Air conditioner regulation and control system and vehicle

Abstract

According to the method and the device, differentiated air conditioner requirements of different users are identified by describing air conditioner use preferences including driver cold and hot preferences, cold and hot sensitivity, circulation preferences and air outlet mode preferences, personalized intelligent recommendation for different users is realized by fusing in-vehicle and out-vehicle environment state characteristics and whole vehicle operation characteristics, and meanwhile, by identifying personal scenes of the users, the user experience is improved. The air conditioner setting is adjusted in time, and more accurate comfort control is achieved.

Description

Technical Field

[0001] This application relates to the field of automotive air conditioning control technology, and in particular to an air conditioning control system and a vehicle. Background Technology

[0002] Traditional air conditioning control schemes have many limitations. Current schemes cannot actively and collaboratively adjust commonly used settings such as temperature, airflow, circulation, and airflow mode, requiring manual operation and thus being inconvenient. Furthermore, making adjustments while driving can easily distract the driver, creating safety hazards. Additionally, under the same operating conditions, different users have different air conditioning needs, and existing schemes cannot provide personalized adjustments for each individual.

[0003] Current technologies that achieve personalized comfort control by learning user habits primarily use data on air conditioning temperature, airflow, and mode settings under various in-vehicle environmental parameters, weather conditions, and geographical information to train a model. The model is then optimized based on whether the user accepts the current settings. However, this approach lacks analysis and preference characterization of user air conditioning usage habits, resulting in weak recommendation targeting. Furthermore, existing solutions rely entirely on user feedback to adjust recommendation probabilities when facing new operating conditions, requiring multiple interactions to learn user habits, leading to a lengthy learning process.

[0004] Current technologies address air conditioning comfort control for specific, complex scenarios, such as recognizing emotions or whether someone is resting in the vehicle, or, as described in the aforementioned patent, only consider automatic control based on environmental conditions. They fail to simultaneously consider environmental factors, user habits, and user scenarios, resulting in current control solutions that cannot meet the diverse and personalized air conditioning usage needs of users. Summary of the Invention

[0005] This application identifies differentiated air conditioning needs by characterizing drivers' preferences for temperature, humidity, air circulation, and airflow mode. It integrates features of the in-vehicle and external environmental conditions with overall vehicle operation characteristics to provide personalized intelligent recommendations. Furthermore, by recognizing individual user scenarios, it adjusts air conditioning settings in real time for more precise comfort control. To protect user privacy, an image processing module is located in the vehicle, creating a vehicle-cloud collaborative personalized intelligent air conditioning recommendation system. This includes: collecting and recording user data such as temperature, fan speed, circulation, and airflow mode settings using the air conditioning multimedia system to analyze user preferences; installing sensors around the vehicle, including but not limited to temperature, humidity, sunlight, and altitude sensors, with their signal outputs connected to the vehicle controller to collect data on external environmental conditions such as weather and environment; and installing sensors inside the vehicle, including but not limited to foot temperature sensors, temperature and humidity sensors at each air vent, and vehicle tilt sensors, to collect information on in-vehicle environmental characteristics and overall vehicle operation characteristics. By integrating the above information and inputting it into a cloud-based personalized intelligent air conditioning recommendation model, personalized air conditioning control is achieved, including control of vehicle-side air conditioning temperature, airflow, circulation mode, and air outlet mode. Furthermore, hardware such as cameras and infrared cameras installed inside the vehicle collects information about the user's individual circumstances. Based on personalized control, the air conditioning settings are dynamically adjusted to achieve more precise comfort control.

[0006] On one hand, an air conditioning control system is proposed, which includes: a multimedia module information acquisition module, an external sensor information acquisition module, an in-vehicle hardware information acquisition module, and an air conditioning personalized recommendation model; the multimedia module is used to collect user air conditioning usage information; the external sensor information acquisition module is used to collect external environmental information; the in-vehicle hardware information acquisition module is used to collect vehicle driving information, in-vehicle environmental information, and user scenario information; the air conditioning personalized recommendation model outputs air conditioning recommendation values, which include temperature, air volume, circulation mode, and air outlet mode.

[0007] Furthermore, the user's air conditioning usage information includes: temperature setting information, air volume setting information, circulation setting information, and air outlet mode setting information.

[0008] Furthermore, the external environmental information includes external temperature, external humidity, altitude, and vehicle tilt angle.

[0009] Furthermore, the vehicle driving information includes: vehicle speed and charging gun connection status; the in-vehicle environment information includes: in-vehicle temperature, air vent channel temperature, in-vehicle humidity, CO2 concentration, particulate matter information, and window status; and the user scenario information includes: in-vehicle camera information and in-vehicle infrared camera information.

[0010] Furthermore, the personalized air conditioner recommendation model includes: a user preference air conditioner recommendation model, a scenario-based air conditioner recommendation model, and a decision fusion model.

[0011] Furthermore, the user inputs the air conditioning information into the corresponding preference judgment module, which outputs preference features; the preference features and external temperature information are simultaneously input into the preference feature recognition module, which outputs preference features under different environmental conditions; the preference features under different environmental conditions are input into the user profile model, which outputs user preference tags.

[0012] Furthermore, the external environment information, weather forecast, and map information are input to the external environment recognition module, which outputs external environment features; the vehicle driving information is input to the vehicle driving feature recognition module, which outputs vehicle driving features; and the in-vehicle environment information is input to the in-vehicle environment recognition module, which outputs in-vehicle environment features.

[0013] Furthermore, the external environmental features, vehicle driving features, and internal environmental features are input into the vehicle operation scene feature recognition module to obtain the vehicle operation scene features.

[0014] Furthermore, the user scenario information is input into the user scenario feature recognition module, and the user scenario features are output; the user scenario feature recognition module includes: clothing feature recognition model, rest feature recognition model, and multi-target feature recognition model.

[0015] Furthermore, the vehicle operation scenario features and user preference tags are input into the user preference air conditioning recommendation model, and the user preference air conditioning recommendation value is output, including air conditioning temperature, air volume, circulation mode, and air outlet mode.

[0016] Furthermore, the user's contextual features are input into the scenario-based air conditioning recommendation model, which outputs multiple sets of scenario-based air conditioning recommendation values, namely the air conditioning temperature, air volume, circulation mode, and air outlet mode corresponding to the current scenario; the user's preferred air conditioning recommendation value and the multiple sets of scenario-based air conditioning recommendation values ​​are input into the decision fusion model, which outputs air conditioning recommendation values, including temperature, air volume, circulation mode, and air outlet mode.

[0017] Furthermore, at least one of the following modules—the preference judgment module, preference feature recognition module, external environment recognition module, internal environment recognition module, vehicle driving feature recognition module, vehicle operation scenario feature recognition module, user scenario-based feature recognition module, and user preference air conditioning recommendation model—is deployed on a remote server. On the other hand, a vehicle is proposed that includes the aforementioned air conditioning control system. Attached Figure Description

[0018] 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 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.

[0019] To gain a more complete understanding of this application and its beneficial effects, the following description will be provided in conjunction with the accompanying drawings, wherein the same reference numerals in the following description denote the same parts.

[0020] Figure 1 User preference feature identification process

[0021] Figure 2 Vehicle External Environment Feature Recognition Process

[0022] Figure 3 In-vehicle environment feature recognition process

[0023] Figure 4 Vehicle driving feature recognition process

[0024] Figure 5 Vehicle operation scenario feature recognition process

[0025] Figure 6 Recommendation process for user-preferred air conditioner models

[0026] Figure 7 Recommendation process for scene-based air conditioning recommendation models

[0027] Figure 8 Recommendation process for decision fusion models

[0028] Figure 9 Inference process for vehicle-cloud collaborative intelligent air conditioning Detailed Implementation

[0029] The technical solutions in this application will now be described with reference to the accompanying drawings.

[0030] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0031] In the embodiments of this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first" and "second" may explicitly or implicitly include one or more of that feature.

[0032] It should be understood that the terminology used in the description of the various examples herein is for the purpose of describing the particular examples only and is not intended to be limiting. As used in the description of the various examples, the singular forms “a” (“a”, “an”) and “the” are intended to include the plural forms as well, unless the context explicitly indicates otherwise.

[0033] In this application, "at least one" means one, two, or more, and "more than" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can mean: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0034] It should also be understood that, in this application, unless otherwise expressly specified and limited, the term "connection" should be interpreted broadly. For example, "connection" can be a fixed connection, a sliding connection, a detachable connection, or an integral part; it can be a direct connection or an indirect connection through an intermediate medium.

[0035] It should also be understood that the term “comprising” (also referred to as “includes”, “including”, “comprises” and / or “comprising”) as used in this specification specifies the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0036] It should be understood that the terms "an embodiment," "another embodiment," and "a possible design" used throughout the specification mean that a specific feature, structure, or characteristic related to an embodiment or implementation is included in at least one embodiment of this application. Therefore, phrases such as "in one embodiment of this application," "in another embodiment of this application," and "a possible design" appearing throughout the specification do not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments.

[0037] It should also be understood that the specific values ​​mentioned in the embodiments of this application are not intended to limit the specific dimensions of particular features or structures. The relevant values ​​may be illustrative examples for ease of understanding, or they may represent the theoretically optimal value for a certain feature. In practice, the relevant dimensions may be a range around the value, for example, the range may be ±50% of the optimal theoretical value, and the actual dimensions should be determined based on the achievement of the corresponding technical effect.

[0038] This application collects and records user air conditioning data such as driver temperature settings, fan speed settings, circulation settings, and air outlet mode settings based on multimedia, in order to identify user air conditioning usage preference information.

[0039] Furthermore, the system implemented in this application employs sensors including: temperature sensors, humidity sensors, altitude sensors, and vehicle tilt angle sensors installed outside the vehicle. The information output terminals of these hardware devices are connected to the vehicle controller to receive information from outside the vehicle.

[0040] In addition, the in-vehicle hardware devices of the system implemented in this application also include temperature sensors, temperature and humidity sensors for each air vent channel, vehicle speed sensors, CO2 concentration sensors, particulate matter sensors, window information sensors, charging gun connection status sensors, vehicle cameras, and vehicle infrared cameras, which are used to collect real-time vehicle operation information and user scenario information.

[0041] On one hand, this application considers the driver's usage preferences and accurately characterizes the driver's air conditioning usage preferences, including temperature preferences, temperature sensitivity, circulation preferences, and airflow mode preferences. The constructed user preference tags are used as input to train the model, achieving personalized comfort control. Compared to existing patent solutions, this approach directly uses user data on air conditioning outlet temperature, airflow, and mode settings under different in-vehicle environmental parameters, weather parameters, and geographical information for training and modeling. Inputting user profile features into the model training makes the recommendation results more targeted. Furthermore, for operating conditions not previously used by a single user, the model can generalize recommendations based on similar tags, reducing the learning time between the model and user feedback.

[0042] On the other hand, this application constructs a scenario-based air conditioning recommendation model for complex scenarios such as clothing, whether the user is resting, and whether there are multiple people present. It then utilizes a decision fusion model to achieve air conditioning comfort control in these complex scenarios. By enhancing the perception of complex scenarios, it meets the diverse and personalized needs of users, further improving their intelligent air conditioning experience.

[0043] The following is a detailed description of an air conditioning control system and a vehicle based on the present application, with reference to the accompanying drawings.

[0044] In some embodiments, as shown in Figure 1, this embodiment judges user preference features based on multi-dimensional data such as the user's historical temperature settings, airflow settings, circulation settings, air outlet settings, and external temperature information. Specifically, when the temperature settings are input into a pre-trained hot / cold preference model and a hot / cold sensitivity model, the models, through deep learning and feature extraction of the user's long-term temperature adjustment habits, can accurately generate temperature preference features reflecting the user's tendency towards high or low temperatures, and temperature sensitivity features reflecting the user's tolerance to temperature changes, effectively avoiding preference judgment bias caused by a single data dimension. Similarly, when the airflow settings are input into the aforementioned hot / cold preference model and hot / cold sensitivity model, the correlation between airflow adjustment and the user's perceived hot / cold needs can be further combined to generate airflow preference features that complement the temperature preferences, and can... The module differentiates users' perception of subtle changes in airflow based on their airflow sensitivity, making the preference feature system more aligned with users' actual tactile needs in real-world usage scenarios. When loop setting information is input into the loop mode preference model, the module can generate scenario-adaptive loop mode preference features based on users' loop mode selection records in different usage scenarios (such as cool environments in spring and autumn, and hot environments in summer), providing accurate loop mode decision-making basis for subsequent automatic device adjustment in different environments. When air outlet setting information is input into the air outlet mode preference model, it can generate exclusive air outlet mode preference features based on users' preferences for air outlet direction and air outlet area, further enhancing the personalized experience of device use. The aforementioned hot / cold preference model, hot / cold sensitivity model, loop mode preference model, and air outlet mode preference model all belong to the preference judgment module.

[0045] Subsequently, the generated preference features and the real-time collected external temperature information are input into the preference feature recognition module. This module, through a multi-feature fusion algorithm, can effectively integrate user preferences with external environmental factors, thereby obtaining preference features with stronger adaptability to different environmental conditions (such as high temperature weather, low temperature weather, and weather with large temperature differences), avoiding the problem of preference feature failure due to ignoring the influence of the external environment. Finally, these preference features adapted to different environmental conditions are input into the user profile model. This model, through in-depth analysis and modeling of multi-dimensional and multi-scenario preference data, can generate more comprehensive, more accurate, and more user-friendly preference tags, providing reliable data support for subsequent automatic adjustment and intelligent control of equipment based on user preferences, significantly improving the intelligence level of the equipment and user satisfaction.

[0046] As shown in Figure 2, this embodiment further activates external vehicle sensors to achieve real-time and continuous acquisition of multi-dimensional environmental data. These sensors have high sensitivity and low latency characteristics, enabling them to accurately and continuously collect information about the external environment. This includes temperature and humidity information that directly affect the user's comfort and the device's operating status (e.g., in high-temperature and high-humidity environments, they can help determine the device's heat dissipation needs, and in low-temperature and low-humidity environments, they can adapt to the user's potential needs for warmth and humidity control). It also covers altitude information that has a significant impact on the vehicle's operating status and environmental adaptability (e.g., in high-altitude areas, it can help adjust the device's operating parameters to adapt to changes in air pressure) and vehicle tilt angle information (e.g., when driving on a slope, it can help optimize the device's stability and user experience). At the same time, through data interaction with the vehicle's multimedia system, it can acquire multimedia information related to the user's travel planning and environmental prediction in real time, including weather forecasts that can detect future environmental changes in advance (e.g., weather warnings such as precipitation and strong winds can help adjust the device's preset modes) and map information that can locate the specific scene where the vehicle is located (e.g., scene information such as urban roads and highways can help match device usage preferences in the corresponding environment). Subsequently, the collected external environment information and multimedia information are input into the external environment recognition module. This module uses a multi-source data collaborative processing algorithm, which can effectively eliminate redundancy and interference between different types of information, realize accurate judgment and feature extraction of the external environment status, and finally generate external environment features that comprehensively reflect the real-time external environment, future environmental trends and the scene in which the vehicle is located. This provides accurate environmental basis for the subsequent dynamic and intelligent adjustment of the device by combining user preference tags, further improving the adaptability of device adjustment and the convenience of user use.

[0047] As shown in Figures 3 and 4, this embodiment further activates the vehicle's internal sensors. These sensors employ high-precision sensing chips and multi-parameter synchronous acquisition technology, offering advantages such as fast response speed, small data errors, and strong continuous working stability. They enable all-weather, uninterrupted, and accurate monitoring of the vehicle's operating status and the in-vehicle environment, continuously collecting vehicle driving information and in-vehicle environmental information. Specifically, the vehicle driving information includes key parameters such as vehicle speed and charging gun connection status. Vehicle speed information reflects real-time changes in vehicle speed (e.g., the impact of airflow on the in-vehicle temperature conduction efficiency at high speeds, and the difference in the rate of heat accumulation at low speeds), providing a basis for dynamic adjustment of the air conditioning fan speed and airflow mode. The charging gun connection status accurately determines whether the vehicle is in charging mode (during charging, the vehicle battery heat may indirectly affect the in-vehicle temperature; in this case, the air conditioning cooling strategy can be adjusted in advance to avoid abnormal fluctuations in in-vehicle temperature while ensuring charging safety and efficiency). The in-vehicle environmental information includes multiple dimensions such as temperature, humidity, CO2 concentration sensor data, particulate matter sensor data, and window information. Temperature and humidity information are directly related to user comfort (e.g., users may feel stuffy in high-temperature and high-humidity environments, so it is necessary to lower the temperature and adjust the humidity appropriately; in low-temperature and low-humidity environments, it is necessary to focus on warmth and humidity replenishment), and are the basis for adjusting the core parameters of the air conditioning. CO2 concentration sensor data can monitor the air quality inside the vehicle in real time (when the CO2 concentration exceeds the comfort threshold, it can trigger the air conditioning external circulation mode to introduce fresh air and ensure the respiratory health of the occupants). Particulate matter sensor data can accurately capture the content of pollutants such as PM2.5 and dust inside the vehicle (if particulate matter exceeds the standard, it can link the air conditioning filter to enhance the filtration function and quickly improve the air quality inside the vehicle). Window information can reflect the ventilation status inside the vehicle (e.g., when the windows are open, the outside airflow will affect the stability of the temperature inside the vehicle, so it is necessary to adjust the air conditioning power to maintain the set temperature and avoid energy waste). Subsequently, the collected vehicle driving information is input into the vehicle driving feature recognition module. This module, through multi-dimensional analysis and feature extraction of data such as vehicle speed change trends and charging duration, can generate vehicle driving features that accurately reflect the current operating conditions of the vehicle (such as constant speed driving, acceleration driving, charging, parking, etc.), effectively avoiding the bias in operating condition judgment caused by a single data dimension. At the same time, the in-vehicle environment information is input into the in-vehicle environment recognition module. This module uses a multi-parameter fusion algorithm to integrate information such as temperature, humidity, air quality, and ventilation status, generating in-vehicle environment features that comprehensively reflect the real-time status and changing trends of the in-vehicle environment. This provides reliable in-vehicle data support for subsequent intelligent air conditioning adjustment based on user preferences and the external environment, further improving the accuracy and adaptability of adjustment decisions.

[0048] As shown in Figure 5, the external environmental features (including external temperature, humidity, altitude, vehicle tilt angle, weather forecast, map and other multi-dimensional environmental and scene information), vehicle driving features (accurately reflecting vehicle speed changes, charging status and other vehicle operating conditions), and internal environmental features (including internal temperature, humidity, air quality, window status and other user-perceived information) generated in this embodiment are all input into the whole vehicle operation scene feature recognition module. This module is equipped with a cross-dimensional feature fusion algorithm and a dynamic scene judgment model, possessing three core advantages: First, it can effectively break down information barriers between different types of features, avoiding the one-sidedness of scene judgment caused by independent analysis of a single feature. For example, when combining the features of high temperature outside the vehicle with the features of high humidity inside the vehicle and the features of high-speed driving, it can accurately identify the composite scene of "high-speed driving in high temperature and high humidity weather," rather than just judging a certain environment or operating state in isolation. Second, it has the ability to adapt to dynamically changing data in real time. When the external environment (such as a sudden increase in humidity due to sudden rainfall), the vehicle's driving state (such as switching from constant speed driving to acceleration driving), or the internal environment (such as a window suddenly opening) changes, the module can quickly integrate the updated three types of features and adjust the scene judgment logic in real time to ensure that the generated features match the actual working conditions synchronously. Third, it can filter redundant and interfering data. Through a feature weight dynamic allocation mechanism, it assigns higher weights to key information that affects scene judgment (such as extreme temperature, special road conditions, etc.) and reasonably weakens secondary or less fluctuating information (such as small humidity changes), further improving the accuracy of scene features. Through the above processing, the vehicle operation scenario features finally generated by this module can not only fully cover the full-dimensional information of "external environment - vehicle operation - in-vehicle status", but also accurately depict the correlation between different factors (such as the correlation between vehicle tilt angle changes and in-vehicle temperature adjustment needs in high-altitude environments, and the correlation between congested road information displayed on the map and heat accumulation in the vehicle when the vehicle is driving at low speed). This provides comprehensive and reliable scenario-based data support for combining scenario features with user preference tags to achieve precise air conditioning adjustment, effectively avoiding the problem of air conditioning adjustment not matching actual needs due to missing scenario information or judgment bias, and further consolidating the scenario perception foundation of vehicle intelligent control.

[0049] As shown in Figure 6, in this embodiment, the vehicle operation scenario features (composite scenario information that comprehensively reflects the external environment, vehicle operation and in-vehicle status) generated in the previous step and the user preference tags (personalized preference features generated based on the user's long-term usage habits) are input into the user preference air conditioning recommendation model. This model employs a scenario-preference deep matching algorithm and a multi-objective optimization strategy, possessing the following core advantages: First, it can accurately map scenario features to user preferences. For example, in a scenario of "high temperature and humidity + high-speed driving," the model can quickly pinpoint the user's preferred adjustment direction of "low temperature + high airflow + external circulation," ensuring that the recommended results not only conform to the user's long-term habits but also adapt to the actual needs of the current scenario. Second, it has dynamic weight allocation capabilities, automatically adjusting weights based on key factors affecting user comfort in different scenarios (such as extreme temperatures and special road conditions), making the recommended results more targeted. Third, it can collaboratively optimize output parameters. When recommending temperature, airflow, circulation mode, and air outlet mode, it not only considers the optimization of a single parameter but also takes into account the mutual influence between multiple parameters (such as increasing airflow may lead to faster temperature adjustment but increase energy consumption), thereby achieving the best balance between comfort, energy consumption, and health and safety. Through the above processing, the model's final output of user-preferred air conditioning recommendations not only includes the user's preferred temperature, airflow, circulation mode, and air outlet mode, but also dynamically fine-tunes these parameters based on real-time scenarios, making the recommendations more closely match the user's actual needs under specific operating conditions. This process not only improves the intelligence and personalization of air conditioning adjustments but also lays a solid foundation for subsequent dynamic and precise control based on real-time user status. It effectively avoids adjustment deviations caused by relying solely on historical preferences or the current scenario, significantly enhancing user experience and equipment operating efficiency.

[0050] As shown in Figure 7, this embodiment further activates in-vehicle cameras and infrared cameras. The in-vehicle camera has high-definition imaging and a wide viewing angle, which can completely capture the movements, postures, and clothing details of the people in the vehicle. The infrared camera can accurately sense changes in the surface temperature and the distribution of sweat through thermal imaging technology (for example, sweat areas will show unique thermal imaging characteristics due to differences in heat conduction). The two work together to achieve multi-dimensional and blind-spot-free monitoring of the state of the people in the vehicle, thereby continuously collecting image information including DMS (Driver Monitoring System) and OMS (Occupant Monitoring System). DMS images can focus on the driver's behavior (such as whether he / she is looking down or closing his / her eyes), while OMS images can cover the overall state of all occupants, providing a comprehensive and accurate image data source for subsequent feature recognition. These image information are then input into the user contextual feature recognition module, which integrates multiple specialized recognition models (including but not limited to clothing feature recognition models, rest feature recognition models, and multi-target feature recognition models). Each model achieves efficient and accurate recognition of specific features through specialized optimization algorithms: the clothing feature recognition model relies on image texture analysis and contour matching technology to accurately distinguish the user's clothing type (such as short sleeves, shirts, jackets, and down jackets), and indirectly judges the user's temperature tolerance through clothing thickness and material (for example, when wearing a down jacket, it can be predicted that the user has a higher tolerance for low temperatures, so there is no need to overheat; when wearing short sleeves, it is necessary to avoid excessive heat that may cause discomfort); the rest feature recognition model can determine whether the user is in a resting state by monitoring the person's limb posture (such as sitting, closing eyes) and movement frequency (when resting, the temperature can be adjusted to be closer to the comfortable sleeping range, while reducing airflow and noise to reduce interference with rest); the multi-target feature recognition model is based on target detection and counting algorithms, which can accurately count the number of people in the vehicle (when the number of people increases, the airflow can be increased and the air distribution optimized to ensure that people in each area can feel a comfortable temperature and avoid local stuffiness caused by an increase in the number of people). Through multi-model collaborative processing, comprehensive, detailed, and real-time-appropriate user scenario-based features are generated. These user scenario-based features are input into a scenario-based air conditioning recommendation model. This model has a built-in scenario-parameter mapping database and dynamic adjustment algorithm, which can match the optimal air conditioning parameters according to different user state combinations (e.g., "short sleeves + not resting + 2 people in the car" scenario corresponds to "low temperature + high air volume + external circulation + wide-angle air outlet", "jacket + resting + 1 person in the car" scenario corresponds to "medium temperature + low air volume + internal circulation + top air outlet"). It then outputs multiple sets of scenario-based air conditioning recommendation values ​​adapted to the current real-time scenario, specifically including the temperature, air volume, circulation mode, and air outlet mode corresponding to the current scenario. This provides accurate real-time scenario decision-making basis for subsequent integration with long-term user preference recommendation values, further improving the dynamic adaptability and humanization level of air conditioning adjustment.

[0051] As shown in Figure 8, in this embodiment, the user's preferred air conditioning recommendation value (generated based on the user's long-term usage habits and vehicle operation scenarios, representing the user's stable core needs, such as the user's daily summer preference for 24℃ temperature, medium fan speed, and automatic circulation mode) and multiple sets of scenario-based air conditioning recommendation values ​​(generated based on the real-time status of occupants, reflecting the user's current dynamic needs, such as recommending 22℃ temperature, high fan speed, and external circulation mode in the "3 people in the car" scenario, and recommending 23℃ temperature, medium fan speed, and internal circulation mode in the "shirt" scenario) are jointly input into the decision fusion model. This model is equipped with a dynamic weight allocation algorithm and a multi-objective collaborative optimization mechanism, possessing three key technical advantages: First, it can intelligently determine the priority compatibility of the two types of recommendation values ​​by analyzing the urgency of the scenario and the rigidity of user needs in real time (e.g., detecting "3 people in the car"). When the in-vehicle heat load increases rapidly, the weight of the scenario-specific air conditioning recommendation value will automatically increase to prioritize meeting immediate comfort needs. If there are no special changes in the scenario, the weight of the user's preferred air conditioning recommendation value will dominate, ensuring long-term consistency of habits and avoiding adjustment deviations caused by relying solely on a certain type of recommendation value. Secondly, it has the ability to reconcile parameter conflicts. When there are parameter differences between two types of recommendation values ​​(such as a user preference of 24℃ and a scenario recommendation of 22℃), the model will combine environmental change trends (such as a moderate shift towards the scenario recommendation value if the in-vehicle temperature continues to rise, and retaining the core range of the preferred value if the in-vehicle temperature is stable) and user feedback logic (such as judging the user's tolerance to temperature fluctuations through historical data and achieving a smooth transition within an acceptable range) to generate compromise optimization parameters that take into account the needs of both parties. Thirdly, it supports multi-dimensional parameter collaborative output. When determining temperature, air volume, circulation mode, and air outlet mode, it will simultaneously consider the linkage effect between parameters (such as increasing the air volume to accelerate cooling efficiency when lowering the temperature, while avoiding excessive air volume causing noise interference; and adjusting the air outlet direction when switching to external circulation to prevent external airflow from blowing directly on the user), ensuring the overall adaptability of the output parameters. Through the above processing, the decision fusion model finally outputs a set of optimal air conditioning recommendation values ​​(including specific temperature values, air volume levels, circulation mode types, and air outlet direction parameters). These recommendation values ​​not only continue the core preferences of users' long-term habits, but also accurately adapt to the real-time needs of the current "human-scenario" situation. They can drive the air conditioning system to dynamically adjust operating parameters according to changes in the scenario. For example, when the vehicle switches from a "constant speed driving in the city + normal user state" scenario to a "parked + user resting" scenario, the air conditioning can automatically and smoothly transition from "24℃ + medium air volume + automatic circulation" to "23℃ + low air volume + external circulation" without manual operation by the user, truly achieving dynamic and precise air conditioning control with deep collaboration between "human-scenario" and "human-scenario" situation.This process not only significantly improves the intelligence and automation of air conditioning control, but also reduces the risk of driver distraction by reducing the frequency of manual adjustments (especially for drivers), while taking into account both comfort and energy consumption optimization (such as prioritizing the use of preferred parameters in non-extreme scenarios to avoid energy waste caused by frequent and large adjustments), further enhancing the user experience and overall performance advantages of the vehicle's intelligent cockpit.

[0052] Furthermore, in some embodiments, the aforementioned modules and models can be selectively deployed on remote servers, i.e., in the cloud. As shown in Figure 9, in this embodiment, the user preference feature recognition module, the vehicle external environment recognition module, the vehicle internal environment recognition module, the vehicle driving feature recognition module, the whole vehicle operation scenario feature recognition module, and the core module of the user preference air conditioning recommendation model are deployed in the cloud. Relying on the high computing power, large-capacity storage, and multi-device collaboration capabilities of the cloud, the limitations of limited computing power and low data processing efficiency of traditional in-vehicle terminals are broken, and efficient integration and in-depth analysis of multi-dimensional data are achieved. In a cloud-deployed architecture, modules can collaborate through real-time data interaction: The user preference feature recognition module (Figure 1) can generate more accurate multi-dimensional preference features such as temperature and airflow preferences based on massive amounts of historical user data stored in the cloud (covering long-term temperature, airflow, and circulation mode settings), thus outputting user preference tags that better match users' long-term habits (compared to in-vehicle terminals relying solely on limited local data, the cloud can continuously improve the accuracy of preference tags through data accumulation and iterative optimization); the vehicle external environment recognition module (Figure 2) can combine real-time updated meteorological and geographic information databases in the cloud to perform secondary verification and supplementation on temperature, humidity, and altitude information collected by external vehicle sensors (e.g., correcting temperature detection deviations caused by local environmental interference from in-vehicle sensors using cloud-based weather forecast data, and improving road condition information corresponding to vehicle tilt angles using cloud-based map information), generating more comprehensive and reliable external environment features; Figures 3 and 4... The corresponding in-vehicle environment recognition module and vehicle driving feature recognition module can upload real-time data such as in-vehicle temperature, CO2 concentration, vehicle speed, and charging gun status to the cloud. With the help of the cloud's high-concurrency data processing capabilities, data cleaning and outlier removal (such as filtering invalid vehicle speed data and abnormal fluctuations in in-vehicle humidity caused by sensor momentary failures) can be completed quickly, thereby generating more accurate and stable vehicle driving features and in-vehicle environment features.

[0053] As shown in Figure 9, in this embodiment, the user scenario-based feature-related modules (including in-vehicle camera and infrared camera data acquisition modules, OMS / DMS image preprocessing modules, and various specialized recognition sub-models such as clothing feature recognition and multi-target feature recognition) are deployed in the cloud. Relying on the high computing power and massive data storage capabilities of the cloud, the limitations of in-vehicle terminals in image data processing speed and sample size are overcome. The cloud can receive in-vehicle image data uploaded by multiple vehicles in real time (covering diverse samples of different people, different clothing, and different in-vehicle scenarios). Through data cleaning and standardization, a large-scale user scenario-based feature sample library with comprehensive scenario coverage is constructed, providing a sufficient data foundation for module training. In a cloud environment, the user-scenario-based feature recognition module can leverage distributed computing resources to efficiently train and iteratively optimize various specialized sub-models. For example, for the clothing feature recognition sub-model, by inputting massive amounts of clothing image samples from different seasons and materials, the texture analysis and contour matching algorithm parameters are optimized, enabling it to more accurately distinguish clothing types such as short sleeves, shirts, jackets, and down jackets, and even identify subtle differences in clothing thickness (such as thin and thick jackets), further improving the accuracy of judging users' temperature tolerance. Through continuous training, the feature extraction accuracy and scene adaptability of the user-scenario-based feature recognition module are significantly improved. Once the module performance reaches preset indicators (such as feature recognition accuracy ≥95% and recognition latency ≤0.5 seconds), the trained module is solidified and deployed in the cloud. This not only enables unified access from multiple vehicles but also allows for incremental training by receiving new image data in real time from the cloud, ensuring that the module always adapts to constantly changing user scenarios and providing more reliable feature input for subsequent scene-based air conditioning recommendation models.

[0054] Meanwhile, as shown in Figure 9, in this embodiment, the relevant modules of the decision fusion model (including the recommendation value preprocessing module, weight allocation algorithm module, and parameter collaborative optimization module) are deployed in the cloud. Leveraging the computing power and data integration capabilities of the cloud, richer samples and more efficient computational support are provided for model training. During training, the cloud can aggregate user preference air conditioning recommendation values ​​(from long-term habit data of different users) uploaded by multiple vehicles and multiple sets of scenario air conditioning recommendation values ​​(from real-time demand data of different scenarios), constructing a diversified training sample library covering "different users - different scenarios - different recommendation value combinations" (such as samples containing various parameter conflicts and adaptations, such as "user preference 24℃ and scenario recommendation 22℃ in high-temperature scenarios" and "user preference 25℃ and scenario recommendation 26℃ in rest scenarios"). Based on this sample database, the decision fusion model continuously adjusts the weight allocation logic and parameter harmonization strategy through optimization algorithms such as gradient descent and cross-validation. For example, for high-priority scenarios such as "multiple people in the car," the model can automatically increase the weight ratio of the scenario recommendation value through training (e.g., from 40% to 60%) to ensure that the user's immediate comfort needs are met first. For the case of "normal driving without special scenarios," the model can optimize the weight allocation so that the user's preferred recommendation value dominates (e.g., a weight ratio of 70%), ensuring the consistency of long-term usage habits. At the same time, the model can also learn the linkage rules between different parameters through training (e.g., the synergistic relationship of increasing the air volume by 1 level when the temperature drops by 1°C), avoiding contradictions between output parameters (e.g., the coexistence of low temperature and low air volume leads to low cooling efficiency). After the model training is completed and passes multiple rounds of scenario testing (e.g., the matching degree between recommended values ​​and actual user satisfaction in different scenarios is ≥92%), the trained decision fusion model is deployed in the cloud. On the one hand, it enables unified access for multiple vehicles, ensuring that different vehicles can obtain recommendation results with the same accuracy. On the other hand, the cloud can receive adjustment effect data from each vehicle in real time (e.g., records of users manually correcting recommended values), continuously iterating and optimizing the model to make the model's recommendation logic more in line with actual user needs. Finally, through the decision fusion model deployed in the cloud, the optimal air conditioning recommendation value that takes into account both "long-term habits" and "real-time scenarios" is provided to all called vehicles, further enhancing the uniformity, accuracy, and scalability of intelligent control of the vehicle's air conditioning.

[0055] As shown in Figure 9, when a user starts and uses the air conditioner, this embodiment uses a vehicle-cloud collaborative architecture to upload the user's air conditioning settings information (such as current temperature setting, fan speed, circulation mode, airflow direction, etc.) generated in real time by the air conditioning system, as well as multi-source data from inside and outside the vehicle (including outside temperature, humidity, altitude, vehicle tilt angle, vehicle speed, charging gun connection status, inside temperature, humidity, CO2 concentration, particulate matter concentration, window status, etc.), combined with external service data (such as real-time weather, map traffic conditions, etc.) to the cloud in real time via a low-latency, high-reliability communication link. After receiving data in the cloud, the system collaboratively processes it through pre-deployed external environment recognition models, vehicle driving feature recognition models, in-vehicle environment recognition models, and vehicle operation scenario feature recognition modules. The external environment recognition model comprehensively analyzes information such as external temperature, humidity, altitude, and weather trends to generate more adaptable external environment features. The vehicle driving feature recognition model characterizes the vehicle's current operating conditions based on parameters such as vehicle speed changes and charging status. The in-vehicle environment recognition model integrates indicators such as temperature, humidity, air quality, and ventilation status to generate feature data that comprehensively reflects the in-vehicle environment. The vehicle operation scenario feature recognition module integrates the above three types of features into complete vehicle operation scenario features through cross-dimensional feature fusion algorithms, such as composite scenarios like "high temperature and humidity + high speed driving + two people in the car + half-open window". Subsequently, the characteristics of the vehicle's operating scenario are deeply integrated with the user preference tags maintained in the cloud, and input into the trained and optimized user preference air conditioning recommendation model. Based on the intelligent matching of historical preferences and the current scenario, the model outputs user preference air conditioning recommendation values ​​that are highly consistent with the user's long-term habits, including temperature, air volume, circulation mode, and air outlet mode, providing a stable personalized benchmark for subsequent decisions.

[0056] In terms of user habit learning, the system relies on the big data storage and continuous computing capabilities of the cloud to process and accumulate features of users' air conditioning adjustment behavior in real time under different scenarios. For example, it records users' manual adjustment preferences in different scenarios such as rainy days in winter, highways in summer, and urban congestion, and analyzes these preferences in relation to the environmental characteristics and vehicle operating characteristics at that time. When a user's car usage time or number of uses reaches a preset tag update cycle (such as accumulating 30 hours or 50 uses), the cloud will automatically trigger the user tag update process. Through incremental learning algorithms, the user preference model is iteratively optimized to make the generated user preference tags more in line with the user's latest usage habits, avoiding preference drift caused by long-term lack of updates, thereby continuously improving recommendation accuracy and user experience.

[0057] As an optional enhancement, when the air conditioning is in use, the system can also activate the vehicle-mounted camera and vehicle-mounted infrared camera to collect real-time image information including OMS and DMS. First, the image is anonymized in the vehicle-side camera preprocessing module. Through privacy protection algorithms such as facial blurring and identity information removal, camera feature data is generated, retaining only clothing outlines, posture, and heat distribution features. This ensures that no personal privacy information is involved during cloud upload, effectively meeting data security and privacy regulations. Subsequently, the anonymized camera features are uploaded to the cloud and input into a trained user scenario-based feature recognition module. This module can accurately identify real-time features such as the user's clothing type, resting state, and the number of people in the vehicle. These features are then input into the scenario-based air conditioning recommendation model, outputting multiple sets of scenario-based air conditioning recommendation values ​​adapted to the current real-time scenario, enabling rapid response to the user's immediate state.

[0058] Finally, the cloud-generated user-preferred air conditioning recommendations (representing long-term stable preferences) and multiple sets of scenario-based air conditioning recommendations (representing real-time dynamic needs) are input into a trained and optimized decision fusion model. This model, through dynamic weight allocation and parameter co-optimization, achieves the optimal balance between long-term habits and real-time needs, outputting a set of optimal air conditioning recommendations that balance comfort, health, and energy consumption, specifically including temperature, airflow, circulation mode, and air outlet mode. The cloud then securely and with low latency distributes these recommendations to the vehicle's multimedia system. The vehicle system interacts in real-time with the air conditioning app, translating the recommended parameters into specific control commands that drive the air conditioning system to automatically perform adjustments, achieving deep collaboration between "human-vehicle-environment" for intelligent air conditioning control. This process not only significantly improves the automation and precision of air conditioning adjustments and reduces the frequency of manual user operation, but also ensures the system's long-term availability, security, and user trust through continuous cloud learning and privacy protection design, further strengthening the overall competitiveness of the vehicle's intelligent cockpit.

[0059] In some embodiments, the vehicle is equipped with the system described in the foregoing embodiments and executes the corresponding methods described in the foregoing embodiments. The vehicle may be a gasoline-powered vehicle, a plug-in hybrid electric vehicle, or a new energy vehicle, etc., and this application does not specifically limit it.

[0060] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

[0061] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0062] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the scope of protection of this application includes the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of this application.

[0063] This document uses specific examples to illustrate the working principle and implementation method of the omnidirectional antenna of this application. The above description of the embodiments is only for the purpose of helping to understand the specific settings and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation method and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

[0064] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An air conditioning control system, characterized in that, The system includes: a multimedia module information acquisition module, an external sensor information acquisition module, an internal hardware information acquisition module, and an air conditioning personalized recommendation model; The multimedia information acquisition module is used to collect user information about air conditioning usage. The external sensor information acquisition module is used to collect information about the external environment of the vehicle. The in-vehicle hardware information acquisition module is used to collect vehicle driving information, in-vehicle environment information, and user scenario information. The personalized air conditioner recommendation model outputs recommended air conditioner values, which include temperature, air volume, circulation mode, and air outlet mode.

2. The air conditioning control system according to claim 1, characterized in that, The user's air conditioning usage information includes: temperature setting information, air volume setting information, circulation setting information, and air outlet mode setting information.

3. An air conditioning control system according to claim 2, characterized in that, The external environmental information includes external temperature, external humidity, altitude, and vehicle tilt angle.

4. An air conditioning control system according to claim 3, characterized in that, The vehicle driving information includes: vehicle speed and charging gun connection status; the in-vehicle environment information includes: in-vehicle temperature, air vent channel temperature, in-vehicle humidity, CO2 concentration, particulate matter information, and window status; the user scenario information includes: in-vehicle camera information and in-vehicle infrared camera information.

5. An air conditioning control system according to claim 4, characterized in that, The personalized air conditioner recommendation model includes: a user preference air conditioner recommendation model, a scenario-based air conditioner recommendation model, and a decision fusion model.

6. An air conditioning control system according to claim 5, characterized in that, The user inputs air conditioning information into the corresponding preference judgment module, which then outputs preference features. The preference features and external temperature information are simultaneously input into the preference feature recognition module, which outputs preference features under different environmental conditions; the preference features under different environmental conditions are input into the user profile model, which outputs user preference labels.

7. An air conditioning control system according to claim 6, characterized in that, The external environment information, weather forecast, and map information are input into the external environment recognition module, which outputs the external environment features. The vehicle driving information is input into the vehicle driving feature recognition module, which outputs the vehicle driving features. The in-vehicle environment information is input into the in-vehicle environment recognition module, which outputs in-vehicle environment characteristics.

8. An air conditioning control system according to claim 7, characterized in that, The external environmental features, vehicle driving features, and internal environmental features are input into the vehicle operation scene feature recognition module to obtain the vehicle operation scene features.

9. An air conditioning control system according to claim 8, characterized in that, The user scenario information is input into the user scenario feature recognition module, which outputs user scenario features. The user scenario-based feature recognition module includes: clothing feature recognition model, emotion feature recognition model, fatigue feature recognition model, sweat feature recognition model, rest feature recognition model, and multi-target feature recognition model.

10. An air conditioning control system according to claim 9, characterized in that, The vehicle operation scenario features and user preference tags are input into the user preference air conditioning recommendation model, which outputs user preference air conditioning recommendation values, including air conditioning temperature, air volume, circulation mode, and air outlet mode.

11. An air conditioning control system according to claim 10, characterized in that, The user's contextual features are input into the scenario-based air conditioning recommendation model, which outputs multiple sets of scenario-based air conditioning recommendation values, namely the air conditioning temperature, air volume, circulation mode, and air outlet mode corresponding to the current scenario. The user's preferred air conditioner recommendation value and multiple sets of scenario air conditioner recommendation values ​​are input into the decision fusion model, and the output air conditioner recommendation value includes temperature, air volume, circulation mode, and air outlet mode.

12. An air conditioning control system according to any one of claims 1-10, characterized in that, At least one of the following modules—preference judgment module, preference feature recognition module, vehicle external environment recognition module, vehicle internal environment recognition module, vehicle driving feature recognition module, whole vehicle operation scenario feature recognition module, user scenario-based feature recognition module, and user preference air conditioning recommendation model—is deployed on a remote server.

13. A vehicle, characterized in that, The vehicle includes the air conditioning control system as described in claims 1-12.

Images