation method based on multi-source data fusion according to claim 1, characterized in that, In the fusion processing step described in step S3, the unstructured user voice is automatically analyzed. The data includes: Natural Language Processing (NLP) technology is used to process the unstructured user voice. The data is cleaned and automatically classified and clustered according to preset dimensions, scenarios, or emotional polarities.

6. The intelligent cockpit evaluation method based on multi-source data fusion according to claim 2, characterized in that, The step of generating the comprehensive analysis results in step S3 further includes: By combining the dynamic weights, the structured scoring data is weighted and calculated to obtain a quantitative comprehensive evaluation score.

7. The intelligent cockpit evaluation method based on multi-source data fusion according to claim 1, characterized in that, The step of generating the smart cockpit evaluation report described in step S4 also includes: The smart cockpit evaluation report provides a range of data from the overall evaluation score to the corresponding bound unstructured user voice data. Drill-down tracing of data paths.

8. The intelligent cockpit evaluation method based on multi-source data fusion according to claim 1, characterized in that, The method also includes the step of forming a feedback loop: Based on the issues revealed in the intelligent cockpit evaluation report, the intelligent cockpit was optimized, and after a preset period, the same evaluation group was invited to perform a retest to verify the optimization effect.

9. The intelligent cockpit evaluation method based on multi-source data fusion according to claim 1, characterized in that, This method also includes steps related to system evolution: In response to external triggering conditions such as changes in user needs, technological iterations, or regulatory updates, the evaluation dimensions or weighting system used in the customized evaluation task steps are dynamically adjusted. The evaluation dimensions or weighting system includes a matrix of 10 primary dimensions and 54 secondary dimensions.

10. An intelligent cockpit evaluation system based on multi-source data fusion, characterized in that, A smart cockpit evaluation method based on multi-source data fusion according to any one of claims 1-9 includes: Task definition module: Used to define customized evaluation tasks for intelligent cockpit evaluation based on selected vehicle models, evaluation scenarios, and user profiles; Data acquisition module: Used to collect multi-source data in response to the execution of the customized evaluation task, including structured scoring data and unstructured user feedback. Data, and implement each of the aforementioned unstructured user voices. The data is bound one-to-one with its corresponding structured scoring data; Fusion Processing Module: Used to fuse the collected multi-source data, and automatically analyze the unstructured user voice. The data is used to interpret the corresponding bound structured scoring data, thereby generating comprehensive analysis results; And a report generation module: used to generate a smart cockpit evaluation report based on the comprehensive analysis results.

Images