DEVICE FOR DETERMINING ANOMALIES

Abstract

Device (40) for anomaly detection for a vehicle comprising a transmission (26) that transmits power using a gear, wherein the device (40) for anomaly detection comprises: an execution device (42); and a storage device (46), wherein the storage device (46) is configured to store map data, wherein the map data define a pre-trained map and the pre-trained map has been trained by machine learning, when a variable representing time-series data of a rotational speed of the gear is supplied to the map as an input variable, the map outputs a state variable representing a state of the gear as an output variable, and the execution device (42) is configured to perform: a capture process that receives the variable representing the time-series data as a value of the input variable;and a determination process which, based on a value of the output variable that is output by the map when the value of the input variable is fed to the map, determines whether there is an anomaly in the transmission (26), characterized in that the input variable comprises one or more of the following features: a variable that represents a dimension of the gear based on a specification of the gear, a variable that represents a meshing error of the gear, a variable that represents kickback when the gear meshes, a variable that represents a fluid pressure in a braking system of the vehicle, and a variable that represents a distance traveled by the vehicle.

Description

BACKGROUND 1. Area

[0001] The present disclosure relates to a device for determining anomalies, which ascertains whether an anomaly exists in a transmission that transmits power via gears. 2. Explanation of the state of the art

[0002] The published Japanese patent application JP 2011-79 489 A discloses a diagnostic device that detects an anomaly in a gear-driven mechanism. The drive mechanism is used to wind up the webbing of a seatbelt. The diagnostic device detects the length of webbing extended and identifies an anomaly in the drive mechanism if the extended length exceeds a predetermined value.

[0003] When determining an anomaly in a manner similar to that of the diagnostic device disclosed in the publication described above, it is possible to detect an anomaly based on the fact that the measured value exceeds a predetermined threshold at a specific time. However, if the measured value remains below the threshold, an anomaly may be present if the range of variation in the measured value is greater or smaller than that in the absence of an anomaly. A measured value anomaly with such characteristics cannot be detected by a device performing the determination in the manner of the diagnostic device disclosed in the publication described above. BRIEF EXPLANATION

[0004] This brief explanation serves to introduce, in simplified form, a selection of concepts that are presented in detail below. This brief explanation is neither intended to represent key features or essential characteristics of the claimed subject matter, nor should it be used as an aid in determining the scope of the claimed subject matter.

[0005] Aspects of the present revelation will now be explained.

[0006] Aspect 1: An anomaly detection device is provided for a vehicle that includes a transmission which transfers power using a gear. The anomaly detection device comprises an execution device and a storage device. The storage device is configured to store map data. The map data defines a pre-trained map. The pre-trained map was trained using machine learning. When a variable representing time-series data of the gear's rotational speed is fed to the map as an input variable, the map outputs a state variable representing a state of the gear as an output variable. The execution device is configured to perform a detection process and a determination process.The acquisition process captures the variable representing the time series data as a value of the input variable. Based on a value of the output variable, which is output by the map when the value of the input variable is fed into the map, the determination process determines whether an anomaly exists in the transmission.

[0007] If an anomaly exists in the gearbox due to a gear failure, it is predicted that the time series data of the gear's rotational speed will exhibit a feature that differs from that observed when there is no anomaly in the gearbox.

[0008] The setup described above feeds the variable representing the time-series data of the gear's rotational speed into the characteristic map, resulting in the output of a state variable representing the gearbox's state. The gearbox's state is thus determined based on the variable representing the time-series data of the gear's rotational speed. This means it's possible to determine whether an anomaly exists in the gearbox by extracting a characteristic that cannot be detected solely from the gear's rotational speed, using the time-series data of that speed.

[0009] Aspect 2: In the anomaly detection device according to the first aspect, the execution device is designed to perform a feature parameter calculation process that calculates a feature parameter obtained by processing the time series data. The determination process includes identifying the feature parameter as the value of the input variable. The feature parameter calculation process includes classifying rotational speed values ​​contained in the time series data into classes according to a rotational speed parameter and calculating the frequency of each class as the feature parameter.

[0010] The setup described above processes the time-series rotational speed data into a feature metric that represents a frequency distribution. It is therefore easy to distinguish between the feature shown by the time-series rotational speed data when there is an anomaly in the gearbox and the feature shown by the time-series rotational speed data when there is no anomaly in the gearbox. This means that the setup described above improves the accuracy of determining whether an anomaly exists in the gearbox.

[0011] Aspect 3: In the device for anomaly detection according to the second aspect, the feature size calculation process includes a process that normalizes the time series data such that a maximum value of the gear speed is 1 and a minimum value of the gear speed is 0.

[0012] The setup described above normalizes the time series data to determine the condition of the gear by capturing fluctuations in rotational speed as a feature shown by the time series data without being affected by the magnitude of the absolute value of the rotational speed.

[0013] Aspect 4: In the anomaly detection device according to Aspect 1, the execution device is configured to perform a feature parameter calculation process that calculates a feature parameter obtained by processing the time series data. The determination process includes determining the feature parameter as the value of the input variable. The feature parameter calculation process includes calculating a distribution of a frequency component as the feature parameter, which is obtained by subjecting the time series data to a Fast Fourier Transform (FFT).

[0014] The setup described above is capable of determining the state of the gear based on the characteristic of the time series data occurring in a frequency range.

[0015] Aspect 5: In the device for anomaly detection according to aspect 4, the feature size calculation process includes the calculation of a primary frequency based on an average value of the rotational speed in the time series data and normalization of the frequency component with respect to the primary frequency.

[0016] The setup described above normalizes the frequency component to determine the state of the gearbox based on the characteristic shown by the time series data of the rotational speed, without being affected by the intensity of the frequency component.

[0017] Aspect 6: In the anomaly detection device according to aspect 4 or 5, the feature size calculation process includes the subdivision of a frequency range into several frequency bands and the use of an average value of an intensity of the frequency component in each frequency band as the intensity of the frequency component in that frequency band.

[0018] The structure described above reduces the number of input variables. This reduces the computational effort required during the determination process.

[0019] Aspect 7: In the device for anomaly detection according to one of aspects 1 to 6, the time series data are calculated based on a detection signal from a speed sensor that detects a rotational speed of the gear.

[0020] Aspect 8: In the device for anomaly detection according to one of aspects 1 to 6, the time series data are calculated based on a driving speed of the vehicle.

[0021] Aspect 9: In the device for anomaly detection according to one of aspects 1 to 8, the input variable includes a variable that represents a magnitude of torque transmitted by the gear.

[0022] In the setup described above, the variable representing the magnitude of the torque is fed into the characteristic map as an input variable in addition to the variable representing the time series data. Feeding different types of variables into the characteristic map improves the accuracy of determining whether an anomaly exists in the transmission.

[0023] Aspect 10: In the device for anomaly detection according to one of aspects 1 to 9, the input variable includes a variable that represents a temperature of a hydraulic fluid in the gearbox.

[0024] The setup described above improves the accuracy of determining whether an anomaly exists in the gearbox.

[0025] Aspect 11: In the device for anomaly detection according to one of aspects 1 to 10, the input variable includes a variable that represents a dimension of the gear based on a specification of the gear.

[0026] The setup described above improves the accuracy of determining whether there is an anomaly in the gearbox.

[0027] Aspect 12: In the device for anomaly detection according to one of aspects 1 to 11, the input variable includes a variable that represents a gear engagement error.

[0028] The setup described above carries out the determination process, taking into account the influence of errors in the tooth shapes on the time-series data of the rotational speed. This improves the accuracy of determining whether an anomaly exists in the gearbox.

[0029] Aspect 13: In the device for anomaly detection according to one of aspects 1 to 12, the input variable includes a variable that represents a backlash or play in the engagement of the gear.

[0030] The setup described above carries out the determination process, taking into account the influence of backlash on the time-series data of the rotational speed. This improves the accuracy of determining whether an anomaly exists in the gearbox.

[0031] Aspect 14: In the device for anomaly detection according to one of aspects 1 to 13, the input variable includes a variable that represents a detection value of a vibration sensor that detects vibrations.

[0032] The setup described above improves the accuracy of determining whether an anomaly exists in the gearbox.

[0033] Aspect 15: In the device for anomaly detection according to one of aspects 1 to 14, the input variable includes a variable that represents a detection value of a sound sensor that detects sound.

[0034] The setup described above improves the accuracy of determining whether an anomaly exists in the gearbox.

[0035] Aspect 16: In the device for anomaly detection according to one of aspects 1 to 15, the input variable includes a variable that represents a fluid pressure in a braking system of the vehicle.

[0036] The setup described above performs the determination process while taking into account the fluctuation in the time-series rotational speed data that accompanies vehicle deceleration. This improves the accuracy of determining whether an anomaly exists in the transmission.

[0037] Aspect 17: In the device for anomaly detection according to one of aspects 1 to 16, the input variable includes a variable that represents a distance traveled by the vehicle.

[0038] The setup described above improves the accuracy of determining whether an anomaly exists in the gearbox.

[0039] Aspect 18: In the device for anomaly detection according to one of aspects 1 to 17, the state variable includes a variable that represents a state in which the gear is damaged.

[0040] The setup described above is capable of determining whether the gear is damaged. Therefore, it is possible to detect anomalies due to causes from different categories.

[0041] Aspect 19: In the device for anomaly detection according to one of aspects 1 to 18, the state variable includes a variable that represents a state in which a rattle is generated when the gear rotates.

[0042] The setup described above is capable of determining whether the gearbox is in a state where rattling is being generated. This makes it possible to identify anomalies due to causes of different categories.

[0043] Aspect 20: In the device for anomaly detection according to one of aspects 1 to 19, the state variable includes a variable representing a state in which a screech is produced when the gear rotates.

[0044] The setup described above is capable of determining whether the gearbox is in a state that generates a squealing noise. This makes it possible to identify anomalies due to causes of different categories.

[0045] Aspect 21: In the anomaly detection device according to one of aspects 1 to 17, the characteristic map is a first characteristic map that outputs a state variable indicating whether the gear is damaged when the input variables are applied. The characteristic map data is first characteristic map data that defines the first characteristic map. The determination process is a first determination process. The storage device is designed to also store second characteristic map data and third characteristic map data. The second characteristic map data and the third characteristic map data each define a pre-trained characteristic map. The pre-trained characteristic map was trained using machine learning. The second characteristic map data define a second characteristic map.When the input variable is entered, the second characteristic map outputs a state variable as an output variable, indicating whether the gear is in a state where chatter is generated when the gear rotates. The third characteristic map data defines a third characteristic map. When the input variable is entered, the third characteristic map outputs a state variable as an output variable, indicating whether the gear is in a state where squealing is generated when the gear rotates. The execution device is designed to perform a second determination process and a third determination process. The second determination process determines whether there is an anomaly in the gearbox, based on a value of the output variable that is output by the second characteristic map when the value of the input variable is entered into the second characteristic map.When the value of the input variable is entered into the third map, the third determination process determines, based on a value of the output variable provided by the third map, whether there is an anomaly in the transmission.

[0046] For example, the time series data will differ when a chatter is generated compared to the time series data when a screech is generated. Using a single characteristic map to distinguish between multiple gear states can increase the computational load on the executing device or decrease the accuracy of the determination.

[0047] The setup described above uses the first characteristic map to detect whether the gearbox is damaged, the second to detect whether chatter is being generated, and the third to detect whether screeching is being generated. This setup thus reduces the computational load for the executing device compared to using a single characteristic map to determine the gear condition. Furthermore, compared to using a single characteristic map to determine the gearbox condition, this setup prevents a reduction in the accuracy of determining whether an anomaly exists in the gearbox.

[0048] Further features and aspects will become apparent from the detailed description, figures, and claims below. List of characters Fig. Figure 1 is a diagram showing the structure of a vehicle and a control device according to a first embodiment. Fig. Figure 2 is a flowchart showing a sequence of processes or procedures that are performed by the control device according to the first embodiment. Fig. Figure 3A is a graph showing time series data of the rotational speed of a gear according to the first embodiment. Fig. Figure 3B is a graph showing time series data of the rotational speed of the gear according to the first embodiment. Fig. 4A is a diagram showing a feature size obtained by processing the time series data of the rotational speed of the gear according to the first embodiment. Fig. Figure 4B is a graph showing a feature size obtained by processing the time series data of the rotational speed of the gear according to the first embodiment. Fig. Figure 5 is a diagram that defines output variables according to the first embodiment. Fig. Figure 6 is a diagram showing a system according to a second embodiment. Fig. Figure 7 is a flowchart showing a sequence of processes that a controller executes according to the second embodiment.

[0049] In the figures and the detailed explanation, the same reference symbols refer to the same elements. The figures may not be to scale, and the relative size, proportions, and representation of elements in the drawings may be exaggerated or distorted for the sake of clarity, illustration, and convenience. DETAILED EXPLANATION

[0050] This explanation provides a comprehensive understanding of the described procedures, devices, and / or systems. Modifications and equivalents of the described procedures, devices, and / or systems are obvious to those skilled in the art. The sequence of operations is exemplary and can be modified by those skilled in the art, except for operations that necessarily follow a specific order. Descriptions of functions and designs or setups that are well known to those skilled in the art may be omitted.

[0051] Exemplary embodiments can take different forms and are not limited to the examples described. However, the examples described are detailed and complete and convey to those skilled in the art the full scope of the disclosure. <Erste Ausführungsform>

[0052] A first embodiment will now be described with reference to the figures.

[0053] As in Fig. As shown in Figure 1, a vehicle VC comprises an internal combustion engine 10, a first motor-generator 22, and a second motor-generator 24. The internal combustion engine 10 includes a crankshaft 12, which is mechanically coupled to a power divider 20. The power divider 20 divides the power of the internal combustion engine 10, the power of the first motor-generator 22, and the power of the second motor-generator 24. The power divider 20 includes a planetary gear mechanism comprising a carrier C, a sun gear S, and a ring gear R. The carrier C is mechanically coupled to the crankshaft 12. The sun gear S is mechanically coupled to a rotating shaft 22a of the first motor-generator 22. The ring gear R is mechanically coupled to a rotating shaft 24a of the second motor-generator 24. An output voltage of a first inverter 23 is applied to the terminals of the first motor-generator 22.The output voltage of a second inverter 25 is applied to the terminals of the second motor generator 24.

[0054] The ring gear R of the power divider 20 is mechanically coupled via a gearbox 26 to driven wheels 30 and to the rotating shaft 24a of the second motor-generator 24. The gearbox 26 is designed to transmit power to the driven wheels 30 by means of combining or engaging gears.

[0055] The carrier C of the power distributor 20 is mechanically coupled to a driven shaft 32a of an oil pump 32. The oil pump 32 is designed to circulate oil in an oil pan 34 as a lubricant to the power distributor 20 and to deliver the oil as hydraulic fluid to the gearbox 26. The pressure of the hydraulic fluid delivered by the oil pump 32 is regulated by a hydraulic control circuit 26a in the gearbox 26.

[0056] A control device 40 is provided for controlling the internal combustion engine 10. The control device 40 corresponds to a device for anomaly detection for the vehicle VC, which includes the transmission 26 that transmits power via gears. The control device 40 operates various actuated units of the internal combustion engine 10 and thereby controls control variables such as the torque and the ratios of exhaust gas components of the internal combustion engine 10. Furthermore, the control device 40 is designed to control the first motor-generator 22. The control device 40 operates the first inverter 23 to control controlled variables such as the torque and speed of the first motor-generator 22. The control device 40 is also designed to control the second motor-generator 24. The control device 40 controls the second inverter 25 to control control variables such as the torque and speed of the second motor-generator 24. Fig. Figure 1 shows operating signals MS that are output by the control device 40 for controlling the internal combustion engine 10 and / or the transmission 26.

[0057] The control device 40, when controlling the control variables described above, refers to an output signal Scr of a crank angle sensor 50, an output signal Sm1 of a first rotation angle sensor 52, which detects a rotation angle of the rotating shaft 22a of the first motor generator 22, and an output signal Sm2 of a second rotation angle sensor 54, which detects a rotation angle of the rotating shaft 24a of the second motor generator 24. The control device 40 also refers to an oil temperature Toil, which is detected by an oil temperature sensor 56.

[0058] The control device 40 comprises a CPU 42, a ROM or read-only memory 44, a storage device 46, which is a non-volatile memory that is electrically overwritable, and peripheral circuits 48. The CPU 42, the ROM 44, the storage device 46, which is a non-volatile memory that is electrically overwritable, and the peripheral circuits 48 can communicate with each other via a local network 49. The peripheral circuits 48 include a circuit that generates a clock signal that controls internal processes, a power supply circuit, and a reset circuit. The control device 40 controls the aforementioned control variables by causing the CPU 42 to execute programs stored in the ROM 44.

[0059] Some of the processes carried out by the control device 40 will now be explained.

[0060] The control device 40 is designed to perform an adjustment process for the drive torque. This adjustment process calculates a setpoint Trq* for the drive torque, which is the target value for the torque to be delivered to the driven wheels 30. The setpoint Trq* for the drive torque is calculated using the actuation value of an accelerator pedal or accelerator lever element of the vehicle VC as input, such that it increases with increasing actuation value.

[0061] The control device 40 is designed to execute a drive force distribution process. Based on the setpoint Trq* for the drive torque, the drive force distribution process defines a torque setpoint Trqe* for the internal combustion engine 10, a torque setpoint Trqm1* for the first motor-generator 22, and a torque setpoint Trqm2* for the second motor-generator 24. The torque setpoints Trqe*, Trqm1*, and Trqm2* are each generated by the internal combustion engine 10, the first motor-generator 22, and the second motor-generator 24, respectively, such that the torque delivered to the driven wheels 30 corresponds to the setpoint Trq* for the drive torque.

[0062] The control device 40 is configured to perform a speed calculation process. This process calculates a gear speed Ngear, which is the rotational speed of a gear in the transmission 26. The gear speed Ngear is calculated based on the output signal Sm1. The gear speed Ngear is calculated at predefined time intervals. Changes in the gear speed Ngear over time are stored in the storage device 46 as time-series data of the gear speed Ngear. For example, the gear speed Ngear is calculated as the rotational speed of a driven gear located at the position closest to the driven gears 30 along the path by which power is transmitted to them. The gear speed Ngear can also be the rotational speed of a driving gear meshing with the driven gear.The gear speed Ngear does not necessarily have to be the speed of the gear located in the position closest to the driven gears 30. That is, the gear speed Ngear can be the speed of any gear in the gearbox 26.

[0063] The control device 40 is designed to execute a process or procedure for determining an anomaly of the gear in the transmission 26. This procedure will now be explained.

[0064] Fig. Figure 2 shows a sequence of processes executed by the control device 40. The in Fig. The processes or operations shown in Figure 2 are executed by the CPU 42 by repeatedly running programs stored in ROM 44 at predetermined time intervals. In the following description, the number of each step is represented by the letter S, followed by a digit.

[0065] In the Fig. In the sequence of processes shown in Figure 2, the CPU 42 first reads the time-series data of the gear speed Ngear and the oil temperature Toil (S101). The CPU 42 then performs a normalization process that normalizes the time-series data of the gear speed Ngear within a specific time period (S102). In addition, the CPU 42 calculates a normalization characteristic NFv, which is obtained by processing the normalized time-series data (S103). The normalization characteristic NFv is calculated as a datum that represents the frequency distribution of the gear speed Ngear during the defined period.

[0066] The processes from S102 and S103 are now being evaluated based on the Fig. 3A, Fig. 3B, Fig. 4A and Fig. 4B described. Fig. 3A shows part of the time series data of the gear speed Ngear when there is no anomaly in the gearbox, i.e. the gearbox is working normally. Fig. Figure 3B shows a portion of the time-series data of the gear speed Ngear when an anomaly is present in the gearbox. In the Fig. 3A and Fig. During the period shown in 3B, the target value Trq* for the drive torque has the same value.

[0067] As in Fig. As shown in 3A, the gear speed Ngear changes in a regular cycle and with a regular amplitude if there is no anomaly in the gearbox. In contrast, as shown in Fig. Figure 3B shows the gear speed Ngear to show a waveform that differs from the one in Fig. 3A detects an anomaly in the gear. For example, the gear speed Ngear fluctuates irregularly. The gear speed Ngear can become significantly high or low.

[0068] In the present embodiment, the time-series data of the gear speed Ngear comprise samples that follow each other in time when the gear speed Ngear is sampled with a constant sampling cycle during the specified period. The filled circles in Fig. 3A and Fig. 3B represents values ​​that are sampled using the constant sampling cycle. The specified period refers to a period during which the gear speed Ngear is sampled.

[0069] In process S102, the CPU 42 normalizes the time-series data of the gear speed Ngear such that the maximum value of the gear speed Ngear during the specified period is 1 and the minimum value of the gear speed Ngear during the specified period is 0. The normalization can be performed, for example, using equation 1 below. N=(n−nmin) / (nmax−nmin)

[0070] In equation 1, the letter N represents the gear speed Ngear after normalization. The letter n stands for the gear speed Ngear to be normalized. The symbol nmax represents the maximum value of the gear speed Ngear before normalization. The symbol nmin represents the minimum value of the gear speed Ngear before normalization.

[0071] In the S103 process, CPU 42 first divides the range from the minimum to the maximum value of the gear speed Ngear during the specified time period evenly into five classes. In other words, the range is divided into classes, each with a width of 0.2. CPU 42 then calculates the number of sampled gear speed values ​​Ngear in each class, referred to as the frequency, as the normalization feature NFv. The five classes are subsequently designated in ascending order as Bin1 (first class), Bin2 (second class), Bin3 (third class), Bin4 (fourth class), and Bin5 (fifth class).

[0072] Fig. Figure 4A shows an example in which the time-series data of the gear speed Ngear are subjected to the processes from S102 and S103 when no anomaly is present in the gearbox. That is, Fig. Figure 4A shows an example of the normalization characteristic value NFv when there is no anomaly in the gearbox. As in Fig. As shown in Figure 4A, if there is no anomaly in the gearbox, the frequency increases in order of frequency from the first class Bin1, through the frequency in the second class Bin2, to the frequency in the third class Bin3. Among the frequencies of the five classes, the frequency of the third class Bin3, which contains the median gear speed Ngear, is the highest. The frequency decreases in order of frequency of the fourth class Bin4 and frequency of frequency of the fifth class Bin5. The frequency in the second class Bin2 is equal to the frequency in the fourth class Bin4, and the frequency in the first class Bin1 is equal to the frequency in the fifth class Bin5. That is, Fig. 4A shows a distribution that is symmetrical on both sides with respect to the third class Bin3.

[0073] Fig. Figure 4B shows an example in which the time-series data of the gear speed Ngear are subjected to the processes from S102 and S103 in the event of an anomaly in the gearbox. That is, Fig. Figure 4B shows an example of the normalization characteristic NFv when an anomaly is present in the gearbox. As in Fig. In 4A, the frequency is highest in the third class Bin3, and lowest in the first class Bin1 and the fifth class Bin5. In contrast to the case of Fig. In 4A, where there is no anomaly in the gearbox, the frequency in the third class, Bin3, is significantly higher than the frequencies in the other classes. This is because a small number of values ​​are sampled, which deviate considerably from the median due to irregular fluctuations in the gear speed, Ngear. To further illustrate this, consider a frequency distribution of data where a small number of values ​​are sampled that deviate significantly from a median. The range from the minimum to the maximum value of the gear speed, Ngear, in the data is divided into classes, the number of which is the same as in the example of Fig. 4A, in which the time-series data of the gear speed Ngear is processed, which changes regularly. Consequently, the class containing the mode or median includes a larger number of samples than in the example shown in Fig. Figure 4A shows that if a small number of values ​​that differ significantly from the median are sampled in this way, the frequency of the class containing the mode or median increases, while the frequency of the other classes decreases.

[0074] As in the Fig. 4A and Fig. Figure 4B shows that the normalization feature size NFv, obtained by processing the time series data of the gear speed Ngear, exhibits different features depending on whether there is no anomaly in the gearbox or whether there is an anomaly in the gearbox.

[0075] Again with reference to Fig. 2. After calculating the normalization characteristic NFv, the CPU 42 replaces the normalization characteristic NFv calculated by the process from S103, and the oil temperature Toil for the input variables x(1) to x(6), which are fed to a characteristic map defined by characteristic map data DM, which is in the Fig. The storage device shown in Figure 1 (S104) contains the following data: The first class Bin1 to the fifth class Bin5 are used as input variables x(1) to x(5). The oil temperature Toil is replaced by the input variable x(6). The average value of the oil temperature Toil during the specified period can be used as the input variable x(6).

[0076] Next, the CPU calculates 42 output variables y(0) to y(4) that represent the state of the gearbox by inserting the values ​​of the input variables x(1) to x(6) into the map (S105).

[0077] In the present embodiment, the characteristic map is a functional approximation. In particular, a fully connected neural feed-forward network with a single middle layer is used as the characteristic map. Specifically, the input variables x(1) to x(6), for which values ​​are inserted by the process of S105, and x(0), which is a bias parameter, are transformed by a linear characteristic map or linear matrix defined by a coefficient wFjk (j = 1 - m, k = 0 - 6) to obtain m values. Each of the m values ​​is inserted into an activation function to determine the values ​​of the nodes of the middle layers. Furthermore, the values ​​of the nodes of the middle layers are transformed by a linear characteristic map defined by a coefficient wSij (i = 0 - 3). The transformed values ​​are each inserted into an activation function g to determine the values ​​of the output variables y(0), y(1), y(2), and y(3).A hyperbolic tangent can be used as the activation function f. A softmax function can be used as the activation function g.

[0078] As in Fig. As shown in Figure 5, the output variables y(0), y(1), y(2), and y(3) are state variables that identify the states of the transmission. The output variable y(0) represents the probability of a state in which there is no anomaly in the transmission, i.e., a state in which the transmission operates normally. The output variable y(1) represents the probability of a state in which the transmission is damaged. The output variable y(2) represents the probability of a state in which a rattle is generated. The output variable y(3) represents the probability of a state in which a screech is generated. In the prior art, rattle is understood to be a noise caused by the contact of tooth surfaces of meshing gears. A screech is a noise that contains higher frequencies than rattle.

[0079] Again with reference to Fig. 2. The CPU 42 selects a maximum value ymax of the output variables y(0) to y(3) (S106). The CPU 42 then determines the state of the gearbox based on an output variable equal to the maximum value ymax of the output variables y(0) to y(3) (S107). Next, the CPU 42 executes a storage process that saves the result of the gearbox state determination in the memory device 46 (S108). For example, if the value of the output variable y(1) is equal to the maximum value ymax, the CPU 42 stores information in the memory device 46 indicating that the gearbox is damaged. In operation S108, the CPU 42 can execute a notification operation to inform about the state of gearbox 26 based on the gearbox state stored in the memory device 46 by sending a message to the storage device 46. Fig. Display 70, shown in Figure 1, is operated. Display 70 is an example of a notification device that provides notification about the status of gearbox 26. The notification device could, for example, be a loudspeaker. In this case, the notification process is carried out by pressing the loudspeaker to output audio signals. When the process of S108 is complete, CPU 42 temporarily suspends the operation in Figure 1. Fig. 2. Sequence of processes shown.

[0080] The DM map data is a pre-trained model. Training the DM map data uses data in which the normalization parameter NFv and the oil temperature Toil are labeled and populated with correct responses and input data, respectively, representing the actual state of the transmission. The normalization parameter NFv in the training data is calculated using procedures similar to those in S102 and S103, based on time-series gear speed data Ngear obtained from test drives of a pre-production vehicle. The DM map data is trained using training data that includes data from a normal transmission state, data from a damaged transmission state, data from a chattering state, and data from a squealing state.

[0081] The functionality and advantages of the present embodiment will now be described.

[0082] The CPU 42 determines the state of the gearbox based on the time-series data of the gear speed Ngear. It is possible to determine, using the time-series data of the gear speed Ngear, whether an anomaly exists in gearbox 26.

[0083] The embodiment described above also includes the following processes and advantages. (1) The normalization characteristic NFv, which is calculated by processing the time-series data of the gear speed Ngear, is used as an input variable that is read into the characteristic map. The time-series data are processed into data that are included in the Fig. 4A and Fig. Figure 4B shows the frequency distribution. This facilitates the differentiation between the feature shown by the time-series data of the gear speed Ngear when an anomaly is present in the gearbox and the feature shown by the time-series data of the gear speed Ngear when no anomaly is present in the gearbox. That is, the present embodiment improves the accuracy of determining whether an anomaly is present in the gearbox. (2) The gear speed Ngear is divided into classes when the time-series data of the gear speed Ngear are processed into the normalization characteristic NFv. This reduces the order of magnitude of the input variables. Accordingly, the computational effort of the process for determining a gearbox anomaly is reduced. (3) The time series data of the gear speed Ngear are normalized or standardized in the same way as the normalization characteristic NFv. This allows the condition of the gearbox to be determined by capturing fluctuations in the gear speed Ngear as a characteristic of the time series data, without being influenced by the absolute value of the gear speed Ngear. (4) Input variables that are fed into the characteristic map simultaneously include, in addition to the normalization characteristic NFv, the mean oil temperature Toil over the specified period. This allows the value of the state variable to be calculated taking the oil temperature into account. (5) After calculating the values ​​of the output variables y(0) to y(3), the CPU 42 determines the state of the gearbox based on the maximum value and stores the result in the storage device 46. This allows not only the detection of an anomaly in the gearbox, but also the cause of the anomaly, if one exists. That is, it is possible to detect gearbox damage, a rattling noise, or a screeching noise. (6) The result of the determination of the transmission condition is stored in the storage device 46. Thus, measures appropriate to the anomaly stored in the storage device 46 can be carried out when the vehicle VC is taken to a repair shop. <Zweite Ausführungsform>

[0084] A second embodiment will now be described with reference to the figures. The main focus will be on the differences compared to the first embodiment.

[0085] Fig. Figure 6 shows the construction of a control device 40 according to the present embodiment. Fig. For illustrative purposes, the same reference symbols are used for the components as those in Section 6. Fig. 1 agree, and further explanations of the same are omitted.

[0086] As in Fig. As shown in Figure 6, the storage device 46 of the vehicle VC stores several sets of map data. For example, the first set of map data, DM (A1), is a pre-trained model that was previously trained using training data based on the normalization feature size NFv with the transmission operating normally and training data based on the normalization feature size NFv with the transmission damaged. The second set of map data, DM (A2), is a pre-trained model that was previously trained using training data based on the normalization feature size NFv with the transmission operating normally and training data based on the normalization feature size NFv in a state where rattling is generated.The third characteristic map data DM (A3) is a pre-trained model that was previously trained using training data based on the normalization characteristic size NFv in a normally operating gearbox and training data based on the normalization characteristic size NFv in a state in which a screech is generated.

[0087] The second embodiment determines the state of the transmission based on the first map data DM (A1), the second map data DM (A2) and the third map data DM (A3).

[0088] Fig. Figure 7 shows a sequence of processes that are determined by the in Fig. The control device 40 shown in section 6 is executed. Fig. The processes shown in the 7 diagrams determine the state of the transmission. Fig. The 7 operations shown are performed by the CPU 42 by repeatedly executing programs stored in ROM 44 at predetermined time intervals.

[0089] In the Fig. In the sequence of processes shown in section 7, the CPU 42 first selects the initial map data DM (A1) as map data to be used for determining the transmission state (S201). The CPU 42 then executes a first determination process (S202).

[0090] The first determination process is carried out according to a Fig. The sequence of processes shown in 2 is executed. The differences between the first determination process and the one in Fig. The processes shown in Figure 2 will now be described. In process S105, the CPU 42 uses a first characteristic map, defined by the first characteristic map data DM (A1), as a characteristic map into which the input variables x(1) to x(6) are inserted. The CPU 42 inserts the values ​​of the input variables x(1) to x(6) into the first characteristic map to calculate, as output variables indicating the state of the transmission, the probability that the transmission is operating normally and the probability that the transmission is damaged. The CPU 42 determines the state of the transmission based on the maximum value of the calculated output variables.

[0091] An example is now described in which the condition of the gearbox is determined based on the output variables in the first determination process. For example, CPU 42 determines that the gear is operating normally if the result indicates that the probability of the gear operating normally is 80% or higher. Conversely, CPU 42 determines that an anomaly of the gear exists if the result indicates that the probability of the gear being damaged is 80% or higher, and classifies the cause of the anomaly as damage to the gear. If the result indicates that both the probability of the gear operating normally and the probability of the gear being damaged are less than 80%, CPU 42 determines that an anomaly has occurred that is not damage to the gear.

[0092] Again with reference to Fig. 7. CPU 42 determines whether the first determination process has detected a gear anomaly (S203). If an anomaly is detected (S203: YES), CPU 42 determines whether the gear anomaly has been classified as a damaged gear (S204).

[0093] If it is determined that the gear anomaly is not a gear damage (S204: NO), the CPU 42 selects the second map data DM (A2) as the map data to be used to determine the gear's condition (S205) and performs a second determination process (S206).

[0094] The second determination process is carried out according to a Fig. The sequence of processes shown in 2 is executed. In contrast to the one in Fig. In the sequence of processes shown in Figure 2, the CPU 42, in the second determination process, inserts the input variables x(1) to x(6) into a second characteristic map, which is defined by the second characteristic map data DM (A2). The CPU 42 calculates, as output variables indicating the state of the transmission, the probability that the transmission is operating normally and the probability of a state in which chatter is generated. The CPU 42 determines the state of the transmission based on the maximum value of the calculated output variables. Following a procedure similar to the first determination process, the CPU 42 determines whether there is an anomaly in the transmission and, if so, classifies its cause.

[0095] Again with reference to Fig. 7. The CPU 42 determines whether the second determination process has identified that a chatter is being generated (S207). If it is determined that the anomaly in the transmission is not the generation of a chatter (S207: NO), the CPU 42 selects the third map data DM (A3) as the map data to be used to determine the gear condition (S208) and executes a third determination process (S209).

[0096] The third determination process is carried out according to a Fig. The sequence of processes shown in 2 is executed. In contrast to the one in Fig. In the sequence of processes shown in Figure 2, the CPU 42, in the third determination process, inserts the input variables x(1) to x(6) into a third characteristic map, which is defined by the third characteristic map data DM (A3). The CPU 42 calculates, as output variables indicating the state of the gearbox, the probability that the gearbox is operating normally and the probability of a state in which a squealing noise is generated. The CPU 42 determines the state of the gearbox based on the maximum value of the calculated output variables. Following a procedure similar to the first determination process, the CPU 42 determines whether there is an anomaly in the gearbox and, if so, classifies its cause.

[0097] When the operation of S209 is completed, when a positive decision is made in the operations in S204 and S207, or when a negative decision is made in the operation of S203, CPU 42 temporarily suspends the process in Fig. 7 shown sequence of processes.

[0098] Furthermore, the control device 40 is capable of referring to measurement values, in addition to the oil temperature detected by the oil temperature sensor 56, which, as in Fig. As shown in Figure 6, various sensors on the vehicle VC can detect the vehicle speed. For example, the control device 40 can use a wheel speed detected by a wheel speed sensor 57. The control device 40 is capable of calculating a vehicle speed SPD based on the wheel speed. The control device 40 is capable of taking into account the vibration VB detected by a vibration sensor 58 on the vehicle VC. The control device 40 is capable of taking into account the sound NZ detected by a noise sensor 59 on the transmission 26.

[0099] Furthermore, the control device 40 is capable of taking into account a value measured by an instrument on the vehicle VC or a state variable received from other control devices. For example, the control device 40 can take into account the total distance traveled OD of the vehicle VC as measured by an odometer 60 on the vehicle VC. The control device 40 can take into account a brake pressure PB provided by a brake control device 80. The brake control device 80 is designed to control the braking system of the vehicle VC. The control device 40 is, for example, capable of using a low-pressure brake actuation element as the brake pressure PB. If the braking system is a fluid pressure braking system, the control device 40 is capable of taking into account a master cylinder pressure as the brake pressure PB.

[0100] The storage device 46 can be used as described in Fig. Figure 6 shows how to store gear data DG. The gear data DG comprises specification data based on the specifications of the gears in gearbox 26. This specification data includes target values ​​representing the dimensions of various parts of the gear, such as the dimensions of the tooth faces. Additionally, a gear mesh error is pre-measured, and this error value is included in the gear data DG. The gear data DG also includes the degree of backlash or recoil when the gears are meshed.

[0101] The operation and advantages of the present embodiment will now be explained.

[0102] The second embodiment utilizes the first characteristic map data DM (A1), which defines the first characteristic map used to detect a condition in which the gear is damaged; the second characteristic map data DM (A2), which defines the second characteristic map used to detect a condition in which chatter is generated; and the third characteristic map data DM (A3), which defines the third characteristic map used to detect a condition in which screeching is generated. Thus, by executing each of the first, second, and third determination processes, the setup reduces the computational load on the execution device compared to a case in which a single characteristic map is used to detect the cause of a gearbox anomaly.Accordingly, the present embodiment prevents the computational load for the execution device from being increased and the accuracy of the determination from being reduced when an anomaly of the gearbox 26 is determined by the execution device. <korrespondenz>

[0103] The correspondence or relationship between the elements of the embodiments explained above and the elements described in the preceding BRIEF EXPLANATION is as follows. The correspondence for each of the numbers in the preceding BRIEF EXPLANATION is shown below.

[0104] [Aspect 1] The device for anomaly detection corresponds to the control device 40 of the Fig. 1 and Fig. 6. The execution device corresponds to the CPU 42 and the ROM 44 of the Fig. 1 and Fig. 6. The storage device corresponds to storage device 46 of the Fig. 1 and Fig. 6. The characteristic map data corresponds to the characteristic map data DM. The acquisition process corresponds to the process of S104 from Fig. 2. The determination process corresponds to the procedures described in sections S105 to S107 of the Fig. 2.

[0105] [Aspects 2 and 3] The characteristic value calculation process corresponds to the processes from S102 and S103 of the Fig. 2.

[0106] [Aspect 21] The first map data corresponds to the first map data DM (A1). The second map data corresponds to the second map data DM (A2). The third map data corresponds to the third map data DM (A3). <Andere Ausführungsformen>

[0107] The embodiments described above can be modified as follows. The embodiments described above and the following modifications can be combined, provided that the combined modifications remain technically consistent. Modification regarding the time series data

[0108] Fig. 3A and Fig. Figure 3B shows an example of the time series data for the gear speed Ngear. In the Fig. 3A and Fig. However, in 3B the number of samples of the gear speed Ngear during the specified period is not limited.

[0109] In the speed calculation method of the embodiments described above, an example is presented in which the gear speed Ngear is calculated based on the output signal Sm1. The device for calculating the gear speed Ngear, which represents the speed of the gear, is not limited to this. For example, the speed of the gear in transmission 26 is related to the vehicle speed SPD. Therefore, the gear speed Ngear can be calculated based on the vehicle speed SPD. Alternatively, a sensor can be used to detect the speed of the gear, and the gear speed Ngear can be calculated based on the value detected by this sensor. Modification relating to the characteristic values ​​as input variables

[0110] The process of S102 in the embodiments described above normalizes the time-series data of the gear speed Ngear such that the maximum value of the gear speed Ngear during the specified period is 1 and the minimum value of the gear speed Ngear during the specified period is 0. The method for normalizing the time-series data of the gear speed Ngear is not limited to this. For example, time-series data of the gear speed Ngear can be normalized such that the mean value of the gear speed Ngear is 0 and the variance is 1.

[0111] The method of S103, in the embodiments described above, divides the range from the minimum to the maximum value of the gear speed Ngear during the specified period evenly into five classes. If the number of classes for the feature variables is constant, the number of classes set in the process of S103 can be changed. In other words, the width of each class can be changed.

[0112] In processes S102 and S103 in the embodiments described above, the time-series data of the gear speed Ngear are normalized, and then the normalized data are processed to calculate the normalization feature NFv, which represents the frequency distribution. This disclosure is not limited to this. For example, a feature Fv can be calculated as a feature representing the frequency distribution by processing the time-series data of the gear speed Ngear, and then the normalization feature NFv can be calculated based on the feature Fv.

[0113] In the embodiments described above, the normalization feature NFv is used as an example of the input variables that are fed into the characteristic map defined by the characteristic map data DM. However, the present disclosure is not limited to this. For example, the feature Fv can be calculated as data representing the frequency distribution from the time-series data of the gear speed Ngear, and the feature Fv can be used as an input variable. That is, the normalization process is not strictly necessary. Even if the widths of the classes of the feature Fv are not adjusted, the feature Fv is still data from features extracted from the time-series data of the gear speed Ngear. With the feature Fv as an input variable, the state of the gear can be determined.

[0114] In the embodiments described above, the data representing the frequency distribution are used as the feature obtained by processing the time-series data of the gear speed Ngear. However, the present disclosure is not limited to this. For example, the distribution of the frequency component obtained by subjecting the time-series data of the gear speed Ngear to a Fast Fourier Transform can be calculated as a feature. Using the feature calculated in this way as an input variable to be fed to the characteristic map, the state of the gearbox can be determined based on the feature of the time-series data occurring in the frequency domain. That is, the state of the gear can be determined based on the feature obtained by analyzing the frequency of the time-series data of the gear speed Ngear.

[0115] The distribution of a frequency component obtained by subjecting time-series data of the gear speed Ngear to a Fast Fourier Transform can be normalized, and the normalized feature can be used as an input variable read into the characteristic map. For example, a primary frequency can be calculated from the mean of the gear speed Ngear during the given period, and the frequency component can be normalized based on this primary frequency. Because the frequency component is normalized, the gear's condition can be determined based on the feature obtained by analyzing the frequency of the time-series data of the gear speed Ngear without being affected by the intensity of the frequency component.

[0116] If the feature variable obtained by analyzing the frequency of the time-series data of the gear speed Ngear is used as an input variable, the frequency range can be divided into several frequency bands, and the mean intensity of the frequency component in each frequency band can be used as the intensity of the frequency component in that frequency band. Accordingly, the number of input variables is reduced when the feature variable obtained by analyzing the frequency of the time-series data of the gear speed Ngear is used as an input variable. That is, this modification reduces the computational load in the process that determines whether there is an anomaly in the gearbox 26.

[0117] The feature quantity representing the frequency distribution, such as the normalization feature quantity NFv in the embodiments described above, and the feature quantity obtained by analyzing the frequency of the time series data in the modification, can both be used as input variables. By using a combination of datasets that have been processed differently, the feature represented by the time series data can be easily identified. This further improves the accuracy of the determination.

[0118] The modifications described above show examples where the feature obtained by analyzing the frequency of the time-series data of the gear speed Ngear is used as an input variable. The present disclosure is not limited to this. For example, a feature obtained by subjecting the time-series data of the gear speed Ngear to a speed ratio analysis can be used as an input variable. Modification regarding the input variables that are read into the characteristic map

[0119] In the embodiments described above, the characteristic value obtained by processing the time-series data of the gear speed Ngear is used as the input variable that is read into the characteristic map. However, the time-series data of the gear speed Ngear can also be used as an input variable. For example, a recorded value can be used as an input variable.

[0120] In the embodiments described above, the average oil temperature (Toil) is used as an input variable fed to the characteristic map. However, the present disclosure is not limited to this. For example, time-series data of the oil temperature (Toil) can also be used as an input variable.

[0121] The oil temperature (Toil) does not necessarily need to be included in the input variables fed to the map defined by the characteristic map data (DM). It is sufficient if the input variables include the time-series data of the gear speed (Ngear).

[0122] The oscillation VB can be one of the input variables that are fed to the characteristic map defined by the characteristic map data DM.

[0123] The input variables that are fed to the characteristic map defined by the characteristic map data DM can include the sound NZ.

[0124] The input variables that are read into the map defined by the characteristic map data DM can include the total distance traveled (OD). Instead of the total distance traveled (OD), the cumulative time during which the vehicle (VC) starts or travels can also be used. With this setup, for example, a particular gear change can be taken into account when determining whether an anomaly exists in transmission 26.

[0125] The input variables that are read into the characteristic map defined by the characteristic map data DM can include the brake pressure PB. In this setup, for example, vibrations due to the deceleration of the vehicle VC and fluctuations in the gear speed Ngear due to the deceleration of the vehicle VC can be taken into account when determining whether an anomaly exists in the transmission 26.

[0126] Input variables that are read into the map defined by the DM map data can also include backlash or gear backlash.

[0127] The input variables, which are read into the characteristic map defined by the characteristic map data DM, can contain an intervention error. This setup allows, for example, the influence of tooth form errors on the gear speed Ngear to be taken into account when determining whether an anomaly exists in the gearbox 26.

[0128] The input variables that are read into the characteristic map defined by the characteristic map data DM can include the target values ​​of the dimensions of various parts of the gearbox, such as the dimensions of the tooth surfaces contained in the gearbox data DG.

[0129] The input variables that are read into the characteristic map defined by the characteristic map data DM can include the magnitude of the torque transmitted by the gear. Modification regarding the characteristic map

[0130] The neural network is not limited to a fully connected neural feed-forward network. For example, a one-dimensional convolutional neural network can be used. Similarly, the machine learning-trained model is not limited to a neural network. For example, the condition of the gearbox can be detected using a support vector machine classification.

[0131] In the procedure according to S105, the number of middle layers in the neural network is one. However, the number of middle layers can also be two or more.

[0132] The characteristic map is not limited to the one with four output variables (the output variables y(0), y(1), y(2), and y(3)). If there are other states of the transmission that can be identified based on feature sizes obtained by processing time-series data, such states can be used as additional state variables and as output variables of the characteristic map. Modification regarding the selection of the characteristic map data

[0133] In the Fig. In the second embodiment, as shown in Figure 7, the first map data DM(A1) to the third map data DM(A3) are selected sequentially, and the determination process is carried out using the selected map data. When selecting one of several sets of map data, the map data do not necessarily have to be selected in the order of the first map data DM(A1) to the third map data DM(A3). For example, if the frequency of the sound NZ detected by sound sensor 59 is as high as the frequency of a screech, the third map data DM(A3) can be selected first. This can make it possible to identify a transmission anomaly in fewer trials.

[0134] In this way, map data can be selected based on detection values ​​that can be acquired from various sensors and state variables of the vehicle VC. Examples of setups for selecting map data are described below.

[0135] If the torque transmitted by the gears is close to zero, the second characteristic map data DM (A2) can be selected first. The torque contains a DC component and an AC component. If the torque transmitted by the gears pulsates in a range that includes zero, the driving gear rotates clockwise and counterclockwise and therefore likely strikes the driven gear. This likely generates chatter. In such a case, the second characteristic map data DM (A2) can be used as a first choice to detect a gearbox anomaly in a smaller number of trials. Whether the torque transmitted by the gears is close to zero can be determined, for example, using the target value Trq* for the drive torque.

[0136] Oil temperature can be used as a criterion for selecting the characteristic map data. Oil temperature affects the oil's viscosity. Depending on the oil's viscosity, it's possible to predict whether the oil-lubricated transmission is likely to produce a squeal or a rattle. If a squeal or rattle is predicted, selecting the second characteristic map data, DM (A2), or the third characteristic map data, DM (A3), can allow for the detection of a transmission anomaly with fewer trials.

[0137] The gear data (DG) can be used as a criterion for selecting the characteristic map data. For example, a large gear backlash is more likely to produce chatter than a small backlash. Selecting characteristic map data based on the gear data (DG) can make it possible to detect a gearbox anomaly with fewer attempts.

[0138] The vibration VB can be used as a criterion for selecting the characteristic map data. The presence of rats can cause the gearbox 26 to vibrate. Selecting characteristic map data based on the vibration VB can make it possible to detect a gearbox anomaly with fewer tests.

[0139] In the second embodiment, three sets of map data are used, namely the first map data DM (A1) to the third map data DM (A3). However, two sets of map data can also be used. Alternatively, four or more sets of map data can be used. Modification regarding the storage process

[0140] In the embodiments described above, the storage device that stores the determination results is the same as the storage device that stores the characteristic map data DM. However, the present disclosure is not limited thereto.

[0141] Instead of executing the save operation to store the results of determining output variables, a transfer operation can be executed to transmit the results to the vehicle manufacturer (VC) and a data analysis center. Both the save and transfer operations can be executed. Modification regarding the use of the output variables

[0142] In the embodiments described above, the calculation results of the output variables are used to address a situation in which the vehicle VC is taken to a repair shop. However, the present disclosure is not limited to this. For example, if the manufacturer of the vehicle VC can identify an operating point at which rattling occurs, the vehicle VC can be designed to avoid generating the rattling. Similarly, if an operating point can be identified at which screeching is generated, the vehicle VC can also be designed to avoid generating the screeching. Modification regarding the execution device

[0143] The execution device is not limited to a device comprising the CPU 42 and the ROM 44 and performing software processing. For example, at least some of the operations performed by the software in the embodiments described above can be performed by hardware circuits designed to carry out these operations (e.g., an application-specific integrated circuit (ASIC)). This means that the execution device can be modified as long as it comprises one of the following configurations (a) to (c): (a) A configuration comprising a processor that performs all the operations described above according to programs, and a program storage device, such as a ROM, that stores the programs.(b) A setup with a processor and a program storage device that perform some of the operations described above according to the programs, and a dedicated hardware circuit that performs the remaining operations; (c) A setup with a dedicated hardware circuit that performs all of the operations described above. Multiple software processing devices, each comprising a processor and a program storage device, and multiple dedicated hardware circuits may be provided. Modification regarding the vehicle

[0144] In the embodiments described above, the vehicle VC is described, comprising the internal combustion engine 10, the first motor-generator 22, and the second motor-generator 24. The control device 40 can be used in any vehicle that contains a transmission which transmits power by means of gears, and it is possible to determine the state of a gear as in the embodiments described above.

[0145] Various modifications of forms and details can be made in the preceding examples without deviating from the scope and extent of the claims and their equivalents. The examples serve only for descriptive purposes and are not to be understood as limiting. Descriptions of features in each example are to be considered applicable to similar features or aspects in other examples. Suitable results can be obtained by performing processes in a different sequence and / or by combining components differently in a described system, architecture, device, or circuit and / or by replacing or supplementing them with other components or their equivalents. The scope of disclosure is defined not by the detailed explanation but by the claims and their equivalents. All modifications within the scope of the claims and their equivalents are included in the disclosure. 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] JP 201179489 A

[0002] < / korrespondenz>

Claims

[1] An abnormality determination device (40) for a vehicle including a transmission (26) that transmits power using a gear, the abnormality determination device (40) comprising: an execution device (42); and a storage device (46), wherein the storage device (46) is configured to store map data, wherein the map data define a pre-trained map and the pre-trained map has been trained by machine learning, when a variable representing time series data of a speed of the gear is supplied to the map as an input variable, the map outputs a state variable representing a state of the gear as an output variable, and the execution device (42) is designed to carry out the following: an acquisition process that receives the variable representing the time series data as a value of the input variable; and a determination process that determines whether there is an abnormality in the transmission (26) based on a value of the output variable output from the map when the value of the input variable is input to the map. [2] An anomaly determination device (40) according to claim 1, wherein the execution device (42) is configured to execute a feature size calculation process that calculates a feature size obtained by processing the time series data, the determination process includes determining the feature size as the value of the input variable, and the feature size calculation process includes classifying speed values included in the time series data into classes according to a size of the speed and calculating a frequency of each class as the feature size. [3] The anomaly determination device (40) according to claim 2, wherein the feature quantity calculation process includes a process that normalizes the time series data so that a maximum value of the gear speed is 1 and a minimum value of the gear speed is 0. [4] Anomaly determination device (40) according to claim 1, wherein the execution device (42) is configured to execute a feature size calculation process that calculates a feature size obtained by processing the time series data, the determination process includes determining the feature quantity as the value of the input variable, and the feature size calculation process includes calculating a distribution of a frequency component obtained by subjecting the time series data to a fast Fourier transform (FFT) as the feature size. [5] The anomaly determination apparatus (40) according to claim 4, wherein the feature size calculation process comprises: Calculating a primary frequency based on an average value of the rotational speed in the time series data; and Normalizing the frequency component relative to the primary frequency. [6] An anomaly determination device (40) according to claim 4 or 5, wherein the feature size calculation process comprises: Dividing a frequency range into several frequency bands; and Using an average value of an intensity of the frequency component in each frequency band as the intensity of the frequency component in that frequency band. [7] The abnormality determination device (40) according to any one of claims 1 to 6, wherein the time series data is calculated based on a detection signal of a rotational speed sensor (52) that detects a rotational speed of the gear. [8] The abnormality determination device (40) according to any one of claims 1 to 6, wherein the time series data is calculated based on a vehicle speed of the vehicle. [9] The abnormality determination device (40) according to any one of claims 1 to 8, wherein the input variable comprises a variable representing a magnitude of a torque transmitted by the gear. [10] The anomaly determination device (40) according to any one of claims 1 to 9, wherein the input variable comprises a variable representing a temperature of a hydraulic fluid in the transmission (26). [11] The abnormality determination device (40) according to any one of claims 1 to 10, wherein the input variable comprises a variable representing a dimension of the gear based on a specification of the gear. [12] An abnormality determination apparatus according to any one of claims 1 to 11, wherein the input variable comprises a variable representing a meshing error of the gear. [13] An abnormality determination device (40) according to any one of claims 1 to 12, wherein the input variable comprises a variable representing a backlash in engagement of the gear. [14] The abnormality determination device (40) according to any one of claims 1 to 13, wherein the input variable comprises a variable representing a detection value of a vibration sensor (58) that detects vibrations. [15] The abnormality determination device (40) according to any one of claims 1 to 14, wherein the input variable includes a variable representing a detection value of a sound sensor (59) that detects sound. [16] An abnormality determination device (40) according to any one of claims 1 to 15, wherein the input variable comprises a variable representing a fluid pressure in a braking system of the vehicle. [17] An abnormality determination device (40) according to any one of claims 1 to 16, wherein the input variable includes a variable representing a travel distance of the vehicle. [18] The abnormality determination device (40) according to any one of claims 1 to 17, wherein the state variable comprises a variable representing a state in which the gear is damaged. [19] The abnormality determination device (40) according to any one of claims 1 to 18, wherein the state variable comprises a variable representing a state in which a chatter is generated when the gear rotates. [20] The abnormality determination device (40) according to any one of claims 1 to 19, wherein the state variable comprises a variable representing a state in which a screech is generated when the gear rotates. [21] Anomaly determination device (40) according to any one of claims 1 to 17, wherein the map is a first map which outputs as an output variable a state variable indicating whether the gear is damaged when the input variables are input, the map data are first map data that define the first map, the determination process is a first determination process, the storage device (46) is configured to further store second map data and third map data, wherein both the second map data and the third map data each define a pre-trained map, and the pre-trained map has been trained by machine learning, the second map data define a second map, when the input variables are input to it, the second map outputs as an output variable a state variable indicating whether the gearbox is in a state in which a rattle is generated when the gear rotates, the third map data define a third map, when the input variables are input to it, the third map outputs as an output variable a state variable indicating whether the gearbox is in a state in which a screech is generated when the gear rotates, and the execution device (42) is designed to carry out the following: a second determination process that determines whether there is an abnormality in the transmission (26) based on a value of the output variable output from the second map when the value of the input variable is supplied to the second map, and a third determination process that determines whether there is an abnormality in the transmission (26) based on a value of the output variable output from the third map when the value of the input variable is input to the third map.

Images