Controlling an air-conditioning device using artificial intelligence

Abstract

The invention relates to a method and a device for controlling an air-conditioning device (1) which air-conditions the interior (15) of a vehicle (120). According to the method, the air-conditioning device (1) is controlled by a control device (30) in order to air-condition the interior (15) of the vehicle (120) using the control process. In order to improve the operation of the air-condition device (1), a computing device (10, 110) forms a model using an artificial neural network (32), said model being designed to ascertain a control variable to be output by the control device (30) using influence variables (2) which have an influence on the air-conditioning of the interior (15). A computer program which represents the model is used (E) on the control device (30) in order to control the air-conditioning device (1).

Description

[0001] Description

[0002] Controlling an air conditioner using artificial intelligence

[0003] The invention relates to a method for controlling an air conditioning unit that air-conditions the interior of a vehicle. The air conditioning unit is controlled by means of a control device in order to air-condition the interior of the vehicle using the control. The invention further relates to a device for controlling an air conditioning unit that air-conditions the interior of a vehicle. The device comprises a control device that is configured to control the air conditioning unit in order to air-condition the interior of the vehicle.

[0004] The air treatment in rail vehicles is described, for example, in the standards DIN EN 13129 (Title: Air treatment in rail vehicles for long-distance transport - Comfort parameters and type tests), DIN EN 14750 (Title: Air treatment in rail vehicles for inner-city and regional transport) and DIN EN 14813 (Title: Air treatment in driver's cabs).

[0005] It is generally known to control an air conditioning unit using a controller. This controller is implemented, for example, as embedded software. Proportional-integral controllers or 2-point controllers are often used. Controllers are also known to respond to input variables with a linear behavior whose parameters are determined before the vehicle is operated.

[0006] Against this background, the object of the invention is to improve the operation of the air conditioning unit.

[0007] This problem is solved by a method of the type mentioned above, in which a computer, using an artificial neural network, creates a model that is configured to determine a control variable for output by the control device based on influencing variables that affect the interior climate. A computer program representing the model is used on the control device to control the air conditioning unit.

[0008] The invention recognizes that influencing factors that influence the climate in the vehicle interior are not taken into account in previous air conditioning control systems, or are only taken into account indirectly - by measuring a temperature deviation between the desired interior temperature and the actual interior temperature. The control system treats the entire system (consisting of vehicle, air conditioning unit and the vehicle's surroundings) as a black box and reacts exclusively to a temperature change within the interior. This leads to inertia, which is accompanied by a long time period until the desired temperature is reached within the interior. In addition, strong influences and the associated reaction of the controller can lead to overshooting of the system, which further reduces the comfort for people inside the interior.

[0009] Furthermore, the invention recognizes that previous control systems only consider those influencing factors known to experts and are therefore taken into account when setting the pre-control parameters. This parameter setting remains static throughout the entire operation of the vehicle.

[0010] The inventive solution solves this problem by creating a model using the artificial neural network. This model generates the control variable for output by the control device, taking into account a large number of influencing variables. In this way, not only can all known influencing variables be taken into account, but previously unknown influencing variables can also be appropriately considered (recognized as relevant influencing variables, so to speak) through the training of the neural network.

[0011] It is also possible to optimize the behavior during vehicle operation.

[0012] The use of an artificial neural network has the advantage that the influence of influencing factors is weighted during training. Furthermore, this influence can be analyzed by examining the weightings (of the network nodes) resulting from training in order to gain more detailed knowledge for future operation of the air conditioning unit.

[0013] A further significant advantage that results from the use of an artificial neural network is that different influencing variables are linked when determining the control variable during processing by the neural network. This linking enables the generation of route profiles that depend on various parameters, for example operating parameters of the vehicle. The route profiles can in turn be used to predict influencing variables: when the vehicle is at a certain location on the route, it is possible to predict which influencing variables can be expected along the route. This is a further advantage because the control system of the air conditioning unit can react in advance of the corresponding influences occurring. This is particularly useful for energy-optimised operation of the air conditioning unit.

[0014] The method according to the invention is preferably a computer-implemented method.

[0015] The air conditioning unit is intended, for example, to provide conditioned air for the interior of the vehicle. The air conditioning unit is preferably controlled by controlling the heating and cooling output. The air conditioning unit is preferably arranged on the roof of the vehicle, where it may be exposed to sunlight during vehicle operation.

[0016] The control device can be a central control device. Alternatively or additionally, the control device can be at least partially part of the air conditioning unit, for example, integrated into the air conditioning unit.

[0017] The vehicle is, for example, a land vehicle (e.g. an automobile), an aircraft (e.g. an airplane) or a watercraft (e.g. a ship).

[0018] The term "influencing variable" is often referred to as "disturbance variable" in the context of control engineering.

[0019] Preferably, the computer program is installed for use on the control device (English: "deployment").

[0020] Preferably, the artificial neural network has one or more layers of neurons that are not input neurons or output neurons. The layers of neurons that are not input neurons or output neurons are often referred to as hidden layers. Preferably, the hidden layers are modified during the training and learning of the artificial neural network. Machine learning involving the artificial neural network with multiple hidden layers is often referred to as deep learning.

[0021] Further preferably, the artificial neural network is trained using training data, with the training taking place in a secure state in which an unwanted data attack is excluded. The secure state is achieved, for example, by using only verified training data for training or by collecting training data during a commissioning and / or testing phase protected from attackers.

[0022] Preferably, when the influencing variables are processed by the artificial neural network to determine the control variable, a plurality of influencing variables are linked to one another. Further preferably, a route profile is generated. The route profile is further preferably generated on the basis of the link. The route profile further preferably depends on a plurality of parameters, for example operating parameters of the vehicle. The route profile is further preferably used to predict influencing variables. When the vehicle is at a certain location on the route, it is foreseeable which influencing variables are to be expected along the route (and at which position along the route). Further preferably, the control of the air conditioning unit reacts to the corresponding influencing variables on the basis of the prediction before they occur. This reaction preferably takes place during energy-optimised operation of the vehicle.

[0023] According to a preferred embodiment of the method according to the invention, the vehicle is a track-bound vehicle, preferably a rail vehicle. The interior of the track-bound vehicle comprises a passenger area for the passengers.

[0024] For example, the rail-bound vehicle is a high-speed train for long-distance public transport, a regional train, or a light rail, tram, or subway for local public transport. The rail vehicle is, for example, a multiple unit train.

[0025] The method according to the invention is particularly suitable for use with track-bound vehicles. This is because vehicles of this type have a large number of relevant influencing factors that can be relatively well detected and predicted due to the track guidance and the associated known route of the vehicle.

[0026] According to a further preferred embodiment of the method according to the invention, the control device comprises

[0027] - a controller for generating the control variable and

[0028] - a pilot control device for influencing a control variable provided for the controller, wherein the computer program representing the model is used on the pilot control device.

[0029] In this way, the behavior of the controller's feedforward control can be determined by the computer program that represents the model. This is particularly useful because previous control systems take external influencing factors, such as outside temperature (outside the vehicle), solar radiation, etc., into account. However, the model can take additional influencing factors into account and influence the controller's manipulated variable accordingly. Furthermore, the behavior of the feedforward control during vehicle operation can be optimized by taking these influencing factors into account.

[0030] In a preferred further development, the computer program which represents the model is used on the pilot control device and the controller.

[0031] In this way, the air conditioning unit's control can be replaced by the computer program representing the model. This allows the typical effects of controllers, such as control deviation, to be avoided or at least reduced.

[0032] In a further preferred embodiment of the method according to the invention, the artificial neural network is trained using training data, wherein the training data is generated based on past operation of the vehicle. In this way, past journeys and the data obtained during these journeys are used to increase the knowledge (through training) of the artificial neural network.

[0033] Alternatively or additionally, the artificial neural network is trained using training data, with the training data being generated based on the past operation of another vehicle of the same type. Since other vehicles of the same type behave in many aspects in the same way or in a similar way to the vehicle itself, other past journeys and the data obtained during these journeys can be used to increase the knowledge (through training) of the artificial neural network. The other vehicle of the same type is, for example, another vehicle in the same vehicle fleet or a test vehicle.

[0034] Alternatively or additionally, the artificial neural network is trained using training data, whereby the training data is generated based on a simulation of a system that represents the vehicle, the air conditioning unit, and at least parts of the vehicle's surroundings. By simulating the system, an expected behavior of the system is determined, and data characterizing this behavior is obtained. This data can be used as training data for training the artificial neural network.

[0035] Alternatively or additionally, the artificial neural network is trained using training data, wherein the training data is generated on the basis of a development process during which the air conditioning unit is developed. This variant is based on the knowledge that data is already generated during the development of the air conditioning unit. For example, tests of the air conditioning unit are carried out during development, during which data is generated that can be used as a basis for training the artificial neural network. According to a further preferred embodiment of the method according to the invention, the model is set up to determine an energy-optimized control variable for output by the control device, wherein the energy-optimized control variable, when processed by the air conditioning unit, brings about energy-optimized operation of the air conditioning unit.This design is based on the realization that a conventional controller is only partially capable of controlling the air conditioning unit in an energy-optimized manner. In contrast, the artificial neural network can be trained for energy-optimized operation and output a corresponding control variable.

[0036] Furthermore, the energy-optimized control variable is preferably a control variable that simultaneously (when processed by the air conditioning unit) results in comfort-optimized operation of the air conditioning unit. In other words, both energy consumption and passenger comfort are taken into account by the artificial neural network, or the artificial neural network is trained on both aspects.

[0037] The neural network is particularly suitable for energy-optimized operation of the air conditioning unit, for example, because a reaction to influencing factors can be preparatory. For this purpose, a route profile, such as the generated route profile described above, can be used, which enables the neural network to take into account foreseeable influencing factors—if they are to be expected along the route. This can prevent, for example, energy-intensive, short-term changes between heating and cooling.

[0038] According to a further preferred embodiment of the method according to the invention, the influencing variables include, as at least one influencing variable, the number of passengers present within the interior. The number of passengers is determined based on a weight signal generated by a braking unit and / or a spring unit of the vehicle.

[0039] The weight signal is determined, for example, based on the braking force and / or braking energy that the braking unit uses for a given braking process.

[0040] If the spring unit is designed as an air spring, the weight signal can be determined, for example, based on the air pressure within the air spring.

[0041] Alternatively or additionally, the number of passengers is determined using a passenger counting system. This variant is particularly advantageous when the vehicle is designed as a rail-bound vehicle, since passenger counting systems are often already present on rail-bound vehicles, and the passenger counting data from these systems is already available.

[0042] Alternatively or additionally, the number of passengers is determined based on the number of mobile devices detected inside the vehicle. This variant is based on the realization that the number of mobile devices is a suitable measure for estimating the number of passengers inside the vehicle.

[0043] Preferably, the weight signal and the number of detected mobile devices can be linked by the neural network or parts of the neural network.

[0044] To determine the number of passengers inside the vehicle, the time of day can also be added in conjunction with the day of the week. This allows the neural network to detect high passenger volumes, caused, for example, by rush hour traffic. According to a further preferred embodiment of the method according to the invention, the influencing variables include, as at least one influencing variable, solar radiation to which the vehicle is exposed during operation. The solar radiation is detected using a location signal and a resulting radiation direction.

[0045] Preferably, the position of the vehicle is determined based on the location signal. If the vehicle is designed as a track-bound vehicle, the direction of travel of the vehicle is preferably determined based on the position and the route traveled. The direction of solar radiation is preferably determined based on the current time (and the associated position of the sun), taking the direction of travel into account.

[0046] Alternatively or additionally, solar radiation is measured using a sun shadow cast onto the vehicle by an object in the vehicle's vicinity. In other words, a reduction in solar radiation is determined based on the sun shadow. Knowing the solar radiation (which would affect the vehicle without a shadow) and the sun shadow, the solar radiation hitting the vehicle can be determined.

[0047] The object is, for example, a building, a planting and / or a tunnel along the route traveled by the vehicle.

[0048] Alternatively or additionally, solar radiation is determined based on a specific point in time, preferably a time of year and day. This variant is based on the knowledge that the position of the sun and, consequently, the angle of incidence of solar radiation depends on the time of day and the time of year. Alternatively or additionally, solar radiation is determined based on weather data. For example, cloud cover can be taken into account when determining solar radiation.

[0049] When the vehicle is designed as a track-bound vehicle, which is usually elongated, the so-called sunny side of the vehicle, which is directly exposed to solar radiation, can be determined particularly easily based on the direction of incidence.

[0050] According to a further preferred embodiment of the method according to the invention, the influencing variables include, as at least one influencing variable, an operating state of an electrical component of the vehicle, which is detected. This variant is based on the knowledge that electrical components radiate heat during operation. This can be taken into account particularly easily by the artificial neural network based on the operating state of the components. The operating state can, for example, include "ON" and "OFF". An example of an electrical component is a lighting device for illuminating the interior.

[0051] In a further preferred embodiment of the method according to the invention, the influencing variables include wind as at least one influencing variable. The wind is detected based on a vehicle's driving speed and / or on weather data. This embodiment is based on the finding that wind influences the heat supplied to the vehicle. It is advantageous that a wind acting on the vehicle can be detected particularly easily based on the vehicle's driving speed and on weather data.

[0052] In a further preferred embodiment of the method according to the invention, the air conditioning unit is controlled based on the control variable output by the control device. The invention further relates to a computer program comprising instructions which, when the program is executed by a computing device, cause the computing device to carry out the method of the type described above. The invention further relates to a computer program product with a computer program of this type. The computing device is preferably at least partially a computing device of the rail-bound vehicle and / or the land-based device.

[0053] The invention further relates to a provision device for the computer program of the type described above, wherein the provision device stores and / or provides the computer program. The provision device is, for example, a storage unit that stores and / or provides the computer program. Alternatively and / or additionally, the provision device is, for example, a network service, a computer system, a server system, in particular a distributed, for example cloud-based computer system and / or virtual computer system, which stores and / or provides the computer program product preferably in the form of a data stream.

[0054] The provision takes place in the form of a program data block as a file, in particular as a download file, or as a data stream, in particular as a download data stream, of the computer program. This provision can also take place, for example, as a partial download consisting of several parts. Such a computer program is, for example, read into a system using the provision device, so that the method according to the invention is executed on a computer.

[0055] The above-mentioned object is further achieved by a device of the type mentioned at the outset. The device comprises a computing device which is configured to form a model with the aid of an artificial neural network, wherein the model is configured to determine a control variable for output by the control device based on influencing variables that influence the climate of the interior. The device further comprises a computer program which represents the model and is configured to be used on the control device for controlling the air conditioning unit.

[0056] The invention further relates to a vehicle, preferably a track-bound vehicle, with a device of the type described above.

[0057] For advantages, embodiments and design details of the computer program according to the invention, the provision device according to the invention, the device according to the invention and the vehicle according to the invention, reference can be made to the above description of the corresponding method features of the method according to the invention.

[0058] Examples of the invention are explained with reference to the drawings. They show:

[0059] Figure 1 shows schematically the structure of an example of a control system for an air conditioning unit,

[0060] Figure 2 shows schematically the structure of an example of a

[0061] Vehicle with an air conditioning unit,

[0062] Figure 3 shows schematically the structure of a

[0063] From an exemplary embodiment of a device according to the invention,

[0064] Figure 4 shows schematically the structure of a

[0065] From an exemplary embodiment of a vehicle according to the invention and

[0066] Figure 5 schematically shows the process of a

[0067] From an exemplary embodiment of a method according to the invention. Figure 1 shows schematically the structure of an example of a control of an air conditioning unit 1. This control comprises a proportional-integral controller 3 of an outer control circuit 5 and a proportional-integral controller 7 of an inner control circuit 9. The outer control circuit 5 contains a temperature sensor 11, which measures a current room temperature Ti n within an interior 15 of a vehicle 20 shown in Figure 2. The inner control circuit 9 contains a temperature sensor 13, which measures a supply air temperature that is provided and supplied to the interior 15 of the vehicle by means of the air conditioning unit 1. The air conditioning unit 1 is controlled by regulating the heating and cooling power that the air conditioning unit 1 delivers to the supply air.

[0068] Any influencing factors 2 that affect the climate of the interior 15 are not taken into account in this control. Instead, the control reacts to any deviations dT of the measured temperature Ti n and a desired setpoint interior temperature Ti c (so-called setpoint interior temperature).

[0069] Figure 2 shows schematically the structure of an example of a vehicle 20. Outside the vehicle 20 there is an outside temperature T e The air conditioning unit 1 is arranged on the roof of the vehicle 20 and has the purpose of air conditioning the interior 15 of the vehicle 20. For this purpose, the air conditioning unit supplies the interior 15 with supply air 17 having a supply air temperature T zu to . The supply air 17 causes an interior temperature Ti n within the interior 15 . The aim of the air conditioning is to achieve the setpoint interior temperature Ti c within the interior space 15 .

[0070] Figure 3 shows a schematic diagram of the structure of an exemplary embodiment of the device according to the invention. Elements that are the same or have the same function are given the same reference numerals as in Figure 1. The device comprises a proportional-integral controller 3 of an outer control loop 5 and a proportional-integral controller 7 of an inner control loop 9. A control device 30 is provided in the direction of action between the outer controller 3 and the inner controller 7 and serves as a pilot control device 31 for pilot controlling the controller 7. A computer program is used on the control device 30 to control the air conditioning unit 1. The computer program represents a model that is formed by an artificial neural network 32.

[0071] Figure 4 schematically shows the structure of an exemplary embodiment of a device according to the invention and of a vehicle 120 according to the invention. Identical and functionally identical elements of the vehicle 120 are provided with the same reference numerals as in relation to the corresponding elements of the vehicle 20 according to Figure 2.

[0072] The vehicle 120 is a track-bound vehicle 121, for example, a rail vehicle 122. The interior 15 of the track-bound vehicle 121 includes a passenger area 16 for the accommodation of passengers 34.

[0073] In a method step A, the artificial neural network 32 is generated or formed on a land-based device 105 by means of a computing device 110. Alternatively, in a method step AA, the artificial neural network 32 is formed by means of a computing device 10 as part of the control device 30 of the vehicle 120.

[0074] The artificial neural network 32 has several layers of neurons that are neither input neurons nor output neurons. The artificial neural network 32 forms a model that is intended to be capable of determining a control variable for output by the control device 30 based on influencing variables 2 that influence the climate of the interior 15.

[0075] One of the influencing variables 2 is, for example, the outside temperature T e. A further influencing variable 2 is, for example, the number Nf of passengers 34 who are inside the interior 15 during operation of the air conditioning unit 1. Furthermore, a further influencing variable 2 is, for example, the solar radiation 36 to which the vehicle 120 is exposed during operation. A further influencing variable 2 is, for example, an operating state 39 of an electrical component 38 of the vehicle 120. A further influencing variable 2 is, for example, wind 40, which is detected based on a driving speed of the vehicle and / or based on weather data.

[0076] In a method step CI, the artificial neural network 32 is trained using training data. Training takes place, for example, in a secure state in which an unwanted data attack is impossible. This secure state is achieved, for example, by using only verified training data for training.

[0077] The training data are generated, for example, on the basis of a past operation of the vehicle 120 in a method step B. For this purpose, for example, during past journeys of the vehicle 120, influencing variables 2, associated output control variables of a pilot control device and the interior temperature Ti achieved thereby nmeasured. These variables can also be measured during past journeys of another vehicle of the same type in a method step BB. In addition, a simulation of a system which represents the vehicle 120, the air conditioning unit 1 and parts of the environment 46 of the vehicle 120 can be a basis for generating training data in a method step BBB. Furthermore, sub-processes of the development process in which the air conditioning unit 1 is developed, for example a test phase, can be the basis for generating training data in a method step BBBB. After the artificial neural network 32 has been trained, a model is formed with the aid of the artificial neural network 32 (method step C2), which model is set up to determine a control variable for output by the control device 30 on the basis of the influencing variables 2.

[0078] The computer program representing the model is installed on the control device 30 in a method step D and used in the further method, in particular in the operation of the vehicle 120, according to a method step E.

[0079] The influencing variable Nf (number of passengers) is recorded during operation of the vehicle 120 in a method step E1, for example based on a weight signal which is generated by a brake unit or a spring unit of the vehicle 120, by means of a passenger counting system of the vehicle 120 and / or based on the number of mobile terminals recorded within the interior 15.

[0080] In addition, the influencing variable 36 (solar radiation) is determined in a method step E2 on the basis of a location signal and a resulting direction of radiation, on the basis of a sun shadow cast onto the vehicle 120 by an object located in the vicinity of the vehicle, on the basis of a point in time, preferably a time of year and day, and / or on the basis of weather data 44.

[0081] Furthermore, the influencing variable 39 (operating state) is detected in a method step E3. For this purpose, it is determined, for example, whether the electrical component 38 of the vehicle 120 is switched on or not.

[0082] In addition, the influencing variable 40 (wind) is recorded in a method step E4 based on a driving speed of the vehicle 120 and / or based on weather data 44. In addition, the outside temperature T e in a method step E5 by means of a temperature sensor and / or based on weather data 44.

[0083] Based on the aforementioned and, if applicable, on the basis of further recorded influencing variables, the computer program used on the control device 30 determines, in a method step E6, the control variable output by the control device 30. In a method step E7, the air conditioning unit 1 is controlled based on the control variable.

[0084] Although the invention has been illustrated and described in detail by the preferred embodiment, the invention is not limited to the disclosed examples and other variations can be derived therefrom by those skilled in the art without departing from the scope of the invention.

Claims

Patent claims 1. Method for controlling an air conditioning device (1) which air-conditions an interior (15) of a vehicle (120), in which - the air conditioning unit (1) is controlled by means of a control device (30) in order to air-condition the interior (15) of the vehicle (120) using the control, characterized in that - a computing device (10, 110) forms a model (C2) with the aid of an artificial neural network (32) which is designed to determine (E6) a control variable for output by the control device (30) on the basis of influencing variables (2) which have an influence on the climate of the interior (15), and - a computer program representing the model is used on the control device (30) to control the air conditioning unit (1) (E).

2. Method according to claim 1, characterized in that - the vehicle (120) is a track-bound vehicle (121), preferably a rail vehicle (122), and - the interior (15) of the track-bound vehicle (121) comprises a passenger area (16) for the stay of passengers (34).

3. Method according to claim 1 or 2, characterized in that the control device (30) - a controller (3, 7) for generating the control variable and - a pilot control device (31) for influencing a manipulated variable provided for the controller (3, 7), and the computer program representing the model is used on the pilot control device (31).

4. Method according to claim 3, characterized in that the computer program representing the model is used on the pilot control device (31) and the controller (3, 7).

5. Method according to at least one of the preceding claims, characterized in that the artificial neural network (32) is trained using training data (CI), wherein the training data - based on past operation of the vehicle (120) , - based on past operation of another vehicle of the same type, - based on a simulation of a system which represents the vehicle (120), the air conditioning unit (1) and at least parts of the environment (46) of the vehicle (120), and / or - are produced on the basis of a development process in which the air conditioning unit (1) is developed (B, BB, BBB, BBBB).

6. Method according to at least one of the preceding claims, characterized in that the model is set up to determine an energy-optimized control variable for output by the control device (30), wherein the energy-optimized control variable, when processed by the air conditioning unit (1), effects energy-optimized operation of the air conditioning unit (1).

7. Method according to at least one of the preceding claims, characterized in that the influencing variables (2) comprise, as at least one influencing variable, a number (Nf) of passengers (34) who are inside the interior (15), and the number (Nf) of passengers (34) - based on a weight signal generated by a brake unit or a spring unit of the vehicle (120), - by means of a passenger counting system and / or - is determined based on the number of mobile devices detected within the interior (15) (El) .

8. Method according to at least one of the preceding claims, characterized in that the influencing variables (2) comprise, as at least one influencing variable, solar radiation (36) to which the vehicle (120) is exposed during operation, and the solar radiation (36) - based on a location signal and a resulting beam direction, - based on a sun shadow cast onto the vehicle (120) by an object in the vicinity of the vehicle (120), - based on a point in time, preferably a time of year and day, and / or - is determined on the basis of weather data (44) (E2) .

9. Method according to at least one of the preceding claims, characterized in that the influencing variables (2) comprise, as at least one influencing variable, an operating state (39) of an electrical component (38) of the vehicle (120), which is detected (E3).

10. Method according to at least one of the preceding claims, characterized in that the influencing variables (2) comprise wind (40) as at least one influencing variable and the wind (40) - based on a driving speed of the vehicle (120) and / or - is recorded on the basis of weather data (44) (E4) .

11. Method according to at least one of the preceding claims, characterized in that the air conditioning unit (1) is controlled (E7) on the basis of the control variable output by the control device (30).

12. A computer program comprising instructions which, when the program is executed by a computing device (38), cause the computing device (38) to carry out the method according to at least one of claims 1 to 11.

13. Provision device for the computer program according to claim 12, wherein the provision device stores and / or provides the computer program.

14. Device for controlling an air conditioning unit (1) which air-conditions an interior (15) of a vehicle (120), comprising: - a control device (30) which is designed to control the air conditioning unit (1) in order to air-condition the interior (15) of the vehicle (120), characterized by - a computing device (10, 110) which is designed to form a model with the aid of an artificial neural network, wherein the model is designed to determine a control variable for output by the control device (30) on the basis of influencing variables (2) which have an influence on the climate of the interior (15), and - a computer program which represents the model and is designed to be used on the control device (30) for controlling the air conditioning device (1).

15. Vehicle, in particular a track-bound vehicle (121), with a device according to claim 14.

Images