A vehicle service control method and system based on intent analysis and constraint optimization

Abstract

The present application relates to the technical field of intelligent vehicle service control, and particularly relates to a vehicle service control method and system based on intention analysis and constraint optimization, the vehicle service control method based on intention analysis and constraint optimization comprising receiving natural language information input by a user; performing intention analysis on the natural language information to obtain a structured user intention vector; determining a candidate vehicle service set matched with the user intention based on an intention-service mapping knowledge graph according to the user intention vector; performing service conflict detection on the candidate vehicle service set to obtain conflict type, conflict object and conflict influence information between vehicle services in the candidate vehicle service set; constructing a constraint optimization model and solving the model according to the result of the service conflict detection to generate a vehicle service control scheme and then executing corresponding vehicle service control; and controlling a target vehicle service according to the vehicle service control scheme.

Description

Technical Field

[0001] This invention relates to the field of intelligent vehicle service control technology, specifically to a vehicle service control method and system based on intent parsing and constraint optimization. Background Technology

[0002] With the rapid development of intelligent connected vehicle technology, vehicle functions are becoming increasingly rich, and users' demand for personalized and scenario-based services has significantly increased. Existing vehicle control systems generally adopt menu-based or parametric configuration interfaces, requiring users to manually select and adjust from numerous engineering options. This operation is complex and lacks semantic understanding capabilities, making it difficult to accurately translate the multidimensional needs expressed by users in natural language into executable service configurations.

[0003] While existing technologies have improved the mapping ability from natural language to vehicle functions to some extent, they still have significant limitations: On the one hand, their intent parsing often focuses on single function triggers or simple label extraction, failing to construct a structured user intent vector that covers multiple dimensions such as driving experience, riding experience, and emotional satisfaction, making it difficult to comprehensively depict the user's comprehensive needs in complex vehicle usage scenarios; on the other hand, existing systems generally ignore the implicit constraints between vehicle services in multiple dimensions, such as safety, physical, resource, and logical constraints. When a user activates multiple services simultaneously (such as "ultra-low power mode" and "high-performance ambient lighting + seat massage"), functional abnormalities, decreased experience, or even safety hazards may occur due to computing power competition, power overload, hardware mutual exclusion, or execution timing conflicts. Because these constraints are highly coupled and not explicitly presented, neither users nor sales personnel can predict conflicts during the configuration phase, making it difficult for personalized configurations to balance intent satisfaction and system security. Summary of the Invention

[0004] To address the aforementioned technical issues, this invention provides a vehicle service control method and system based on intent parsing and constraint optimization. This method deeply integrates natural language intent parsing, knowledge graph-driven service matching, and multi-dimensional constraint-based optimization decision-making mechanisms to provide users with highly available, secure, and high-experience intelligent vehicle services.

[0005] To achieve the above objectives, the present invention adopts the following technical solution.

[0006] The first aspect of this invention provides a vehicle service control method based on intent parsing and constraint optimization, comprising: Receive natural language information input by the user, which is used to describe the user's car usage scenario, functional preferences and / or emotional needs; The natural language information is parsed to obtain a structured user intent vector; Based on the intent-service mapping knowledge graph, a set of candidate vehicle services matching the user intent is determined according to the user intent vector; Based on the multi-dimensional constraint graph between services, service conflict detection is performed on the candidate vehicle service set to obtain the conflict type, conflict object and conflict impact information between vehicle services in the candidate vehicle service set. Based on the service conflict detection results, a constraint optimization model is constructed and solved, and a vehicle service control scheme is generated and the corresponding vehicle service control is executed. According to the vehicle service control scheme, the target vehicle service is enabled, disabled, parameter adjusted, execution timing adjusted, resource scheduled, working mode switched and / or linked control is performed.

[0007] As an optional implementation, the step of receiving natural language information input by the user, wherein the natural language information is used to describe the user's car usage scenario, functional preferences, and / or emotional needs, includes: Receive natural language information input by the user via voice or text; Identify at least one vehicle usage scenario from the natural language information, and extract the time conditions, travel purpose, user role, functional preferences and / or emotional needs corresponding to each vehicle usage scenario.

[0008] As an optional implementation, the step of parsing the natural language information to obtain a structured user intent vector includes: Using a natural language understanding model trained or fine-tuned with vehicle-domain corpus, we analyze the functional preferences and / or emotional demands corresponding to each vehicle usage scenario, and generate user intent vectors for each scenario.

[0009] As an optional implementation, the step of parsing the natural language information to obtain a structured user intent vector includes: Generate user intent vectors that include driving experience, riding experience, and emotional satisfaction dimensions; The driving experience dimension is used to characterize the user's preference intensity for power, handling, economy and / or intelligent driving; the riding experience dimension is used to characterize the user's preference intensity for comfort, entertainment and / or health; and the emotional satisfaction dimension is used to characterize the user's preference intensity for personality expression, social sharing and / or emotional resonance.

[0010] As an optional implementation, the step of determining a set of candidate vehicle services matching the user intent based on the intent-service mapping knowledge graph and the user intent vector includes: Based on the intent-service mapping knowledge graph including intent nodes, vehicle service nodes, and associated edges, the vehicle service node corresponding to the user intent vector is determined. The associated edge is used to characterize the association strength and / or contribution direction between the intent node and the vehicle service node. Each vehicle service node is configured with a service contribution vector, which is used to characterize the degree of contribution of the corresponding vehicle service to each dimension of the user intent vector.

[0011] As an optional implementation, the step of determining a set of candidate vehicle services matching the user intent based on the intent-service mapping knowledge graph and the user intent vector includes: Calculate the matching degree between the user intent vector and the service contribution vector corresponding to each vehicle service node; Vehicle services that meet the preset matching criteria are identified as candidate vehicle services, and the candidate vehicle services form the candidate vehicle service set.

[0012] As an optional implementation, the step of performing service conflict detection on the candidate vehicle service set based on a multi-dimensional constraint graph between services to obtain conflict types, conflict objects, and conflict impact information between vehicle services in the candidate vehicle service set includes: Based on a multi-dimensional constraint graph between services with vehicle services as nodes and service constraints as edges, the system detects whether there are security constraints, physical constraints, resource constraints and / or logical constraints between vehicle services in the candidate vehicle service set. When the inter-service constraint relationship is detected to be triggered, it is determined that a service conflict exists, and the corresponding conflict type, conflict object, and conflict impact information are determined.

[0013] As an optional implementation, the step of performing service conflict detection on the candidate vehicle service set based on a multi-dimensional constraint graph between services to obtain conflict types, conflict objects, and conflict impact information between vehicle services in the candidate vehicle service set includes: When two or more candidate vehicle services have conflicts such as data acquisition frequency conflict, communication bandwidth competition, computing power resource competition, power resource competition, hardware condition limitations, functional dependency failure, functional mutual exclusion, execution order contradiction, or impact on driving safety, it is determined that there is a service conflict between the two or more candidate vehicle services.

[0014] As an optional implementation, the step of constructing and solving a constraint optimization model based on the service conflict detection results, generating a vehicle service control scheme, and then executing the corresponding vehicle service control includes: The matching degree between the three-dimensional intent vector of user driving experience, riding experience, and emotional satisfaction and the total contribution vector of the candidate service set is used as the optimization objective, and the safety constraints, physical constraints, resource constraints, and logical constraints in the multi-dimensional constraint graph between services are used as constraints. When a feasible solution exists that satisfies all constraints, a vehicle service control scheme containing the optimal service selection result and conflict resolution strategy is generated; when no fully feasible solution exists, at least two differentiated experience package schemes are generated, with each scheme corresponding to the quantified gains and losses of intent satisfaction in three dimensions.

[0015] As an optional implementation, the step of enabling, disabling, adjusting parameters, adjusting execution timing, scheduling resources, switching working modes, and / or coordinating control of the target vehicle service according to the vehicle service control scheme includes: When there are multiple vehicle service control schemes that meet the constraints, output the driving experience dimension, riding experience dimension, emotional satisfaction dimension and / or safety compliance impact information corresponding to each vehicle service control scheme; The target vehicle service control scheme is determined based on user selection or preset priority. The system executes the target vehicle service control scheme based on time conditions, vehicle status, user instructions, and / or historical usage behavior, and updates the user intent vector, the intent-service mapping knowledge graph, and / or the multi-dimensional constraint graph between services based on the user's confirmation, modification, or cancellation of the execution result.

[0016] A second aspect of the present invention provides a vehicle service control system based on intent parsing and constraint optimization, comprising: The receiving unit is at least used to receive natural language information input by the user, wherein the natural language information is used to describe the user's vehicle usage scenario, functional preferences and / or emotional needs; The parsing unit is at least used to perform intent parsing on the natural language information to obtain a structured user intent vector; The mapping unit is at least used to determine a set of candidate vehicle services that match the user intent based on the user intent vector, according to the intent-service mapping knowledge graph. The detection unit is at least used to perform service conflict detection on the candidate vehicle service set based on the multi-dimensional constraint graph between services, and to obtain the conflict type, conflict object and conflict impact information between vehicle services in the candidate vehicle service set. The control unit is at least used to construct and solve a constraint optimization model based on the results of the service conflict detection to generate a vehicle service control scheme. The execution unit is at least used to enable, disable, adjust parameters, adjust execution timing, schedule resources, switch working modes, and / or control linkage of the target vehicle service in accordance with the vehicle service control scheme.

[0017] A third aspect of the present invention provides an electronic device, comprising: at least one processor; and at least one memory communicatively connected to the processor, wherein: the memory stores program instructions executable by the processor, and the processor, by invoking the program instructions, can perform the steps of the method described in the first aspect of the present invention.

[0018] A fourth aspect of the present invention provides a readable storage medium storing a computer program that is executed by a processor as described in the first aspect of the present invention.

[0019] Compared with the prior art, the present invention has the following advantages: 1. This invention enables users to express their vehicle usage needs in natural language without having to operate complex engineering parameters, thus significantly reducing the learning cost and operational threshold of personalized vehicle configuration; 2. This invention, by constructing a three-dimensional user intent vector and a dual knowledge graph architecture, can accurately quantify and match users' multi-dimensional needs, resulting in higher service matching accuracy; 3. This invention comprehensively detects multi-dimensional hidden conflicts between vehicle services and adopts exclusive resolution strategies for different conflict types, thereby avoiding functional abnormalities, resource contention, and driving safety hazards from the source. 4. This invention supports automatic switching between multiple vehicle usage scenarios and closed-loop iteration of user behavior, continuously optimizing intent recognition and knowledge graph content, adapting to the long-term usage habits of different users, and improving the effect of personalized services. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart illustrating a vehicle service control method based on intent parsing and constraint optimization according to a specific embodiment of the present invention.

[0022] Figure 2 This is a flowchart illustrating another vehicle service control method based on intent parsing and constraint optimization according to a specific embodiment of the present invention.

[0023] Figure 3 This is a block diagram of another vehicle service control system based on intent parsing and constraint optimization according to a specific embodiment of the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Furthermore, it should be understood that the specific embodiments described herein are only for illustration and explanation of this application and are not intended to limit this application.

[0025] It should be noted that the order of description of the following embodiments is not intended to limit the preferred order of the embodiments of this application. Furthermore, the descriptions of each embodiment in the following embodiments have their own emphasis; for parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0026] like Figure 1 and Figure 2 As shown, the first aspect of the present invention provides a vehicle service control method based on intent parsing and constraint optimization, including the following:

[0027] Step S100: Receive natural language information input by the user, wherein the natural language information is used to describe the user's car usage scenario, functional preferences and / or emotional needs.

[0028] In one embodiment of the present invention, receiving natural language information input by the user includes: Receive natural language information input by the user via voice or text; Identify at least one vehicle usage scenario from the natural language information, and extract the time conditions, travel purpose, user role, functional preferences and / or emotional needs corresponding to each vehicle usage scenario.

[0029] For example, the system receives natural language input from a user via voice: "I drive this car to work from Monday to Friday. As a working person, I need to clock in on time and can't be late. I also want the commute to be less boring. On weekends, I want to take my wife and kids out for some fun and relaxation. I also want to avoid driving too tiring on weekends." This natural language input identifies two usage scenarios: a weekday commuting scenario and a weekend family outing scenario. The weekday commuting scenario corresponds to the morning and evening rush hours from Monday to Friday. The purpose of the trip is to arrive at work on time. The user role is a solo commuter, and the functional preferences include efficient travel and alleviating the boredom of commuting. The emotional appeal is the importance of time certainty. The weekend family outing scenario corresponds to all day on Saturday and Sunday. The purpose of the trip is family leisure and entertainment. The user role is a family driver, and the functional preferences include reducing driving burden and improving passenger comfort. The emotional appeal is creating a warm family atmosphere.

[0030] Step S200: Perform intent parsing on the natural language information to obtain a structured user intent vector.

[0031] Specifically, a natural language understanding model trained or fine-tuned with vehicle-related corpus is used to analyze the functional preferences and / or emotional demands corresponding to each vehicle usage scenario, and generate user intent vectors corresponding to each vehicle usage scenario.

[0032] Specifically, the user intent vector is a user intent vector that includes a driving experience dimension, a riding experience dimension, and an emotional satisfaction dimension; wherein, the driving experience dimension is used to characterize the user's preference intensity for power, handling, economy, and / or intelligent driving, the riding experience dimension is used to characterize the user's preference intensity for comfort, entertainment, and / or health, and the emotional satisfaction dimension is used to characterize the user's preference intensity for personality expression, social sharing, and / or emotional resonance.

[0033] Preferably, the natural language understanding model is a natural language understanding model fine-tuned with vehicle domain corpus. The basic pre-trained model is selected from at least one of the large pre-trained models such as Chinese BERT, which needs to be able to recognize typical car use scenarios (such as "family trip" and "business reception"), emotional appeals (such as "romantic", "cool" and "entertainment") and functional preferences (such as "energy saving", "quiet" and "performance").

[0034] Specifically, the vehicle domain corpus includes structured labeled triplet samples. Each sample is labeled with three categories of tags: usage scenario tags, user emotional appeal tags, and vehicle function preference tags. The usage scenario tags cover eight standard scenarios: short-distance commuting, family travel, long-distance self-driving, business reception, outdoor camping, picking up and dropping off children, track driving, and highway travel, supporting multi-scenario overlay recognition. The emotional appeal tags are divided into eight emotional dimensions: comfort, romance, coolness, home, sports, tranquility, entertainment, and stability, supporting multi-emotion weight output for a single sentence. The vehicle function preference tags cover range saving, in-car tranquility, power performance, and other aspects. The system comprises eight core vehicle functional dimensions: intelligent audio-visual system, spacious interior, fast charging, off-road capability, and comfortable seating. The fine-tuning of the vehicle domain corpus includes: freezing the bottom encoding layer of the pre-trained model and fine-tuning only the top classifier head and the domain-adaptive attention layer; synchronously training scene, sentiment, and preference classification tasks using a multi-label cross-entropy loss function; and the natural language understanding model's inference process includes: receiving user natural language input text → text cleaning, segmentation, and encoding → model encoding and outputting semantic vectors → multi-branch classification heads outputting scene confidence, sentiment weight, and functional preference scores respectively → filtering valid recognition results according to preset thresholds and outputting a ternary feature set.

[0035] This embodiment fully discloses the entire process of training, annotation, and inference using the BERT fine-tuning scheme as an example. Technicians in the relevant field can fully reproduce the model and implement the three types of semantic recognition functions based on this section.

[0036] Construction of a corpus specifically for fine-tuning in the vehicle field.

[0037] A dedicated corpus of labeled texts representing vehicle user needs was constructed, with a total labeled sample size of ≥15,000. Sample sources included real user reviews, in-vehicle interaction dialogues, car purchase consultation scripts, and texts representing usage scenarios. All samples were manually labeled with three tags simultaneously, with each text corresponding to a three-element tag: scenario tag, emotional appeal tag, and functional preference tag. Multi-tag labeling of single texts was supported.

[0038] The complete set of car usage scenario tags includes: short commuting, family trips, long-distance self-driving, business reception, outdoor camping, picking up and dropping off children, track driving, highway long-distance driving, city commuting, and mountain off-roading.

[0039] The complete set of emotional appeal tags includes: romantic atmosphere, cool sporty style, homey comfort, quiet leisure, audio-visual entertainment, stable business, warm family atmosphere, and aggressive driving.

[0040] The complete set of vehicle function preference tags includes: energy saving and low energy consumption, quiet interior sound insulation, strong acceleration performance, large audio-visual entertainment, spacious passenger space, fast charging, high off-road capability, soft seats, and intelligent driver assistance.

[0041] The preprocessing rules for the corpus include: removing garbled characters, meaningless words, and text irrelevant to advertisements; standardizing vehicle terminology (range, quietness, energy consumption per 100 kilometers, 0-100 km / h acceleration, etc.); standardizing sentence segmentation for short sentences and colloquial vehicle-to-machine dialogues; and uniformly truncating text to the maximum preset length, such as 64 tokens, and filling in placeholders if the length is insufficient.

[0042] Fine-tuning the complete training process in the vehicle domain.

[0043] We load publicly available pre-trained BERT-base Chinese weights, freeze the first 10 encoding Transformer layers, and only open the last 2 Transformer layers plus a new three-branch multi-label classification head for vehicle semantic recognition, reducing the computational cost of fine-tuning while retaining general language understanding capabilities; we divide the corpus into a training set:validation set:test set ratio of 8:1:1; we use AdamW as the optimizer, with a base learning rate of 2e-5 and a linear decay learning rate strategy, and a batch size of 16; we simultaneously calculate scene classification loss, sentiment classification loss, and function preference classification loss, with the total loss being a weighted sum of the three losses in a weight ratio of 1:1:1, and the loss function using multi-label binary cross-entropy (BCEWithLogitsLoss); we iterate through 20 training epochs, checking the recognition accuracy on the validation set every 2 epochs; when the scene recognition accuracy is ≥92%, sentiment recognition accuracy is ≥90%, and function preference recognition accuracy is ≥93%, we stop fine-tuning and save the optimal model weight file; after fine-tuning, we perform lightweight processing on the model: we quantize and compress the encoding layer weights to adapt to the real-time inference requirements of automotive embedded chips.

[0044] Real-time inference and recognition logic of the model.

[0045] The finely tuned vehicle NLU model is deployed on the vehicle's infotainment domain controller. The complete inference steps are as follows: Input layer: Receives natural language text transcribed from user speech / manual input. Example input texts include "Taking the whole family on a road trip this weekend, I want a quiet car with long battery life" and "Using it to receive clients, the car needs to be stable and good-looking, with a high-end audio-visual system." Encoding layer: Converts the cleaned text into a sequence of input tokens for the model. The finely tuned BERT encoding outputs a 768-dimensional semantic vector. Parallel computation of multi-branch recognition heads: Outputs confidence scores for all usage scenario labels, taking labels with a confidence score > 0.6 as the recognized usage scenario; Outputs weight values ​​for each sentiment dimension, with the top two weights representing the core sentiment appeal; Outputs vehicle function preference scores, with a score threshold of 0.5 used to filter user-focused functions; Outputs a structured ternary recognition result: {set of usage scenarios, set of core sentiment appeals, set of user function preferences}, for use by downstream vehicle configuration recommendation and infotainment system personalization adjustment modules.

[0046] Specific identification implementation cases.

[0047] Case 1: User input text "The whole family goes on a road trip during holidays, hoping for a quiet car with low power consumption"; Model recognition output: Car usage scenario = [family trip, long-distance road trip]; Emotional needs = [comfortable at home, quiet and relaxing]; Functional preferences = [quiet, power saving and low energy consumption].

[0048] Case 2: User input text "Used for picking up and dropping off clients, looks dignified and stable, and has good audio-visual effects"; Model recognition output: Usage scenario = [Business reception]; Emotional needs = [Stable business, audio-visual entertainment]; Functional preferences = [Intelligent audio-visual, spacious seating].

[0049] Case 3: User input text "Weekend mountain road self-driving, want fast acceleration and good passability"; Model recognition output: Usage scenario = [mountain off-road, short track driving]; Emotional appeal = [cool and sporty style]; Functional preference = [strong acceleration performance, high off-road passability].

[0050] For example, in conjunction with the example scenario described in step S100, the user intent vector corresponding to the weekday commuting scenario includes a driving experience dimension, a riding experience dimension, and an emotional satisfaction dimension, with specific values ​​of [0.9, 0.7, 0.5]. The driving experience dimension represents the user's strong preference for punctual and efficient travel, the riding experience dimension represents the user's need for entertainment content during the commute, and the emotional satisfaction dimension represents the user's moderate emotional expression needs as a commuting tool user. The user intent vector corresponding to the weekend family travel scenario is [0.8, 0.9, 0.8]. The driving experience dimension represents the user's preference for easy and effortless driving, the riding experience dimension represents the user's high demand for family comfort and children's entertainment, and the emotional satisfaction dimension represents the user's high demand for family emotional resonance.

[0051] Those skilled in the art can set specific parameters according to the actual situation.

[0052] Step S300: Based on the intent-service mapping knowledge graph, determine the set of candidate vehicle services that match the user intent according to the user intent vector.

[0053] Specifically, the step of determining a set of candidate vehicle services matching the user intent based on the intent-service mapping knowledge graph and the user intent vector includes: Based on the intent-service mapping knowledge graph including intent nodes, vehicle service nodes, and associated edges, the vehicle service node corresponding to the user intent vector is determined; wherein, the associated edges are used to characterize the association strength and / or contribution direction between the intent node and the vehicle service node, and each vehicle service node is configured with a service contribution vector, which is used to characterize the degree of contribution of the corresponding vehicle service to each dimension of the user intent vector.

[0054] Furthermore, the matching degree between the user intent vector and the service contribution vector corresponding to each vehicle service node is calculated; vehicle services whose matching degree meets the preset conditions are identified as candidate vehicle services, and the candidate vehicle services form the candidate vehicle service set.

[0055] Specifically, this intent-service mapping knowledge graph includes intent nodes, vehicle service nodes, and association edges representing the strength and direction of association. Each vehicle service node is configured with a service contribution vector, which represents the degree of contribution of the service to each dimension of the user intent vector. For the user intent vector [0.9, 0.7, 0.5] in the weekday commuting scenario, the matching degree between it and the service contribution vectors of each vehicle service node is calculated. Services with a matching degree greater than 0.75 are identified as candidate services, forming a candidate vehicle service set, including real-time traffic intelligent navigation (matching degree...). For the user intent vector [0.8, 0.9, 0.8] for the weekend family travel scenario, the matching degree was also calculated. Services with a matching degree greater than 0.75 were selected to form a candidate vehicle service set, including full-speed adaptive cruise control (matching degree 0.81), rear child entertainment system (matching degree 0.89), comfort suspension mode (matching degree 0.80), warm ambient lighting (matching degree 0.83), and family travel mode (matching degree 0.75).

[0056] This invention relies on an intent-service mapping knowledge graph to store the relationship between intents and vehicle services. It quantifies the degree of adaptation through vector matching, accurately filters suitable services, and balances the interpretability of matching with the efficiency of filtering. It can quickly output candidate services to adapt to different car use scenarios.

[0057] Step S400: Based on the multi-dimensional constraint graph between services, perform service conflict detection on the candidate vehicle service set to obtain the conflict type, conflict object and conflict impact information between vehicle services in the candidate vehicle service set.

[0058] Specifically, in step S400, based on the multi-dimensional constraint graph between services with vehicle services as nodes and service constraint relationships as edges, it is detected whether there are security constraints, physical constraints, resource constraints and / or logical constraints between vehicle services in the candidate vehicle service set; when the service constraint relationship is detected to be triggered, it is determined that there is a service conflict, and the corresponding conflict type, conflict object and conflict impact information are determined.

[0059] Furthermore, when two or more candidate vehicle services have conflicts such as data acquisition frequency conflict, communication bandwidth competition, computing power resource competition, power resource competition, hardware condition limitations, functional dependency failure, functional mutual exclusion, execution order contradiction, or impact on driving safety, it is determined that there is a service conflict between the two or more candidate vehicle services.

[0060] Specifically, this multi-dimensional constraint graph between services uses vehicle services as nodes and inter-service constraint relationships as edges. The constraint types include safety constraints, physical constraints, resource constraints, and logical constraints. Conflict detection was performed on the candidate service set for weekday commuting, revealing communication bandwidth contention and data collection frequency conflicts between real-time traffic intelligent navigation and driving efficiency analysis reports. This is a resource constraint conflict, specifically involving the shared high-frequency GPS and network data channels. The impact of this conflict may lead to navigation response delays or missing data in the analysis reports. Conflict detection was performed on the candidate service set for weekend family outings, revealing a safety constraint conflict between ambient lighting and full-speed adaptive cruise control. The conflict involves the interference of colored ambient lighting on nighttime dashboard visual perception, potentially reducing the driver's ability to recognize critical driving information.

[0061] This invention relies on a multi-dimensional constraint graph to comprehensively detect resource and security conflicts between vehicle services, identify conflict types, objects and impacts, and proactively avoid functional abnormalities and driving risks, thereby ensuring the stable operation of multiple in-vehicle services and driving safety.

[0062] Step S500: Based on the service conflict detection results, construct and solve the constraint optimization model, generate a vehicle service control scheme, and then execute the corresponding vehicle service control.

[0063] Furthermore, the construction of the constraint optimization model includes: using the matching degree between the three-dimensional intent vector of user driving experience, riding experience, and emotional satisfaction and the total contribution vector of the candidate service set as the optimization objective, and using the security constraints, physical constraints, resource constraints, and logical constraints in the multi-dimensional constraint graph between services as the constraint conditions.

[0064] Furthermore, solving the constrained optimization model includes: when there is a feasible solution that satisfies all constraints, generating a vehicle service control scheme that includes the optimal service selection result and conflict resolution strategy; when there is no fully feasible solution, generating at least two differentiated experience package schemes, each scheme corresponding to the quantified gains and losses of intent satisfaction in three dimensions.

[0065] Furthermore, when resource constraint conflicts are detected among candidate services, the corresponding conflict resolution strategies include: assigning differentiated operating priorities to services with resource competition, retaining the original operating parameters for high-priority services, adjusting resource usage parameters for low-priority services, and supplementing missing data during system idle periods.

[0066] Furthermore, when a conflict of completely mutually exclusive security constraints is detected between candidate services, the corresponding conflict resolution strategy includes: setting service linkage trigger rules, and automatically switching the conflicting low-security-priority service to the preset security operation mode when the high-security-level service is enabled.

[0067] Furthermore, when a non-mutually exclusive conflict of parameter threshold tension is detected between candidate services, the corresponding conflict resolution strategies include: generating a fusion composite service mode while retaining the core functional parameters of both conflicting services; adding a new operation status monitoring trigger rule, and when a preset risk trigger condition is detected, progressively adjusting the operation parameters of the conflicting services to a safe threshold range, and simultaneously outputting the status prompt information of the current mode.

[0068] Furthermore, when generating conflict resolution solutions, multi-dimensional impact quantification information corresponding to each solution is generated simultaneously, including the change value of user intent satisfaction in each dimension and the change value of security compliance level, and displayed to the user.

[0069] Furthermore, it receives user selection instructions for multiple conflict resolution schemes and solidifies the control strategy corresponding to the user-selected scheme into the final vehicle service control scheme.

[0070] For example, in weekday commuting scenarios with resource constraint conflicts: a conflict resolution strategy is adopted by adjusting the data collection frequency: real-time traffic intelligent navigation maintains a high-frequency data mode, while the driving efficiency analysis report switches to a low-frequency data mode and supplements data collection during off-peak hours at night; in weekend family outing scenarios with completely mutually exclusive safety constraint conflicts: a linkage control strategy is adopted: when full-speed adaptive cruise control is activated, the ambient lighting automatically switches to a safety mode, i.e., monochrome, low-brightness display, to avoid interfering with the driver's vision; at the same time, the impact information of the two conflict resolution schemes is displayed to the user, where scheme one reduces the emotional satisfaction dimension score but increases safety compliance. Option 2 maintains the emotional satisfaction score but reduces safety compliance; for performance driving scenarios with non-mutually exclusive conflicts in parameter threshold tension: when a parameter threshold tension conflict is detected between the sport performance mode and the high-level safety assistance mode, a fusion safety performance composite mode is generated: the aggressive power response and steering feel parameters of the sport performance mode are retained, while new rules for monitoring driver attention and road condition complexity are added; when the risk triggering conditions of driver distraction or road condition complexity are detected, the conservative intervention threshold of the electronic stability system is temporarily and gradually restored, and a voice output prompt is given: "Performance mode is activated, and the safety assistance system remains highly vigilant."

[0071] Step S600: According to the vehicle service control scheme, enable, disable, adjust parameters, adjust execution timing, schedule resources, switch working modes and / or control linkage of the target vehicle service.

[0072] Specifically, when there are multiple vehicle service control schemes that meet the constraints, output the driving experience dimension, riding experience dimension, emotional satisfaction dimension and / or safety compliance impact information corresponding to each vehicle service control scheme; The target vehicle service control scheme is determined based on user selection or preset priority. The system executes the target vehicle service control scheme based on time conditions, vehicle status, user instructions, and / or historical usage behavior, and updates the user intent vector, the intent-service mapping knowledge graph, and / or the multi-dimensional constraint graph between services based on the user's confirmation, modification, or cancellation of the execution result.

[0073] For example, the weekday commuting plan automatically takes effect from 07:00-09:00 and 17:00-19:00 from Monday to Friday, enabling real-time traffic intelligent navigation (high-frequency mode), commuting mode (automatic congestion avoidance), and driving efficiency analysis report (low-frequency mode); the weekend family travel plan takes effect all day on Saturday and Sunday, enabling full-speed adaptive cruise control (comfort following mode), rear child entertainment system (content filtered to be suitable for children), comfort suspension mode (automatic adjustment), warm ambient lighting (safety mode linked to the cruise system), and family travel mode (automatically recommending family-friendly destinations such as parks and museums).

[0074] Preferably, the present invention can also automatically switch between the two schemes according to time conditions, and users can also temporarily switch modes through voice commands; the system records the user's temporary switching behavior, confirmation or modification of the execution result, and dynamically updates the user intent vector, intent-service mapping knowledge graph and multi-dimensional constraint graph between services accordingly, so as to achieve continuous optimization of vehicle services.

[0075] This invention lowers the user configuration threshold by introducing a natural language understanding model finely tuned with vehicle-domain corpus and a three-dimensional structured user intent vector. By constructing a multi-dimensional constraint graph between services and incorporating it as a hard constraint into the optimization model, it achieves automatic identification and resolution of all implicit conflicts, avoiding safety and resource conflicts such as "ambient lighting interfering with nighttime cruise perception" and "high-frequency navigation and driving analysis competing for communication bandwidth," ensuring that the configuration scheme 100% meets automotive-grade functional safety requirements. Through dual-knowledge graph collaborative reasoning and multi-objective constraint optimization, it improves the user experience. Furthermore, by standardizing the definition of service intent contribution vectors and constraint relationships during the engineering service development phase, it promotes the computable, scalable, and sustainable evolution of the intelligent vehicle service ecosystem.

[0076] like Figure 2 and Figure 3 As shown, a second aspect of the present invention provides a vehicle service control system based on intent parsing and constraint optimization, comprising: The receiving unit is at least used to receive natural language information input by the user, wherein the natural language information is used to describe the user's vehicle usage scenario, functional preferences and / or emotional needs; The parsing unit is at least used to perform intent parsing on the natural language information to obtain a structured user intent vector; The mapping unit is at least used to determine a set of candidate vehicle services that match the user intent based on the user intent vector, according to the intent-service mapping knowledge graph. The detection unit is at least used to perform service conflict detection on the candidate vehicle service set based on the multi-dimensional constraint graph between services, and to obtain the conflict type, conflict object and conflict impact information between vehicle services in the candidate vehicle service set. The control unit is at least used to construct and solve a constraint optimization model based on the results of the service conflict detection to generate a vehicle service control scheme. The execution unit is at least used to enable, disable, adjust parameters, adjust execution timing, schedule resources, switch working modes, and / or control linkage of the target vehicle service in accordance with the vehicle service control scheme.

[0077] The present invention will be further described in detail below with reference to specific embodiments.

[0078] Example of a configuration for a mixed scenario of weekday commuting and weekend family outings.

[0079] This embodiment illustrates the complete execution flow of this method in a multi-vehicle scenario.

[0080] The user inputs natural language information: "I drive this car to work from Monday to Friday. As a working person, I have to clock in on time and can't be late. I hope the journey isn't too boring. On weekends, I can take my wife and kids out to play and relax. I hope the driving on weekends isn't too tiring."

[0081] Step 1: Scene recognition and intent parsing.

[0082] The system identifies two typical car usage scenarios from the above natural language information and parses them to obtain the corresponding user intent vectors.

[0083] Scenario A: Weekday commute scenario.

[0084] In this scenario, the user's core needs are punctual travel and relief from the monotony of commuting, with the user primarily driving alone. The values ​​of each dimension of the obtained intent vector are: driving experience 0.9, riding experience 0.7, and emotional satisfaction 0.5, represented in vector form as U_weekday = [0.9, 0.7, 0.5].

[0085] Scenario B: Weekend family outing.

[0086] In this scenario, the core user needs are relaxation, reduced driving burden, and improved family riding experience. The values ​​of each dimension of the obtained intent vector are: driving experience 0.8, riding experience 0.9, and emotional satisfaction 0.8, which are represented in vector form as U_weekend = [0.8, 0.9, 0.8].

[0087] Step 2: Complete scenario-based service matching based on the intent-service mapping knowledge graph.

[0088] For weekday commuting scenarios, the intent-service mapping knowledge graph is invoked to calculate the matching degree between each vehicle service and the user intent vector: The service contribution vector for real-time traffic intelligent navigation is [0.9, 0.1, 0], with a matching degree of approximately 0.88; the service contribution vector for commuting mode is [0.8, 0.1, 0.1], with a matching degree of 0.79; the service contribution vector for personalized podcast / music recommendation is [0.1, 0.8, 0.1], with a matching degree of 0.66; and the service contribution vector for driving efficiency analysis report is [0.7, 0.2, 0.1], with a matching degree of 0.82.

[0089] In this embodiment, a matching degree threshold of 0.75 is set, and services with a matching degree greater than 0.75 are filtered to form a candidate vehicle service set for weekday commuting scenarios: {real-time traffic intelligent navigation, commuting mode, driving efficiency analysis report}.

[0090] For the weekend family travel scenario, the intent-service mapping knowledge graph is invoked to calculate the matching degree between each vehicle service and the user intent vector: The service contribution vector of the full-speed adaptive cruise control is [0.8, 0.1, 0.1], with a matching degree of 0.81; the service contribution vector of the rear child entertainment system is [0, 0.9, 0.1], with a matching degree of 0.89; the service contribution vector of the family travel mode is [0.2, 0.3, 0.5], with a matching degree of 0.75; the service contribution vector of the comfort suspension mode is [0.1, 0.8, 0.1], with a matching degree of 0.80; and the service contribution vector of the ambient lighting is [0, 0.3, 0.7], with a matching degree of 0.83.

[0091] Based on a matching threshold of 0.75, a set of candidate vehicle services for weekend family outings was formed: {full-speed adaptive cruise control, rear child entertainment system, comfort suspension mode, ambient lighting, family outing mode}.

[0092] Step 3: Conduct scenario-specific conflict detection based on the multi-dimensional constraint graph between services.

[0093] Conflict detection was performed on candidate vehicle services for weekday commuting scenarios. The results showed that there were conflicts in data collection frequency and communication bandwidth between real-time traffic intelligent navigation and driving efficiency analysis reports. The two shared high-frequency GPS and network data channels, which is a resource constraint conflict.

[0094] Conflict detection was conducted on candidate vehicle services for weekend family outings. The results showed that there was a safety constraint conflict between the ambient lighting and the full-speed adaptive cruise control. The colored ambient lighting could easily interfere with the driver's visual recognition of the instrument panel, posing a driving safety hazard.

[0095] Step 4: Perform intelligent conflict resolution according to different scenarios.

[0096] To address resource constraint conflicts during weekday commutes, the system determines that the two services can be made compatible through resource scheduling and proposes two resolution schemes: Scheme A1 prioritizes the operation of real-time traffic intelligent navigation, maintaining its high-frequency data collection mode, while switching the driving efficiency analysis report to a low-frequency data collection mode and supplementing data collection during system idle periods; Scheme A2 displays the conflict issue to the user, allowing the user to choose whether to reduce the real-time performance of the driving efficiency analysis report. In this embodiment, the system automatically executes Scheme A1 without manual user intervention.

[0097] To address the safety constraints and conflicts encountered during weekend family outings, the system generates two differentiated solutions and displays the corresponding dimensional scores and safety compliance information for each solution to the user. Option B1 is the safety-first mode. When full-speed adaptive cruise control is activated, the ambient lighting automatically switches to a single color and low brightness safety mode. Under this option, the emotional satisfaction score drops from 0.83 to 0.70, and the safety compliance is 100%. Option B2 is the experience-first mode, which retains all the functions of the warm ambient lighting, while also displaying driving risk warnings. Under this option, the emotional satisfaction score remains unchanged at 0.83, while the safety compliance score drops to 80%.

[0098] In this embodiment, the user selects scheme B1 as the final resolution strategy.

[0099] Step 5: Configure output and automatic switching between scenarios.

[0100] Based on the conflict resolution results, the system generates two independent vehicle service configuration schemes and configures automatic switching rules.

[0101] Weekday commuting configuration plan.

[0102] The effective time is set to 07:00-09:00 and 17:00-19:00, Monday to Friday. The system can adaptively adjust the time period based on the user's historical travel data. This solution enables real-time traffic intelligent navigation (high-frequency data mode), commuting mode, and driving efficiency analysis reports (low-frequency data mode, with data supplemented at night).

[0103] Weekend family trip planning ideas.

[0104] The program is set to take effect every Saturday and Sunday, all day. It includes full-speed adaptive cruise control (comfort following mode), rear-seat child entertainment system (children's content filtering mode), comfort suspension mode, ambient lighting (cruise-linked safety mode), and family travel mode.

[0105] The system automatically switches to the corresponding configuration scheme based on the system time; it also supports users to temporarily switch configuration modes via voice commands, such as saying "switch to my work mode" to complete the temporary switch. The system will fully record the user's temporary switching behavior and continuously optimize the scene recognition model based on the behavioral data.

[0106] Example of a scenario configuration that balances personalized performance preferences and security constraints.

[0107] This embodiment is used to illustrate the processing flow of this method for non-mutually exclusive conflicts of parameter threshold tension.

[0108] The user entered the following natural language information: "I like to drive fast and occasionally want to experience the feeling of being pushed back in my seat, but safety comes first most of the time, so it's best not to distract me."

[0109] Intent parsing.

[0110] The system identified that the user simultaneously had demands for both driving performance and driving safety, and analyzed the data to obtain the corresponding user intent vector: driving experience 0.9, passenger experience 0.6, and emotional satisfaction 0.7. This vector indicates that the user seeks vehicle power and handling performance while requiring all functions to operate within safe boundaries.

[0111] Service matching.

[0112] The intent-service mapping knowledge graph is invoked for matching, and the candidate vehicle services obtained are: Sport+ mode, intelligent safety assistance system, and driver attention monitoring. Among them, the intent contribution vector of Sport+ mode is [0.9, 0.1, 0.7], the intent contribution vector of intelligent safety assistance system is [0.3, 0.2, 0.8], and the intent contribution vector of driver attention monitoring is [0.1, 0.1, 0.9].

[0113] Conflict detection and intelligent resolution.

[0114] Conflict detection is performed based on a multi-dimensional constraint graph between services. It is determined that there is a non-mutually exclusive conflict of parameter threshold tension between the motion+ mode and safety services: when the motion+ mode is running, it will lower the intervention threshold of the electronic stability system, which creates constraint tension with the user's demand for "safety first".

[0115] To address this type of conflict, the system generates a hybrid security performance service model, with the following specific rules: First, the core parameters of Sport+ mode are retained to maintain aggressive power response and steering feel, meeting users' performance experience needs; Second, continuously operate the driving attention monitoring and road condition recognition functions, and establish risk triggering rules; Third, when risk scenarios such as driver distraction or complex road conditions are detected, the electronic stability system will be temporarily and gradually restored to a conservative intervention threshold. Fourth, during service operation, voice status prompts will be provided simultaneously, with the voice message: "Performance mode is activated, and the safety assistance system remains on high alert." This solution balances user performance requirements while maintaining driving safety boundaries through dynamic monitoring and parameter adjustment, achieving a harmonious integration of these two opposing demands.

[0116] Final configuration.

[0117] The system sets the safety performance composite service mode as the default operating mode. During vehicle operation, the system continuously collects user driving behavior data and dynamically optimizes parameters and adjusts thresholds through machine learning algorithms to match user-personalized usage habits with driving performance and driving safety.

[0118] This invention can solve the problems of high configuration threshold, frequent functional abnormalities and security risks caused by the semantic gap between user expression and engineering configuration in the current personalized configuration of intelligent vehicles, as well as the implicit constraints of four types of relationships (security, physical, resource and logical) among hundreds of engineering services in the vehicle. It can achieve the technical effects of significantly reducing the user configuration threshold, improving configuration security and reliability, enhancing the accuracy of intent fulfillment and consistency of experience, and supporting the sustainable evolution of the service ecosystem.

[0119] A third aspect of the present invention provides an electronic device, comprising: at least one processor; and at least one memory communicatively connected to the processor, wherein: the memory stores program instructions executable by the processor, and the processor invokes the program instructions to perform the steps of the method as described in any of the above embodiments.

[0120] A fourth aspect of the present invention discloses a readable storage medium storing a computer program, which is executed by a processor as described in any of the above embodiments.

[0121] Computer-readable storage media can include: any entity or device capable of carrying computer programs, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory, random access memory (RAM), and software distribution media, etc. Computer programs include computer program code. Computer program code can be in the form of source code, object code, executable files, or some intermediate form, etc. Computer-readable storage media can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory, random access memory (RAM), and software distribution media, etc.

[0122] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0123] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus or device (such as a computer-based system, a system including a processing module or other system that can fetch and execute instructions from, an instruction execution system, apparatus or device).

[0124] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A vehicle service control method based on intent parsing and constraint optimization, characterized in that, include: Receive natural language information input by the user, which is used to describe the user's car usage scenario, functional preferences and / or emotional needs; The natural language information is parsed to obtain a structured user intent vector; Based on the intent-service mapping knowledge graph, a set of candidate vehicle services matching the user intent is determined according to the user intent vector; Based on the multi-dimensional constraint graph between services, service conflict detection is performed on the candidate vehicle service set to obtain the conflict type, conflict object and conflict impact information between vehicle services in the candidate vehicle service set. Based on the service conflict detection results, a constraint optimization model is constructed and solved, and a vehicle service control scheme is generated and the corresponding vehicle service control is executed. According to the vehicle service control scheme, the target vehicle service is enabled, disabled, parameter adjusted, execution timing adjusted, resource scheduled, working mode switched and / or linked control is performed.

2. The vehicle service control method according to claim 1, characterized in that, The receipt of natural language information input by the user, wherein the natural language information is used to describe the user's car usage scenario, functional preferences and / or emotional needs, includes: Receive natural language information input by the user via voice or text; Identify at least one vehicle usage scenario from the natural language information, and extract the time conditions, travel purpose, user role, functional preferences and / or emotional needs corresponding to each vehicle usage scenario.

3. The vehicle service control method according to claim 2, characterized in that, The process of parsing the natural language information to obtain a structured user intent vector includes: Using a natural language understanding model trained or fine-tuned with vehicle-domain corpus, we analyze the functional preferences and / or emotional demands corresponding to each vehicle usage scenario, and generate user intent vectors for each scenario.

4. The vehicle service control method according to claim 3, characterized in that, The process of parsing the natural language information to obtain a structured user intent vector includes: Generate user intent vectors that include driving experience, riding experience, and emotional satisfaction dimensions; The driving experience dimension is used to characterize the user's preference intensity for power, handling, economy and / or intelligent driving; the riding experience dimension is used to characterize the user's preference intensity for comfort, entertainment and / or health; and the emotional satisfaction dimension is used to characterize the user's preference intensity for personality expression, social sharing and / or emotional resonance.

5. The vehicle service control method according to claim 1, characterized in that, The intent-service mapping knowledge graph-based method for determining a set of candidate vehicle services matching the user intent based on the user intent vector includes: Based on the intent-service mapping knowledge graph including intent nodes, vehicle service nodes, and associated edges, the vehicle service node corresponding to the user intent vector is determined. The associated edge is used to characterize the association strength and / or contribution direction between the intent node and the vehicle service node. Each vehicle service node is configured with a service contribution vector, which is used to characterize the degree of contribution of the corresponding vehicle service to each dimension of the user intent vector.

6. The vehicle service control method according to claim 5, characterized in that, The intent-service mapping knowledge graph-based method for determining a set of candidate vehicle services matching the user intent based on the user intent vector includes: Calculate the matching degree between the user intent vector and the service contribution vector corresponding to each vehicle service node; Vehicle services that meet the preset matching criteria are identified as candidate vehicle services, and the candidate vehicle services form the candidate vehicle service set.

7. The vehicle service control method according to claim 1, characterized in that, The method involves performing service conflict detection on the candidate vehicle service set based on a multi-dimensional constraint graph between services, thereby obtaining information on the conflict types, conflict objects, and conflict impacts among the vehicle services in the candidate vehicle service set, including: Based on a multi-dimensional constraint graph between services with vehicle services as nodes and service constraints as edges, the system detects whether there are security constraints, physical constraints, resource constraints and / or logical constraints between vehicle services in the candidate vehicle service set. When the inter-service constraint relationship is detected to be triggered, it is determined that a service conflict exists, and the corresponding conflict type, conflict object, and conflict impact information are determined.

8. The vehicle service control method according to claim 7, characterized in that, The method involves performing service conflict detection on the candidate vehicle service set based on a multi-dimensional constraint graph between services, thereby obtaining information on the conflict types, conflict objects, and conflict impacts among the vehicle services in the candidate vehicle service set, including: When two or more candidate vehicle services have conflicts such as data acquisition frequency conflict, communication bandwidth competition, computing power resource competition, power resource competition, hardware condition limitations, functional dependency failure, functional mutual exclusion, execution order contradiction, or impact on driving safety, it is determined that there is a service conflict between the two or more candidate vehicle services.

9. The vehicle service control method according to claim 1, characterized in that, The step of constructing and solving a constraint optimization model based on the service conflict detection results, generating a vehicle service control scheme, and then executing the corresponding vehicle service control includes: The matching degree between the three-dimensional intent vector of user driving experience, riding experience, and emotional satisfaction and the total contribution vector of the candidate service set is used as the optimization objective, and the safety constraints, physical constraints, resource constraints, and logical constraints in the multi-dimensional constraint graph between services are used as constraints. When a feasible solution exists that satisfies all constraints, a vehicle service control scheme containing the optimal service selection result and conflict resolution strategy is generated; when no fully feasible solution exists, at least two differentiated experience package schemes are generated, with each scheme corresponding to the quantified gains and losses of intent satisfaction in three dimensions.

10. The vehicle service control method according to claim 1, characterized in that, The process of enabling, disabling, adjusting parameters, adjusting execution timing, scheduling resources, switching working modes, and / or coordinating control of the target vehicle service according to the vehicle service control scheme includes: When there are multiple vehicle service control schemes that meet the constraints, output the driving experience dimension, riding experience dimension, emotional satisfaction dimension and / or safety compliance impact information corresponding to each vehicle service control scheme; The target vehicle service control scheme is determined based on user selection or preset priority. The system executes the target vehicle service control scheme based on time conditions, vehicle status, user instructions, and / or historical usage behavior, and updates the user intent vector, the intent-service mapping knowledge graph, and / or the multi-dimensional constraint graph between services based on the user's confirmation, modification, or cancellation of the execution result.

11. A vehicle service control system based on intent parsing and constraint optimization, characterized in that, include: The receiving unit is at least used to receive natural language information input by the user, wherein the natural language information is used to describe the user's vehicle usage scenario, functional preferences and / or emotional needs; The parsing unit is at least used to perform intent parsing on the natural language information to obtain a structured user intent vector; The mapping unit is at least used to determine a set of candidate vehicle services that match the user intent based on the user intent vector, according to the intent-service mapping knowledge graph. The detection unit is at least used to perform service conflict detection on the candidate vehicle service set based on the multi-dimensional constraint graph between services, and to obtain the conflict type, conflict object and conflict impact information between vehicle services in the candidate vehicle service set. The control unit is at least used to construct and solve a constraint optimization model based on the results of the service conflict detection to generate a vehicle service control scheme. The execution unit is at least used to enable, disable, adjust parameters, adjust execution timing, schedule resources, switch working modes, and / or control linkage of the target vehicle service in accordance with the vehicle service control scheme.

12. An electronic device, characterized in that, include: At least one processor; And at least one memory communicatively connected to the processor, wherein: the memory stores program instructions executable by the processor, and the processor invokes the program instructions to perform the steps of the vehicle service control method based on intent parsing and constraint optimization as described in any one of claims 1-10.

13. A readable storage medium storing a computer program, characterized in that, The computer program is executed by a processor using the steps of the vehicle service control method based on intent parsing and constraint optimization as described in any one of claims 1-10.

Images