JP2024069897A - Sensor failure determination device and sensor failure determination method - Google Patents

Abstract

[Problem] To provide a sensor failure determination device and a sensor failure determination method capable of determining a failure of a sensor provided in a tire. [Solution] A sensor failure determination device 40 includes a data acquisition unit 41, a calculation processing unit 42, and a determination unit 43. The data acquisition unit 41 acquires data of physical quantities measured by a sensor 20 attached to a tire 10. The calculation processing unit 42 has an encoding unit that generates feature amount data by performing a convolution operation on the data acquired by the data acquisition unit 41, and a decoding unit that regenerates the data by performing an inverse operation on the feature amount data. The determination unit 43 determines whether or not the sensor 20 has failed based on the data acquired by the data acquisition unit 41 and the data regenerated by the calculation processing unit 42. [Selected Figure] Figure 4

Description

The present invention relates to a sensor failure determination device and a sensor failure determination method.

Recently, research has been conducted into systems that input information measured on tires and vehicles into a learning-based computational model to estimate tire physical information such as tire force.

Patent Document 1 describes a conventional tire physical information estimation system. The tire physical information estimation system includes a physical information estimation unit and a data acquisition unit. The physical information estimation unit has a learning-type computation model extending from an input layer to an output layer in order to estimate physical information related to a tire that is generated by the motion of the tire. The data acquisition unit acquires input data for the input layer. The computation model has a feature extraction unit that extracts feature amounts by performing a convolution computation in the intermediate computation from the input layer to the output layer.

JP 2021-46080 A

In the tire physical information estimation system described in Patent Document 1, tire physical information is estimated based on physical quantities measured by a sensor provided on the tire. The inventors considered it necessary to determine whether the sensor has failed, because if the sensor fails, the tire physical information estimation system will continue to estimate erroneous tire physical information.

The present invention was made in consideration of the above circumstances, and its purpose is to provide a sensor failure determination device and a sensor failure determination method that can determine failure of a sensor installed in a tire.

One aspect of the present invention is a sensor failure determination device. The sensor failure determination device includes a data acquisition unit that acquires data on physical quantities measured by a sensor attached to a tire, an arithmetic processing unit having an encoding unit that generates feature amount data by using a convolution operation on the data acquired by the data acquisition unit and a decoding unit that regenerates data by using an inverse operation on the feature amount data, and a determination unit that determines whether or not the sensor is malfunctioning based on the data acquired by the data acquisition unit and the data regenerated by the arithmetic processing unit.

Another aspect of the present invention is a sensor failure determination method. The sensor failure determination method includes a data acquisition step of acquiring data of a physical quantity measured by a sensor attached to a tire, an encoding process of generating feature data using a convolution operation on the data acquired in the data acquisition step, and a decoding process of regenerating the feature data using an inverse operation, and a determination step of determining whether the sensor is malfunctioning based on the data acquired in the data acquisition step and the data regenerated in the calculation process.

The present invention makes it possible to determine if a sensor installed in a tire is faulty.

1 is a schematic diagram for explaining an overview of a tire physical information estimating system including a sensor failure determination device according to an embodiment; FIG. 2 is a block diagram showing a functional configuration of the tire physical information estimation device. FIG. 2 is a schematic diagram showing a configuration of a computation model. 2 is a block diagram showing a functional configuration of a sensor failure determination device; FIG. FIG. 2 is a schematic diagram showing a configuration of a computation model. 4 is a flowchart showing a procedure of a sensor failure determination process performed by the sensor failure determination device. 4 is a graph showing input data of acceleration measured by a sensor during normal operation. 8 is a graph showing reproduced data calculated by a computation model for the input data of FIG. 7; 4 is a graph showing input data of acceleration measured by a faulty sensor 20. 10 is a graph showing reproduced data calculated by a computation model for the input data of FIG. 9 .

The present invention will be described below based on a preferred embodiment with reference to Figures 1 to 10. The same or equivalent components and parts shown in each drawing are given the same reference numerals, and duplicated explanations will be omitted as appropriate. The dimensions of the parts in each drawing are enlarged or reduced as appropriate to facilitate understanding. Some of the parts that are not important for explaining the embodiment are omitted in each drawing.

(Embodiment)
1 is a schematic diagram for explaining an overview of a tire physical information estimation system 100 including a sensor failure determination device 40 according to an embodiment. The tire physical information estimation system 100 includes a sensor 20 disposed in a tire 10, a tire physical information estimation device 30, and a sensor failure determination device 40. The tire physical information estimation system 100 may also include a server device 80 for acquiring and storing tire physical information, such as tire force F estimated by the tire physical information estimation device 30 and moments about three axes acting on the tire 10, via a communication network 91 and monitoring the tire physical information.

The sensor 20 measures physical quantities of the tire 10, such as the acceleration and strain of the tire 10, the tire air pressure, and the tire temperature, and outputs the measured data to the tire physical information estimation device 30 and the sensor failure determination device 40.

The tire physical information estimation device 30 estimates tire physical information based on data measured by the sensor 20. The tire physical information estimation device 30 uses data measured by the sensor 20 in the calculation to estimate the tire physical information, but may also obtain information from the vehicle side, such as vehicle acceleration, from the vehicle control device 90, etc., and use the information in the calculation to estimate the tire physical information.

The tire physical information estimation device 30 outputs tire physical information such as the estimated tire force F and moments around three axes acting on the tire 10 to, for example, a vehicle control device 90. The vehicle control device 90 uses the tire physical information input from the tire physical information estimation device 30, for example, to estimate braking distance, apply it to vehicle control, and further notify the driver of information related to safe vehicle driving. The vehicle control device 90 can also provide information related to safe vehicle driving in the future using map information, weather information, and the like. Furthermore, when the vehicle control device 90 has a function of automatically driving the vehicle, the tire physical information estimation system 100 provides the estimated tire physical information to the vehicle control device 90 as data to be used for vehicle speed control in automatic driving.

The sensor failure determination device 40 determines whether the sensor 20 is malfunctioning based on the data measured by the sensor 20, and notifies the tire physical information estimation device 30, the server device 80, etc. of the determination result. The sensor failure determination device 40 determines whether the sensor 20 is malfunctioning based on a calculation model that performs encoding and decoding using a convolution operation on the data measured by the sensor 20.

Figure 2 is a block diagram showing the functional configuration of the tire physical information estimation device 30. The sensor 20 has an acceleration sensor 21, a strain gauge 22, a pressure gauge 23, a temperature sensor 24, etc., and measures physical quantities in the tire 10. These sensors measure physical quantities related to the deformation and movement of the tire 10 as physical quantities of the tire 10.

The acceleration sensor 21 and strain gauge 22 move mechanically together with the tire 10, and measure the acceleration and strain amount occurring in the tire 10, respectively. The acceleration sensor 21 is disposed, for example, in the tread, side, bead, wheel, etc. of the tire 10, and measures the acceleration in three axes of the tire 10: the circumferential direction, the axial direction, and the radial direction.

The strain gauges 22 are disposed in the tread, sides, beads, etc. of the tire 10, and measure the strain at the locations where they are disposed. The pressure gauge 23 and temperature sensor 24 are disposed, for example, in the air valve of the tire 10, and measure the tire air pressure and tire temperature, respectively. The temperature sensor 24 may be disposed directly on the tire 10 in order to accurately measure the temperature of the tire 10. The tires 10 may be fitted with, for example, an RFID 11 or the like to which unique identification information is assigned in order to identify each tire.

The tire physical information estimation device 30 has a data acquisition unit 31, a physical information estimation unit 32, and a communication unit 33. The tire physical information estimation device 30 is an information processing device such as a PC (personal computer). Each unit in the tire physical information estimation device 30 can be realized in hardware terms by electronic elements and mechanical parts such as a computer CPU, and in software terms by a computer program, but here, functional blocks realized by the cooperation of these are depicted. Therefore, it will be understood by those skilled in the art that these functional blocks can be realized in various forms by combining hardware and software.

The data acquisition unit 31 acquires information on acceleration, strain, air pressure, and temperature measured by the sensor 20 through wireless communication or the like. The communication unit 33 communicates with external devices such as the vehicle control device 90 and the server device 80 through wired or wireless communication or the like. The communication unit 33 transmits the physical quantities of the tire 10 measured by the sensor 20 and tire physical information estimated about the tire 10, etc., to the external device via a communication line, for example a control area network (CAN), the Internet, etc.

The physical information estimation unit 32 has a calculation model 32a, inputs information from the data acquisition unit 31 into the calculation model 32a, and estimates tire physical information such as tire force F and moments about three axes acting on the tire 10. As shown in FIG. 2, the tire force F has three axial components: a longitudinal force Fx in the longitudinal direction of the tire 10, a lateral force Fy in the lateral direction, and a load Fz in the vertical direction. The physical information estimation unit 32 may calculate all of these three axial components, or may calculate at least one of the components or two components in any combination.

The computation model 32a uses a learning model such as a neural network. FIG. 3 is a schematic diagram showing the configuration of the computation model 32a. The computation model 32a is a CNN (Convolutional Neural Network) type, and is a learning model equipped with the convolution and pooling operations used in its prototype, the so-called LeNet. FIG. 3 shows an example in which acceleration data in three axial directions is used as input data to the computation model 32a, and tire forces in three axial directions are output.

The computational model 32a includes an input layer 50, a feature extraction unit 51, an intermediate layer 52, a full connection unit 53, and an output layer 54. The input layer 50 receives time series data of acceleration in three axial directions acquired by the data acquisition unit 31. The acceleration data is measured in time series by the sensor 20, and data for a certain time interval is extracted using a window function to be used as input data.

The acceleration measured by the tire 10 is periodic for each rotation of the tire 10. The time interval of the input data extracted by the window function may be, for example, a time period corresponding to the rotation period of the tire 10, and the input data itself may be periodic. The window function may extract input data in a time interval shorter or longer than one rotation of the tire 10, and the calculation model 32a can be trained as long as the extracted input data contains at least periodic information.

The feature extraction unit 51 extracts features using a convolution operation, a pooling operation, etc., and transmits them to each node in the intermediate layer 52. The feature extraction unit 51 performs a convolution operation on the input data using multiple filters. The convolution operation is performed while moving the filter on the time-series input data such as acceleration data. The pooling operation is performed on the data after the convolution operation by performing a maximum value pooling operation that selects the larger of two values arranged in a time series, for example. The feature extraction unit 51 repeats the convolution operation, the pooling operation, etc., to extract features.

The full connection unit 53 fully connects data from each node in the intermediate layer 52 at one or more levels, and outputs tire forces Fx, Fy, and Fz to each node in the output layer 54. The full connection unit 53 performs calculations using a full connection path that performs linear calculations using weighting, but in addition to linear calculations, it may also perform nonlinear calculations using activation functions, etc.

Each node of the output layer 54 may output tire physical information such as tire forces in the three axial directions and moments about three axes acting on the tire 10. The output layer 54 may output one type of tire physical information or multiple types of tire physical information in any combination of tire forces in the three axial directions and moments about three axes acting on the tire 10.

The calculation model 32a can learn by mounting a tire 10 with specifications appropriate to the vehicle on an actual vehicle and running the vehicle on a test run. The specifications of the tire 10 include information related to the performance of the tire, such as tire size, tire width, aspect ratio, tire strength, tire outer diameter, load index, and date of manufacture.

The server device 80 acquires tire physical information, such as the physical quantities of the tire 10 measured by the sensor 20, as well as tire force F estimated for the tire 10 and moments around three axes acting on the tire 10, from the tire physical information estimation device 30. The server device 80 may accumulate the physical quantities measured for the tire 10 and the tire physical information estimated by the tire physical information estimation device 30 from multiple vehicles.

Figure 4 is a block diagram showing the functional configuration of the sensor failure determination device 40. As described above, the sensor 20 has an acceleration sensor 21, a strain gauge 22, a pressure gauge 23, a temperature sensor 24, etc., and measures physical quantities in the tire 10.

The sensor failure determination device 40 has a data acquisition unit 41, a calculation processing unit 42, a determination unit 43, and a communication unit 44. The sensor failure determination device 40 is an information processing device such as a PC (personal computer). Each unit in the sensor failure determination device 40 can be realized in hardware terms by electronic elements and mechanical parts such as a computer CPU, and in software terms by a computer program, but here, functional blocks realized by the cooperation of these are depicted. Therefore, it will be understood by those skilled in the art that these functional blocks can be realized in various forms by combining hardware and software.

The data acquisition unit 41 acquires data on acceleration, strain, air pressure, and temperature measured by the sensor 20 via wireless communication or the like, and outputs the data to the pre-processing unit 42a of the calculation processing unit 42. The communication unit 44 communicates with the tire physical information estimation device 30 and external devices such as the server device 80 via wired or wireless communication or the like. The communication unit 44 transmits the determination result as to whether or not the sensor 20 is malfunctioning to the external device.

The calculation processing unit 42 has a pre-processing unit 42a and a calculation model 42b. The pre-processing unit 42a normalizes the data input from the data acquisition unit 41 and outputs the normalized data to the calculation model 42b. The normalization in the pre-processing unit 42a will be described using an example in which the input data is acceleration data Tx, Ty, and Tz in three axial directions. The acceleration data Tx, Ty, and Tz are the acceleration in the circumferential direction, axial direction (axle direction), and radial direction of the tire 10, respectively.

For example, if the acceleration data Tx generated while the vehicle is traveling is greater than or equal to -0.2 G and less than or equal to 0.2 G, the pre-processing unit 42a divides the acceleration data Tx by a constant value of 0.2 and outputs normalized acceleration data Txn. If the acceleration data Ty generated while the vehicle is traveling is greater than or equal to -0.1 G and less than or equal to 0.1 G, the pre-processing unit 42a divides the acceleration data Ty by a constant value of 0.1 and outputs normalized acceleration data Tyn.

When the acceleration data Tz generated while the vehicle is traveling is greater than or equal to -0.8 G and less than or equal to 1 G, the pre-processing unit 42a subtracts the median value 0.1 from the acceleration data Tz, divides it by a constant value 0.9, and outputs normalized acceleration data Tzn. Through these processes, the pre-processing unit 42a outputs the normalized acceleration data Txn, Tyn, and Tzn as values in the range from greater than or equal to -1 to less than 1.

Normalization of physical quantities measured in the tire 10, such as acceleration and strain in three axial directions, is not limited to the above example, and can be set appropriately depending on the properties of the physical quantities measured in the tire 10, the range of possible values, etc.

The pre-processing unit 42a does not need to perform normalization processing, for example, when only acceleration data for one of the three axes is input data to the calculation model 42b. The pre-processing unit 42a performs normalization processing when physical quantities are measured in at least two of the three axial directions of the tire 10 and input to the calculation model 42b.

The computation model 42b uses an autoencoder type learning model that uses convolution operations and the like. FIG. 5 is a schematic diagram showing the configuration of the computation model 42b. The computation model 42b receives normalized acceleration data in three axial directions, for example, generates feature data using convolution operations, and restores the input data by inverse operations.

The computational model 42b includes an input layer 60, an encoding unit 61, a decoding unit 62, and an output layer 63. The input layer 60 receives normalized acceleration time series data in three axial directions output from the preprocessing unit 42a. The acceleration data is measured in a time series manner by the sensor 20, and data for a certain time interval is extracted using a window function to be used as input data.

The acceleration measured by the tire 10 is periodic for each rotation of the tire 10. The time interval of the input data extracted by the window function may be, for example, a time period corresponding to the rotation period of the tire 10, and the input data itself may be periodic. Note that the window function may extract input data in a time interval shorter or longer than one rotation of the tire 10, as long as the extracted input data at least contains periodic information.

The encoding unit 61 generates feature data using a convolution operation, a pooling operation, and the like. The encoding unit 61 performs a convolution operation on input data using multiple filters. The convolution operation is performed while moving a filter on time-series input data such as acceleration data. The pooling operation is performed on the data after the convolution operation by performing a maximum value pooling operation that selects the larger of two values arranged in time series, for example. The encoding unit 61 repeats the convolution operation, the pooling operation, and the like to generate feature data.

The decoding unit 62 reproduces the feature data generated by the encoding unit 61 using the inverse operation of the encoding unit 61, and outputs the data to the output layer 63. The inverse operation performed by the decoding unit 62 is called a transpose convolution operation or a deconvolution operation.

In the encoding unit 61, the input data is downsampled to generate feature data, whereas the decoding unit 62 performs the reverse operation of upsampling the feature data to reproduce the input data. The decoding unit 62 may use an up convolution technique as an operation equivalent to the transpose convolution operation.

The computation model 42b is capable of so-called unsupervised learning by training the data input to the input layer 60 to be similar to the data output from the output layer 63. The computation model 42b is trained in a state in which the sensor 20, such as the acceleration sensor 21, is operating normally. If the sensor 20 breaks down due to excessive vibration, for example, the data input to the input layer 60 will differ from the data during normal operation, and the trained computation model 42b will reproduce and output data with an increased error compared to the input data.

Returning to FIG. 4, the determination unit 43 determines whether the sensor 20 is malfunctioning based on the input data to the input layer 60 of the computation model 42b and the reproduced data output from the output layer 63. The determination unit 43 determines that the sensor 20 is operating normally when the sum of the absolute values of the errors between the input data to the input layer 60 and the reproduced data output from the output layer 63 is less than a predetermined threshold. The determination unit 43 determines that the sensor 20 is malfunctioning when the sum of the absolute values of the errors between the input data to the input layer 60 and the reproduced data output from the output layer 63 is equal to or greater than a predetermined threshold.

For example, when the input data is acceleration data in three axial directions of the tire 10, the determination unit 43 may calculate the sum of the absolute values of the errors between the input data to the input layer 60 and the reproduced data output from the output layer 63 for each axial direction, and make a determination for each axis. The determination unit 43 may also calculate the sum of the absolute values of the errors between the input data to the input layer 60 and the reproduced data output from the output layer 63 for each axial direction, and further add up all the calculated sums for each axis to make a single overall determination. When the input data is acceleration data in two of the three axial directions of the tire 10, the determination unit 43 may make a similar determination for each axis, or may make a single overall determination.

The server device 80 may acquire and store the judgment result of whether the sensor 20 is malfunctioning from the sensor malfunction judgment device 40. The tire physical information estimation device 30 acquires the judgment result of whether the sensor 20 is malfunctioning, and if the sensor 20 is malfunctioning, stops the tire physical information estimation process, or switches to an alternative estimation process if there is one. The tire physical information estimation device 30 may output the sensor 20 malfunction to the vehicle control device 90, and the vehicle control device 90 may notify the driver of the sensor 20 malfunction.

Next, the operation of the sensor failure determination device 40 will be described. FIG. 6 is a flowchart showing the procedure of the sensor failure determination process performed by the sensor failure determination device 40. The sensor failure determination device 40 acquires data on physical quantities such as acceleration, strain, tire pressure, and tire temperature of the tire 10 measured by the sensor 20 using the data acquisition unit 41 (S1).

The computation processing unit 42 executes a process of extracting input data for a certain time interval from the data acquired by the data acquisition unit 41 (S2). The preprocessing unit 42a executes a normalization process on the input data (S3). The normalized input data is input to the input layer 60 of the computation model 42b.

The encoding unit 61 of the computation model 42b executes a process of generating feature data by performing a convolution operation and a pooling operation on the input data (S4). The decoding unit 62 of the computation model 42b executes an inverse operation of the encoding on the feature data generated by the encoding unit 61 to reproduce the data (S5).

The determination unit 43 determines whether the sensor 20 is faulty based on the input data to the computation model 42b and the regenerated data output by the decoding unit 62 (S6), and ends the process.

Figure 7 is a graph showing the input data of acceleration measured by the sensor 20 during normal operation, and Figure 8 is a graph showing the reproduced data calculated by the calculation model 42b for the input data of Figure 7. The input data shown in Figure 7 is normalized acceleration data Tx, Ty, and Tz in the three axial directions. In the calculation model 42b, feature data is generated by a convolution calculation in the encoding unit 61, and the data is reproduced by an inverse calculation. Therefore, as shown in Figure 8, when the input data is normal, the characteristics of the waveform fluctuation in each input data are reflected in the reproduced data.

Figure 9 is a graph showing the input data of acceleration measured by the faulty sensor 20, and Figure 10 is a graph showing the regenerated data calculated by the computation model 42b for the input data of Figure 9. The input data shown in Figure 9 is normalized acceleration data Tx, Ty, and Tz in the three axial directions. The regenerated data output by the computation model 42b for the input data from the faulty sensor 20 exhibits waveform fluctuations that are different from the characteristics of the input data, and has increased errors compared to the input data.

Rc is the normal range of the value obtained by adding the absolute value of the error in the input data from the sensor 20 during normal operation and the regenerated data calculated by the calculation model 42b. Re is the abnormal range of the value obtained by adding the absolute value of the error in the input data from the faulty sensor 20 and the regenerated data calculated by the calculation model 42b. The determination unit 43 sets a threshold D between the normal range Rc and the abnormal range Re to determine whether the sensor 20 is faulty. Even if the upper limit of the normal range Rc and the lower limit of the abnormal range Re partially overlap, the threshold D can be set taking into account the distribution of the normal range Rc and the abnormal range Re.

The sensor failure determination device 40 encodes the input data in the calculation model 42b using a convolution operation and regenerates the data using an inverse operation, and can determine whether the sensor 20 is faulty or not by making a judgment based on the input data and the regenerated data.

The calculation processing unit 42 normalizes the data acquired by the data acquisition unit 41 in the pre-processing unit 42a, thereby making it possible to equalize the influence of the data in each axial direction on the fault judgment by the judgment unit 43.

The sensor 20 may measure physical quantities in at least two of the three axial directions of the tire 10, and the determination unit 43 may determine a fault for each axial direction. The sensor fault determination device 40 may provide information that a portion of the sensor 20 is in a faulty state. For example, even if a portion of the sensor 20 is in a faulty state, the tire physical information estimation device 30 may continue to estimate tire physical information based on data acquired from a sensor 20 that is operating normally.

The sensor 20 may also measure physical quantities in at least two of the three axial directions in the tire 10, and the determination unit 43 may make a single overall determination for the data in each axial direction. The sensor failure determination device 40 may determine whether the input data in each axial direction to the calculation model 42b is in a state suitable for extracting feature quantities overall.

Next, features of the sensor failure determination device 40 and the sensor failure determination method according to the embodiment will be described.
The sensor failure determination device 40 according to the embodiment includes a data acquisition unit 41, a calculation processing unit 42, and a determination unit 43. The data acquisition unit 41 acquires data of physical quantities measured by the sensor 20 attached to the tire 10. The calculation processing unit 42 has an encoding unit 61 that generates feature amount data by performing a convolution operation on the data acquired by the data acquisition unit 41, and a decoding unit 62 that reproduces the data by performing an inverse operation on the feature amount data. The determination unit 43 determines whether or not the sensor 20 has a failure, based on the data acquired by the data acquisition unit 41 and the data reproduced by the calculation processing unit 42. This allows the sensor failure determination device 40 to determine a failure of the sensor 20 provided in the tire 10.

The calculation processing unit 42 also includes a pre-processing unit 42a that normalizes the data acquired by the data acquisition unit 41. This allows the sensor failure determination device 40 to equalize the influence of data in each axial direction on failure determination by the determination unit 43.

The sensor 20 also measures physical quantities in at least two of the three axial directions in the tire 10. The determination unit 43 determines whether the sensor 20 is malfunctioning for each axial direction. This allows the sensor malfunction determination device 40 to provide information that a portion of the sensor 20 is in a malfunctioning state.

The determination unit 43 also comprehensively determines whether the sensor 20 is faulty or not for the data in each axial direction. This allows the sensor fault determination device 40 to determine whether the input data in each axial direction to the calculation model 42b is in a state suitable for extracting feature quantities overall.

The sensor failure determination method includes a data acquisition step, a calculation processing step, and a determination step. The data acquisition step acquires data of physical quantities measured by the sensor 20 attached to the tire 10. The calculation processing step performs an encoding process that generates feature data using a convolution operation on the data acquired in the data acquisition step, and a decoding process that reproduces the feature data using an inverse operation. The determination step determines whether the sensor 20 is malfunctioning or not based on the data acquired in the data acquisition step and the data reproduced in the calculation processing step. According to this sensor failure determination method, it is possible to determine whether the sensor 20 provided in the tire 10 is malfunctioning.

The above describes the embodiments of the present invention. These embodiments are merely examples, and it will be understood by those skilled in the art that various modifications and changes are possible within the scope of the claims of the present invention, and that such modifications and changes are also within the scope of the claims of the present invention. Therefore, the descriptions and drawings in this specification should be treated as illustrative rather than restrictive.

10 Tire, 20 Sensor, 40 Sensor failure determination device,
41 data acquisition unit, 42 calculation processing unit,
42a pre-processing unit, 42b computation model, 43 determination unit,
61 encoding unit, 62 decoding unit.

Claims (5)

a data acquisition unit that acquires data of physical quantities measured by sensors attached to the tires;
a processor having an encoding unit that generates feature data by performing a convolution operation on the data acquired by the data acquisition unit, and a decoder that reproduces data by performing an inverse operation on the feature data;
a determination unit that determines whether or not the sensor is malfunctioning based on the data acquired by the data acquisition unit and the data reproduced by the arithmetic processing unit;
A sensor failure determination device comprising: The sensor failure determination device according to claim 1, characterized in that the calculation processing unit includes a pre-processing unit that normalizes the data acquired by the data acquisition unit. The sensor measures physical quantities in at least two of three axial directions of the tire,
3. The sensor failure determination device according to claim 1, wherein the determination unit performs a determination for each axial direction. The sensor measures physical quantities in at least two of three axial directions of the tire,
3. The sensor failure determination device according to claim 1, wherein the determination unit performs a comprehensive determination on data in each axial direction. A data acquisition step of acquiring data of physical quantities measured by sensors attached to the tires;
an encoding process for generating feature data by using a convolution operation on the data acquired in the data acquisition step, and a decoding process for reproducing the data by using an inverse operation on the feature data;
a determination step of determining whether or not the sensor is malfunctioning based on the data acquired in the data acquisition step and the data reproduced in the arithmetic processing step;
A sensor failure determination method comprising:

Images