Intelligent cabin personalized adaptive control method and system

Abstract

The application discloses an intelligent cabin personalized adaptive control method and system, and the method comprises the following steps: identifying the identity of a current driver or passenger to index a corresponding local preference model; collecting current environmental parameters in real time to construct a current context vector; inputting the current context vector into the preference model to obtain an optimal control vector and calculating the similarity between the current context vector and a historical context vector of the driver or passenger; when the similarity exceeds a preset threshold, automatically performing a thermal comfort pre-adjustment operation according to the optimal control vector; monitoring the adjustment behavior of the driver or passenger within a preset time window after the pre-adjustment operation, calculating an instant reward according to the intervention amplitude of the adjustment behavior relative to the pre-adjustment operation; and updating the strategy parameters of the preference model online. The application improves the personalized operation efficiency and the automation level of the intelligent cabin.

Description

Technical Field

[0001] This invention relates to the automotive field, and in particular to a personalized adaptive control method and system for an intelligent cockpit. Background Technology

[0002] As vehicles evolve towards electrification, intelligence, and connectivity, the cabin has transformed from a simple transportation space into a user-centric intelligent service platform. Cabin thermal comfort, as a key factor influencing the driving experience, directly impacts the user's perception of the overall vehicle quality through the intelligence level of its control strategies.

[0003] Traditional vehicle cabin thermal comfort control primarily employs rule-based or simple PID control strategies. While these methods are stable, they lack flexibility and adaptability. Current attempts at personalization often involve preset fixed modes (such as energy-saving mode, comfort mode, etc.) or relying on large cloud models for processing, but these suffer from response latency, data privacy issues, and limitations in offline scenarios. Summary of the Invention

[0004] In view of the above situation, it is necessary to provide a personalized adaptive control method and system for intelligent cockpits to address the problems in the existing technology.

[0005] This invention provides a personalized adaptive control method for an intelligent cockpit, comprising the following steps: Identify the current driver and passengers to index the corresponding local preference model; Real-time collection of current environmental parameters to construct the current context vector; The current context vector is input into the preference model to obtain the optimal control vector, and the similarity between the current context vector and the driver's historical context vector is calculated. When the similarity exceeds a preset threshold, thermal comfort pre-adjustment is automatically performed according to the optimal control vector. The system monitors the adjustment behavior of drivers and passengers within a preset time window after the thermal comfort pre-adjustment operation, calculates an immediate reward based on the intervention magnitude of the adjustment behavior relative to the thermal comfort pre-adjustment operation, and inputs the current context vector, the optimal control vector, and the immediate reward as a training sample into the preference model to update the policy parameters of the preference model online.

[0006] Furthermore, in the above method, the formula for calculating the instant reward r in the reward calculation step is: , in, The variable Δ represents the magnitude of intervention and indicates whether the driver or passenger intervenes. The estimated energy consumption penalty term resulting from the thermal comfort pre-adjustment operation is represented by w1, w2, and w3, which are adaptively adjustable weighting coefficients.

[0007] Furthermore, in the above method, the step of inputting the current context vector, the optimal control vector, and the immediate reward as a training sample into the preference model, and updating the policy parameters of the preference model online, specifically includes: The training samples are simultaneously input into the short-term memory model and the long-term memory model; The parameters of the short-term memory model are updated with a first learning rate to capture temporary changes in the driver's or passenger's preferences during the current journey or in the near future; and, The parameters of the long-term memory model are updated with a second learning rate lower than the first learning rate to solidify the stable preferences of drivers and passengers across trips.

[0008] Furthermore, the above method also includes a cold start step: When new drivers or passengers use the system for the first time or cannot obtain a complete local preference model, a general initial model generated based on group statistics or factory calibration data is loaded as the preference model.

[0009] Furthermore, in the above method, after the step of automatically performing thermal comfort pre-adjustment operation according to the optimal control vector when the similarity exceeds a preset threshold, the method further includes: If any of the following abnormal events are detected, the thermal comfort pre-adjustment operation will be immediately interrupted and the system will revert to the preset safety control mode: Receives a reverse adjustment operation from the driver or passenger that exceeds the preset range; The cabin temperature was detected to exceed the preset safety threshold. Or it may receive a device malfunction signal from the on-board diagnostic system.

[0010] Furthermore, in the above method, in the step of calculating the similarity between the current context vector and the driver's historical context vector, a weighted Euclidean distance algorithm is used to calculate the similarity. The weight coefficients of each environmental factor in the weighted Euclidean distance algorithm are determined based on the driver's sensitivity to each environmental factor in historical feedback, wherein environmental factors with higher sensitivity are assigned higher weights.

[0011] This invention also discloses an intelligent cockpit personalized adaptive control system, comprising: The indexing module is used to identify the identity of the current driver and passengers in order to index the corresponding local preference model; The context vector construction module is used to collect current environmental parameters in real time and construct the current context vector; The similarity calculation module is used to input the current context vector into the preference model to obtain the optimal control vector, and to calculate the similarity between the current context vector and the driver's historical context vector. The execution module is used to automatically perform thermal comfort pre-adjustment operations based on the optimal control vector when the similarity exceeds a preset threshold. The update module is used to monitor the adjustment behavior of drivers and passengers within a preset time window after the thermal comfort pre-adjustment operation, calculate the instant reward based on the intervention magnitude of the adjustment behavior relative to the thermal comfort pre-adjustment operation, and input the current context vector, the optimal control vector, and the instant reward as a training sample into the preference model to update the policy parameters of the preference model online.

[0012] Furthermore, the aforementioned system also includes a security control module for: If any of the following abnormal events are detected, the thermal comfort pre-adjustment operation will be immediately interrupted and the system will revert to the preset safety control mode: Receives a reverse adjustment operation from the driver or passenger that exceeds the preset range; The cabin temperature was detected to exceed the preset safety threshold. Or it may receive a device malfunction signal from the on-board diagnostic system.

[0013] The present invention also discloses a readable storage medium having a program stored thereon, which, when executed by a processor, implements any of the methods described above.

[0014] The present invention also discloses an electronic device, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described in any of the above-mentioned embodiments. This invention utilizes online learning and context matching mechanisms to automatically adapt to user habits and environmental changes, significantly reducing the need for manual intervention and improving operational efficiency and automation. The user preference model constructed in this invention not only optimizes personalized services in real time but also employs a local modeling and edge computing architecture. This achieves millisecond-level response and low-latency services while ensuring user data is processed locally, effectively protecting privacy and compliance. Attached Figure Description

[0015] Figure 1 A flowchart of the intelligent cockpit personalized adaptive control method provided in the first embodiment of the present invention; Figure 2 This is a structural block diagram of the intelligent cockpit personalized adaptive control system provided in the second embodiment of the present invention; Figure 3This is a schematic diagram of the structure of an electronic device in an embodiment of the present invention. Detailed Implementation

[0016] Embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0017] These and other aspects of the embodiments of the invention will become clear from the following description and accompanying drawings. In these descriptions and drawings, some specific embodiments of the invention are specifically disclosed to illustrate some ways of implementing the principles of the embodiments of the invention; however, it should be understood that the scope of the embodiments of the invention is not limited thereto. Rather, the embodiments of the invention include all variations, modifications, and equivalents falling within the spirit and scope of the appended claims.

[0018] Please see Figure 1 The first embodiment of the present invention is a smart cockpit personalized adaptive control method, which includes steps S11 to S15.

[0019] Step S11: Identify the identity of the current driver / passenger to index the corresponding local preference model.

[0020] After the vehicle starts, user identification is achieved through account login, biometrics (fingerprint, facial recognition, seat pressure), or mobile phone Bluetooth binding. Based on the identified identity, the corresponding preference model is retrieved from local storage. Specifically, in one implementation of this invention, facial images of the driver and passengers can be captured by an in-vehicle camera, and facial feature vectors are extracted using a convolutional neural network (CNN) with a feature dimension of 512. Simultaneously, the voice signals of the driver and passengers are captured by an in-vehicle microphone, and voice features are extracted using Mel-frequency cepstral coefficients (MFCC), then speaker identification is performed using a deep neural network (DNN). The facial recognition results and voice recognition results are then weighted and fused; for example, the facial recognition weight is 0.7, and the voice recognition weight is 0.3, to obtain the final identity recognition result.

[0021] Furthermore, when a new driver or passenger uses the model for the first time or cannot access the complete local preference model, a general initial model generated based on group statistics or factory calibration data is loaded as the preference model. In addition, during the first N interactions of the driver or passenger, an exploration rate higher than that during the stable operation period is adopted to accelerate the convergence of the preference model to the individual preferences of the new driver or passenger.

[0022] Understandably, the local preference model in this embodiment aims to be lightweight, interpretable, and low-latency, and prioritizes the use of online learning algorithms that can run on the vehicle side, combined with a context-aware multi-armed bandit framework to achieve the exploration-exploitation balance.

[0023] The multi-armed slot machine framework views each control decision as selecting an arm (control vector) given a context. The arm's reward is calculated by considering whether the user subsequently intervenes, the magnitude of the intervention, and the direction of the intervention. A strategy with upper confidence bounds (UCB) or Bayesian posterior sampling (Thompson Sampling) is employed to balance exploration and exploitation.

[0024] Step S12: Collect current environmental parameters in real time and construct the current context vector.

[0025] Current environmental parameters mainly include interior / exterior temperature, humidity, solar radiation intensity, seat temperature, vehicle speed, window status, occupant distribution, and time information. These can be collected by sensors installed on the vehicle, such as interior temperature sensors, exterior temperature sensors, humidity sensors, solar radiation sensors, window / sunroof status sensors, seat surface temperature sensors, seat pressure distribution sensors, wind speed / direction sensors, and vehicle speed sensors.

[0026] Collect environmental parameters to construct the current context vector. An example format for the current context vector C_t is: C_t = {T_in: 26.3, T_out: 32.1, RH: 45%, Solar: 800W / m 2 , SeatTemp: 28.0, CarSpeed:60km / h, Window: closed, TimeOfDay: 14:30}, where T_in and T_out are the interior and exterior temperatures of the vehicle, RH is the humidity, Solar is the light intensity, SeatTemp is the seat temperature, and CarSpeed ​​is the vehicle speed.

[0027] Step S13: Input the current context vector into the preference model to obtain the optimal control vector, and calculate the similarity between the current context vector and the driver's historical context vector.

[0028] The collected current context vector C_T is input into the preference model, and the model infers and outputs the optimal control vector. A* ,like A* = {T:25°C, Fan:2, SeatHeat:0}.

[0029] Next, the similarity between the current context vector C_T and the driver's historical context vector C_{hist} is calculated, S = sim(C_t, C_{hist}). Specifically, the existing weighted Euclidean distance algorithm is used to calculate this similarity. The weight coefficients of each environmental factor in this weighted Euclidean distance algorithm are determined based on the driver's sensitivity to each environmental factor in their historical feedback. Among them, environmental factors with higher sensitivity are assigned higher weights (for example, if a user is sensitive to sunlight, then sunlight has a high weight).

[0030] Step S14: When the similarity exceeds a preset threshold, thermal comfort pre-adjustment operation is automatically performed according to the optimal control vector.

[0031] When an authorized user is identified and the similarity between the current environmental context and the user's historical preferences exceeds a preset threshold, the system generates a control vector (target temperature, airflow, airflow direction, seat heating / ventilation level, etc.) and performs thermal comfort pre-adjustment, directly adjusting the device parameters. That is, when the similarity S ≥ (Preset threshold) Perform thermal comfort pre-adjustment operation.

[0032] Understandably, when the similarity is less than the threshold, a pre-adjustment prompt is executed. This prompt can be displayed on the central control panel or HUD and the user must confirm. For example, the central control panel could display a prompt saying, "The preset temperature is 26°C. Do you accept?" followed by a short confirmation period.

[0033] In this embodiment, the thermal comfort pre-adjustment operation and the prompt pre-adjustment operation are two execution modes that can be determined by user preferences and safety policies. The thermal comfort pre-adjustment operation is used when the confidence level is high and the historical acceptance rate is high (such as when the similarity exceeds the threshold), while the prompt pre-adjustment operation is used when the confidence level is moderate or the user preference is uncertain, and a user confirmation interface is provided.

[0034] Step S15: Monitor the adjustment behavior of the driver and passengers within a preset time window after the thermal comfort pre-adjustment operation, calculate the immediate reward based on the intervention magnitude of the adjustment behavior relative to the thermal comfort pre-adjustment operation, and input the current context vector, the optimal control vector, and the immediate reward as a training sample into the preference model to update the policy parameters of the preference model online.

[0035] User adjustment behaviors mainly include: adjustment events made through voice commands, air conditioning panel, touch screen, steering wheel shortcut keys, etc. (such as temperature, air volume, air direction, seat heating / ventilation, steering wheel heating, etc.).

[0036] The system monitors the adjustment behavior of drivers and passengers within a preset time window after a pre-adjustment operation. If the user does not intervene and accepts the pre-adjustment, it is considered positive feedback and a high reward r_pos is given. If the user adjusts in the opposite direction within the preset time, the intervention magnitude is calculated. = |to - from| and the direction of intervention, mapped to negative or low reward r_{neg}.

[0037] The formula for calculating the instant reward r is, for example:

[0038] in, The variable Δ represents the magnitude of intervention and indicates whether the driver or passenger intervenes. The estimated energy consumption penalty term resulting from the thermal comfort pre-adjustment operation is represented by w1, w2, and w3, which are adaptively adjustable weighting coefficients.

[0039] The reward function takes into account factors such as the number of user interventions (the fewer the better), the magnitude of intervention (the smaller the better), the user confirmation rate (the proportion of pre-adjustment that is accepted), and the impact of energy consumption (whether pre-adjustment increases energy consumption). It adopts a weighted linear or non-linear combination and can adaptively adjust the weights over time.

[0040] The system monitors user actions within a preset time frame (e.g., adjusting the temperature from the system's preset 25°C to 26°C), recording the event E = {userID, timestamp, actionType: setTemp, from:25, to:26, context:C_t}. Each manual adjustment by the user is considered a feedback signal to the current strategy, and the system maps this feedback to an immediate reward. For example, if a user increases the temperature from the system's preset value to 26°C within 2 minutes (an intervention increment of 1°C), it is considered that the system underestimates the preference, and a negative or low reward is given. The reward is calculated as r = -0.2, and the model is updated online, with a high short-term memory learning rate for rapid adaptation.

[0041] The steps of using the current context vector, the optimal control vector, and the immediate reward as a training sample and inputting them into the preference model to update the policy parameters of the preference model online specifically include: The training samples are simultaneously input into the short-term memory model and the long-term memory model; The parameters of the short-term memory model are updated with a first learning rate to capture temporary changes in the driver's or passenger's preferences during the current journey or in the near future; and, The parameters of the long-term memory model are updated with a second learning rate lower than the first learning rate to solidify the stable preferences of drivers and passengers across trips.

[0042] The current context vector, optimal control vector, and immediate reward are used as a training sample (C_t, A*, r) and input into the preference model. The policy parameters of the preference model are updated online, and the exploration rate is adjusted. Simultaneously, short-term memory and long-term memory are updated with different learning rates. The short-term memory window (e.g., the last 7 days) is used to capture temporary preference changes (e.g., seasonal changes, temporary physical discomfort), while long-term memory is used to stabilize preferences. A dual learning rate is used during model updates to update short-term and long-term parameters separately. After updating, a model summary (parameter hash, version, statistical indicators) is written to local encrypted storage. When the synchronization strategy is met, the summary or aggregated statistics are reported to the BI platform for product optimization, group preference analysis, and service innovation. Differential privacy or aggregated reporting is used during the synchronization process to reduce privacy risks.

[0043] Furthermore, after the step of automatically performing thermal comfort pre-adjustment operation according to the optimal control vector when the similarity exceeds a preset threshold, the method further includes: If any of the following abnormal events are detected, the thermal comfort pre-adjustment operation will be immediately interrupted and the system will revert to the preset safety control mode: Receives a reverse adjustment operation from the driver or passenger that exceeds the preset range; The cabin temperature was detected to exceed the preset safety threshold. Or it may receive a device malfunction signal from the on-board diagnostic system.

[0044] This embodiment sets safe upper and lower limits for cabin temperature and equipment operation (to prevent extreme pre-adjustments from causing passenger discomfort or equipment overload), and immediately reverts to safe mode when strong reverse operation or abnormal physiological signals from the user are detected.

[0045] The safety control mode includes: resetting the target temperature setting to the factory-calibrated safety reference temperature (e.g., 24℃ in summer and 22℃ in winter), reducing the blower speed to level 2, and turning off all high-level seat heating and ventilation functions; at the same time, freezing the online update process of the local preference model until the abnormal state is eliminated and a manual recovery command from the user is received.

[0046] This embodiment utilizes online learning and context matching mechanisms to automatically adapt to user habits and environmental changes, significantly reducing the need for manual intervention and improving operational efficiency and automation. The user preference model constructed in this embodiment not only optimizes personalized services in real time but also employs a local modeling and edge computing architecture. This achieves millisecond-level response and low-latency services while ensuring user data is processed locally, effectively protecting privacy, security, and compliance.

[0047] Please see Figure 2The intelligent cockpit personalized adaptive control system in the second embodiment of the present invention includes: Index module 21 is used to identify the identity of the current driver or passenger in order to index the corresponding local preference model; Context vector construction module 22 is used to collect current environmental parameters in real time and construct the current context vector; The similarity calculation module 23 is used to input the current context vector into the preference model to obtain the optimal control vector, and to calculate the similarity between the current context vector and the driver's historical context vector; Execution module 24 is used to automatically perform thermal comfort pre-adjustment operation according to the optimal control vector when the similarity exceeds a preset threshold; The update module 25 is used to monitor the adjustment behavior of drivers and passengers within a preset time window after the thermal comfort pre-adjustment operation, calculate the instant reward based on the intervention magnitude of the adjustment behavior relative to the thermal comfort pre-adjustment operation, and input the current context vector, the optimal control vector and the instant reward as a training sample into the preference model to update the policy parameters of the preference model online.

[0048] Furthermore, the aforementioned system also includes a security control module for: If any of the following abnormal events are detected, the thermal comfort pre-adjustment operation will be immediately interrupted and the system will revert to the preset safety control mode: Receives a reverse adjustment operation from the driver or passenger that exceeds the preset range; The cabin temperature was detected to exceed the preset safety threshold. Or it may receive a device malfunction signal from the on-board diagnostic system.

[0049] The intelligent cockpit personalized adaptive control system provided in this embodiment of the invention has the same implementation principle and technical effects as the aforementioned method embodiment. For the sake of brevity, any parts not mentioned in the device embodiment can be referred to the corresponding content in the aforementioned method embodiment.

[0050] In another aspect, the present invention also proposes an electronic device, please refer to [link to relevant documentation]. Figure 3 The image shows an electronic device according to an embodiment of the present invention, including a processor 10, a memory 20, and a computer program 30 stored in the memory and executable on the processor. When the processor 10 executes the computer program 30, it implements the intelligent cockpit personalized adaptive control method as described above.

[0051] The electronic device may be a controller installed in the vehicle, such as an integrated body controller or a cockpit domain controller. In some embodiments, the processor 10 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip, used to run program code stored in the memory 20 or process data.

[0052] The memory 20 includes at least one type of readable storage medium, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 20 can be an internal storage unit of an electronic device, such as the hard disk of the electronic device. In other embodiments, the memory 20 can also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. Furthermore, the memory 20 can include both internal and external storage units of the electronic device. The memory 20 can be used not only to store application software and various types of data installed on the electronic device, but also to temporarily store data that has been output or will be output.

[0053] Optionally, the electronic device may further include a user interface, a network interface, a communication bus, etc. The user interface may include a display, an input unit such as a keyboard, and optionally, a standard wired interface or a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device and to display a visual user interface. The network interface may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface), typically used to establish communication connections between the device and other electronic devices. The communication bus is used to enable communication between these components.

[0054] It should be pointed out that, Figure 3 The structure shown does not constitute a limitation on the electronic device. In other embodiments, the electronic device may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0055] The present invention also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the intelligent cockpit personalized adaptive control method described above. Those skilled in the art will understand that the logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequential 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 or apparatus (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from or in conjunction with such an instruction execution system or apparatus). For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit programs for use by or in conjunction with an instruction execution system or apparatus.

[0056] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0057] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0058] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0059] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A personalized adaptive control method for an intelligent cockpit, characterized in that, Includes the following steps: Identify the current driver and passengers to index the corresponding local preference model; Collect current environmental parameters in real time and construct the current context vector; The current context vector is input into the preference model to obtain the optimal control vector, and the similarity between the current context vector and the driver's historical context vector is calculated. When the similarity exceeds a preset threshold, thermal comfort pre-adjustment is automatically performed according to the optimal control vector. The system monitors the adjustment behavior of drivers and passengers within a preset time window after the thermal comfort pre-adjustment operation, calculates an immediate reward based on the intervention magnitude of the adjustment behavior relative to the thermal comfort pre-adjustment operation, and inputs the current context vector, the optimal control vector, and the immediate reward as a training sample into the preference model to update the policy parameters of the preference model online.

2. The method according to claim 1, characterized in that, In the reward calculation step, the formula for calculating the instant reward is: ， Where r represents the instant reward. The variable Δ represents the magnitude of intervention and indicates whether the driver or passenger intervenes. The estimated energy consumption penalty term resulting from the thermal comfort pre-adjustment operation is represented by w1, w2, and w3, which are adaptively adjustable weighting coefficients.

3. The method according to claim 1, characterized in that, The step of using the current context vector, the optimal control vector, and the immediate reward as a training sample and inputting them into the preference model to update the policy parameters of the preference model online specifically includes: The training samples are simultaneously input into the short-term memory model and the long-term memory model; The parameters of the short-term memory model are updated with a first learning rate to capture temporary changes in the driver's preferences during the current trip or in the near future, and the parameters of the long-term memory model are updated with a second learning rate lower than the first learning rate to solidify the driver's stable preferences across trips.

4. The method according to claim 1, characterized in that, It also includes a cold start procedure: When new drivers or passengers use the system for the first time or cannot obtain a complete local preference model, a general initial model generated based on group statistics or factory calibration data is loaded as the preference model.

5. The method according to claim 1, characterized in that, When the similarity exceeds a preset threshold, the step of automatically performing thermal comfort pre-adjustment based on the optimal control vector further includes: If any of the following abnormal events are detected, the thermal comfort pre-adjustment operation will be immediately interrupted and the system will revert to the preset safety control mode: Receives a reverse adjustment operation from the driver or passenger that exceeds the preset range; The cabin temperature was detected to exceed the preset safety threshold. Or it may receive a device malfunction signal from the on-board diagnostic system.

6. The method according to claim 1, characterized in that, In the step of calculating the similarity between the current context vector and the driver's historical context vector, the similarity is calculated using a weighted Euclidean distance algorithm. The weight coefficients of each environmental factor in the weighted Euclidean distance algorithm are determined based on the driver's sensitivity to each environmental factor in historical feedback, wherein environmental factors with higher sensitivity are assigned higher weights.

7. A personalized adaptive control system for an intelligent cockpit, characterized in that, include: The indexing module is used to identify the identity of the current driver and passengers in order to index the corresponding local preference model; The context vector construction module is used to collect current environmental parameters in real time and construct the current context vector; The similarity calculation module is used to input the current context vector into the preference model to obtain the optimal control vector, and to calculate the similarity between the current context vector and the driver's historical context vector. The execution module is used to automatically perform thermal comfort pre-adjustment operations based on the optimal control vector when the similarity exceeds a preset threshold. The update module is used to monitor the adjustment behavior of drivers and passengers within a preset time window after the thermal comfort pre-adjustment operation, calculate the instant reward based on the intervention magnitude of the adjustment behavior relative to the thermal comfort pre-adjustment operation, and input the current context vector, the optimal control vector, and the instant reward as a training sample into the preference model to update the policy parameters of the preference model online.

8. The system according to claim 7, characterized in that, It also includes a safety control module for: If any of the following abnormal events are detected, the thermal comfort pre-adjustment operation will be immediately interrupted and the system will revert to the preset safety control mode: Receives a reverse adjustment operation from the driver or passenger that exceeds the preset range; The cabin temperature was detected to exceed the preset safety threshold. Or it may receive a device malfunction signal from the on-board diagnostic system.

9. A readable storage medium having a program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-6.

10. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method as described in any one of claims 1-6.

Images