A defogging method, model training method, and apparatus based on a fusion model.

Abstract

This invention relates to a defogging method, model training method, and apparatus based on a fusion model, relating to the field of vehicle technology. The method determines the irradiance power of a first vehicle window based on light intensity and the window structure information. Then, based on a preset first model, the interior temperature, exterior temperature, and irradiance power of the first window are processed to obtain the glass temperature of the first window. Finally, based on a preset second model, environmental information, window information, the glass temperature of the first window, and a pre-collected glass temperature of a second window are processed to obtain defogging recommendation information. In this process, a fusion model combining the first model used to predict window glass temperature and the second model used to recommend defogging information generates defogging recommendations applicable to different windows, replacing the simple application of the windshield treatment method and significantly optimizing the defogging effect of each window in the vehicle.

Description

Technical Field

[0001] This invention relates to the field of vehicle technology, and specifically to a defogging method, model training method, and apparatus based on a fusion model. Background Technology

[0002] In modern automotive design, window clarity is crucial for driving safety and passenger comfort. Window fogging is a common problem, especially in environments with high humidity or large temperature differences.

[0003] In related technologies, defogging is mainly performed on the windshield. Typically, a temperature sensor is installed on the windshield to collect its temperature, and this temperature reading is used to predict whether fogging will occur and then perform defogging. For other windows (such as side windows or rear windows), temperature sensors are usually not installed, and the defogging process is the same as that used for the windshield.

[0004] However, since the temperature and structure of these windows are usually different from those of the windshield, directly applying the same defogging strategy as the windshield will result in poor defogging performance for other windows. Summary of the Invention

[0005] The purpose of this invention is to provide a defogging method, model training method and device based on a fusion model, so as to optimize the defogging effect of vehicle windows.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] According to a first aspect of the present invention, a defogging method based on a fusion model is provided, comprising:

[0008] Acquire vehicle environmental information and window information; wherein, the environmental information includes vehicle interior temperature, vehicle exterior temperature and light intensity, and the window information includes window glass humidity and / or window structure information, and the window includes a first window and a second window;

[0009] Based on the light intensity and the window structure information of the first window, the light radiation power of the first window is determined;

[0010] According to the preset first model, the vehicle interior temperature, vehicle exterior temperature and light radiation power of the first vehicle window are processed to obtain the glass temperature of the first vehicle window. The first model is trained based on the historical glass temperature data of the second vehicle window and is used to predict the glass temperature of the first vehicle window.

[0011] Based on the preset second model, the environmental information, the window information, the glass temperature of the first window, and the pre-collected glass temperature of the second window are processed to obtain defogging recommendation information; wherein, the second model is a model trained based on historical defogging data and used to generate defogging recommendation information.

[0012] In one embodiment, the first window includes a side window of the vehicle, and the second window includes the windshield of the vehicle.

[0013] In one embodiment, the window structure information includes the installation angle and surface area of ​​the window glass; determining the illuminance power of the first window based on the light intensity and the window structure information of the first window includes:

[0014] The relative light intensity of the first vehicle window is determined based on the installation angle of the first vehicle window and the light intensity.

[0015] The illuminance power of the first vehicle window is determined based on the relative illuminance and the surface area.

[0016] In one implementation, the method further includes:

[0017] The ambient thermal radiation power of the first window is determined based on the interior temperature, the exterior temperature, and the surface area.

[0018] Determining the illuminance power of the first vehicle window based on the relative illuminance and the surface area includes:

[0019] The initial illuminance power of the first vehicle window is determined based on the relative illuminance and the surface area.

[0020] The illumination radiation power is obtained based on the ambient thermal radiation power and the initial illumination radiation power.

[0021] In one implementation, the second model includes a preprocessing module, a risk prediction module, and a recommendation module; the step of processing the environmental information, the window information, the glass temperature of the first window, and the pre-collected glass temperature of the second window according to the preset second model to obtain defogging recommendation information includes:

[0022] The preprocessing module processes the glass temperature and humidity of the first window and the glass temperature and humidity of the second window to obtain the temperature and humidity changes of the first window and the second window.

[0023] The risk prediction module processes at least one of the glass temperature, glass humidity and temperature and humidity change information of the first window and at least one of the glass temperature, glass humidity and temperature and humidity change information of the second window to determine the fogging level of the first window and the second window respectively.

[0024] The recommendation module generates defogging recommendation information for the first and second windows based on the fogging level, environmental information, and window information.

[0025] In one embodiment, the processing of the glass temperature and humidity of the first vehicle window and the glass temperature and humidity of the second vehicle window respectively includes:

[0026] The temperature and humidity of the first car window are differentiated by the first derivative to obtain the temperature and humidity change information of the first car window.

[0027] The temperature and humidity of the first car window are differentiated by first-order derivatives to obtain the temperature and humidity change information of the second car window.

[0028] The processing of at least one of the glass temperature, glass humidity, and temperature and humidity change information of the first vehicle window, and at least one of the glass temperature, glass humidity, and temperature and humidity change information of the second vehicle window, includes:

[0029] The fogging level of the first vehicle window is determined based on at least one of the glass temperature, glass humidity, and temperature and humidity change information of the first vehicle window; and the fogging level of the second vehicle window is determined based on at least one of the glass temperature, glass humidity, and temperature and humidity change information of the second vehicle window.

[0030] In one embodiment, determining the fogging level of the first vehicle window based on its glass temperature and temperature change information, and its humidity and humidity change information includes:

[0031] Based on the temperature and humidity change information of the first vehicle window, the fogging probability change coefficient corresponding to the temperature and humidity change information is looked up from a preset mapping table to obtain the fogging probability change coefficient of the first vehicle window; wherein, the mapping table includes fogging probability change coefficients corresponding to different temperature and humidity change information.

[0032] The fogging level of the first vehicle window is determined based on at least one of the glass temperature, glass humidity, and the fogging probability variation coefficient of the first vehicle window.

[0033] In one embodiment, determining the fogging level of the second vehicle window based on temperature and humidity change information includes:

[0034] Based on the temperature and humidity change information of the second window, the fogging probability change coefficient corresponding to the temperature and humidity change information is found in the preset mapping table to obtain the fogging probability change coefficient of the second window.

[0035] The fogging level of the second window is determined based on at least one of the glass temperature, glass humidity, and the fogging probability variation coefficient of the second window.

[0036] In one implementation, the method further includes:

[0037] Based on the defogging recommendation information, the target actuator is controlled to perform the corresponding defogging operation; wherein, the defogging recommendation information includes one of the following recommendations: opening the first window and / or the second window, adjusting the ventilation settings, and adjusting the air conditioning settings; the target actuator includes at least one of the following: a window control system, a ventilation system, and an air conditioning system.

[0038] In one implementation, controlling the target actuator to perform a corresponding defogging operation based on the defogging recommendation information includes:

[0039] The defogging recommendation information is displayed on the central control screen, and / or the defogging recommendation information is prompted through an audio device;

[0040] In response to a touch operation on the central control screen and / or a voice input operation on the audio device, defogging setting information is determined, which may be the same as or different from the defogging recommendation information;

[0041] Based on the defogging settings information, the target actuator is controlled to perform the corresponding defogging operation.

[0042] In one implementation, the method further includes:

[0043] Real-time monitoring of the humidity and / or temperature trends of the vehicle windows;

[0044] In response to the detected trend information of humidity change and / or temperature change being lower than a preset threshold, the target actuator is controlled to stop the current defogging operation.

[0045] In one embodiment, the environmental information further includes vehicle driving information and / or rainfall information, and the window information further includes window position.

[0046] According to a second aspect of the present invention, a model training method is provided, comprising:

[0047] Collect first historical sample data, which includes historical interior temperature, historical exterior temperature, historical light intensity, and historical glass temperature of the second window within the preset historical time period.

[0048] Based on the historical light intensity and the window structure information of the second window, the light radiation power of the second window is determined;

[0049] The first model is trained by taking the historical interior temperature, the historical exterior temperature, and the historical irradiance of the second window as inputs, and the historical glass temperature of the second window as output.

[0050] According to a third aspect of the present invention, a model training method is provided, comprising:

[0051] Collect second historical sample data, which includes historical environmental information, pre-collected historical glass temperature of the second vehicle window, and corresponding historical defogging data. The historical defogging data includes the user's expected defogging settings under the corresponding historical conditions.

[0052] Based on the historical light intensity and the window structure information of the first window, the light radiation power of the first window is determined, and the historical glass temperature of the first window is predicted by the first model based on the light radiation power of the first window.

[0053] The historical environmental information, window information, historical glass temperature of the first window, and historical glass temperature of the second window are used as inputs, and the desired defogging setting information is used as output to train a second model.

[0054] According to a fourth aspect of the present invention, a defogging treatment apparatus based on a fusion model is provided, comprising:

[0055] The acquisition module is used to acquire the vehicle's environmental information and window information; wherein, the environmental information includes the vehicle interior temperature, the vehicle exterior temperature, and the light intensity, and the window information includes the humidity of the window glass, and the window includes a first window and a second window;

[0056] The first determining module is used to determine the light radiation power of the first vehicle window based on the light intensity and the window structure information of the first vehicle window;

[0057] The first model processing module is used to process the vehicle interior temperature, vehicle exterior temperature and light radiation power of the first vehicle window according to the preset first model to obtain the glass temperature of the first vehicle window. The first model is trained based on the historical glass temperature data of the second vehicle window and is used to predict the glass temperature of the first vehicle window.

[0058] The second model processing module is used to process the environmental information, the window information, the glass temperature of the first window, and the pre-collected glass temperature of the second window according to the preset second model to obtain defogging recommendation information; wherein, the second model is a model trained based on historical defogging data and used to generate defogging recommendation information.

[0059] According to a fifth aspect of the present invention, a model training apparatus is provided, comprising:

[0060] The first acquisition module is used to acquire first historical sample data within a preset historical time period. The first historical sample data includes the historical interior temperature, historical exterior temperature, historical light intensity, and historical glass temperature of the second window within the preset historical time period.

[0061] The second determining module is used to determine the light radiation power of the second window based on the historical light intensity and the window structure information of the second window;

[0062] The first training module is used to train a first model by taking the historical interior temperature, the historical exterior temperature, and the historical irradiance power of the second window as inputs and the historical glass temperature of the second window as outputs.

[0063] According to a sixth aspect of the present invention, a model training apparatus is provided, comprising:

[0064] The second acquisition module is used to acquire second historical sample data. The second historical sample data includes historical environmental information, the historical glass temperature of the pre-acquired second vehicle window, and the corresponding historical defogging data. The historical defogging data includes the user's expected defogging settings information under the corresponding historical conditions.

[0065] The third determining module is used to determine the light radiation power of the first vehicle window based on the historical light intensity and the window structure information of the first vehicle window, and to predict the historical glass temperature of the first vehicle window based on the light radiation power of the first vehicle window through the first model.

[0066] The second training module is used to train a second model by taking the historical environmental information, window information, historical glass temperature of the first window and historical glass temperature of the second window as inputs and the desired defogging setting information as outputs.

[0067] According to a seventh aspect of the present invention, a defogging system based on a fusion model is provided, comprising a cloud, a central controller communicatively connected to the cloud, a vehicle controller communicatively connected to the central controller, and at least one sensor communicatively connected to the vehicle controller; wherein,

[0068] The cloud is used to store historical data of the vehicle, including historical glass temperature data and / or historical defogging data of the second window;

[0069] The at least one sensor is used to collect environmental information and window information of the vehicle;

[0070] The central controller is used to train a first model based on the historical glass temperature data of the second window, and to train a second model based on the historical defogging data.

[0071] The vehicle controller is used to execute the defogging method based on the fusion model provided in any of the first aspects above.

[0072] In one embodiment, the system further includes a central control screen and / or an audio device that are communicatively connected to the vehicle controller; wherein the central control screen is used to display the defogging recommendation information, and / or the audio device is used to prompt the defogging recommendation information.

[0073] According to an eighth aspect of the present invention, a vehicle is provided, including a memory, a processor, and a communication interface;

[0074] The memory stores computer-executed instructions;

[0075] The communication interface receives historical data transmitted from the cloud, including historical glass temperature data and / or historical defogging data of the second window.

[0076] The processor executes computer execution instructions stored in the memory, causing the processor to perform a defogging method based on a fusion model as provided in any of the first aspects above, or a model training method as provided in the second or third aspects above.

[0077] The present invention provides a defogging method, model training method, and apparatus based on a fusion model. This fusion model combines a first model for predicting vehicle window temperature and a second model for recommending defogging information. The first model accurately predicts the window temperature based on information such as interior and exterior vehicle temperatures and solar radiation power. The second model processes the predicted temperature and other relevant information to generate defogging recommendations suitable for different windows, replacing the simple application of the windshield method and significantly optimizing the defogging effect for all windows in the vehicle. Furthermore, this fusion model-based defogging method can quickly respond to environmental changes and adjust the defogging strategy in a timely manner, thereby improving defogging efficiency. Attached Figure Description

[0078] Figure 1 This is one of the possible scenario diagrams provided by an embodiment of the present invention;

[0079] Figure 2 This is a second possible scenario illustration provided by an embodiment of the present invention;

[0080] Figure 3 This is a schematic diagram illustrating another possible scenario provided by an embodiment of the present invention;

[0081] Figure 4 A schematic flowchart of a defogging method based on a fusion model provided in an embodiment of the present invention;

[0082] Figure 5 This is a schematic diagram of the process for predicting the glass temperature of a vehicle window in an embodiment of the present invention.

[0083] Figure 6 This is an exemplary structural diagram of the second model in an embodiment of the present invention;

[0084] Figure 7 This is a schematic diagram illustrating the process of model training and processing on the vehicle side in an embodiment of the present invention;

[0085] Figure 8 A schematic flowchart of a defogging method based on a fusion model provided for an exemplary embodiment of the present invention;

[0086] Figure 9 A schematic flowchart of a model training method provided in an embodiment of the present invention;

[0087] Figure 10 A flowchart illustrating another model training method provided in an embodiment of the present invention;

[0088] Figure 11 A schematic diagram of the structure of the defogging method based on the fusion model provided in an embodiment of the present invention;

[0089] Figure 12 This is a schematic diagram of the structure of a model training device provided in an embodiment of the present invention;

[0090] Figure 13 This is a schematic diagram of another model training device provided in an embodiment of the present invention;

[0091] Figure 14 A schematic diagram of a defogging system based on a fusion model provided in an embodiment of the present invention;

[0092] Figure 15 This is a structural schematic diagram of a vehicle provided in an embodiment of the present invention. Detailed Implementation

[0093] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0094] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0095] In the description of this invention, it should be understood that the terms "first," "second," "third," etc., are used only to distinguish similar objects and are not necessarily used to describe a specific order or sequence, nor should they be construed as indicating or implying relative importance. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances. Furthermore, in the description of this invention, unless otherwise stated, "multiple" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0096] It should be noted that, due to space limitations, this specification does not exhaustively list all possible implementation methods. Those skilled in the art, after reading this specification, should be able to deduce that any combination of technical features can constitute an optional implementation method, provided that the technical features do not contradict each other. The following provides a detailed description of each embodiment.

[0097] To facilitate understanding of the embodiments of the present invention, the embodiments of this application will first be explained in conjunction with application scenarios. The defogging method based on the fusion model provided in the embodiments of this application can be applied to intelligent driving application scenarios, and more specifically, to autonomous driving application scenarios based on vehicle cloud computing. For example, the execution subject of the method provided in the embodiments of this application can be a vehicle or a server such as a cloud server (i.e., the cloud). The following description, in conjunction with possible application scenarios, uses vehicles and servers as the execution subjects of the method provided in the embodiments of this application:

[0098] Figure 1 This is one possible application scenario provided by the embodiments of the present invention, such as... Figure 1As shown, this application scenario includes a cloud server 110 and a vehicle 120. The cloud server 110 and the vehicle 120 are connected. The cloud server 110 can be used to store historical data of the vehicle (corresponding to the first historical sample data and the second historical sample data mentioned later). This historical data includes, but is not limited to, historical glass temperature data of the vehicle windows, historical outside temperature, historical inside temperature, historical light intensity, historical glass temperature, historical glass humidity, historical vehicle driving information (such as navigation information), historical user information, historical network environment data, historical defogging data (such as the defogging settings information expected by the user under the corresponding historical conditions mentioned above), etc. The vehicle 120 can be used to obtain relevant historical data from the cloud server 110 to train a first model for predicting the window glass temperature and a second model for generating defogging recommendation information. By combining the first model and the second model, the vehicle can control the execution of corresponding defogging operations based on the defogging recommendation information output by the model, and can upload the processed data (such as all data from input to output) to the cloud server 110. For example, combined with Figure 2 As shown, vehicle 120 may include a central controller 121, a vehicle controller 122, at least one sensor 123 (such as a windshield temperature sensor, humidity sensor, sunlight intensity sensor, interior temperature sensor, exterior temperature sensor, etc.), a central control screen 124, an audio device 125, and multiple actuators 126 (such as a window control system, ventilation system, and air conditioning system, etc.). Vehicle 120 can communicate with cloud server 110 via the central controller 121. The central controller 121 is used to train a first model and a second model based on relevant historical data from the cloud server 110. The vehicle controller 122 can use the first model to predict the glass temperature of windows (e.g., side windows without glass temperature sensors) and use the glass temperature of the window and related sensor data as input to the second model to predict recommended information for defogging each window. After predicting the recommended information for defogging the vehicle windows, the system can interact with the user via the central control screen 124 and / or audio device 125 to determine the final defogging settings (e.g., the defogging settings from the recommended information, such as opening the side windows). The system then identifies the target actuator corresponding to this defogging setting from multiple actuators 126, and the vehicle controller 122 controls one or more target actuators to perform the defogging operation. In this application scenario, the model training process is performed on the vehicle side, and the trained model is used for defogging processing.

[0099] Figure 3 Another possible application scenario provided by the embodiments of the present invention, such as Figure 3As shown, this application scenario can also include a cloud server 110 and a vehicle 120 communicating with the cloud server 110. The cloud server 110 stores historical data of the vehicle and can use the relevant historical data to train a first model for predicting the window glass temperature of the vehicle 120 and a second model for generating defogging recommendation information for the vehicle 120. The first and second models are then sent to the vehicle 120, which uses them to perform defogging. In this application scenario, the model training process is performed by the cloud, and by transmitting the model to the vehicle, the vehicle can use the trained model for defogging. Optionally, the cloud can simultaneously train the first and second models corresponding to each vehicle for multiple vehicles.

[0100] Optionally, the cloud server 110 may also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms. In some embodiments, the cloud server 110 may also be replaced by a standalone physical server, or a server cluster or distributed system composed of multiple physical servers; this embodiment does not impose any particular limitation on this.

[0101] The defogging processing scheme based on a fusion model provided in this invention acquires vehicle environmental information and window information. The environmental information includes interior temperature, exterior temperature, and light intensity. The window information includes glass humidity and / or window structure information. The windows include a first window and a second window. Based on the light intensity and the window structure information of the first window, the light radiation power of the first window is determined. According to a preset first model, the interior temperature, exterior temperature, and light radiation power of the first window are processed to obtain the glass temperature of the first window. This first model is trained based on historical glass temperature data of the second window and is used to predict the glass temperature of the first window. Furthermore, according to a preset second model, the environmental information, window information, glass temperature of the first window, and pre-collected glass temperature of the second window are processed to obtain defogging recommendation information. This second model is trained based on historical defogging data and is used to generate the defogging recommendation information. In this process, a fusion model is used to combine a first model for predicting window glass temperature and a second model for recommending defogging information. The first model accurately predicts the window glass temperature based on information such as the temperature inside and outside the vehicle and the power of light radiation. The second model processes these predicted temperatures and other relevant information to generate defogging recommendations applicable to different windows, replacing the simple application of the windshield treatment method and significantly optimizing the defogging effect of each window in the vehicle. Furthermore, the defogging processing method based on the aforementioned fusion model can quickly respond to environmental changes and adjust the defogging strategy in a timely manner, thereby improving defogging efficiency.

[0102] The application scenarios of the embodiments of the present invention have been briefly introduced above. The following will use... Figure 1 Using a specific application scenario as an example, with vehicle 120 as the executing entity, this embodiment provides a detailed description of a defogging method based on a fusion model. Figure 4 As shown, the method includes steps S401-S404.

[0103] Step S401: Obtain the vehicle's environmental information and window information; wherein, the environmental information includes the vehicle interior temperature, the vehicle exterior temperature, and the light intensity, and the window information includes the glass humidity and / or window structure information, and the windows include the first window and the second window.

[0104] For example, environmental and window information can be collected through various sensors installed in the vehicle, such as windshield temperature sensors, humidity sensors, sunlight intensity sensors, interior temperature sensors, and exterior temperature sensors. In some embodiments, the vehicle can also obtain relevant data by communicating and interacting with other devices. For instance, the vehicle can exchange data with a smartphone or other electronic devices with sensor functions located inside the vehicle via a wireless communication module to obtain environmental and window information.

[0105] Optionally, the first window can be a side window of the vehicle, and the second window can be the windshield of the vehicle. In some embodiments, the first window can also be other windows such as the rear window that do not have a temperature sensor installed or whose temperature measurement results are inaccurate (such as a faulty temperature sensor), and the second window can also be other windows that have a temperature sensor installed (such as the left window having a temperature sensor installed and the right window having a temperature sensor installed). The embodiments of the present invention do not particularly limit the specific type of windows that the first and second windows are.

[0106] Step S402: Determine the light radiation power of the first window based on the light intensity and the window structure information of the first window.

[0107] For example, the window structure information may include the window glass's installation angle, surface area, glass thickness, and other information. Optionally, it may also include the glass's material properties. Different material properties may affect the glass's light absorption, thereby affecting the glass's light radiation power. Specifically, the corresponding absorption coefficient can be obtained by combining the glass's material properties.

[0108] In related technologies, the effect of light intensity on glass temperature is usually not considered during the defogging process of vehicle windows, especially the different light radiation power caused by different window structures, resulting in inaccurate window temperature readings. This embodiment uses light intensity and window structure information to calculate the light radiation power of the window, so that this light radiation power can be applied in subsequent steps to correct the window glass temperature, thereby providing a more accurate glass temperature prediction value for the first window.

[0109] In one approach, the window structure information includes the installation angle and surface area of ​​the window glass. The determination of the luminous radiation power of the first window based on the illuminance and its structural information can be achieved as follows: determine the relative illuminance of the first window based on its installation angle and illuminance; then determine the luminous radiation power of the first window based on its relative illuminance and surface area.

[0110] For example, the installation angle may include the horizontal angle and the circumferential angle (north) of the glass. The installation angle of the glass affects the angle at which light is incident on the surface of the window, thereby changing the effective intensity of the light. This embodiment takes into account the angle of light incidence, allowing for a more accurate assessment of the actual impact of light on the window.

[0111] Optionally, the relative illuminance hsolar of the first window can be calculated as follows:

[0112] h solar =h*cosθ

[0113] cosθ=sinγ*cosβ+cosγ*sinβ*cosδ

[0114] In the formula, h represents the light intensity, cosθ represents the cosine of the angle of incidence of sunlight relative to the glass surface, γ represents the horizontal angle of the sun, α represents the angle of orientation of the sun, β represents the horizontal angle of the glass, ξ represents the circumferential angle of the glass, and δ represents the angular difference between the direction of sunlight incidence and the direction of the glass surface, where δ = α - ξ. The horizontal angle of the sun and the angle of orientation of the sun can be calculated using time and latitude / longitude, and are therefore known quantities.

[0115] In calculating the relative light intensity h solar Then, by combining the surface area of ​​the glass, the luminous radiation power of the vehicle window can be calculated. Optionally, to further improve the accuracy of the luminous radiation power, i.e., the solar radiation power absorbed by the glass, this embodiment considers the ambient thermal radiation power, i.e., the air thermal radiation power, in the calculation of the luminous radiation power. Specifically, the method may also include the following steps: determining the ambient thermal radiation power of the first vehicle window based on the interior temperature, exterior temperature, and surface area.

[0116] Optionally, the ambient thermal radiation work q rad The rate can be calculated as follows: q rad = 0.0000000567 * 0.95 *area*(T2^4-T1^4).

[0117] In the formula, T1 is the outside temperature of the vehicle, T2 is the inside temperature of the vehicle (as an approximation of the glass surface temperature), and area is the surface area of ​​the glass.

[0118] The above method of determining the luminous radiation power of the first vehicle window based on relative illuminance and surface area can be adopted as follows: determine the initial luminous radiation power of the first vehicle window based on relative illuminance and surface area; obtain the luminous radiation power based on the ambient thermal radiation power and the initial luminous radiation power.

[0119] For example, the initial illumination radiant power q solar The calculation method is as follows: q solar =h solar *area. Glass absorbs solar radiation power P solar The calculation method is as follows: P sola =abscoeff*q solar +q rad In the formula, abscoeff represents the absorption coefficient corresponding to the material properties of the glass.

[0120] The above technical solution can accurately calculate the solar radiation power absorbed by the glass, thereby improving the accuracy of the prediction of the window glass temperature in the subsequent model processing.

[0121] Step S403: Based on the preset first model, process the interior temperature, exterior temperature and light radiation power of the first window to obtain the glass temperature of the first window. The first model is trained based on the historical glass temperature data of the second window and is used to predict the glass temperature of the first window.

[0122] Taking a vehicle with a side window as the first window and a windshield as the second as an example, vehicle design typically doesn't install temperature sensors (or other temperature measuring instruments) at every window location. Defogging is primarily focused on the windshield, with temperature sensors (or other temperature measuring instruments) only installed on the windshield to accurately measure its temperature. However, clear side window visibility is also crucial for lane changes, turning, and observing blind spots. If the side windows are not effectively defogged, it will also affect the driver's judgment of the surrounding environment, increasing the risk of accidents.

[0123] This approach replaces the existing unified defogging method based on the windshield, or directly using the windshield temperature to predict the temperature of the side windows (or other vehicle windows), which often results in inaccurate predictions or poor defogging effects. This embodiment utilizes a first model, combining the vehicle's interior temperature, exterior temperature, and the radiant power of the side windows' illumination, to predict the side window glass temperature. Specifically, this embodiment is based on the principles of heat transfer and energy balance (i.e., equivalent heat transfer). As the equivalent heat transfer principle (similar to that of the windshield and side windows) shows, changes in glass temperature are primarily influenced by the heat exchange process of the surrounding environment. The vehicle's interior temperature and exterior temperature represent the environmental conditions on either side of the glass, influencing heat absorption and dissipation through convection and conduction. Specifically, the interior temperature of a vehicle can affect the inner surface temperature of the glass through convective heat transfer between the interior air and the glass, while the exterior temperature can affect the outer surface temperature through convective heat transfer between the exterior air and the glass. Solar radiation power is another important factor, as solar radiation directly affects the energy absorption of the glass; the solar energy absorbed by the glass leads to an increase in its temperature, an effect particularly pronounced under strong sunlight. This embodiment comprehensively considers these factors to establish a model for predicting glass temperature. This model is based on the fundamental principles of heat transfer, combining ambient temperature and radiant energy to effectively simulate the dynamic temperature changes of glass under different conditions. The model is trained using measured values ​​of the windshield's temperature, enabling accurate prediction of glass temperature.

[0124] Taking the windshield and side windows as examples, both follow the same basic physical principles in terms of heat transfer. Their temperatures are affected by ambient temperature, vehicle interior temperature, and solar radiation. The equivalent heat transfer model describes the heat exchange process of the glass through convection, conduction, and radiation, processes that are consistent across different types of vehicle windows. Therefore, in this embodiment, to address the lack of accurate temperature data due to the absence of temperature sensors on side windows (or other vehicle windows), the model is trained using actual temperature data from the windshield (or other windows equipped with temperature sensors), and this model is used to predict the temperature of the side windows. During model training, input parameters include ambient temperature, vehicle interior temperature, and solar radiation power. Since the temperatures inside and outside the vehicle tend to be balanced, the interior temperatures of the windshield and side windows are roughly the same within the same vehicle, as are the exterior temperatures of the windshield and side windows. However, when calculating the solar radiation power of the windows, this embodiment considers different window structures, ensuring that the relationship between solar radiation power and glass temperature accurately reflects the characteristics of different windows. In this way, by training the model using temperature data from the windshield, the temperature of the side windows can be predicted more accurately without direct temperature sensors. It can be understood that the equivalent heat transfer model for glass is dT / dt = Ptotal / cp*mass, where dT / dt represents the rate of temperature change of the glass, cp*mass is a constant, cp represents the specific heat capacity of the glass, mass represents the mass, and P... 总 =P solar +P ambient +P cabin P 总 P represents the total heat transfer power of the glass. solar P represents solar radiant power. ambient P represents the heat transfer power between the glass and the external environment. cabin This indicates the heat exchange capacity between the glass and the vehicle's interior environment. Typically, the specific heat capacity of glass (such as the windshield and side windows) in the same vehicle model is consistent. However, the side windows and windshield may differ in mass or area. To further improve the accuracy of side window temperature prediction, since the mass and area of ​​the windshield and side windows are known parameters, the model parameters of the windshield model can be adjusted based on the differences in their mass or area (e.g., by proportionally adjusting the weights in the trained first model according to the mass ratio and / or area ratio between the windshield and side windows). The adjusted model can then accurately predict the temperature of the side windows, which lack temperature sensors.

[0125] To facilitate understanding of the embodiments of this application, further explanation is provided in this embodiment. This embodiment is based on the principle of equivalent heat transfer:

[0126]

[0127] In the formula, dT / dt is the rate of change of glass temperature, and cp and mass are the specific heat capacity and mass of glass, respectively, which are known constants; The heat transfer coefficient inside the glass vehicle; The external heat transfer coefficient is given by [reference to a specific value]. The heat transfer coefficients for the side windows and windshield are approximately equal.

[0128]

[0129]

[0130] In the formula, T 车内 Indicates the interior temperature, T 玻璃 T represents the glass temperature. 车外 Representing the ambient temperature, we can revise the above formula to obtain:

[0131]

[0132] The above formula is transformed to obtain:

[0133]

[0134] Based on the above formula, the left side can be... As a function representing the glass temperature, it corresponds to the model output, while the right side can be used as the model training input parameters: solar radiation power (i.e., The model represents the glass temperature (e.g., windshield temperature) and the interior and exterior temperatures of the vehicle. , , The relationship between them.

[0135] Therefore, even if the A (area), mass (mass), and even cp (specific heat capacity) of the side windows and the windshield are different, for these known parameters, the training parameters obtained by using historical windshield glass temperature, historical ambient temperature, historical vehicle interior temperature, and historical solar radiation power can be adjusted proportionally or otherwise. By fine-tuning the above-mentioned known parameters, more accurate training parameters can be obtained, thereby improving the prediction accuracy of the side window temperature.

[0136] For example, let's take training a first model to predict side window temperatures using windshield temperature data of the same vehicle model. The model is trained by collecting historical windshield data, which can be data from the windshield under different operating conditions (e.g., historical data collected under various conditions such as different blower speeds, airflow modes, and vehicle speeds). Based on the above formula, the function used to represent the model can be expressed as follows, for example, w1, w2, and w3 multiplied by... Solar radiation intensity, vehicle interior temperature, and vehicle exterior temperature.

[0137]

[0138] Where weight w1 corresponds to the formula above. / ( When there are differences in mass and area between the windshield and the side windows, given the area A1 and mass1 of the windshield, and the area A2 and mass2 of the side windows, the proportional adjustment method for the weighting coefficients can be w1' = w1 / A1 / mass2*A2*mass1.

[0139] Similarly, weight w2 corresponds to the weight in the above formula. / ( When there are differences in mass and area between the windshield and the side windows, the weighting coefficient can be adjusted proportionally as follows: w2' = w2 / A1 / mass2*A2*mass1.

[0140] Similarly, weight w3 corresponds to the weight in the above formula. In the case where there are differences in mass and area between the windshield and the side windows, the weighting coefficient w3 can be adjusted proportionally as follows: w3' = w3 / A1 / mass2*A2*mass1.

[0141] It should be understood that the above model is applicable to the prediction of side window temperatures in different vehicle models. During the model training process, for each vehicle model, the above model training method, combined with the historical data of that vehicle model, can be used to train a side window temperature prediction model for that vehicle model, thus achieving the prediction of side window temperatures for different vehicle models.

[0142] As can be seen, since the windshield and side windows follow the same basic physical principles in terms of heat transfer, and parameters such as the specific heat capacity and mass of the windows are known, even if they are not exactly the same, the model can be fine-tuned based on these known parameters to accurately predict the temperature of the side windows, which lack temperature sensors. Thus, even without direct temperature data for the side windows, the model can still accurately predict the temperature of the side windows using historical input data from the windshield.

[0143] For example, the training process of the first model can involve collecting first historical sample data, which includes historical interior temperature, historical exterior temperature, historical light intensity, and historical glass temperature of the second window within a preset historical time period. Based on the historical light intensity and the window structure information of the second window, the illuminance power of the second window is determined. Using the historical interior temperature, historical exterior temperature, and historical illuminance power of the second window as inputs, and the historical glass temperature of the second window as output, the first model is trained.

[0144] Optionally, the first model can be an artificial intelligence (AI) model, such as a machine learning model.

[0145] Optionally, during the data acquisition phase, first historical sample data is collected, which may include in-vehicle temperature data recorded within a preset time period, historical out-of-vehicle temperature data recorded within the same time period, historical light intensity data recorded within the same time period, and second window glass temperature data recorded within the same time period. Before model training, the above data can be preprocessed, such as handling missing values ​​and outliers, to improve data accuracy and align these data with the same timestamp to facilitate model training. Next, based on the historical light intensity and the window structure information of the second window, the illuminance power of the second window is calculated. The calculation process is similar to that of the first window, and related explanations will not be repeated. By using the historical in-vehicle temperature, historical out-of-vehicle temperature, and historical illuminance power of the second window as input features of the model, and the historical glass temperature of the second window as the output target of the initial model (such as an AI model), the model is trained using the training dataset. The model parameters are adjusted to minimize the error between the predicted output and the actual output (such as gradient descent optimization algorithm), or the first model is obtained when the maximum number of iterations is reached.

[0146] In some embodiments, the first model can also be used for predicting the glass temperature of a second window (such as the windshield). For example, if the windshield temperature sensor is not activated, the method is similar to that for predicting the glass temperature of the first window, and related details will not be elaborated here. Optionally, the methods for predicting the glass temperature of the windshield and the side windows can be as follows: Figure 5 As shown. This method allows for a relatively accurate acquisition of the glass temperature of a car window when the window lacks a temperature measurement function or when the temperature measurement function is not activated, providing data support for subsequent defogging procedures.

[0147] Step S404: Based on the preset second model, process the environmental information, window information, glass temperature of the first window and glass temperature of the pre-collected second window to obtain defogging recommendation information; wherein, the second model is a model trained based on historical defogging data and used to generate defogging recommendation information.

[0148] In this embodiment, the second model is trained based on historical defogging data and can identify defogging needs and corresponding optimal defogging strategies (such as the user's desired defogging settings) under different conditions. By inputting environmental information (such as interior and exterior temperatures), window information (such as window humidity), and the glass temperatures of the first and second windows into the second model, the model can output defogging recommendations for the first and second windows. Optionally, the window recommendations can be specific to different windows or to a particular window, such as suggestions to open side windows, activate wipers, adjust window (first and / or second) heating, activate the defroster, turn on or adjust the defogging mode of the air conditioning system, etc. In addition, other operational suggestions may be provided, such as adjusting the in-vehicle air circulation mode or opening vents, etc. In some embodiments, feedback information on actual defogging effects can also be collected to evaluate the effectiveness of the recommendations. Based on this feedback, the second model is further optimized to improve its prediction accuracy and recommendation effect under different environmental conditions.

[0149] In some embodiments, environmental information may also include one or more of the following: vehicle driving information (navigation information, vehicle speed, etc.), rainfall information, current air conditioning settings, network environment data, user information, etc. Window information may also include window position, etc.

[0150] Understandably, the pre-collected glass temperature of the second car window can be obtained from the temperature measurement function of the second car window, or it can be predicted using the first model.

[0151] For example, the training method of the second model can be achieved by collecting second historical sample data, which includes historical environmental information, pre-collected historical glass temperature of the second vehicle window, and corresponding historical defogging data. The historical defogging data includes the user's desired defogging settings under the corresponding historical conditions. Based on historical light intensity and the window structure information of the first vehicle window, the light radiation power of the first vehicle window is determined, and the historical glass temperature of the first vehicle window is predicted by the first model based on the light radiation power of the first vehicle window. The second model is trained by using historical environmental information, window information, the historical glass temperature of the first and second vehicles window as inputs, and the desired defogging settings as output. Optionally, the second model can be trained using the same initial model as the first model or a different initial model; this embodiment does not impose any particular limitation on this. Accordingly, the training process of the second model can be similar to that of the first model; please refer to the training process of the first model.

[0152] In some embodiments, to further optimize defogging efficiency, this embodiment incorporates changes in temperature and humidity around the vehicle windows for fogging prediction during the defogging recommendation process, thereby providing better fogging recommendation information. For example... Figure 6 As shown, the second model may include an input module 601, a preprocessing module 602, a risk prediction module 603, a recommendation module 604, and an output module 605. In the above steps, based on the second model, environmental information, window information, the glass temperature of the first window, and the pre-collected glass temperature of the second window are processed to obtain defogging recommendation information. Specifically, the input module 601 receives the input environmental information, window information, the glass temperature of the first window, and the pre-collected glass temperature of the second window. The preprocessing module 602 processes the glass temperature and humidity of the first window and the glass temperature and humidity of the second window respectively to obtain the temperature and humidity changes of the first and second windows. The risk prediction module 603 processes at least one of the glass temperature, glass humidity, and their temperature and humidity changes of the first and second windows to determine the fogging levels corresponding to the first and second windows respectively. The recommendation module 604 generates defogging recommendation information for the first and second windows based on the fogging level, environmental information, and window information. This defogging recommendation information is then output through the output module 605.

[0153] For example, the preprocessing module can calculate the glass temperature and humidity to obtain the temperature and humidity changes. This preprocessing module can directly calculate the temperature and humidity changes using a first-order derivative algorithm, or it can use time series analysis methods to extract features to reflect the temperature and humidity change trends. Specifically, the calculation of temperature and humidity changes using a first-order derivative algorithm involves: performing first-order derivatives on the glass temperature and humidity of the first window to obtain the temperature and humidity change information of the first window; and then performing first-order derivatives on the glass temperature and humidity of the first window to obtain the temperature and humidity change information of the second window.

[0154] The glass humidity can refer to the ambient humidity inside the vehicle, which can be collected using a humidity sensor installed inside the vehicle, or obtained using other existing technologies. This embodiment does not specifically limit the method of obtaining the glass humidity.

[0155] For example, the risk prediction module can predict the fog level based on the temperature and humidity changes obtained from the preprocessing module. A classification model (such as a decision tree) can be used to classify the fog level into multiple categories (such as no fog, foggy).

[0156] In related technologies, the accuracy of fog prediction based on air moisture content and dew point temperature is often addressed by determining the number of people inside the vehicle, which can be done using seats or in-vehicle cameras. However, some vehicle models lack rear seat weight sensors and in-vehicle cameras, making it impossible to determine the number of people inside. Furthermore, simply determining the number of people makes it difficult to accurately assess their physical characteristics, such as differences in height, weight, and metabolism. Relying on water vapor generated by occupants or changes in water vapor inside the vehicle caused by environmental changes has low accuracy and is complex to implement. Research has found that changes in temperature and humidity (including both temperature and humidity variations) affect the air's moisture content and dew point temperature, thus influencing the formation or dissipation of fog. This embodiment replaces the above-mentioned technical solution by incorporating temperature and humidity changes into the model for fogging prediction. Specifically, the humidity change inside the vehicle is affected by the number of occupants and their physical characteristics, while the glass temperature change can also represent sudden external weather changes, such as sudden rainfall. This embodiment uses temperature and humidity changes to solve the problem of inaccurate judgment of the automatic defogging function caused by the number of people in the vehicle, their different physical characteristics, and changes in ambient humidity. It not only eliminates the need for in-vehicle cameras or seat gravity sensors, but also effectively improves the accuracy of fogging prediction.

[0157] For example, the recommendation module generates defogging recommendations based on the predicted fog level and environmental information, window information, etc. The recommendation module can use a rule engine or a generative model to provide specific operational suggestions (such as adjusting air conditioning settings, turning on the defroster, etc.). Specifically, the fog level of the first window can be determined based on at least one of the glass temperature, glass humidity, and temperature and humidity changes of the first window. And, the fog level of the second window can be determined based on at least one of the glass temperature, glass humidity, and temperature and humidity changes of the second window.

[0158] By incorporating the temperature and humidity changes of the car windows into the defogging recommendation process, the prediction and recommendation effects of car window defogging are further optimized.

[0159] Furthermore, based on the glass temperature and its temperature change information, and the glass humidity and its humidity change information of the first vehicle window, the fogging level of the first vehicle window can be determined as follows: Based on the temperature and humidity change information of the first vehicle window, the fogging probability change coefficient corresponding to the temperature and humidity change information is looked up from a preset mapping table to obtain the fogging probability change coefficient of the first vehicle window; wherein, the mapping table includes fogging probability change coefficients corresponding to different temperature and humidity change information; the fogging level of the first vehicle window is determined based on at least one of the glass temperature, glass humidity, and the fogging probability change coefficient of the first vehicle window. In this embodiment, the first and second derivatives of the glass temperature and glass humidity of the vehicle window can be calculated respectively to obtain four new variables: h1 (derivative of humidity) represents the humidity change, t1 (derivative of glass temperature) represents the glass temperature change, h2 (second derivative of humidity) represents the rate of humidity change, and t2 (second derivative of temperature) represents the rate of temperature change. The second derivative can be used as an exit condition for defogging, as detailed in the following embodiment. During the calculation process, relative humidity can be converted into moisture content to avoid the influence of temperature, thus making the two parameters independent. In related technologies, dew point temperature is calculated using glass temperature and humidity. When the glass temperature is lower than the dew point temperature, fogging can be determined. However, due to factors such as sensor placement, the measuring point only covers one area. This embodiment introduces a fogging probability variation coefficient. By establishing a two-dimensional mapping table, the first derivatives of the newly introduced humidity and temperature are used as inputs to output the fogging probability variation coefficient (wherein, this two-dimensional mapping table can be determined based on a large amount of experimental data or prior data). The faster the humidity increases, the larger the coefficient; the faster the temperature decreases, the larger the coefficient. For example, setting a threshold of 0.3 (or 30%), if the fogging probability variation coefficient is lower than 0.3, the car window is considered fog-free. Or, if the fogging probability variation coefficient is higher than or equal to 0.3, the car window is considered fogged. Alternatively, the fogging probability coefficient can be obtained by taking the fogging probability coefficient corresponding to the temperature and humidity, such as the fogging probability coefficient corresponding to the difference between glass temperature and dew point temperature (the fogging probability coefficient can be determined using existing methods), and then by calculating the weighted average of the fogging probability coefficient and the original fogging probability coefficient, etc.

[0160] In some embodiments, the fogging level of the first vehicle window can also be determined by the glass temperature, for example, by comparing the glass temperature with the dew point temperature. For example, if the glass temperature is below the dew point temperature, but is more than 2°C above the dew point temperature, the fogging level is determined to be no fog; or if the glass temperature is more than 2°C below or equal to the dew point temperature, the fogging level is determined to be foggy. In some embodiments, the fogging level of the first vehicle window can also be determined by the glass humidity; for example, if the glass humidity is below 30%, the fogging level is determined to be no fog; or if the glass humidity is above or equal to 30%, fogging is considered to be present.

[0161] In some embodiments, the glass temperature and humidity of the first window, along with the corresponding fogging probability variation coefficient, can be combined to determine a more accurate fogging level. For example, a weighted algorithm can be used to comprehensively calculate a fogging level index, and this index can be used to determine the fogging level, as shown in the following formula: Y = AX1 + BX2 + CX 3， In the formula, Y is the fogging level index, X1 is the glass temperature of the first window, X2 is the glass humidity of the first window, X3 is the fogging probability variation coefficient of the first window, and A, B, and C represent the weighting coefficients corresponding to the glass temperature, glass humidity, and fogging probability variation coefficient, respectively (these weighting coefficients can be the same or different, and those skilled in the art can adapt them by combining practical applications or empirical values). For example, if the fogging level index reaches a preset index threshold, it is considered fogging; if it is below the preset index threshold, it is considered fog-free.

[0162] It should be noted that those skilled in the art can pre-establish this mapping table based on practical applications or a large amount of empirical data. This mapping table serves as part of the model, enabling the model to efficiently determine and adjust the fog probability, thereby improving prediction accuracy and real-time response capabilities.

[0163] Accordingly, the above-mentioned method of determining the fogging level of the second window based on the temperature and humidity change information of the second window can be adopted as follows: based on the temperature and humidity change information of the second window, look up the fogging probability change coefficient corresponding to the temperature and humidity change information in a preset mapping table to obtain the fogging probability change coefficient of the second window; determine the fogging level of the second window based on at least one of the glass temperature, glass humidity and the fogging probability change coefficient of the second window.

[0164] It should be noted that the method for determining the fogging level of the second window is similar to that for the first window, and the relevant explanations will not be repeated here.

[0165] In some embodiments, after generating defogging recommendation information, the target actuator can be directly controlled to perform corresponding defogging operations using the defogging recommendation information to improve defogging efficiency. Specifically, the method may further include the following steps: controlling the target actuator to perform corresponding defogging operations according to the defogging recommendation information; wherein, the defogging recommendation information includes one of the following recommendations: opening the first window and / or the second window, adjusting ventilation settings, and adjusting air conditioning settings; the target actuator includes at least one of the following: a window control system, a ventilation system, and an air conditioning system.

[0166] In this embodiment, the target actuator, the actuator corresponding to the defogging recommendation information, may include other actuators besides those mentioned above, such as heating, ventilation, and air conditioning (HVAC) systems, damper systems, sunroof systems, sunroof systems, etc., which can be used to perform operations corresponding to the defogging recommendation information according to control, such as opening / closing windows, ventilation, etc.

[0167] For example, when the model predicts fogging, defogging recommendations can be generated based on the predicted fog level (e.g., both the first and second windows are predicted to be foggy). These recommendations might suggest opening the windows to promote air circulation, adjusting ventilation settings to increase airflow, and adjusting the air conditioning to reduce humidity. Multiple target actuators can be automatically controlled to perform the corresponding defogging operations. For instance, the window control system could partially open the first window to accelerate air circulation, while the windshield wipers could wipe the second window, and so on.

[0168] In a further example of this embodiment, defogging can also be implemented through user interaction to optimize the user experience. Specifically, controlling the target actuator to perform the corresponding defogging operation based on defogging recommendation information can be done in the following way.

[0169] Method 1: Display defogging recommendation information on the central control screen, and in response to touch operation on the central control screen, determine the defogging setting information. If the defogging setting information is the same as or different from the defogging recommendation information, control the target actuator to perform the corresponding defogging operation.

[0170] In this method, by displaying defogging recommendations on the central control screen, the driver or passengers can view system-generated defogging recommendations, such as suggesting opening windows, adjusting ventilation settings, or adjusting the air conditioning. Users can respond to these recommendations via touchscreen operation on the central control screen. Based on the user's touch input, the system determines the final defogging settings, which can be the same as the recommendations (if the user accepts them) or different (if the user makes custom adjustments). The vehicle then further controls the target actuators to perform the corresponding defogging operations based on these settings, such as opening windows, adjusting ventilation, or adjusting air conditioning settings.

[0171] This approach allows users to adjust settings to their individual needs, enabling personalized defogging control and enhancing user experience and driving safety.

[0172] Method 2: Defogging recommendation information is prompted through an audio device. In response to voice input operation on the audio device, defogging setting information is determined. If the defogging setting information is the same as or different from the defogging recommendation information, the target actuator is controlled to perform the corresponding defogging operation.

[0173] It should be noted that the interaction method through the audio device is similar to that of the central control screen mentioned above. For relevant explanations and effects, please refer to Method 1 above. It will not be elaborated further here.

[0174] Method 3: It can also be combined with the central control screen and audio device to achieve specific defogging operation. The relevant principle is similar to that of Method 1 and Method 2 above, and will not be elaborated here.

[0175] The following explanation uses the vehicle-side model training and processing (omitting the temperature prediction process of the first model) and voice interaction as an example to illustrate the defogging process. Figure 7 As shown, this includes training data input, model training, model deployment, (current) data input, model processing, and defogging recommendation information output. Among these,

[0176] The model training and deployment phase begins with inputting training data, including historical data such as outside temperature, inside temperature, light intensity, glass temperature, glass humidity, current air conditioning settings, network environment data, navigation information, and user information, as well as corresponding historical defogging data (defogging operation results and user's desired defogging settings), which corresponds to the second historical sample data discussed later. This training data can be stored in the cloud and retrieved from the cloud for model training when needed on the vehicle, or it can be directly input into the vehicle for training. After the vehicle completes model training, model deployment can proceed. For example, the model can be converted into a form that can run on the vehicle's local system (such as the in-vehicle controller), making it compatible with the vehicle's hardware and other software modules.

[0177] During the model processing phase, the (current) data is first input, such as defogging-related data, into a second model for processing to output defogging recommendations. Optionally, the input data used for model processing can be transmitted as historical data to the cloud for storage, facilitating model optimization. One possible approach is to combine the model with machine learning algorithms (such as decision trees, particle swarm optimization, etc.). Taking the decision tree algorithm as an example, it can determine the fogging level of each window based on various real-time input data (outside temperature, inside temperature, light intensity, glass temperature, glass humidity, current air conditioning settings, network environment data, navigation information, user information, etc.), and then generate corresponding defogging recommendations by combining other factors such as temperature, such as window opening / closing, air conditioning settings (such as air conditioning switch, temperature setting, automatic switch (AUTO), air conditioning compressor switch (AC) switch, mode setting), ventilation settings, internal / external circulation settings, defogging through window / sunroof position (such as slightly opening the window to allow air circulation between the inside and outside of the vehicle and reduce the humidity difference), and window heating. Particle swarm optimization (PSO) can be used to optimize parameters in the defogging process, such as finding combinations of air conditioning airflow and temperature to further improve the defogging effect. It should be noted that this process is merely one possible example of this embodiment and is not intended to limit the scope of this embodiment.

[0178] Optionally, the model output can also include a voice interaction stage. Specifically, based on the defogging recommendation information output by the model, the user is prompted through an audio device, such as a voice message saying "It is recommended to turn on the defogging mode and activate the wipers on the first window." The system can also receive user feedback on the defogging recommendation information to achieve the final defogging settings, thereby optimizing the user experience and improving driving safety and comfort.

[0179] In some embodiments, considering the lack of a reasonable exit mechanism in the defogging process in related technologies, frequent exits or entrys of the defogging process may occur. To further optimize the user's driving experience, the trend information of humidity changes and / or temperature changes of the vehicle window can also be monitored in real time. In response to the detected trend information of humidity changes and / or temperature changes falling below a preset threshold, the target actuator is controlled to stop the current defogging operation.

[0180] In this embodiment, the trend information of humidity change can be represented by the derivative of humidity change, that is, the second derivative of glass humidity mentioned above, which can characterize the rate of change of humidity inside the vehicle. The trend information of temperature change is similar.

[0181] It should be noted that those skilled in the art can adapt the preset threshold by combining it with actual application or experience value, and the specific value of the preset threshold is not particularly limited in the embodiments of the present invention.

[0182] By introducing the aforementioned exit mechanism, the problem of frequent start-up and exit operations that may result from the unreasonable time interval between the current automatic defogging function's exit and re-entry, which affects the comfort of the passenger cabin, has been addressed, thus further optimizing the user experience.

[0183] To facilitate understanding of the embodiments of the present invention, an exemplary embodiment will be further described below. For example... Figure 8 As shown, the process includes the following:

[0184] Acquire input parameters (including environmental information, window information, etc.). For example, environmental information may include outside temperature, inside temperature, light intensity, current air conditioning settings, network environment data, navigation information, user information, etc. Window information may include window glass temperature, glass humidity, etc.

[0185] The relevant parameters from the input parameters (including direct input or processed input; the relevant parameters and specific processing procedures can be understood in conjunction with the above embodiments) are input into the trained first model for processing to predict the glass temperature of the car window (where glass temperature measurement results are missing).

[0186] The glass temperature of the car windows predicted by the first model, along with relevant parameters from the input parameters, are fed into a trained second model for processing to predict defogging recommendations. Specifically, the second model may include a fogging level prediction module and a recommendation module. The prediction module analyzes the model input information to determine the fogging level of each car window. The recommendation module combines the fogging level, environmental information, and car window information to output defogging recommendations.

[0187] After outputting defogging recommendations, safety rule boundaries can be introduced to judge the specific content of the recommendations. For example, the recommendations might suggest air conditioning defogging, but the recommended temperature exceeds the reasonable temperature range adjustable by the vehicle's air conditioning system (e.g., the lower limit for cooling is around 16°C, and the upper limit for heating is around 32°C). If the recommended temperature is below the lower limit or above the upper limit, it is determined to violate the safety rule boundaries. If it does not violate the safety rule boundaries, the process can end, and the user can be recommended defogging information determined by existing technology, such as recommendations based on rule algorithms. For defogging recommendations that comply with the safety rule boundaries, the recommendations can be broadcast via audio or displayed on a large screen. If the user confirms the use, the defogging operation is performed according to the recommendations; otherwise, the original state remains unchanged. Throughout the entire defogging process, the cloud can record data for subsequent model optimization and analysis of user preferences, etc.

[0188] Figure 9 This is a flowchart illustrating a model training method provided in an embodiment of the present invention, as shown below. Figure 9 As shown, the method includes:

[0189] Step S901: Collect the first historical sample data, which includes the historical interior temperature, historical exterior temperature, historical light intensity, and historical glass temperature of the second window within a preset historical time period.

[0190] Step S902: Determine the light radiation power of the second window based on the historical light intensity and the window structure information of the second window;

[0191] Step S903: Using the historical interior temperature, historical exterior temperature, and historical irradiance power of the second window as inputs, and the historical glass temperature of the second window as output, train the first model.

[0192] It should be noted that the model training process has been described in the above model processing example. You can refer to the above example for understanding here, and the relevant explanations will not be repeated.

[0193] Figure 10 This is a flowchart illustrating a model training method provided in an embodiment of the present invention, as shown below. Figure 10 As shown, the method includes:

[0194] Step S1001: Collect second historical sample data. The second historical sample data includes historical environmental information, the historical glass temperature of the pre-collected second vehicle window, and the corresponding historical defogging data. The historical defogging data includes the user's expected defogging settings information under the corresponding historical conditions.

[0195] Step S1002: Based on the historical light intensity and the window structure information of the first window, determine the light radiation power of the first window, and predict the historical glass temperature of the first window based on the light radiation power of the first window using the first model;

[0196] Step S1003: Take historical environmental information, window information, historical glass temperature of the first window and historical glass temperature of the second window as input, and take the desired defogging setting information as output to train the second model.

[0197] It should be noted that the model training process has been described in the above model processing example. You can refer to the above example for understanding here, and the relevant explanations will not be repeated.

[0198] Figure 11 This is a schematic diagram of a defogging device based on a fusion model provided in an embodiment of the present invention, as shown below. Figure 11 As shown, the device 1100 includes:

[0199] The acquisition module 1101 is used to acquire the vehicle's environmental information and window information; wherein, the environmental information includes the vehicle interior temperature, the vehicle exterior temperature and the light intensity, and the window information includes the glass humidity of the windows, and the windows include the first window and the second window.

[0200] The first determining module 1102 is used to determine the light radiation power of the first vehicle window based on the light intensity and the window structure information of the first vehicle window;

[0201] The first model processing module 1103 is used to process the vehicle interior temperature, vehicle exterior temperature and light radiation power of the first vehicle window according to the preset first model to obtain the glass temperature of the first vehicle window. The first model is trained based on the historical glass temperature data of the second vehicle window and is used to predict the glass temperature of the first vehicle window.

[0202] The second model processing module 1104 is used to process environmental information, window information, glass temperature of the first window and glass temperature of the pre-collected second window according to the preset second model to obtain defogging recommendation information; wherein, the second model is a model trained based on historical defogging data and used to generate defogging recommendation information.

[0203] In one embodiment, the first window includes a side window of the vehicle, and the second window includes the windshield of the vehicle.

[0204] In one embodiment, the window structure information includes the installation angle and surface area of ​​the window glass; the first determining module 1102 includes: a relative light intensity determining unit, used to determine the relative light intensity of the first window based on the installation angle and light intensity of the first window; and a first power determining unit, used to determine the light radiation power of the first window based on the relative light intensity and surface area.

[0205] In one embodiment, it further includes: a second power determination module, used to determine the ambient thermal radiation power of the first window based on the interior temperature, exterior temperature and surface area;

[0206] The first power determination unit is specifically used to: determine the initial illuminance power of the first window based on the relative illuminance and surface area; and obtain the illuminance power based on the ambient thermal radiation power and the initial illuminance power.

[0207] In one implementation, the second model includes a preprocessing module, a risk prediction module, and a recommendation module. The second model processing module 1104 includes:

[0208] The first processing unit is used to process the glass temperature and humidity of the first vehicle window and the glass temperature and humidity of the second vehicle window respectively through the preprocessing module to obtain the temperature and humidity changes of the first vehicle window and the temperature and humidity changes of the second vehicle window.

[0209] The second processing unit is used to process at least one of the glass temperature, glass humidity and temperature and humidity change information of the first window and the glass temperature, glass humidity and temperature and humidity change information of the second window through the risk prediction module, so as to determine the fogging level of the first window and the second window respectively.

[0210] The recommendation generation unit is used to generate defogging recommendation information for the first and second windows based on the fog level, environmental information, and window information through the recommendation module.

[0211] In one embodiment, the first processing unit is specifically used to: perform first-order derivatives on the glass temperature and glass humidity of the first vehicle window to obtain temperature and humidity change information of the first vehicle window; and perform first-order derivatives on the glass temperature and glass humidity of the first vehicle window to obtain temperature and humidity change information of the second vehicle window.

[0212] The second processing unit is specifically used to: determine the fogging level of the first vehicle window based on at least one of the glass temperature, glass humidity, and temperature and humidity change information of the first vehicle window; and to determine the fogging level of the second vehicle window based on at least one of the glass temperature, glass humidity, and temperature and humidity change information of the second vehicle window.

[0213] In one embodiment, the fogging level of the first vehicle window is determined based on the glass temperature and its temperature change information, and the glass humidity and its humidity change information. Specifically, based on the temperature and humidity change information of the first vehicle window, the fogging probability change coefficient corresponding to the temperature and humidity change information is looked up from a preset mapping table to obtain the fogging probability change coefficient of the first vehicle window; wherein, the mapping table includes fogging probability change coefficients corresponding to different temperature and humidity change information; the fogging level of the first vehicle window is determined based on at least one of the glass temperature, glass humidity, and the fogging probability change coefficient of the first vehicle window.

[0214] In one embodiment, the fogging level of the second window is determined based on the temperature and humidity change information of the second window. Specifically, the fogging probability change coefficient of the second window is obtained by looking up the fogging probability change coefficient corresponding to the temperature and humidity change information in a preset mapping table based on the temperature and humidity change information of the second window. The fogging level of the second window is determined based on at least one of the glass temperature, glass humidity and the fogging probability change coefficient of the second window.

[0215] In one embodiment, the device further includes:

[0216] The control module is used to control the target actuator to perform corresponding defogging operations based on the defogging recommendation information; wherein, the defogging recommendation information includes one of the following recommendations: opening the first window and / or the second window, adjusting the ventilation settings, and adjusting the air conditioning settings; the target actuator includes at least one of the following: the window control system, the ventilation system, and the air conditioning system.

[0217] In one implementation, the control module includes:

[0218] An interactive unit is used to display defogging recommendation information via a central control screen and / or to provide defogging recommendation information via an audio device;

[0219] The response unit is used to respond to touch operations on the central control screen and / or voice input operations on the audio device to determine defogging setting information, which may be the same as or different from the recommended defogging information.

[0220] The defogging unit is used to control the target actuator to perform the corresponding defogging operation based on the defogging setting information.

[0221] In one embodiment, the device further includes:

[0222] The monitoring module is used to monitor the trend information of humidity change and / or temperature change of the vehicle window in real time.

[0223] The control module is also used to control the target actuator to stop the current defogging operation in response to the detected trend information of humidity change and / or temperature change being lower than a preset threshold.

[0224] In one implementation, the environmental information also includes vehicle driving information and / or rainfall information, and the window information also includes window position.

[0225] Figure 12 This is a schematic diagram of the structure of a model training device provided in an embodiment of the present invention, as shown below. Figure 12 As shown, the device 1200 includes:

[0226] The first acquisition module 1201 is used to acquire first historical sample data within a preset historical time period. The first historical sample data includes historical interior temperature, historical exterior temperature, historical light intensity, and historical glass temperature of the second window within the preset historical time period.

[0227] The second determining module 1202 is used to determine the light radiation power of the second window based on the historical light intensity and the window structure information of the second window.

[0228] The first training module 1203 is used to train the first model by taking the historical interior temperature, historical exterior temperature, and historical light radiation power of the second window as inputs and the historical glass temperature of the second window as outputs.

[0229] Figure 13 Another model training apparatus provided in the embodiments of the present invention, such as Figure 13 As shown, the device 1300 includes:

[0230] The second acquisition module 1301 is used to acquire second historical sample data. The second historical sample data includes historical environmental information, the historical glass temperature of the pre-acquired second vehicle window and the corresponding historical defogging data. The historical defogging data includes the user's expected defogging settings information under the corresponding historical conditions.

[0231] The third determining module 1302 is used to determine the light radiation power of the first vehicle window based on the historical light intensity and the window structure information of the first vehicle window, and to predict the historical glass temperature of the first vehicle window based on the light radiation power of the first vehicle window through the first model.

[0232] The second training module 1302 is used to train the second model by taking historical environmental information, window information, historical glass temperature of the first window and historical glass temperature of the second window as inputs, and expected defogging setting information as outputs.

[0233] Figure 14 This is a schematic diagram of a defogging system based on a fusion model provided in an embodiment of the present invention, as shown below. Figure 14 As shown, the system 1400 includes a cloud 1401, a central controller 1402 communicatively connected to the cloud 1401, a vehicle controller 1403 communicatively connected to the central controller 1402, and at least one sensor 1404 communicatively connected to the vehicle controller 1403; wherein,

[0234] Cloud 1401 is used to store the vehicle's historical data, including historical glass temperature data and / or historical defogger data for the second window.

[0235] At least one sensor 1404 is used to collect environmental information and window information of the vehicle;

[0236] The central controller 1402 is used to train a first model based on historical glass temperature data of the second window and a second model based on historical defogging data.

[0237] The vehicle controller 1403 is used to execute the defogging method based on the fusion model provided in the above method embodiment.

[0238] Optionally, it also includes a central control screen and / or an audio device that are communicatively connected to the vehicle controller; wherein the central control screen is used to display defogging recommendation information, and / or the audio device is used to provide defogging recommendation information.

[0239] Figure 15 This is a schematic diagram of the structure of a vehicle provided in an embodiment of the present invention, such as... Figure 15 As shown, the vehicle 1500 includes a memory 1501, a processor 1502, and a communication interface 1503;

[0240] Memory 1501 stores computer-executed instructions;

[0241] Communication interface 1503 receives historical data transmitted from the cloud, including historical glass temperature data and / or historical defogging data of the second window;

[0242] The processor 1502 executes computer execution instructions stored in the memory 1501, causing the processor 1502 to execute the defogging processing method based on the fusion model provided in the above method embodiments, or the model training method provided in the above method embodiments.

[0243] The present invention also provides a computer storage medium, which stores computer-executable instructions that, when executed by a processor, are used to implement the methods corresponding to any of the above embodiments.

[0244] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0245] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an application-specific integrated circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0246] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0247] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0248] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0249] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0250] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0251] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

[0252] The above embodiments are merely preferred embodiments provided to fully illustrate the present invention, and the scope of protection of the present invention is not limited thereto. Equivalent substitutions or modifications made by those skilled in the art based on the present invention are all within the scope of protection of the present invention.

Claims

1. A defogging method based on a fusion model, characterized in that, include: Acquire vehicle environmental information and window information; wherein, the environmental information includes vehicle interior temperature, vehicle exterior temperature and light intensity, and the window information includes window glass humidity and / or window structure information, and the window includes a first window and a second window; Based on the light intensity and the window structure information of the first window, the light radiation power of the first window is determined; According to the preset first model, the vehicle interior temperature, vehicle exterior temperature and light radiation power of the first vehicle window are processed to obtain the glass temperature of the first vehicle window. The first model is trained based on the historical glass temperature data of the second vehicle window and is used to predict the glass temperature of the first vehicle window. Based on the preset second model, the environmental information, the window information, the glass temperature of the first window, and the pre-collected glass temperature of the second window are processed to obtain defogging recommendation information; wherein, the second model is a model trained based on historical defogging data and used to generate defogging recommendation information.

2. The method according to claim 1, characterized in that, The first window includes the side window of the vehicle, and the second window includes the windshield of the vehicle.

3. The method according to claim 1 or 2, characterized in that, The window structure information includes the installation angle and surface area of ​​the window glass; determining the illuminance power of the first window based on the light intensity and the window structure information includes: The relative light intensity of the first vehicle window is determined based on the installation angle of the first vehicle window and the light intensity. The illuminance power of the first vehicle window is determined based on the relative illuminance and the surface area.

4. The method according to claim 3, characterized in that, Also includes: The ambient thermal radiation power of the first window is determined based on the interior temperature, the exterior temperature, and the surface area. Determining the illuminance power of the first vehicle window based on the relative illuminance and the surface area includes: The initial illuminance power of the first vehicle window is determined based on the relative illuminance and the surface area. The illumination radiation power is obtained based on the ambient thermal radiation power and the initial illumination radiation power.

5. The method according to claim 1 or 2, characterized in that, The second model includes a preprocessing module, a risk prediction module, and a recommendation module; the process of processing the environmental information, the window information, the glass temperature of the first window, and the pre-collected glass temperature of the second window according to the preset second model to obtain defogging recommendation information includes: The preprocessing module processes the glass temperature and humidity of the first window and the glass temperature and humidity of the second window to obtain the temperature and humidity changes of the first window and the second window. The risk prediction module processes at least one of the glass temperature, glass humidity and temperature and humidity change information of the first window and at least one of the glass temperature, glass humidity and temperature and humidity change information of the second window to determine the fogging level of the first window and the second window respectively. The recommendation module generates defogging recommendation information for the first and second windows based on the fogging level, environmental information, and window information.

6. The method according to claim 5, characterized in that, The process of processing the glass temperature and humidity of the first vehicle window and the glass temperature and humidity of the second vehicle window includes: The temperature and humidity of the first car window are differentiated by the first derivative to obtain the temperature and humidity change information of the first car window. The temperature and humidity of the first car window are differentiated by first-order derivatives to obtain the temperature and humidity change information of the second car window. The processing of at least one of the glass temperature, glass humidity, and temperature and humidity change information of the first vehicle window, and at least one of the glass temperature, glass humidity, and temperature and humidity change information of the second vehicle window, includes: The fogging level of the first vehicle window is determined based on at least one of the glass temperature, glass humidity, and temperature and humidity change information of the first vehicle window; and the fogging level of the second vehicle window is determined based on at least one of the glass temperature, glass humidity, and temperature and humidity change information of the second vehicle window.

7. The method according to claim 6, characterized in that, The step of determining the fogging level of the first vehicle window based on its glass temperature and temperature change information, and its glass humidity and humidity change information includes: Based on the temperature and humidity change information of the first vehicle window, the fogging probability change coefficient corresponding to the temperature and humidity change information is looked up from a preset mapping table to obtain the fogging probability change coefficient of the first vehicle window; wherein, the mapping table includes fogging probability change coefficients corresponding to different temperature and humidity change information. The fogging level of the first vehicle window is determined based on at least one of the glass temperature, glass humidity, and the fogging probability variation coefficient of the first vehicle window.

8. The method according to claim 6, characterized in that, The step of determining the fogging level of the second vehicle window based on temperature and humidity change information includes: Based on the temperature and humidity change information of the second window, the fogging probability change coefficient corresponding to the temperature and humidity change information is found in the preset mapping table to obtain the fogging probability change coefficient of the second window. The fogging level of the second window is determined based on at least one of the glass temperature, glass humidity, and the fogging probability variation coefficient of the second window.

9. The method according to any one of claims 1, 2, 4, 6-8, characterized in that, Also includes: Based on the defogging recommendation information, the target actuator is controlled to perform the corresponding defogging operation; wherein, the defogging recommendation information includes one of the following recommendations: opening the first window and / or the second window, adjusting the ventilation settings, and adjusting the air conditioning settings; the target actuator includes at least one of the following: a window control system, a ventilation system, and an air conditioning system.

10. The method according to claim 9, characterized in that, The step of controlling the target actuator to perform the corresponding defogging operation based on the defogging recommendation information includes: The defogging recommendation information is displayed on the central control screen, and / or the defogging recommendation information is prompted through an audio device; In response to a touch operation on the central control screen and / or a voice input operation on the audio device, defogging setting information is determined, which may be the same as or different from the defogging recommendation information; Based on the defogging settings information, the target actuator is controlled to perform the corresponding defogging operation.

11. The method according to claim 9, characterized in that, Also includes: Real-time monitoring of the humidity and / or temperature trends of the vehicle windows; In response to the detected trend information of humidity change and / or temperature change being lower than a preset threshold, the target actuator is controlled to stop the current defogging operation.

12. The method according to any one of claims 1, 2, 4, 6-8, 10, and 11, characterized in that, The environmental information also includes vehicle driving information and / or rainfall information, and the window information also includes window position.

13. A model training method, characterized in that, include: Collect first historical sample data, which includes historical interior temperature, historical exterior temperature, historical light intensity, and historical glass temperature of the second window within the preset historical time period. Based on the historical light intensity and the window structure information of the second window, the light radiation power of the second window is determined; The first model is trained by taking the historical interior temperature, the historical exterior temperature, and the historical irradiance of the second window as inputs, and the historical glass temperature of the second window as output.

14. A model training method, characterized in that, include: Collect second historical sample data, which includes historical environmental information, pre-collected historical glass temperature of the second vehicle window, and corresponding historical defogging data. The historical defogging data includes the user's expected defogging settings under the corresponding historical conditions. Based on historical light intensity and the window structure information of the first window, the light radiation power of the first window is determined, and the historical glass temperature of the first window is predicted by the first model based on the light radiation power of the first window. The historical environmental information, window information, historical glass temperature of the first window, and historical glass temperature of the second window are used as inputs, and the desired defogging setting information is used as output to train a second model.

15. A defogging treatment device based on a fusion model, characterized in that, include: The acquisition module is used to acquire the vehicle's environmental information and window information; wherein, the environmental information includes the vehicle interior temperature, the vehicle exterior temperature, and the light intensity, and the window information includes the humidity of the window glass, and the window includes a first window and a second window; The first determining module is used to determine the light radiation power of the first vehicle window based on the light intensity and the window structure information of the first vehicle window; The first model processing module is used to process the vehicle interior temperature, vehicle exterior temperature and light radiation power of the first vehicle window according to the preset first model to obtain the glass temperature of the first vehicle window. The first model is trained based on the historical glass temperature data of the second vehicle window and is used to predict the glass temperature of the first vehicle window. The second model processing module is used to process the environmental information, the window information, the glass temperature of the first window, and the pre-collected glass temperature of the second window according to the preset second model to obtain defogging recommendation information; wherein, the second model is a model trained based on historical defogging data and used to generate defogging recommendation information.

16. A model training device, characterized in that, include: The first acquisition module is used to acquire first historical sample data within a preset historical time period. The first historical sample data includes the historical interior temperature, historical exterior temperature, historical light intensity, and historical glass temperature of the second window within the preset historical time period. The second determining module is used to determine the light radiation power of the second window based on the historical light intensity and the window structure information of the second window; The first training module is used to train a first model by taking the historical interior temperature, the historical exterior temperature, and the historical irradiance power of the second window as inputs and the historical glass temperature of the second window as outputs.

17. A model training device, characterized in that, include: The second acquisition module is used to acquire second historical sample data. The second historical sample data includes historical environmental information, the historical glass temperature of the pre-acquired second vehicle window, and the corresponding historical defogging data. The historical defogging data includes the user's expected defogging settings information under the corresponding historical conditions. The third determining module is used to determine the light radiation power of the first window based on the historical light intensity and the window structure information of the first window, and to predict the historical glass temperature of the first window based on the light radiation power of the first window through the first model. The second training module is used to train a second model by taking the historical environmental information, window information, historical glass temperature of the first window and historical glass temperature of the second window as inputs and the desired defogging setting information as outputs.

18. A defogging system based on a fusion model, characterized in that, This includes a cloud platform, a central controller communicatively connected to the cloud platform, a vehicle controller communicatively connected to the central controller, and at least one sensor communicatively connected to the vehicle controller; wherein, The cloud is used to store historical data of the vehicle, including historical glass temperature data and / or historical defogging data of the second window; The at least one sensor is used to collect environmental information and window information of the vehicle; The central controller is used to train a first model based on the historical glass temperature data of the second window, and to train a second model based on the historical defogging data. The vehicle controller is used to execute the defogging process based on the fusion model as described in any one of claims 1-12.

19. The system according to claim 18, characterized in that, It also includes a central control screen and / or an audio device that are communicatively connected to the vehicle controller; wherein the central control screen is used to display the defogging recommendation information, and / or the audio device is used to prompt the defogging recommendation information.

20. A vehicle, characterized in that, Includes memory, processor, and communication interface; The memory stores computer-executed instructions; The communication interface receives historical data transmitted from the cloud, including historical glass temperature data and / or historical defogging data of the second window. The processor executes computer execution instructions stored in the memory, causing the processor to perform the defogging processing method based on the fusion model as described in any one of claims 1-12, or the model training method as described in claim 13 or 14.

Images