Determining the cooling behavior of a multi-plate clutch using a neural network

Abstract

The present invention relates to a method (50) for training a neural network (22) by means of which a cooling coefficient of a multi-plate clutch of a transmission (20) can be determined. The neural network (22) is trained with a training data set (50) to learn a relationship between an operating parameter and a cooling coefficient. The invention further relates to a method for determining a heat dissipation from a multi-plate clutch of a transmission (20) using the neural network (22) trained in this way. The invention also relates to a method for determining a clutch temperature of a multi-plate clutch of a transmission (20), in which a heat input is determined analytically and the heat dissipation is modeled by means of the neural network (22).

Description

Technical field

[0001] The present invention relates to a method for training a neural network, whereby the cooling coefficient of a multi-plate clutch of a transmission can be determined using the trained neural network. Furthermore, the invention relates to a method for determining the heat dissipation from a multi-plate clutch of a transmission and a method for determining the clutch temperature of a multi-plate clutch of a transmission. State of the art

[0002] CN 103 967 963 A describes a method for predicting the temperature of a wet clutch using a neural network. This method directly determines the temperature using the neural network. The possibility of temperature prediction using analytical methods remains unused. As a result, the neural network is complex to train and requires significant computing power for its application. Description of the invention

[0003] A first aspect concerns methods for training a neural network that can determine the cooling coefficient of a multi-plate clutch in a transmission. The transmission could be, for example, a vehicle transmission or a transmission in a stationary system, such as an industrial plant. A neural network, also known as an artificial neural network, can be a universal function approximator. The neural network has, for example, nodes and connections. It can be single-layered or multi-layered. The neural network can be a feedforward neural network or a recurrent neural network. The connections between nodes can have weights. This allows output data to be calculated from input data.The weights and connections can be modified by training the neural network. For this purpose, the neural network can be provided with synthetic or experimental training datasets. Synthetic training datasets can be generated, for example, in simulations. The training can be performed iteratively.

[0004] A neural network can be understood as a mathematical model that at least partially replicates the structure of neurons in the human brain. A neural network can be created using a computer. It can have input nodes, output nodes, and multiple intermediate nodes located between the input and output nodes. The input nodes can be, for example, data interfaces through which input data can be fed into the neural network. The output nodes can be, for example, data interfaces through which output data can be sent from the neural network. The input nodes can be connected to the intermediate nodes, and the intermediate nodes can be connected to each other. The intermediate nodes can also be connected to the output nodes.

[0005] Intermediate nodes can be used to temporarily store information. At least one arithmetic operation can be performed on each intermediate node. Input data can be transmitted from the input nodes to the output nodes via the intermediate nodes. During this transmission, the input data can be mathematically processed, for example, converted into output data. The intermediate nodes of the neural network can be arranged in one or more layers. Intermediate nodes within a layer can be interconnected. Additionally, intermediate nodes within a layer can be connected to intermediate nodes in other layers. The individual connections between input nodes, intermediate nodes, and output nodes can be assigned mathematical weights.Depending on the purpose of the neural network, the individual connection weights can vary. These weights can be modified during the neural network's training. By adjusting the mathematical weights of the connections between individual nodes during training, the neural network can learn a relationship between the input and output data. When the neural network is used for its intended purpose, this learned relationship can be applied to the input data to generate output data according to the network's defined function.

[0006] A transmission can be a component of a motor vehicle. A motor vehicle can be, for example, a passenger car, truck, construction machine, forestry machine, or agricultural machine. The transmission can connect the vehicle's drive unit, such as a vehicle engine, to an output unit, such as a vehicle axle, for torque transmission. The drive unit can be the vehicle engine. The output unit can be the vehicle axle. By actuating various switching elements of the transmission, such as multi-plate clutches, the power flow between the drive unit and the output unit can be established or disconnected. Alternatively or additionally, the gear ratio of the torque transmission can be changed. For example, the transmission can be designed to provide different gear ratios. During gear changes in operation, the switching elements can heat up.The temperature of switching elements results from heat input and heat dissipation to the surrounding environment. Heat input can be caused, for example, by the switching process itself. The cooler environment can be, for example, air or oil within the switching element.

[0007] A multi-plate clutch is a coupling used to connect two components in a rotationally fixed manner. A multi-plate clutch can be designed, for example, as a wet clutch. It can be adjustable between an open and a closed state, or it can be operated in a slipping state. Typically, significant heating occurs during slipping operation. In the open state, torque transmission in the multi-plate clutch is interrupted. In the closed state, torque transmission is possible. In the slipping state, torque transmission is possible, with the individual plates still rotating relative to each other. A multi-plate clutch usually has two sets of plates, each containing one or more plates. When the multi-plate clutch is engaged, the sets of plates are pressed together, creating a frictional connection.The plates of a clutch pack are compressed, for example, by a hydraulic cylinder or an electric actuator. At a differential speed, heat can be generated, which causes heat input into the clutch. The plates of one clutch pack are usually made of metal, for example, steel. These metal plates have high thermal conductivity. Through the flow of oil in the clutch, these metal plates can dissipate a significant amount of heat. The plates of the other clutch pack have a friction lining, for example, made of cellulose. These lining plates conduct heat poorly and therefore contribute little to cooling the clutch. Generally, heat dissipation from a clutch pack can occur, for example, through heat radiation and convection to the air, and alternatively or additionally through the oil flowing around the clutch.A significant factor is the temperature difference between the clutch and the oil. Depending on the condition of the transmission, the vehicle itself, and other parameters, this difference can be large and difficult to determine analytically. The cooling coefficient can serve as a measure of heat dissipation from the multi-plate clutch.

[0008] The method includes a step of recording the clutch temperature under various operating conditions. This recording can be performed, for example, using a temperature sensor. The temperature sensor might be installed solely for training the neural network. It could also be part of a test bench. The clutch temperature could be, for example, the temperature of the metallic plates of the multi-plate clutch. It could also be the temperature at one or more locations where the multi-plate clutch is particularly susceptible to overheating. The clutch temperature could be, for example, the maximum temperature or the average temperature of spaced-ahead temperature measurements. The clutch temperature is recorded, for example, as a trend. This recording can be performed at discrete time intervals or continuously.It may be provided that the operating state is not changed for a certain period of time, which then allows a profile of the coupling temperature to be recorded when switching from a previous operating state to a current operating state.

[0009] An operating state can correspond to various usage states of the multi-plate clutch or the entire transmission. The usage state of the multi-plate clutch can be, for example, open, closed, or slipping. The usage state can include further parameters such as applied torque, applied load, and, alternatively or additionally, rotational speed. For example, the clutch temperature and / or other parameters described below can be recorded at various operating points of the transmission or just the multi-plate clutch. The operating states can, for example, correspond to a range of expected or permissible usage of the multi-plate clutch. For example, the parameters to be recorded can be determined in all gears or shift states at the planned driving speeds and, alternatively or additionally, under loads in the vehicle's drivetrain.The determination of the quantities to be measured can, for example, be done experimentally.

[0010] The procedure includes a step of measuring the oil temperature under various operating conditions. The oil temperature could be the temperature of the oil in a transmission sump. It could also be the temperature of the oil before it enters the multi-plate clutch. The oil temperature is measured, for example, as a trend. Measurements can be taken at discrete time intervals or continuously. The operating condition can be kept constant for a specific period, allowing the measurement of the oil temperature trend during a transition from a previous to the current operating condition. The oil temperature can be measured by a sensor on the test bench.

[0011] The oil temperature can be measured and provided by existing sensors or a measuring system in the vehicle or transmission. This data can then be made available in the vehicle's bus system. Unlike a multi-plate clutch, for example, an oil sump can be easily accessed for this temperature measurement.

[0012] The process includes a step of acquiring at least one operating parameter that characterizes the various operating states in each of them. For example, the rotational speed applied to an input of the multi-plate clutch can be measured or calculated as an operating parameter. An applied load and the currently engaged gear can also be such an operating parameter. Other operating parameters include, for example, the switching state of the multi-plate clutch or the pump speed of an oil pump that supplies the multi-plate clutch with oil. The respective operating parameters can be acquired and provided, for example, by existing sensors or a measurement system in the vehicle or transmission. These can then be made available in a vehicle bus system. The operating parameter is recorded, for example, as a trend line.Data collection can be performed at discrete time intervals or continuously.

[0013] A multitude of operating parameters can be recorded and used in other steps of the process. For example, the operating state can be characterized by the rotational speed applied to the input of the multi-plate clutch, the current gear, and the clutch's engagement state. An operating state vector can be recorded. The following explanations for a single operating parameter also apply to multiple operating parameters, where applicable. When referring to the "operating parameter," this can always refer to multiple operating parameters. For example, the training dataset described below can contain multiple operating parameters, and the neural network can be provided with a corresponding multitude of operating parameters as input data. For simplicity, however, only the "operating parameter" will be used.

[0014] The procedure includes a step of determining the cooling coefficient as a function of the measured clutch temperature and the measured oil temperature under the various operating conditions. This determination can be performed, for example, without using the neural network that is to be trained. The cooling coefficient can be derived analytically from the measured values. In this way, a corresponding cooling coefficient can be determined for all experimentally investigated operating conditions without requiring averaging or estimation. The cooling coefficient can be determined for each measurement point in time. The cooling coefficient can be a ratio of the cooling rate of the multi-plate clutch to the temperature difference between the clutch temperature and the oil temperature. The cooling rate of the multi-plate clutch can be a gradient of the clutch temperature at the measurement point.The cooling rate of the multi-plate clutch can be determined, for example, based on two or more consecutive clutch temperatures.

[0015] The method includes a step of creating a training dataset for the neural network. This dataset uses the cooling coefficients determined for different operating states as output data and the operating parameters determined for different operating states as input data. The training dataset can be used to train the neural network. The method can also include a step of training the neural network with this training dataset to establish a relationship between the operating parameters and the cooling coefficients. For this purpose, the training dataset is iteratively provided to the neural network until it can determine the cooling coefficient with sufficient accuracy as a function of the measured operating parameters. Sufficient accuracy can be defined, for example, by a threshold value.Accuracy can be determined, for example, as mean squared deviation. If this deviation is smaller than the threshold, the accuracy in determining the cooling coefficient as a function of the recorded operating parameter by the neural network may be sufficient. The training can then be terminated. The neural network thus learns to determine the current cooling coefficient in the future when using the multi-plate clutch, depending on the operating state or the operating parameter corresponding to that state. This allows the heat dissipation to be modeled by a neural network. This heat dissipation can then be combined with the analytically determined heat input into the multi-plate clutch. In this way, the temperature of the multi-plate clutch can be determined. Alternatively or additionally, a prediction of the future temperature of the multi-plate clutch can also be made.Based on the cooling coefficient, the heat input, and alternatively or additionally the heat output, a future coupling temperature can be predicted. This determination is precise and requires little computational effort.

[0016] In another embodiment of the method, the neural network can be configured as a forward-facing neural network. In forward-facing neural networks, each layer is always connected only to the next higher layer. The neural network is then not configured as a recurrent network. This reduces the neural network's computing power requirements. Furthermore, a buffer can be omitted or at least made very small. This allows the neural network to be implemented on very simple, low-power computing devices, such as a transmission control unit.

[0017] In a further embodiment of the method, the input data in the various operating states can contain the specific operating parameters for a current operating state as well as for a previous operating state. This allows a history of the operating data to be considered by a forward-facing neural network. The operating parameters of previous operating states can be prior operating parameters. For example, the previous operating state can have a predetermined time interval from the current operating state. The previous operating state can also be the last operating state before the operating state was changed. For example, the operating state can be changed by altering the actuation state of the multi-plate clutch and, alternatively or additionally, by shifting gears.The input data can also contain the operating parameters of several previous operating states, which may, for example, date back to different periods. Corresponding operating parameters can be recorded for each operating state.

[0018] In another embodiment of the method, the operating parameter acquired for the previous operating state can be provided to the neural network as input data via a first-order delay transfer element. For example, a first-order low-pass filter can be used as the first-order delay transfer element. This is also known as a first-order PT1 element. The first-order delay transfer element eliminates the need for intermediate storage of operating parameters from previous operating states to provide them as input data. Instead, the operating parameters of previous operating states can be directly connected to the operating parameters of the current operating state as input data via a suitable electrical connection.

[0019] In a further embodiment of the method, at least one operating parameter can be selected from at least one of the following parameters. Several of the operating parameters described below can also be selected. The operating parameter can be the torque of a motor shaft of a vehicle engine, which is mechanically connected to the vehicle clutch. The motor shaft can be a shaft at which the vehicle engine provides torque for propulsion. The torque can, for example, be measured or estimated from an operating state of the vehicle. The operating parameter can be a speed difference between two clutch elements of the multi-plate clutch rotating relative to each other. For example, the clutch plates can rotate relative to each other when disengaged. This differential speed can be a measure of the load when the clutch engages.The operating parameter can be an actuation parameter of the multi-plate clutch, such as a setpoint for pressure or current flow for its actuation. The operating parameter can be the rotational speed of a rotating clutch element of the multi-plate clutch, such as an input or output shaft. The operating parameter can be the transmission sump temperature at the beginning of a multi-plate clutch engagement. The sump temperature can be the oil temperature, for example, before the multi-plate clutch engages. The oil temperature, and alternatively or additionally, a previous clutch temperature, can be an operating parameter. The operating parameter can be an actuation state of the multi-plate clutch, for example, whether the clutch is open, closed, or slipping. The operating parameter can be an oil flow rate, for example, the current oil flow through the clutch. The operating parameter can be the oil pump speed.The oil pump speed can be a measure of the oil flow rate and is easy to measure. The operating parameter(s) can be measured by sensors and made available, either alternatively or additionally, in a vehicle bus system.

[0020] In another embodiment of the method, the training data can be generated on a test bench. This allows for the predefined, controlled state. Additional sensors can also be used. The test bench can be a vehicle test bench, a transmission test bench, or even just a clutch test bench. During the generation of the training data on the test bench, a defined quantity of oil can flow through the multi-plate clutch. For example, the multi-plate clutch or the transmission in which the multi-plate clutch is installed can be sealed. Generally, a state can be created in which leakage is minimal or completely eliminated. Oil flow to other clutches, consumers, or the like can also be prevented. This allows the oil flow through the multi-plate clutch to be known very precisely in every operating condition, unlike, for example, in real-world use.This allows the neural network to be trained with particularly precise data.

[0021] A second aspect concerns a method for determining the heat dissipation from a multi-plate clutch of a transmission. This method includes a step of acquiring at least one operating parameter characterizing an operating state, for example, from the transmission or just the multi-plate clutch. The method according to the second aspect includes a step of determining a cooling coefficient of the multi-plate clutch as a function of the acquired operating parameter using a neural network. The neural network was trained using a method according to the first aspect. Therefore, the method according to the second aspect can utilize the neural network trained using the method according to the first aspect. The respective advantages and further features can be found in the description of the first aspect, whereby embodiments of the first aspect also constitute embodiments of the second aspect and vice versa.

[0022] For determining the cooling coefficient, the recorded operating parameter is provided to the neural network as input data. In the method described in the second aspect, the cooling coefficient is thus determined using the neural network. The neural network can be provided with the determined operating parameters from a current operating state as input data. Alternatively or additionally, the neural network can be provided with the determined operating parameters from one or more previous operating states as input data. The selection of input data can correspond to the training data used during training. The method described in the second aspect includes a step of determining the heat dissipation from the multi-plate clutch as a function of the determined cooling coefficient. The cooling coefficient can be a current value.The heat dissipation can be the current heat dissipation. Alternatively, a heat dissipation curve can be determined. Determining the heat dissipation can be based on a current or previous clutch temperature and, alternatively or additionally, on a currently measured or calculated oil temperature.

[0023] In a further embodiment of the method for determining heat dissipation, the determination of heat dissipation from the multi-plate clutch can additionally consider the mass of the metallic plates of the clutch as a characteristic parameter. For example, the mass of only the plates without friction linings or the mass of only the metallic components of all plates can be considered. Alternatively or additionally, the specific heat capacity of the metallic plates of the multi-plate clutch can be considered as a characteristic parameter. Alternatively or additionally, the surface area of ​​the plates of the multi-plate clutch can be considered as a characteristic parameter. For each of the aforementioned characteristic parameters, friction lining plates can be included or excluded. Depending on the characteristic parameters, a heat transfer coefficient for the heat dissipation can be determined.The defined parameters allow the neural network to be adapted to different multi-plate clutches. For example, in a series of multi-plate clutches, the number or size of the plates can be adjusted to meet different performance requirements. The neural network can thus be trained generically. The results can then be applied to different multi-plate clutches. This also allows the heat dissipation of clutch derivatives to be determined without having to regenerate training data or re-adapt the neural network.

[0024] A third aspect concerns a method for determining the clutch temperature of a multi-plate clutch in a transmission. This method includes a step for analytically determining the heat input into the multi-plate clutch. The heat input can, for example, be simulated, predicted, or calculated from operating parameters. For this purpose, a physical temperature model can be implemented on a transmission control unit or another vehicle control unit. The method also includes a step for determining the heat dissipation from the multi-plate clutch using the method described in the second aspect. The neural network trained using the method described in the first aspect can be used for this step.The respective advantages and further features can be found in the descriptions of the first and second aspects, whereby embodiments of the first and second aspects also constitute embodiments of the third aspect, and vice versa. The method for determining the coupling temperature also includes a step of determining the coupling temperature as a function of the specified heat input and heat output. For example, the heat output can be subtracted from the heat input, and the difference can be used to infer a temperature change. This allows for the prediction of a current or future coupling temperature. Brief description of the drawings Fig. Figure 1 illustrates a lamellar coupling in a schematic sectional view. Fig. Figure 2 schematically illustrates a determination of the clutch temperature of a multi-plate clutch in a transmission designed as a vehicle transmission. Fig. Figure 3 schematically illustrates a method for determining the coupling temperature. Detailed description of embodiments

[0025] Fig. Figure 1 illustrates a multi-plate clutch in a schematic sectional view. The multi-plate clutch has a radially inner plate carrier 10 and a radially outer plate carrier 10 as coupling elements that rotate relative to each other. A pack of friction lining plates 12 is attached to the radially inner plate carrier 10. The friction lining plates 12 have a paper lining. A pack of metallic plates 14 is attached to the radially outer plate carrier 10. When the multi-plate clutch is actuated by applying pressure, the two plate packs 12, 14 are pressed together, thus creating a frictional connection for torque transmission. If the two coupling elements rotate relative to each other beforehand, heat is generated. Simultaneously, heat is also dissipated within the multi-plate clutch. Heat is primarily dissipated by transferring heat to an oil flowing through the multi-plate clutch.Furthermore, heat is dissipated to the environment through thermal radiation and convection. If the clutch plates 12, 14 are not compressed, a spring element 16 pushes them apart. This returns the clutch from a closed to an open state, thus interrupting torque transmission.

[0026] Fig. Figure 2 schematically illustrates a determination of the clutch temperature of the multi-plate clutch of a transmission 20, which in the present embodiment is considered a vehicle transmission. The transmission 20 has the multi-plate clutch of Fig. 1. A method for determining the coupling temperature is shown schematically in Fig. 3 illustrated.

[0027] To determine the clutch temperature, a neural network 22 is first trained in step 50. The neural network 22 is designed as a forward-directed neural network. Step 50 includes a substep 52 for acquiring a clutch temperature under various operating conditions. Step 50 also includes a substep 54 for acquiring an oil temperature under the various operating conditions. Furthermore, step 50 includes a substep 56 for acquiring at least one operating parameter characterizing the different operating conditions under the various operating conditions. In the example shown, substeps 52 to 56 are performed on a transmission test bench. In substep 58, a generic cooling coefficient for multi-plate clutches is determined as a function of the acquired clutch temperature and the acquired oil temperature under the various operating conditions.The generic cooling coefficient is determined analytically using recorded measurements. In substep 60, a training dataset is created for the neural network 22. The training dataset contains the cooling coefficient determined for the various operating states as output data. The training dataset contains the operating parameters determined for the various operating states as input data. In substep 62, this training dataset is provided to the neural network 22 to train a relationship between the operating parameters and the cooling coefficient. The input data in the various operating states contain the determined operating parameters for a current and a previous operating state. For this purpose, recorded operating parameters are provided to the neural network 22 directly and via a first-order delay transfer element 24 as input data.Providing the input and output data for training 50 of the neural network 22 is also included in . Fig. 2 illustrated.

[0028] To determine the clutch temperature, the trained neural network 22 is used in step 64 to determine the heat dissipation from a multi-plate clutch of the transmission 20. In substep 56, at least one operating parameter characterizing an operating state is recorded during the operation of the specific multi-plate clutch. In substep 66, the cooling coefficient of the multi-plate clutch is determined as a function of the recorded operating parameter using the previously trained neural network 22. For this purpose, the trained neural network 22 is provided with the determined operating parameters for a current operating state directly, as well as for a previous operating state via the transmission element 24 with a first-order delay. The cooling coefficient for generic multi-plate clutches is then generated as output data by the neural network 22.The cooling coefficient here is the ratio of the cooling rate of the multi-plate clutch to the temperature difference between the clutch temperature and the oil temperature.

[0029] In substep 68, the heat dissipation is determined by converting the cooling coefficient calculated in substep 66. This is done by multiplying the cooling coefficient by the mass of the metallic plates 14 of the multi-plate clutch and by the specific heat capacity of the metallic plates 14 of the multi-plate clutch. This value is then divided by the total surface area of ​​the metallic plates 14 of the multi-plate clutch to calculate the heat dissipation.

[0030] In a further step 70, a physical temperature model is used to analytically determine the heat input into the multi-plate clutch. In a subsequent step 72, the clutch temperature is then determined as a function of the determined heat input and heat dissipation. All calculations, including the neural network 22, are performed on a transmission control unit of the transmission 20. Reference sign 10 lamella carriers 12 friction lining lamellae 14 metallic slats 16 spring element 20 gearboxes 22 neural network 24 First-order delay transmission element Step 50: Training the Neural Network 52 Substep: Detecting a clutch temperature 54. Step below: Recording an oil temperature 56 Substep: Recording an operating parameter 58. Step below: Determining a cooling coefficient Step 60: Creating a training dataset Step 62: Providing the training dataset Step 64: Determining a heat drain 66. Step below: Determining the cooling coefficient 68. Step below: Converting the cooling coefficient Step 70: Analytical determination of heat input Step 72: Determining the clutch temperature QUOTES INCLUDED IN THE DESCRIPTION

[0000] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited patent literature

[0000] CN 103 967 963 A

[0002]

Claims

[1] Method for training (50) a neural network (22) by means of which a cooling coefficient of a multi-plate clutch of a transmission (20) can be determined, wherein the method comprises at least the following steps: - Recording (52) a coupling temperature in different operating states; - Recording (54) an oil temperature in the different operating conditions; - Recording (56) at least one operating parameter characterizing the different operating states in the different operating states; - Determining (58) the cooling coefficient as a function of the recorded clutch temperature and the recorded oil temperature in the different operating conditions; - Creating (60) a training data set for the neural network (22) with the cooling coefficients determined for the different operating states as output data and the operating parameter determined for the different operating states as input data. [2] Method according to claim 1, characterized by , that the neural network (22) is configured as a forward-directed neural network (22) and the input data in the different operating states have the specified operating parameters for a current operating state as well as for a previous operating state. [3] Method according to claim 2, characterized by , that the operating parameter recorded for the previous operating state is provided to the neural network (22) as input data by a transfer element (24) with a first-order delay. [4] Method according to any one of the preceding claims, characterized by, that at least one operating parameter is selected from one of the following parameters: - a torque of a drive axle of a vehicle engine, which is in mechanical operative connection with the multi-plate clutch; - a speed difference between two rotating clutch elements of the multi-plate clutch; - an actuation parameter of the multi-plate clutch; - a rotational speed of a rotating clutch element of the multi-plate clutch; - a sump temperature of a gearbox (20) at the beginning of a shifting operation of the multi-plate clutch; - an actuation state of the multi-plate clutch; - an oil flow rate; and - an oil pump speed. [5] Method according to any one of the preceding claims, characterized by that the training data is generated on a test bench in which the multi-plate clutch is supplied with a defined amount of oil. [6] Method for determining (64) a heat dissipation from a multi-plate clutch of a transmission (20), wherein the method comprises at least the following steps: - Recording (56) at least one operating parameter characterizing an operating state; - Determining (66) a cooling coefficient of the multi-plate clutch as a function of the detected operating parameter using a neural network (22) trained with a method according to one of the preceding claims; and - Determining (64) the heat dissipation from the multi-plate clutch as a function of the determined cooling coefficient. [7] Method according to claim 6, characterized by , that when determining (64) the heat dissipation from the multi-plate clutch, at least one of the following characteristics of the multi-plate clutch is additionally taken into account: - a mass of metallic plates (14) of the multi-plate clutch; - a specific heat capacity of the metallic plates (14) of the multi-plate clutch; - a size of a surface area of ​​the plates (12, 14) of the multi-plate clutch. [8] Method for determining (72) a clutch temperature of a multi-plate clutch of a transmission (20), wherein the method comprises at least the following steps: - Analytical determination (70) of heat input into the multi-plate clutch; - Determining (64) a heat dissipation from the multi-plate clutch using a method according to claim 6 or 7; and - Determining (72) the coupling temperature as a function of the specified heat input and heat output.

Images