Vehicle personalized decision-making system optimization method based on multi-source implicit feedback and application
By collecting multi-source heterogeneous feedback signals and assigning confidence weights, the vehicle personalized decision-making system is optimized, solving the problem of insufficient utilization of implicit feedback and achieving efficient and accurate personalized decision-making model updates and improved user experience.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ANHUI JIANGHUAI AUTOMOBILE GRP CORP LTD
- Filing Date
- 2026-03-03
- Publication Date
- 2026-05-19
AI Technical Summary
Existing technologies in vehicle personalization decision-making systems lack effective utilization of implicit user feedback, resulting in biased learning direction, slow learning cycles, and low learning efficiency, failing to provide a real-time and accurate personalized experience.
By collecting multi-source heterogeneous feedback signals, including direct manipulation signals, physiological regression signals, and behavioral continuation signals, and combining them with confidence assessment rules to assign weights, the personalized decision-making model is incrementally updated, thereby optimizing the decision-making system.
It enables the accurate capture of users' long-term preferences and short-term state characteristics without any active user intervention, thereby improving learning quality and system adaptability, and providing more accurate collaborative control services.
Smart Images
Figure CN122064362A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of automotive intelligent control system technology, and in particular to an optimization method and application of a vehicle personalized decision-making system based on multi-source implicit feedback. Background Technology
[0002] In the design of personalized decision-making systems for vehicles, continuous learning is key to achieving intelligent adaptation. However, traditional machine learning paradigms heavily rely on explicit feedback provided by users, such as active ratings, likes, or comments, to drive model optimization. But in continuous, focused, and safety- and immersive closed environments like inside a car, frequently requesting explicit feedback from users can severely disrupt their normal experience, resulting in stiff interactions and low user compliance. Therefore, effectively utilizing implicit feedback generated unconsciously by users has become a key research direction for improving the intelligence level of in-vehicle systems.
[0003] Currently, existing technologies have significant limitations in handling implicit feedback, mainly in the following three aspects:
[0004] 1. One-dimensional feedback signal: Existing solutions typically only capture a binary result: whether the user manually overridden the system's automatic settings. This one-dimensional signal fails to reveal the complex reasons behind the overriding behavior—whether it's a conflict between the system's decision and the user's preferences, or a temporary adjustment made by the user due to temporary changes in the context. The lack of analysis of behavioral motivation in existing solutions may lead to biases in the learning direction.
[0005] 2. Lack of a Quantification Mechanism for Feedback Intensity: Different user behaviors imply significantly different levels of preference intensity and emotional attitude. For example, the intensity of dissatisfaction or desire for correction conveyed by a user quickly and drastically turning off an auto-start function is drastically different from that conveyed by slowly and subtly tweaking the same function. Current technology lacks the ability to perform fine-grained analysis and quantification of such behavioral patterns, such as operation speed, magnitude, and latency, and cannot incorporate the key dimension of feedback intensity into the learning model. This results in all covered behaviors being treated with equal weight, reducing learning efficiency and accuracy.
[0006] 3. Slow and Delayed Learning Cycle: Most systems rely on mining statistical patterns from massive amounts of historical data for batch-processing offline model updates. This approach results in long learning cycles, making it difficult to respond quickly to recent user feedback. Consequently, the system is slow to adapt and struggles to provide users with real-time, accurate, and personalized experiences, especially when dealing with shifting user preferences or new scenarios.
[0007] Furthermore, while some advanced technologies involve implicit feedback and personalized models, they are not suitable for personalized decision-making systems in vehicles. For example, patent application CN111177580A discloses a method for personalized recommendation using multiple implicit feedback. The steps are as follows: 1. Define a new adoption method, defining items from target feedback as positive examples and items that only interact with the user through support feedback as negative examples; 2. Apply matrix factorization to model the linear interaction relationship between users and items. By mapping the interaction relationship between users and items to a potential shared space of dimension d, the matrix factorization model models the user's preference for items as the inner product of the corresponding latent factor vectors in this space; 3. Apply a multilayer perceptron to model the nonlinear interaction relationship between users and items, as well as the interrelationship between different types of implicit feedback; 4. Integrate a network architecture that combines the matrix factorization module and the multilayer perceptron module. This method provides an end-to-end recommendation model that comprehensively simulates multiple implicit feedbacks between users and items, improving recommendation quality. However, this method is suitable for e-commerce, video and other platforms. If it is used for in-vehicle platforms, the above problems still exist. The heavy neural network architecture of this method is difficult to support the low latency and lightweight online incremental updates required by in-vehicle platforms. Summary of the Invention
[0008] In view of the shortcomings of the prior art, the present invention provides a method and application for optimizing a vehicle personalized decision-making system based on multi-source implicit feedback. By analyzing the implicit feedback of users to the automatic decision-making of the system, the personalized decision-making system can be continuously optimized, and efficient and accurate incremental model optimization can be achieved.
[0009] To achieve the above and related objectives, the present invention adopts the following technical solution:
[0010] The first aspect of this invention provides an optimization method for a vehicle personalized decision-making system based on multi-source implicit feedback, comprising the following steps:
[0011] Step S100: When the vehicle executes a collaborative control strategy initiated by the personalized decision-making system, record the metadata of the collaborative control strategy;
[0012] Step S200: During the preset feedback collection period after the execution of the collaborative control strategy, at least two types of multi-source heterogeneous feedback signals reflecting the user's non-command interaction behavior are collected simultaneously.
[0013] Step S300: Associate the multi-source heterogeneous feedback signals with metadata, parse them into structured feedback events, and assign weights to each structured feedback event according to the preset confidence evaluation rules to obtain a weighted set of feedback events.
[0014] Step S400: Based on the weighted feedback event set, the personalized decision-making model for generating collaborative control strategies is incrementally updated to optimize the personalized decision-making model's strategy generation capability for corresponding users and triggering scenarios.
[0015] Furthermore, the metadata includes the trigger scenario identifier, the set of execution parameters, and the expected user state target.
[0016] Furthermore, in step S200, the multi-source heterogeneous feedback signal includes:
[0017] Direct operation signals are operation event signals that allow users to manually reverse the parameters that are automatically adjusted by the system.
[0018] Physiological regression signal is the deviation trend signal between the user's actual physiological state indicators and the expected user state target.
[0019] Behavioral continuation signals are subsequent operational event signals performed by the user that are consistent with the intent of the collaborative control strategy.
[0020] Further, in step S300, the confidence assessment rule is calculated based on the strength of user intent represented by the structured feedback event, wherein...
[0021] Feedback events that represent a user's explicit negative intent toward the collaborative control strategy are assigned a high level of confidence weight;
[0022] Feedback events that characterize a user’s significant physiological response to a collaborative control strategy or their active compounding action are assigned a medium confidence weight.
[0023] Feedback events that characterize a user's fine-tuning or fuzzy response to a collaborative control strategy are assigned a low confidence weight.
[0024] Furthermore, in step S400, the personalized decision-making model includes a dynamic user profile sub-model, used to store and learn the user's long-term preference features and short-term state features; and a scenario policy matrix, used to store the mapping relationship between user-scenario pairs and policy parameters.
[0025] Furthermore, the personalized decision-making model also includes a feedback learning sub-model, which is configured to: directly adjust the policy parameters of a specific user-scenario pair based on the positive or negative feedback events corresponding to the high-level confidence weights; and update the user's long-term preference features based on the positive feedback events corresponding to the high-level confidence weights.
[0026] Furthermore, the method also includes triggering a re-evaluation and calibration of the collaborative control strategy for the same user-scenario when the number of feedback events collected and assigned high confidence weights under the same user-scenario reaches a preset number.
[0027] A second aspect of the present invention provides an optimization system for a vehicle personalized decision-making system based on multi-source implicit feedback, comprising:
[0028] The strategy recording model is used to record metadata of the collaborative control strategy when the vehicle executes a collaborative control strategy initiated by the personalized decision-making system.
[0029] The feedback acquisition module is used to simultaneously acquire at least two types of multi-source heterogeneous feedback signals reflecting non-command interaction behaviors of users during a preset feedback acquisition period after the execution of the collaborative control strategy.
[0030] The feedback analysis module is used to associate multi-source heterogeneous feedback signals with metadata, parse them into structured feedback events, and assign weights to each structured feedback event according to preset confidence evaluation rules to obtain a weighted set of feedback events.
[0031] The incremental optimization module is used to incrementally update the personalized decision-making model that generates collaborative control strategies based on a weighted set of feedback events, so as to optimize the personalized decision-making model's strategy generation capability for corresponding users and triggering scenarios.
[0032] A third aspect of the present invention provides a computer-readable storage medium storing computer-readable instructions thereon, which, when executed by a computer's processor, cause the computer to execute the aforementioned optimization method for a vehicle personalized decision-making system based on multi-source implicit feedback.
[0033] A fourth aspect of the present invention provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described optimization method for a vehicle personalized decision-making system based on multi-source implicit feedback.
[0034] The beneficial technical effects of this invention are as follows:
[0035] This invention replaces the traditional explicit feedback method that relies on user-initiated ratings or settings by collecting at least two types of multi-source heterogeneous feedback signals that reflect the user's non-instructional interaction behavior. This enables the personalized decision-making system to accurately capture the user's long-term preference characteristics and short-term state characteristics without the user's awareness or additional operation, thereby providing more accurate collaborative control services and improving the user experience.
[0036] This invention analyzes multi-source heterogeneous feedback signals into structured feedback events with confidence weights, which can effectively distinguish between users' explicit objections, positive inclinations, and random noise, significantly improve learning quality, greatly reduce the risk of mislearning, accelerate the convergence process, and improve the accuracy of the optimization direction.
[0037] The collaborative control strategy of the personalized decision-making model of this invention can dynamically adapt to changes in user preferences over time and real-time needs in different scenarios, and the system has strong self-adaptability.
[0038] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description
[0039] The accompanying drawings, incorporated in and forming part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without inventive effort. In the drawings:
[0040] Figure 1 This is a flowchart of the optimization method for the vehicle personalized decision-making system based on multi-source implicit feedback proposed in this application;
[0041] Figure 2 This is a framework diagram of the vehicle personalization decision-making system optimization system based on multi-source implicit feedback in this application;
[0042] Figure 3 A schematic diagram of the structure of a computer system suitable for an embodiment of this application is shown. Detailed Implementation
[0043] Unless otherwise defined, all technical and / or scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should be understood that certain features of the invention (described in the context of separate embodiments for clarity) may also be provided in a single embodiment. Conversely, multiple features of the invention (described in the context of a single embodiment for brevity) may also be provided separately or in any suitable combination or, where appropriate, in any other described embodiment of the invention. Certain features described in the context of various embodiments will not be considered essential features of those embodiments unless the embodiment is inoperable without those elements. The invention is further illustrated below by specific examples; however, it should be noted that the specific process conditions and results described in the embodiments of the invention are merely illustrative and should not be construed as limiting the scope of protection of the invention. All equivalent changes or modifications made in accordance with the spirit and essence of the invention should be covered within the scope of protection of the invention.
[0044] Please see Figure 1 The flowchart of the vehicle personalization decision-making system optimization method based on multi-source implicit feedback in this application is described in detail below:
[0045] Step S100: When the vehicle executes a collaborative control strategy initiated by the personalized decision-making system, the metadata of the collaborative control strategy is recorded.
[0046] Specifically, the metadata of this application includes a trigger scenario identifier, a set of execution parameters, and a target user state. The trigger scenario identifier is the specific situation when the collaborative control strategy is triggered, such as urban traffic congestion, highway cruising, nighttime mountain driving, parking and charging, weather, time, road conditions, driving mode, and user state. The set of execution parameters consists of the parameters of the in-vehicle system actually controlled by the collaborative control strategy and their set values, such as driver assistance parameters, cabin environment parameters, and infotainment parameters. The target user state is the physiological or psychological state of the user that is expected to be achieved by executing the above parameter set, such as the driver's heart rate returning to normal, improved concentration, and reduced tension.
[0047] Step S200: During the preset feedback acquisition period after the execution of the collaborative control strategy, at least two types of multi-source heterogeneous feedback signals reflecting the user's non-command interaction behavior are collected simultaneously.
[0048] Specifically, the multi-source heterogeneous feedback signals based on implicit feedback in this application include: direct operation signals, which are operation event signals in which the user manually reverses the parameters automatically adjusted by the system; physiological regression signals, which are deviation trend signals between the user's actual physiological state indicators and the expected user state target; and behavioral continuation signals, which are subsequent operation event signals executed by the user that are consistent with the intention of the collaborative control strategy.
[0049] Specifically, this application presupposes a feedback collection period of one observation window, such as 5 minutes, with the specific value set according to actual needs. Direct operation signals include: signals indicating that the user manually adjusts any parameter adjusted by the personalized decision-making system in reverse, recording the magnitude, speed, and delay time of the reverse adjustment operation, such as quickly turning a knob, slowly sliding a knob, adjusting within 2 seconds of execution, or adjusting after 2 minutes. Physiological regression signals include: continuously monitoring the user's physiological indicators after executing the collaborative control strategy, such as facial tension and heart rate trends, using sensors such as visual sensors, photoelectric sensors, and infrared sensors. Comparing these with the expected physiological goals of the strategy, such as reducing tension and stabilizing heart rate, yields a deviation trend signal. If the deviation trend deviates from the expected goal, it is considered negative feedback; if it meets the expected goal, it is considered positive feedback. Behavioral continuation signals include: the user does not modify the system settings but performs a series of subsequent operations that may indicate their preferences. For example, the system plays a relaxation playlist, and the user not only does not change the song but also increases the volume and turns on the chair massage; this is a strong positive feedback composite signal.
[0050] In step S300, the multi-source heterogeneous feedback signals are associated with metadata, parsed into structured feedback events, and weights are assigned to each structured feedback event according to the preset confidence evaluation rules to obtain a weighted set of feedback events.
[0051] Specifically, this application associates multi-source heterogeneous feedback signals with metadata to clarify which specific instance of the collaborative control strategy was generated by these feedback signals. This application transforms raw, continuous physical feedback signals into discrete, structured feedback events, enabling algorithms to understand and process them. These include reverse adjustment events corresponding to direct manipulation signals, deviation trend events corresponding to physiological regression signals, and intent-consistent events corresponding to behavioral continuation signals. The confidence assessment rule of this application is calculated based on the intensity of user intent represented by the structured feedback events. Feedback events representing a clear negative intent from the user towards the collaborative control strategy are assigned a high confidence weight; feedback events representing a significant physiological response or active compound operation from the user towards the collaborative control strategy are assigned a medium confidence weight; and feedback events representing a fine-tuning or ambiguous response from the user towards the collaborative control strategy are assigned a low confidence weight.
[0052] Specifically, the confidence weights for high-level operations in this application are calculated based on the magnitude, speed, and delay time of the reverse adjustment operation. For example, rapid, large-amplitude, and short-delay reverse adjustment operations receive the highest weight, such as a weight value of 0.9, indicating a system decision error. The confidence weights for medium-level operations are calculated based on the significance and persistence of the physiological deviation trend, such as weight values of 0.6–0.8, with a weight value of 0.7 for proactive composite operations. The confidence weights for low-level operations are calculated based on the consistency between subsequent operations and the intent of the collaborative control strategy. For example, slow, small-amplitude, and long-delay adjustments receive a weight value of 0.3, indicating a user's fine-tuning or temporary change in needs.
[0053] Specifically, through confidence assessment and weighting, this application can effectively distinguish between users' explicit objections, positive inclinations, and random noise, ensuring that the subsequent personalized decision-making model is driven only by high-quality feedback events.
[0054] Step S400: Based on the weighted feedback event set, the personalized decision-making model for generating collaborative control strategies is incrementally updated to optimize the personalized decision-making model's strategy generation capability for corresponding users and triggering scenarios.
[0055] Specifically, the personalized decision-making system of this application includes a personalized decision-making model, which is used to generate collaborative control strategies based on user profiles and scenario information. The personalized decision-making model includes a dynamic user profile sub-model, used to store and learn the user's long-term preference characteristics and short-term state characteristics; and a scenario policy matrix, used to store the mapping relationship between user-scenario pairs and policy parameters. The personalized decision-making model also includes a feedback learning sub-model, which is configured to: directly adjust the policy parameters of a specific user-scenario pair based on positive or negative feedback events corresponding to high-level confidence weights; and update the user's long-term preference characteristics based on positive feedback events corresponding to high-level confidence weights.
[0056] Specifically, the personalized decision-making model of this application incorporates online learning or mini-batch incremental learning algorithms. For high-weight negative feedback events, the policy parameters of the corresponding user-scene pairs in the scenario policy matrix are adjusted to correct them in real time. For example, in a rainy commuting scenario, the model automatically turns on the focus mode for the user, but the user quickly turns off the ambient light each time. The model learns that the user is sensitive to the ambient light in this scenario and will remove or disable the ambient light adjustment by default in the collaborative control strategy for this user in this scenario in the future. For high-weight positive feedback events, the policy parameters of the corresponding user-scene pairs in the scenario policy matrix are strengthened, and the user's long-term preference features in the dynamic user profile sub-model are updated, which may be generalized to similar scenarios.
[0057] Specifically, the personalized decision-making model update in this application is based on user-scenario pairs, which can ensure that the optimization is highly personalized without affecting the experience of other users.
[0058] Specifically, the method of this application also includes triggering a re-evaluation and calibration of the collaborative control strategy for the same user-scenario pair when a preset number of feedback events with high confidence weights are collected. The preset number is set according to requirements. This application only filters feedback events with high confidence weights to ensure that the data used to trigger the re-evaluation is high-quality and strongly represents the user's intent, thus excluding low-weighted, ambiguous, and noisy data. The re-evaluation can include data backtracking analysis, strategy effectiveness analysis, and attribution analysis; calibration can include resetting strategy parameters, resetting model confidence, and co-calibrating related strategies. In this way, this application can ensure that the personalized decision-making system can not only respond to users in the short term but also maintain the effectiveness and accuracy of the collaborative control strategy as users evolve over the long term.
[0059] Please see Figure 2 This is a framework diagram of the vehicle personalization decision-making system optimization system 200 based on multi-source implicit feedback in this application, including:
[0060] Strategy recording model 210 is used to record metadata of the collaborative control strategy when the vehicle executes a collaborative control strategy initiated by the personalized decision-making system;
[0061] The feedback acquisition module 220 is used to simultaneously acquire at least two types of multi-source heterogeneous feedback signals reflecting the user's non-command interaction behavior during a preset feedback acquisition period after the execution of the collaborative control strategy.
[0062] The feedback analysis module 230 is used to associate multi-source heterogeneous feedback signals with metadata, parse them into structured feedback events, and assign weights to each structured feedback event according to preset confidence evaluation rules to obtain a weighted set of feedback events.
[0063] The incremental optimization module 240 is used to incrementally update the personalized decision-making model that generates collaborative control strategies based on a weighted set of feedback events, so as to optimize the personalized decision-making model's strategy generation capability for corresponding users and triggering scenarios.
[0064] Specifically, this application's system does not rely on a single signal source but simultaneously collects at least two types of heterogeneous signals, including direct operational signals, physiological regression signals, and behavioral continuation signals. Different feedback signals are weighted using confidence assessment rules; for example, high weights are assigned to explicit negative feedback, while medium weights are assigned to deviation trend signals. This effectively addresses the shortcomings of traditional personalized decision-making systems, such as high noise and susceptibility to misjudgment from single-sensor data. The rule-based data fusion mechanism of this application's system can significantly improve the accuracy and robustness of feedback analysis, ensuring that subsequent updates to the personalized decision-making model are based on high-confidence user intent rather than random noise. Furthermore, this application's system can achieve dynamic adaptation of the personalized decision-making model through closed-loop incremental optimization.
[0065] It should be noted that the vehicle personalization decision-making system optimization system based on multi-source implicit feedback provided in the above embodiments and the vehicle personalization decision-making system optimization method based on multi-source implicit feedback provided in the above embodiments belong to the same concept. The specific operation methods of each module and unit have been described in detail in the method embodiments and will not be repeated here. In practical applications, the vehicle personalization decision-making system optimization system based on multi-source implicit feedback provided in the above embodiments can be assigned to different functional modules as needed, that is, the internal structure of the system can be divided into different functional modules to complete all or part of the functions described above. This is not a limitation here.
[0066] Embodiments of this application also provide a computer device, including: one or more processors; and a storage device for storing one or more programs, which, when executed by the one or more processors, cause the computer device to implement the vehicle personalized decision-making system optimization method based on multi-source implicit feedback provided in the above embodiments.
[0067] Figure 3 A schematic diagram of the structure of a computer system suitable for an embodiment of this application is shown. It should be noted that... Figure 3 The computer system 300 of the electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.
[0068] like Figure 3 As shown, the computer system 300 includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage section 308 into a random access memory (RAM) 303, such as performing the methods described in the above embodiments. Various programs and data required for system operation are also stored in the RAM 303. The CPU 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304. The following components are connected to the I / O interface 305: an input section 306 including a keyboard, mouse, etc.; an output section 307 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN (local area network) card, modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A driver 310 is also connected to the I / O interface 305 as needed. Removable media 311, such as disks, optical discs, magneto-optical discs, semiconductor memories, etc., are installed on drive 310 as needed so that computer programs read from them can be installed into storage section 308 as needed.
[0069] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer tool programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 309, and / or installed from removable medium 311. When the computer program is executed by central processing unit (CPU) 301, it performs various functions defined in the system of this application.
[0070] It should be noted that the computer-readable medium shown in the embodiments of this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, flash memory, an optical fiber, a portable compact disk read-only memory, an optical storage device, a magnetic storage device, or any suitable combination thereof. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable computer program. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. Computer programs contained on computer-readable media can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.
[0071] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. Each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.
[0072] The units described in the embodiments of this application can be implemented by tools or by hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the unit itself.
[0073] Another aspect of this application provides a computer-readable storage medium storing a computer program thereon, which, when executed by a computer's processor, causes the computer to perform the aforementioned optimization method for a vehicle personalization decision-making system based on multi-source implicit feedback. This computer-readable storage medium may be included in the computer device described in the above embodiments, or it may exist independently and not incorporated into the computer device.
[0074] Another aspect of this application provides a computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the vehicle personalization decision-making system optimization method based on multi-source implicit feedback provided in the various embodiments above.
[0075] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.
Claims
1. An optimization method for a vehicle personalized decision-making system based on multi-source implicit feedback, characterized in that, Includes the following steps: Step S100: When the vehicle executes a collaborative control strategy initiated by the personalized decision-making system, the metadata of the collaborative control strategy is recorded. Step S200: During the preset feedback acquisition period after the execution of the collaborative control strategy, at least two types of multi-source heterogeneous feedback signals reflecting the user's non-command interaction behavior are collected simultaneously. Step S300: Associate the multi-source heterogeneous feedback signal with the metadata, parse it into structured feedback events, and assign weights to each structured feedback event according to a preset confidence evaluation rule to obtain a weighted feedback event set. Step S400: Based on the weighted feedback event set, incrementally update the personalized decision model that generates the collaborative control strategy to optimize the personalized decision model's strategy generation capability for the corresponding user and triggering scenario.
2. The optimization method according to claim 1, characterized in that, In step S100, the metadata includes the trigger scenario identifier, the set of execution parameters, and the expected user state target.
3. The optimization method according to claim 2, characterized in that, In step S200, the multi-source heterogeneous feedback signal includes: Direct operation signals are operation event signals that allow users to manually reverse the parameters that are automatically adjusted by the system. Physiological regression signal, which is the deviation trend signal between the user's actual physiological state indicators and the expected user state target; Behavior continuation signals are subsequent operation event signals performed by the user that are consistent with the intent of the collaborative control strategy.
4. The optimization method according to claim 3, characterized in that, In step S300, the confidence assessment rule is calculated based on the strength of the user intent represented by the structured feedback event, wherein... Feedback events that represent a user's explicit negative intent toward the collaborative control strategy are assigned a high level of confidence weight; Feedback events that characterize a user’s significant physiological response to the collaborative control strategy or their active compounding actions are assigned a medium level of confidence weight. Feedback events that characterize a user's fine-tuning or fuzzy response to the collaborative control strategy are assigned a low level of confidence weight.
5. The optimization method according to claim 4, characterized in that, In step S400, the personalized decision-making model includes a dynamic user profile sub-model, which is used to store and learn the user's long-term preference features and short-term state features. The scenario strategy matrix is used to store the mapping relationship between user-scenario pairs and strategy parameters.
6. The optimization method according to claim 5, characterized in that, The personalized decision-making model further includes a feedback learning sub-model, which is configured to: directly adjust the strategy parameters of a specific user-scenario pair based on the positive or negative feedback events corresponding to the high-level confidence weights; and update the user's long-term preference features based on the positive feedback events corresponding to the high-level confidence weights.
7. The optimization method according to claim 6, characterized in that, The method further includes triggering a re-evaluation and calibration of the collaborative control strategy in the same user-scenario when the number of feedback events collected and assigned the high confidence weight in the same user-scenario reaches a preset number.
8. A vehicle personalized decision-making system optimization system based on multi-source implicit feedback, characterized in that, include: A strategy recording model is used to record metadata of a collaborative control strategy initiated by a personalized decision-making system when a vehicle executes such a strategy. The feedback acquisition module is used to simultaneously acquire at least two types of multi-source heterogeneous feedback signals reflecting non-command interaction behavior of the user during a preset feedback acquisition period after the execution of the collaborative control strategy. The feedback analysis module is used to associate the multi-source heterogeneous feedback signals with the metadata, parse them into structured feedback events, and assign weights to each structured feedback event according to a preset confidence evaluation rule to obtain a weighted set of feedback events. The incremental optimization module is used to incrementally update the personalized decision model that generates the collaborative control strategy based on the weighted feedback event set, so as to optimize the strategy generation capability of the personalized decision model for the corresponding user and triggering scenario.
9. A computer-readable storage medium, characterized in that, It stores computer-readable instructions, which, when executed by the computer's processor, cause the computer to perform the vehicle personalization decision-making system optimization method based on multi-source implicit feedback as described in any one of claims 1 to 7.
10. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, when the processor executes the computer program, it implements the steps of the vehicle personalized decision-making system optimization method based on multi-source implicit feedback as described in any one of claims 1 to 7.