Method for setting up air conditioning system

Abstract

The invention relates to a method for setting up an air-conditioning system, in particular a method (1) for setting up an air-conditioning system (2) of a vehicle (3) for a user (13) in a user-adapted manner by means of a control architecture (4). In this case, model-dependent data packets (5) are collected, and then a comfort model (9) is trained in an AI unit (8). Furthermore, setting-dependent data packets (11) are collected and transmitted to the trained comfort model (9). On the basis thereof, a user-adapted setting (12) is predicted by means of a comfort model (9), and the air-conditioning installation (2) is set. The invention also relates to a control architecture (4) for implementing the method (1).

Description

Technical Field

[0001] The invention relates to a method for adjusting the air conditioning system of a vehicle for a user in a user-adapted manner by means of a control architecture according to the preamble of claim 1. The invention also relates to a control architecture for carrying out the method. Background Art

[0002] The user of the air-conditioning facility can set the air-conditioning facility according to his thermal perception by changing multiple parameters. Through air-conditioning control based on comprehensive characteristic curves, the user can be exempted from the setting of the air-conditioning facility, and the operation of the air-conditioning facility can be simplified. However, the thermal comfort of each user is very individual. Disadvantageously, the traditional air-conditioning control based on comprehensive characteristic curves is designed for "ordinary" people and cannot be adapted to each user individually. If the user feels uncomfortable in the case of air-conditioning control based on comprehensive characteristic curves, the user must laboriously reset the air-conditioning facility manually. AI-assisted comfort models (AI: artificial intelligence) provide another solution, which can be adapted to the personal preferences of individual users. It is hoped that the existing comfort models can also be used for driving other vehicles.

[0003] For example, DE 10 2020 107178A1, DE 10 2015 221416A1, DE 102018 113074A1 disclose solutions for setting up air conditioning facilities in a manner adapted to the user. In addition, DE 10 2018 129417A1 discloses AI-supported regulation of any device. Summary of the invention

[0004] The object of the present invention is therefore to specify an improved specific embodiment or at least an alternative specific embodiment for a method of this type in which the described disadvantages are overcome. It is also an object of the present invention to provide a corresponding control architecture for implementing the method.

[0005] According to the invention, this object is achieved by the subject matter of the independent claim. Advantageous embodiments are the subject matter of the dependent claims.

[0006] The general idea on which the invention is based is to train the AI ​​unit and create an individual comfort model for the user.

[0007] The method according to the invention is provided for setting the air conditioning system of a vehicle for a user in a manner adapted to the user and is implemented with the aid of a control architecture. In this regard, in the method, a modeling cycle and a setting cycle are repeatedly implemented. In the modeling cycle, at least one model-related data packet of the user is first collected. Then, in an AI unit (AI: artificial intelligence), a comfort model is trained using the data of at least one model-related data packet of the user's model-related data packets. In the setting cycle, setting-related data packets of the vehicle are collected and transmitted to the trained comfort model. Then, with the aid of the comfort model, user-adapted settings for setting the air conditioning system in a manner adapted to the user are predicted based on the data of the setting-related data packets. The air conditioning system is then set according to these user-adapted settings.

[0008] The modeling cycle is repeatedly implemented to train the comfort model with the data of the model-related data packets. Here, once the comfort model has been trained once, the respective modeling cycle can be ended. Then, a new modeling cycle can be started. Here, in the respective modeling cycle, a plurality of model-related data packets can be transmitted. Here, the model-related data packets can be repeatedly formed when the air conditioning system is used and can be used for further training of the comfort model. As a result, the comfort model can always be further adapted to the user or always further individualized when the air conditioning system is used. The setting cycle is repeatedly implemented to predict the user-adapted settings based on the data of the setting-related data packets and to set the air conditioning system. These user-adapted settings are necessary for the use of the air conditioning system and are predicted with the aid of the trained comfort model. In other words, the comfort model is always further individualized when the air conditioning system is used, and the air conditioning system is set more individually or better and better in a way that is adapted to the user with the aid of the individualized comfort model. Here, the comfort model is trained with the data of the user's model-related data packets, so that the computing power in the vehicle is reduced. In addition, the trained comfort model can be transplanted to other vehicles of the same user. Therefore, with the aid of this method, individualized comfort models can be created and centrally trained. Here, it should be understood that using the described method, multiple comfort models can also be trained for multiple users.

[0009] In conjunction with the present invention, user-adapted settings of the air conditioning facility should be understood as settings that are adapted to the user or that set the air conditioning facility according to the user's desired feeling. The user-adapted settings of the air conditioning facility may include target settings of the air conditioning facility of the vehicle. Here, the target settings of the air conditioning facility may be adapted to the user. In other words, the trained comfort model may map all of the settings to be set to the target settings of the air conditioning facility. Here, the target settings of the air conditioning facility correspond to the settings of the air conditioning facility that are optimal for a specific user under given environmental conditions.

[0010] In conjunction with the present invention, the data of the model-related data packet should be understood as data or parameter values ​​that are important for the training of the comfort model or can be taken into account when training the comfort model. In conjunction with the present invention, the data of the setting-related data packet should be understood as data or parameter values ​​that are important for the setting of the air-conditioning facility and / or can be taken into account in the comfort model when predicting settings that adapt to the user. The model-related data packet and the setting-related data packet may include data or parameter values ​​that are partially identical to each other. The model-related data packet and the setting-related data packet may include data or parameter values ​​that are partially different from each other. In particular, the model-related data packet may contain a setting-related data packet and other information.

[0011] The model-related data packets and / or the setting-related data packets may include situational data detected by sensors of the vehicle. Here, these situational data may include at least one situational parameter of the vehicle. Preferably, the at least one situational parameter may be the temperature inside the vehicle and / or the temperature outside the vehicle and / or the air humidity and / or the position of the sun and / or the light intensity and / or the air pressure. It should be understood that the mentioned enumeration is not closed. Thus, these situational data may be taken into account when training a comfort model and / or when setting up an air conditioning facility, so that the comfort model may be individualized and / or the air conditioning facility may be set according to the individual needs of the user.

[0012] The model-related data packet and / or the setting-related data packet may include user data. Here, these user data may include at least one user parameter of the user. Preferably, the at least one user parameter may be user physiological information, such as sitting position and / or user identity and / or skin temperature and / or pulse and / or breathing rate and / or age and / or gender and / or clothing condition and / or emotional state and / or biometric input and / or physiological input. It should be understood that the enumeration mentioned is not closed. These user data may be detected, for example, by sensors of the vehicle. The sensors of the vehicle may be shaped by cameras installed in the vehicle or infrared cameras installed in the vehicle. In addition, these user data may be provided, for example, by the user's mobile device. These user data may be learned, for example, from information stored in the mobile device. In this way, it may be learned, for example, from calendar data whether the user has just performed a planned sports activity.

[0013] The creation of a model-related data package can be initiated when the user manually makes a user input. In this case, the user input can be provided for changing the current user-adapted settings of the air-conditioning system. The user input can then be transmitted in a model-related data package, if necessary together with other data, and can thus be taken into account when training the comfort model. In this case, in particular, the creation of a model-related data package can be initiated with each user input. In this case, the user input is made manually by the user and is intended to change the current settings of the air-conditioning system. The user can make this user input, for example, when the desired feeling of the user is not achieved by the predicted user-adapted settings and therefore by the current comfort model.

[0014] Then, the creation of a model-related data packet can be initiated by means of a user input, and the data of the current model-related data packet can be taken into account in the comfort model. In this case, the current model-related data packet includes the user input currently performed by the user and, if necessary, further data, so that the comfort model can be adapted to the individual user and his or her individual senses. In addition to the user input, the model-related data packet can include the above-mentioned situation data and / or the above-mentioned user data, so that the comfort model can be trained based on the user input in the currently existing situation data and / or in combination with the currently existing user data. Since the user input is performed manually by the user, the model-related data packets can be created correspondingly without a predefinable frequency and / or without a predefinable frequency.

[0015] Here, the user input may include setting data input by the user. Then, these setting data may include at least one setting parameter. Preferably, the at least one setting parameter may be the desired temperature and / or the fan gear level to be set and / or the seat heating to be set and / or the heating surface to be set and / or the heating intensity to be set. It should be understood that the mentioned enumeration is not closed. Alternatively or additionally, the user input may include optimization data input by the user. Then, these optimization data may include at least one optimization parameter. The at least one optimization parameter may preferably include input about user-specific sensations. In this way, through these optimization data, such information as "head is too cold", "right leg is too hot", "air supply is uncomfortable", etc. can be taken into account.

[0016] Alternatively or additionally, if no user input occurs within a predefined time period, the creation of a model-related data package can be initiated after the predefined time period. As already explained above, the user input is performed manually by the user and causes a change to the current user-adapted settings of the air-conditioning facility. Therefore, if the user does not change the user-adapted settings of the air-conditioning facility, the desired feeling of the user is already achieved with the help of the current comfort model, or the user-specific settings are predicted well enough under the existing environmental conditions. This can serve as positive feedback for the user. The predefined time period can be, for example, between 1 minute and 30 minutes, in particular between 5 minutes and 20 minutes.

[0017] The setting cycle can be executed at a predefined frequency. The predefined frequency can preferably be between 0.1 Hz and 10 Hz, particularly preferably 1 Hz. Here, the setting cycle can be started independently of the user input. As a result, the air conditioning facility can be set according to changing environmental conditions without the user actively performing manual operation. Therefore, the modeling cycle and the setting cycle can be executed independently of each other and / or at different time points. As already explained above, the modeling cycle can be started without a predefined frequency and / or without a predefined frequency. Correspondingly, the modeling cycle and the setting cycle can be executed with different frequencies or frequencies.

[0018] Based on the data of the model-related data packets, offset settings can be created. These offset settings can then be stored in the vehicle and additionally used to set the air conditioning facility. However, once these offset settings have been taken into account in the trained comfort model, they can be deleted. Thus, when setting the air conditioning facility, data of the model-related data packets that have not yet been taken into account in the current comfort model can also be taken into account. These offset settings can, for example, take into account user input that has not yet been taken into account in the comfort model. These offset settings can, in particular, include offsets that the user desires relative to the current settings of the air conditioning facility or relative to the currently set setting parameters of the air conditioning facility. In this way, if the user, for example, wants to reduce or increase the in-vehicle temperature set by the air conditioning facility, these offset settings can, for example, include an offset of the in-vehicle temperature set by the air conditioning facility. If multiple model-related data packets that cannot yet be taken into account in the comfort model are formed, all of these data packets can be used in sequence to create these offset settings.

[0019] In the method, the comfort model can be trained periodically at predefined time points, such as every hour or every day or every week. Here, all model-related data packets of the user can be stored between two consecutive time points, and these model-related data packets can be used to train the comfort model. In other words, multiple model-related data packets of the user can be stored, such as stored in the vehicle and / or stored in the cloud, and the data of multiple data packets can be used to train the comfort model at the same time. In other words, the creation of a model-related data packet does not automatically mean the training of the comfort model. These model-related data packets can be stored before training the comfort model, such as stored in the vehicle and / or stored in the cloud, and can be used once the comfort model is trained. It is conceivable that not all model-related data packets are used to train the comfort model. In other words, the comfort model can also be trained using some model-related data packets. Here, the comfort model can be trained periodically at predefined time points, such as every hour or every day or every week. Here, it should be understood that if there is no model-related data packet, the comfort model is not trained.

[0020] These model-related data packets can be weighted when training the comfort model. For example, the time period between two consecutive model-related data packets can be used for this weighting. If the time period is small, the data of the first data packet does not seem to achieve the user's desired feeling and is not so important when training the comfort model. If the time period is large, the data of the first data packet seems to achieve the user's desired feeling and is highly important when training the comfort model. The terms "small" and "large" should be viewed relative to each other. A "small" time period can be, for example, 1 minute to 5 minutes, while a "large" time period can be, for example, 10 minutes to 30 minutes.

[0021] In a possible first embodiment of the method, the modeling cycle can be performed in the vehicle. The comfort model can then be trained in the vehicle's AI unit. Here, the method can be implemented completely locally in the vehicle without the need for the cloud.

[0022] In a first embodiment of the method, the comfort model can be stored in the vehicle after training in a modeling cycle, and the comfort model can be implemented in the vehicle in a setting cycle. In this case, the settings adapted to the user can be predicted locally in the vehicle. For this purpose, it is self-evident that the comfort model trained in the vehicle can be used.

[0023] In a possible second embodiment of the method, the modeling cycle can be performed in the cloud. Then, after collection, the user's model-related data packets can be transmitted to the cloud and the comfort model can be trained in the AI ​​unit of the cloud. In this case, the cloud is a network that provides computer resources such as servers and / or data storage and / or applications.

[0024] In a second embodiment of the method, the model-related data packets can be stored in the memory of the vehicle if the user's model-related data packets cannot be transmitted to the cloud. The model-related data packets can then be transmitted to the cloud as soon as the user's model-related data packets can be transmitted to the cloud. Thus, when training the comfort model, the data of the model-related data packets that cannot be transmitted immediately can also be taken into account. If a plurality of model-related data packets are formed, which cannot be transmitted immediately to the cloud, all of these data packets can be stored in the memory and then transmitted together to the cloud. Correspondingly, when training the comfort model, all of these data packets can be taken into account without data loss.

[0025] As already described above, the comfort model can be trained periodically at predefined points in time. In this case, all model-related data packets of a user can be stored in the vehicle or in the cloud between two consecutive points in time, and these model-related data packets can be used to train the comfort model. If all model-related data packets of a user are stored in the vehicle, these model-related data packets can be periodically transmitted from the vehicle to the cloud at predefined points in time. In other words, a plurality of model-related data packets of a user can be transmitted to the cloud simultaneously. Thus, the comfort model can be trained simultaneously using all transmitted model-related data packets.

[0026] In a second embodiment of the method, the comfort model can be stored in a model pool in the cloud. Then, a comfort model that has been created can be called from the model pool and trained to retrain the comfort model. Then, the comfort model stored in the model pool can be replaced or updated with a new or current or trained comfort model. Additionally, the comfort model can be transmitted from the model pool to the vehicle and stored in the vehicle. Here, the comfort model can be periodically transmitted from the model pool to the vehicle at predefined time points. In this way, for example, the comfort model can be transmitted to the vehicle every hour, every day, or every week. In particular, the comfort model can be transmitted from the model pool in the cloud to the vehicle after each training is completed.

[0027] The method can be performed in different ways in its second embodiment. In the following, three alternatives that differ from each other are described in more detail. It should be understood that it is conceivable to expand the described alternative with the above features. It should also be understood that other alternatives can also be implemented in the method.

[0028] In a first alternative, in the modeling cycle, the comfort model can be stored in a model pool in the cloud after training. In addition, the comfort model can be transferred from the model pool to the vehicle and additionally stored in the vehicle. Then, in the setting cycle, the setting-related data packets can be transferred to the comfort model stored in the vehicle. Correspondingly, the comfort model can be implemented in the vehicle and these user-adapted settings can be predicted locally in the vehicle. Therefore, in the first alternative, the cloud is used only in the modeling cycle. As described above, the modeling cycle and the setting cycle can be executed with different frequencies or frequencies from each other. In particular, the frequency or frequency of the modeling cycle can be several times smaller than the frequency or frequency of the setting cycle. As a result, the amount of data transmitted to and received from the cloud can be significantly reduced.

[0029] In a second alternative, in the modeling cycle, the comfort model can be stored in a model pool in the cloud after training. Then, in the setting cycle, the setting-related data packets can be transferred from the vehicle to the comfort model stored in the model pool. The comfort model can then be implemented in the cloud and these user-adapted settings can be predicted in the cloud and then transferred to the vehicle. Thus, in the second alternative, the comfort model is not stored in the vehicle and the modeling cycle and the setting cycle are performed via the cloud.

[0030] In a third alternative, in the modeling loop, the comfort model can be stored in a model pool in the cloud after training. In addition, the comfort model can be transferred from the model pool to the vehicle and additionally stored in the vehicle. If the vehicle's setting-related data packets cannot be transferred to the cloud, then in the setting loop, the setting-related data packets are transferred to the comfort model stored in the vehicle. The comfort model is then implemented in the vehicle and the user-adapted settings are predicted locally in the vehicle. However, if the vehicle's setting-related data packets can be transferred to the cloud, the setting-related data packets are transferred from the vehicle to the comfort model stored in the model pool. The comfort model is then implemented in the cloud, and these user-adapted settings are predicted in the cloud and then transmitted to the vehicle. The third alternative is therefore a hybrid solution of the first alternative and the second alternative of the method.

[0031] The present invention also relates to a control architecture for implementing the above method. Here, the control architecture is designed to implement the method. Specifically, the control architecture is designed to repeatedly collect model-related data packets of the user; to repeatedly train the comfort model using the data of the model-related data packets of the user in the AI ​​unit; to repeatedly collect setting-related data packets of the vehicle and to repeatedly transmit the setting-related data packets to the comfort model; to repeatedly predict user-adapted settings for setting air-conditioning facilities in a user-adapted manner based on the data of the setting-related data packets with the help of the comfort model; and to repeatedly set the air-conditioning facilities according to the user-adapted settings. It should be understood that the control architecture has the hardware and / or the software required for implementing the method. Here, the required hardware and / or software can be part of the vehicle and / or part of the cloud. In order to avoid repetition, the above statements are quoted here.

[0032] Further important features and advantages of the invention emerge from the dependent claims, from the drawings and from the associated description of the figures based on the drawings.

[0033] It goes without saying that the features mentioned above and yet to be explained below can be used not only in the respectively specified combination but also in other combinations or alone, without departing from the scope of protection of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Preferred exemplary embodiments of the invention are shown in the drawings and are explained in more detail in the following description, wherein identical reference numerals refer to identical or similar or functionally identical components.

[0035] In which respectively schematically:

[0036] Figure 1 A schematic sequence of the method according to the invention is shown in a first alternative with a cloud;

[0037] Figure 2 shows a schematic sequence of the method according to the invention in a second alternative with a cloud;

[0038] Figure 3 A schematic sequence of the method according to the invention is shown in a third alternative without clouds. DETAILED DESCRIPTION

[0039] Figure 1 A schematic diagram of the process of the method 1 according to the invention in a first alternative is shown. The method 1 is provided for setting an air conditioning system 2 of a vehicle 3 in a manner adapted to a user and is carried out by means of a control architecture 4. In the method 1, a modeling cycle MS and a setting cycle are repeatedly carried out.

[0040] Here, in a modeling cycle MS, model-related data packets 5 are collected and transmitted to a cloud 6. In the cloud 6, the model-related data packets 5 are stored in a memory 7. Then, in the AI ​​unit 8 of the cloud 6, a comfort model 9 is trained with the data of the model-related data packets 5. The trained comfort model 9 is then stored in a model pool 10. In a first alternative of the method 1, the comfort model 9 is additionally transmitted to the motor vehicle 3 and stored in the motor vehicle 3. In a setting cycle ES, setting-related data packets 11 of the vehicle 3 are collected and transmitted to the comfort model 9 stored in the vehicle 3. With the aid of the comfort model 9, user-adapted settings 12 are predicted based on the data of the setting-related data packets 11. These user-adapted settings 12 are then used to set the air conditioning system 2 in a user-adapted manner.

[0041] The setting-related data packet 11 may include situational data of the vehicle 3, such as the temperature inside the vehicle and / or the temperature outside the vehicle and / or the air humidity and / or the position of the sun and / or the light intensity and / or the air pressure. Alternatively or additionally, the setting-related data packet 11 may include user data, such as user physiological information, such as the seating position and / or the user identity and / or the skin temperature and / or the pulse and / or the breathing rate and / or the age and / or the gender and / or the clothing condition and / or the emotional state and / or the biometric input and / or the physiological input. Based on the data of the setting-related data packet 11, the setting 12 of the air conditioning facility 2 can be predicted with the aid of the comfort model 9, and the air conditioning facility 2 can be set. As a result, the air conditioning facility 2 is set individually in a user-defined manner or according to the needs of the user 13.

[0042] In the method 1 , a modeling cycle MS is started by a user input 14. If the desired thermal sensation of a user 13 of the vehicle 3 is not achieved with the current user-adapted settings 12 of the air conditioning system 2 , the user manually performs a user input 14. The user input 14 may be setting data entered by the user 13, such as a desired temperature and / or a fan level to be set and / or a seat heating to be set and / or a heating surface to be set and / or a heating intensity to be set. Alternatively or additionally, the user input 14 may include optimization data entered by the user 13, such as “head too cold”, “right leg too hot”, “air supply uncomfortable”, etc.

[0043] Next, the user input 14 is combined with the setting-related data packet 11 containing the situation data and the user data to form a model-related data packet 5 and transmitted to the cloud 6. The model-related data packet 5 is stored in the cloud 6 and used for further training of the comfort model 9. For this purpose, the currently applicable comfort model 9 is called from the model pool 10 and further trained. After this training, the comfort model 9 in the model pool 10 is updated and the updated comfort model 9 is stored in the vehicle 3. Correspondingly, in the updated comfort model 9, the user input 14 is taken into account in association with the data of the setting-related data packet 11.

[0044] If the model-dependent data package 5 cannot be transmitted, the model-dependent data package 5 is stored in the vehicle 3. In the method 1 according to the first alternative, offset settings 15 are also created based on the data of the model-dependent data package 5 that has not been transmitted. The offset settings 15 are stored in the vehicle 3 and are used to correct the user-adapted settings 12. These offset settings 15 can, for example, include offsets of the setting variables to be set for the air conditioning system 2. As soon as the transmission is possible, the model-dependent data package 5 is transmitted to the cloud 6 and the offset settings 15 are deleted.

[0045] Figure 2 The schematic sequence of the method 1 according to the invention in the second alternative is shown. The method 1 in the second alternative differs from the method 1 in the first alternative in that the comfort model 9 is not transmitted to the vehicle 3 and is implemented in the cloud 6. Correspondingly, the setting-related data packets 11 are transmitted to the cloud 6 and the comfort model 9 is implemented in the cloud 6. Thus, user-adapted settings 12 are predicted in the cloud 6 and then transmitted to the vehicle 3, in which the air conditioning system 2 is now set. Here, these offset settings 15 can also include offsets of the setting variables to be set for the air conditioning system 2.

[0046] refer to Figure 1 and Figure 2 , the method 1 can also be performed according to another alternative. In this alternative, the trained comfort model 9 is stored in a model pool 10 and in the vehicle 3. If there is a data transmission connection to the cloud 6, the comfort model 9 in the model pool 10 is used, and in accordance with Figure 2 If there is no data transmission connection to the cloud 6, the comfort model 9 in the vehicle 3 is used and in accordance with Figure 1 Method 1 is used to set up loops.

[0047] Figure 3The schematic process of the method 1 according to the invention is shown in a third alternative without clouds. Here, the modeling cycle MS and the setting cycle ES are performed locally in the vehicle 3. Otherwise, the method 1 in the third alternative corresponds to the method 1 in the second alternative.

Claims

1. A method (1) for setting a climate control system (2) of a vehicle (3) for a user (13) in a manner adapted to the user by means of a control architecture (4), in, In the iterative modeling cycle (MS): - collecting at least one model-related data package (5) of the user (13); - in the AI ​​unit (8), training a comfort model (9) using data from at least one of the model-related data packets (5) of the user (13); Among them, in the repeated setup cycle (ES): - collecting data packets (11) related to the settings of the vehicle (3) and transmitting the data packets related to the settings to the trained comfort model (9); - predicting user-adapted settings (12) for setting the air conditioning system (2) in a user-adapted manner based on the data of the settings-related data package (11) by means of the comfort model (9); - Setting the air conditioning system (2) according to the user-adapted settings (12).

2. The method according to claim 1, It is characterized in that - upon manual user input (14) by the user (13) to change the current user-adapted settings (12) of the air conditioning installation (2), the creation of the model-related data package (5) is initiated; and / or - if no manual user input (14) by the user (13) occurs within a predefined time period to change the current user-adapted settings (12) of the air-conditioning system (2), the creation of the model-related data package (5) is initiated after the predefined time period.

3. The method according to claim 1 or 2, It is characterized in that The setup cycle (ES) is executed with a predefined frequency, which is preferably between 0.5 Hz and 10 Hz, particularly preferably 1 Hz.

4. The method according to any one of the preceding claims, It is characterized in that - creating an offset setting (15) based on the data of the model-related data package (5), storing the offset setting in the vehicle (3) and additionally serving to set the air conditioning system (2); and - once the offset setting (15) has been taken into account in the trained comfort model (9), the offset setting (15) is deleted.

5. The method according to any one of the preceding claims, It is characterized in that - periodically training the comfort model (9) at predefined time points, wherein all model-related data packets (5) of the user (13) are stored between two consecutive time points and all model-related data packets are used to train the comfort model (9); and / or The data of the model-related data packets (5) are weighted when training the comfort model (9).

6. The method according to any one of the preceding claims, It is characterized in that - executing said modelling cycle (MS) in said vehicle (3); and - Training the comfort model (9) in an AI unit (8) of the vehicle (3).

7. The method according to claim 6, It is characterized in that In the Modeling Cycle (MS): - storing the comfort model (9) in the vehicle (3) after the training; In the setup loop(ES): - implementing the comfort model (9) in the vehicle (3) and predicting the user-adapted settings (12) locally in the vehicle (3).

8. The method according to any one of the preceding claims, It is characterized in that - executing said modelling cycle (MS) in the cloud (6); - after collection, transmitting the user's (13) model-related data package (5) to the cloud (6); and - Training the comfort model (9) in the AI ​​unit (8) of the cloud (6).

9. The method according to claim 8, It is characterized in that - if the model-related data package (5) of the user (13) cannot be transmitted to the cloud (6), storing the model-related data package (5) in the vehicle (3); and - As soon as the model-related data package (5) of the user (13) can be transmitted to the cloud (6), the model-related data package (5) is transmitted to the cloud (6) and the model-related data package is deleted in the vehicle (3).

10. The method according to claim 8 or 9, It is characterized in that - storing all model-related data packets (5) of the user (13) in the vehicle (3) and transmitting them periodically at predefined points in time from the vehicle (3) to the cloud (6); and / or - storing the comfort model (9) in a model pool (10) of the cloud (6); and / or - the comfort model (9) is stored in a model pool (10) of the cloud (6) and is additionally transmitted from the model pool (10) to the vehicle (3) and stored in the vehicle (3); and / or - storing the comfort model (9) in a model pool (10) of the cloud (6), and periodically transmitting the current comfort model (9) from the model pool (10) to the vehicle (3) at predefined time points; and / or The comfort model (9) is stored in a model pool (10) of the cloud (6) and is transmitted from the model pool (10) of the cloud (6) to the vehicle (3) after each training is completed.

11. The method according to any one of claims 8 to 10, It is characterized in that In the Modeling Cycle (MS): - storing the comfort model (9) in a model pool (10) of the cloud (6) after the training; - transmitting the comfort model (9) from the model pool (10) to the vehicle (3) and additionally storing it in the vehicle (3); In the setup loop(ES): - transmitting the setting-related data packet (11) to a comfort model (9) stored in the vehicle (3); - implementing the comfort model (9) in the vehicle (3) and predicting the user-adapted settings (12) locally in the vehicle (3).

12. The method according to any one of claims 8 to 10, It is characterized in that In the Modeling Cycle (MS): - storing the comfort model (9) in a model pool (10) of the cloud (6) after the training; In the setup loop(ES): - transmitting the setting-related data packet (11) from the vehicle (3) to the comfort model (9) stored in the model pool (10); The comfort model (9) is implemented in the cloud (6), and the user-adapted settings (12) are predicted in the cloud (6) and then transmitted to the vehicle (3).

13. The method according to any one of claims 8 to 10, It is characterized in that In the Modeling Cycle (MS): - storing the comfort model (9) in a model pool (10) of the cloud (6) after the training; - transmitting the comfort model (9) from the model pool (10) to the vehicle (3) and additionally storing it in the vehicle (3); In the setup loop(ES): - if the data packet (11) related to the settings of the vehicle (3) cannot be transmitted to the cloud (6), the data packet (11) related to the settings is transmitted to a comfort model (9) stored in the vehicle (3), the comfort model (9) is implemented in the vehicle (3), and the user-adapted settings (12) are predicted locally in the vehicle (3); and - If it is possible to transmit a data packet (11) related to the settings of the vehicle (3) to the cloud (6), the data packet (11) related to the settings is transmitted from the vehicle (3) to a comfort model (9) stored in the model pool (10), the comfort model (9) is implemented in the cloud (6), the user-adapted settings (12) are predicted in the cloud (6) and then the user-adapted settings are transmitted to the vehicle (3).

14. A control architecture (4) for implementing the method (1) according to any one of the preceding claims, wherein: The control architecture (4) is designed to: - for repeatedly collecting model-related data packets (5) of users (13); - for repeatedly training a comfort model (9) in an AI unit (8) using data from a model-related data package (5) of the user (13); - for repeatedly collecting setting-related data packets (11) of the vehicle (3) and for repeatedly transmitting the setting-related data packets (11) to the comfort model (9); - for repeatedly predicting user-adapted settings (12) for setting the air conditioning system (2) of the vehicle (3) in a user-adapted manner by means of the comfort model (9) based on the data of the settings-related data packet (11); - for repeatedly setting the air conditioning device (2) according to the user-adapted settings (12).

Images