CN-121979610-A - RAG-based vehicle-to-machine function knowledge base management system and method for vehicle-mounted desktop generation

CN121979610ACN 121979610 ACN121979610 ACN 121979610ACN-121979610-A

Abstract

The invention relates to the technical field of vehicle-mounted intelligent cabins, and discloses a vehicle-mounted desktop generation-based vehicle-to-machine function knowledge base management system and method. The system comprises a vehicle function knowledge base, a vehicle-mounted HMI design specification knowledge base and a user behavior habit knowledge base, a dynamic updating module, a search output module and a search output module, wherein the dynamic updating module is used for triggering a search instruction according to the difference between new user behavior characteristics and historical portraits, the search output module is used for responding to the instruction, constructing a compound query based on the behavior characteristics, combining the three knowledge bases and searching, and outputting a corresponding function definition set, interaction specification constraints and related user behavior modes so as to drive a large model to generate or update a vehicle-mounted desktop application. According to the invention, through behavior driving and multisource knowledge fusion retrieval, individuation, precision and compliance generation of the vehicle-mounted desktop application are realized.

Inventors

ZHOU ENZE
HUANG XIN
YU WEN
PENG ZHAOXIN
LIANG JINGBO
JIN ZHENYU
YAN ZEKUN

Assignees

上海艾拉比智能科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260330

Claims (10)

1. A vehicle-to-machine function knowledge base management system based on RAG for vehicle-mounted desktop generation, comprising: The vehicle function knowledge base is used for storing definition data of vehicle functions; the vehicle-mounted HMI design specification knowledge base is used for storing interface design constraints; The user behavior habit knowledge base is used for storing a user behavior feature set taking a statistical period as a unit; The dynamic updating module is used for receiving new user behavior characteristic data, calculating the difference between the new user behavior characteristic data and the historical behavior portrait, and triggering a retrieval instruction when the difference exceeds a preset generation threshold value; and the retrieval output module is used for responding to the retrieval instruction, constructing a compound query based on the new user behavior feature set and the historical behaviors to obtain a query result, and jointly retrieving the vehicle function knowledge base, the vehicle-mounted HMI design specification knowledge base and the user behavior habit knowledge base based on the compound query to output a function definition set, interaction specification constraints and related user behavior modes for the large model to generate or update the vehicle-mounted desktop application.
2. The RAG-based vehicle-mounted desktop generation system of claim 1, wherein the user behavior habit knowledge base is configured to archive user behavior data in layers according to user identifiers and statistical periods, the layers including recent hot data for real-time retrieval, mid-term temperature data for pull-on-demand, and cold data for long-term trend analysis.
3. The RAG-based vehicle-to-machine function knowledge base management system for vehicle desktop generation of claim 1, wherein the dynamic update module is specifically configured to: receiving Zhou Du a set of user behavior characteristics; calculating the difference between the current feature set and the long, medium and short term behavior portraits in the knowledge base; based on the differences, the behavior weights, the high frequency function rankings, and the behavior confidence levels are updated.
4. The RAG-based vehicle-mounted desktop generation system of claim 3, wherein the dynamic update module triggers the search instruction only when the discrepancy exceeds an application generation threshold, and only updates the user behavior habit knowledge base and records a change history if the discrepancy does not exceed the application generation threshold.
5. The RAG-based vehicle-mounted desktop generation system of claim 1, wherein the composite query constructed by the search output module includes information including a summary of current behavior changes of the user, long-term behavior images, vehicle model information, and vehicle-mounted system software version information.
6. The RAG-based vehicle-to-machine function knowledge base management system for vehicle desktop generation of claim 1, wherein the search output module obtains the query result through a joint search mechanism comprising structured exact match search for function identification or specification entry and vector similarity search for behavior pattern or usage scenario.
7. The system for RAG-based vehicle-to-machine function knowledge base management for vehicle desktop generation of claim 1, further comprising a knowledge base version management module for: In response to the completion of OTA updating by the vehicle-mounted system, synchronously updating the vehicle function knowledge base and the vehicle-mounted HMI design specification knowledge base, and establishing a version mapping relation; triggering compatibility evaluation and processing of historical behavior characteristics related to updated contents in the user behavior habit knowledge base.
8. The vehicle set function knowledge base management system based on RAG for vehicle desktop generation as recited in claim 1, wherein the content output by the retrieval output module based on the query result comprises a set of available function definitions, interaction and interface design specification constraints that force validation, and a long-term user behavior pattern that is highly relevant to current behavior.
9. The system of claim 1, wherein the interface design constraints stored in the on-board HMI design specification knowledge base are defined by a scenario and a function class, and the content of the interface design constraints comprises at least one of a screen adaptation specification, an operation safety specification, a function interaction specification, and a font component specification.
10. A vehicle-mounted desktop generation-based vehicle-mounted function knowledge base management method, characterized by being used for realizing the vehicle-mounted desktop generation-based vehicle-mounted function knowledge base management system according to any one of claims 1 to 9, the method comprising the following steps: maintaining a vehicle function knowledge base, a vehicle-mounted HMI design specification knowledge base and a user behavior habit knowledge base; receiving new user behavior feature data; Calculating the difference between the new user behavior characteristic data and the historical behavior portraits; Triggering a search instruction when the difference exceeds a preset generation threshold value; The search instruction specifically comprises constructing a compound query based on the new user behavior feature set and the historical behavior portraits; According to the compound query, the vehicle function knowledge base, the vehicle-mounted HMI design specification knowledge base and the user behavior habit knowledge base are searched in a combined mode to obtain a corresponding function definition set, interaction specification constraint and related user behavior modes; And outputting the function definition set, the interaction specification constraint and the related user behavior mode for generating or updating the vehicle-mounted desktop application by the large model.

Description

RAG-based vehicle-to-machine function knowledge base management system and method for vehicle-mounted desktop generation Technical Field The invention relates to the technical field of vehicle-mounted intelligent cabins and artificial intelligence, in particular to a vehicle-mounted desktop generation-based vehicle-to-machine function knowledge base management system and method. Background With the rapid development of automobile intellectualization, the functions of an in-vehicle infotainment system (IVI) are increasingly abundant, and how to provide personalized, efficient and safe in-vehicle desktop interaction experience for different users becomes an important subject. Currently, most of application layout and function entry of a vehicle-mounted desktop are static preset or recommended based on simple rules, and it is difficult to deeply adapt to different use habits of different users. Meanwhile, when the personalized interface is generated or recommended, the presented functional control needs to be ensured to accord with the actual functional definition of the vehicle, and the interactive design of the functional control needs to strictly follow the safety and consistency specifications of the rule level of the daemon (such as touch control hot zone, driving anti-interference and the like). The existing scheme usually processes user behavior analysis, function management and design specifications in isolation, lacks a multi-dimensional knowledge capable of cooperatively utilizing vehicle function knowledge, interactive design specification, user behavior habit and the like, and can ensure an automatic knowledge management and application generation system with accurate and compliant output results, so that the generated desktop has limited individuation degree and risk of mismatch with vehicle functions or violating the design specification. Disclosure of Invention In view of the above-mentioned shortcomings in the technical field of vehicle-mounted intelligent cabins, the invention provides a vehicle-mounted desktop generation-based vehicle-to-machine function knowledge base management system and method, which can achieve the effects of deep personalized adaptation, function matching and design compliance assurance, and automatic driving interface generation. In order to achieve the above purpose, the embodiment of the present invention adopts the following technical scheme: a RAG-based vehicle-to-machine function knowledge base management system for vehicle-mounted desktop generation, comprising: The vehicle function knowledge base is used for storing definition data of vehicle functions; the vehicle-mounted HMI design specification knowledge base is used for storing interface design constraints; The user behavior habit knowledge base is used for storing a user behavior feature set taking a statistical period as a unit; The dynamic updating module is used for receiving new user behavior characteristic data, calculating the difference between the new user behavior characteristic data and the historical behavior portrait, and triggering a retrieval instruction when the difference exceeds a preset generation threshold value; and the retrieval output module is used for responding to the retrieval instruction, constructing a compound query based on the new user behavior feature set and the historical behaviors to obtain a query result, and jointly retrieving the vehicle function knowledge base, the vehicle-mounted HMI design specification knowledge base and the user behavior habit knowledge base based on the compound query to output a function definition set, interaction specification constraints and related user behavior modes for the large model to generate or update the vehicle-mounted desktop application. In accordance with one aspect of the invention, the user behavior habit knowledge base is configured to archive user behavior data in layers according to user identifiers and statistical periods, the layers including recent hot data for retrieval in real-time, mid-term temperature data for pull-on-demand, and cold data for long-term trend analysis. According to one aspect of the invention, the dynamic update module is specifically configured to: receiving Zhou Du a set of user behavior characteristics; calculating the difference between the current feature set and the long, medium and short term behavior portraits in the knowledge base; based on the differences, the behavior weights, the high frequency function rankings, and the behavior confidence levels are updated. According to one aspect of the invention, the dynamic update module triggers the search instruction only when the difference exceeds an application generation threshold, and only updates the user behavior habit knowledge base and records a change history if the difference does not exceed the application generation threshold. According to one aspect of the invention, the compound query constructed by the retrieval output module compri