Anomaly detection device and method for steering motor, and computer readable storage medium

Abstract

The invention discloses a steering motor anomaly detection device and method and a computer readable storage medium. A steering motor abnormality detection device that detects an abnormality of a steering motor that is disposed in a steer-by-wire system of a vehicle and that provides a reaction force with respect to an operation of a steering wheel includes: a memory that stores one or more instructions; and a processor executing one or more instructions, the processor inputting an input value related to driving of the rack to the artificial neural network model by executing the one or more instructions, and acquiring one or more estimated values related to steering output by the artificial neural network model, and a rack that receives a driving force from a wheel actuator that is driven in accordance with the operation of the steering wheel and moves wheels of the vehicle, and that detects whether or not an abnormality has occurred in a steering motor by comparing the one or more estimated values with one or more measured values relating to steering.

Description

Technical Field

[0001] The present invention relates to a steering motor abnormality detection device, method and a computer-readable storage medium storing a program for executing the method, and more specifically, to a steering motor abnormality detection device capable of detecting performance degradation of a steering motor, a steering motor abnormality detection method and a non-temporary computer-readable storage medium storing a program for executing the steering motor abnormality detection method, wherein the steering motor is configured in a vehicle's steer-by-wire system and provides a reaction force to the operation of a steering wheel. Background Art

[0002] The steer-by-wire system is a steering system that transmits the driver's steering intention to the vehicle's wheels through electrical signals without the need for a mechanical connection between the vehicle's steering wheel and the vehicle's wheels. The steer-by-wire system minimizes mechanical connections to prevent unnecessary vibrations and improves steering accuracy and responsiveness. In addition, the steer-by-wire system provides a high degree of freedom in layout, making it easy to share parts.

[0003] The steer-by-wire system includes a steering feeling actuator (SFA) that provides a steering wheel reaction force to the driver of the vehicle, and a road wheel actuator (RWA) that transmits the driver's steering intention to the vehicle's wheels, thereby causing the wheels to rotate. Usually, the SFA and RWA are composed of motors.

[0004] The motor constituting the SFA may be a permanent magnet synchronous motor (PMSM). The output torque of the PMSM gradually decreases due to demagnetization caused by degradation and other reasons.

[0005] When the output torque of the SFA decreases, the driver will feel an unusual driving feeling that is different from his or her steering intention. This unusual feedback may lead to an accident caused by steering error. Therefore, in order to improve the safety of the steer-by-wire system, it is necessary to develop a technology that can predict (prognostics) motor abnormalities in advance rather than diagnose them after they occur.

[0006] Prior art literature

[0007] Patent Literature

[0008] Korean Patent No. 1962647 "Steering-by-wire system and control method thereof", granted on March 21, 2019 Summary of the invention

[0009] Problem that the invention aims to solve

[0010] The present invention is used to solve the above-mentioned problem. The purpose of the present invention is to provide a steering motor abnormality detection device capable of detecting the performance degradation of a steering motor, a steering motor abnormality detection method and a non-temporary computer-readable storage medium storing a program for executing the steering motor abnormality detection method. The steering motor is configured in a vehicle's steer-by-wire system and provides a reaction force to the operation of the steering wheel.

[0011] Another object of the present invention is to provide a steering motor abnormality detection device, a steering motor abnormality detection method and a non-temporary computer-readable storage medium storing a program for executing the steering motor abnormality detection method, which can effectively detect the output torque reduction of the steering motor in advance even without using additional sensors other than the sensors configured in the vehicle and in the existing vehicle.

[0012] The technical problems of the present invention are not limited to the technical problems mentioned above, and ordinary technicians in the technical field to which the present invention belongs can clearly understand other technical problems not mentioned through the following description.

[0013] Means used to solve problems

[0014] According to one aspect of the present invention, a steering motor abnormality detection device can be provided, which is used to detect abnormalities in a steering motor, wherein the steering motor is configured in a steer-by-wire system of a vehicle and provides a reaction force to a steering wheel operation, and includes: a memory for storing one or more instructions; and a processor for executing the one or more instructions, wherein the processor inputs input values ​​related to the drive of a rack into an artificial neural network model by executing the one or more instructions, and obtains one or more estimated values ​​related to steering output by the artificial neural network model, and detects whether an abnormality has occurred in the steering motor by comparing one or more of the estimated values ​​with one or more measured values ​​associated with steering, wherein the rack receives a driving force from a wheel actuator driven in response to the operation of the steering wheel, so that the wheels of the vehicle move.

[0015] According to the steering motor abnormality detection device according to one aspect of the present invention, the input value may be a propulsion force of the rack.

[0016] According to a steering motor abnormality detection device according to one aspect of the present invention, the one or more estimated values ​​include a first estimated value for a first residual, wherein the first residual is a difference between an instruction and a measurement value for a steering torque, wherein the steering torque is a steering torque transmitted to a steering wheel of the vehicle corresponding to the input value, and the one or more measured values ​​include a first measured value which is a measured value of the first residual.

[0017] According to a steering motor abnormality detection device according to one aspect of the present invention, the one or more estimated values ​​also include a second estimated value for a second residual, wherein the second residual is a difference between an instruction and a measured value for a motor torque, wherein the motor torque is a motor torque transmitted to the steering motor corresponding to the input value, and the one or more measured values ​​also include a second measured value which is a measured value of the second residual.

[0018] According to the steering motor abnormality detection device according to one aspect of the present invention, the input value may be acquired through a controller area network (CAN) of the vehicle.

[0019] According to a steering motor abnormality detection device according to one aspect of the present invention, the artificial neural network model can be composed of a generative adversarial network (GAN), and the generative adversarial network includes: a generator, which receives the input value and generates one or more estimated values; and a discriminator, which receives the input value and measured data including one or more measured values ​​and outputs a discrimination value for the measured data.

[0020] According to the steering motor abnormality detection device according to one aspect of the present invention, the generator may be composed of a multivariate transformer.

[0021] According to a steering motor abnormality detection device according to one aspect of the present invention, the processor can input error data into an abnormality detection model and determine whether an abnormality occurs in the steering motor by executing more than one of the instructions, and the error data is associated with the difference between the one or more estimated values ​​and the one or more measured values.

[0022] According to the steering motor abnormality detection device according to one aspect of the present invention, the abnormality detection model may use a One-Class Support Vector Machine (OCSVM) algorithm.

[0023] According to a steering motor abnormality detection device according to one aspect of the present invention, there are multiple data sets, each of which includes the input value, one or more estimated values ​​and one or more measured values, and the error data may include: the average value and standard deviation of the error between the measured value and the estimated value obtained from each of the multiple data sets; the maximum absolute error between the measured value and the estimated value obtained from each of the multiple data sets; and the identification value of the identifier for the measured data obtained from each of the multiple data sets.

[0024] According to the steering motor abnormality detection device of one aspect of the present invention, the discriminator may additionally receive the input value and estimation data including one or more of the estimation values, and additionally output a discrimination value for the estimation data.

[0025] According to a steering motor abnormality detection device according to one aspect of the present invention, the artificial neural network model can be constructed by the generator and the discriminator alternately performing learning with each other, and the input value and the measured value used when performing the learning can be obtained when the vehicle and the steering motor are in a normal state.

[0026] According to another aspect of the present invention, a steering motor abnormality detection method is provided, which is used to detect abnormalities in a steering motor, wherein the steering motor is arranged in a steer-by-wire system of a vehicle and provides a reaction force to the operation of a steering wheel, and comprises: a step in which a processor inputs input values ​​related to the drive of a rack into an artificial neural network model, and obtains one or more estimated values ​​related to steering output by the artificial neural network model, wherein the rack receives a driving force from a wheel actuator driven corresponding to the operation of the steering wheel, so that the wheels of the vehicle move; and a step in which the processor detects whether an abnormality has occurred in the steering motor by comparing one or more of the estimated values ​​with one or more measured values ​​related to steering.

[0027] According to the steering motor abnormality detection method according to one aspect of the present invention, the input value may be a propulsion force of the rack.

[0028] According to a steering motor abnormality detection method according to one aspect of the present invention, the one or more estimated values ​​include a first estimated value for a first residual, wherein the first residual is a difference between a command and a measurement value for a steering torque, wherein the steering torque is a steering torque transmitted to a steering wheel of the vehicle corresponding to the input value, and the one or more measured values ​​include a first measured value which is a measured value of the first residual.

[0029] According to a steering motor abnormality detection method according to one aspect of the present invention, the one or more estimated values ​​also include a second estimated value for a second residual, wherein the second residual is a difference between an instruction and a measured value for a motor torque, wherein the motor torque is the motor torque transmitted to the steering motor corresponding to the input value, and the one or more measured values ​​also include a second measured value which is a measured value of the second residual.

[0030] According to a steering motor abnormality detection method according to one aspect of the present invention, the artificial neural network model can be composed of a generative adversarial network (GAN), which includes: a generator, which receives the input value and generates one or more estimated values; and a discriminator, which receives the input value and measured data including one or more measured values ​​and outputs a discrimination value for the measured data.

[0031] According to a steering motor abnormality detection method according to one aspect of the present invention, the step of detecting whether an abnormality occurs in the steering motor may include: the processor inputs the measured data into the discriminator, and obtains an identification value for the measured data generated by the discriminator; and the processor inputs the identification value for the measured data and error data including a numerical value related to the difference between one or more of the estimated values ​​and one or more of the measured values ​​into an abnormality detection model and obtains the output of the abnormality detection model.

[0032] According to a steering motor abnormality detection method according to one aspect of the present invention, there are multiple data sets, the data sets include the input value, one or more estimated values ​​and one or more measured values, and the error data may include: the average value and standard deviation of the error between the measured value and the estimated value obtained from each of the multiple data sets; the maximum absolute error between the measured value and the estimated value obtained from each of the multiple data sets; and the identification value of the identifier for the measured data obtained from each of the multiple data sets.

[0033] According to still another aspect of the present invention, there is provided a non-transitory computer-readable storage medium storing a program, the program including at least one instruction for executing the steering motor abnormality detection method.

[0034] Effects of the Invention

[0035] According to the above configuration, a steering motor abnormality detection device, a method, and a non-transitory computer-readable storage medium storing a program for executing the method according to one aspect of the present invention can predict the performance degradation of a steering motor in advance before a steering motor fails based on an artificial intelligence-based digital twin algorithm. The steering motor is configured in a vehicle's steer-by-wire system and provides a reaction force to the operation of the steering wheel.

[0036] In addition, according to one aspect of the present invention, a steering motor abnormality detection device, method and a non-temporary computer-readable storage medium storing a program for executing the method can use the vehicle's controller area network (CAN) signal to predict the performance degradation of the steering motor without the need for additional separate sensors.

[0037] The effects of the present invention are not limited to the above-mentioned effects, and should be understood to include all effects that can be derived from the constitution of the invention described in the detailed description of the present invention or the appended claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 The diagram schematically shows the configuration of a steer-by-wire system of a vehicle.

[0039] Figure 2 The diagram shows the detailed configuration and operation of a steering motor controller of a steer-by-wire system of a vehicle.

[0040] Figure 3 : is a diagram showing a configuration of a steering motor abnormality detection device according to one embodiment of the present invention.

[0041] Figure 4 : is a diagram showing a model for acquiring an estimated value or for detecting abnormality in a steering motor abnormality detection device according to an embodiment of the present invention.

[0042] Figure 5 is a diagram showing a detailed configuration of an artificial neural network model.

[0043] Figure 6 is a diagram showing a detailed configuration of a generator of an artificial neural network model.

[0044] Figure 7 1 is a diagram showing the operation of the steering motor abnormality detection device according to one embodiment of the present invention.

[0045] Figure 8 is a sequence diagram of a steering motor abnormality detection method according to an embodiment of the present invention.

[0046] Fig. 9 is a detailed sequence diagram of steps for detecting whether an abnormality occurs in a steering motor in a steering motor abnormality detection method according to an embodiment of the present invention.

[0047] Description of Reference Numerals

[0048] 100: Steering motor abnormality detection device

[0049] 110: Memory

[0050] 120: Processor DETAILED DESCRIPTION

[0051] Hereinafter, embodiments of the present invention are described in detail with reference to the accompanying drawings so that a person skilled in the art can easily implement the present invention. The present invention can be implemented in various forms and is not limited to the embodiments described herein. In order to clearly explain the present invention, parts not related to the description are omitted in the accompanying drawings, and the same reference numerals are given to the same or similar components throughout the specification.

[0052] The words and terms used in this specification and the appended claims should not be interpreted as being limited to their usual or dictionary meanings, but in order to describe one's invention in the best way, they should be interpreted as meanings and concepts that are consistent with the technical ideas of the present invention in accordance with the principle that the inventor is able to define terms and concepts.

[0053] In this specification, the terms "include" or "have" are used to illustrate the existence of features, numbers, steps, actions, constituent elements, parts or their combinations recorded in the specification, rather than excluding the existence or additional possibilities of one or more other features, numbers, steps, actions, constituent elements, parts or their combinations in advance.

[0054] Figure 1 The diagram schematically shows the configuration of a steer-by-wire system of a vehicle.

[0055] The steer-by-wire system 200 is a steering system that transmits the driver's steering intention to the vehicle's wheels W through electrical signals without mechanical connection between the vehicle's steering wheel 210 and the vehicle's wheels W. In the steer-by-wire system, the steering column 220 combined with the steering wheel 210 does not have a mechanical connection structure with the vehicle's wheels W.

[0056] When the driver of the vehicle operates the steering wheel 210, the steering sensor 230 senses any one or more of the steering torque and the steering angular velocity. In other words, the steering sensor 230 may include any one or more of the torque sensor and the steering angle sensor.

[0057] In addition, the steer-by-wire system 200 includes: a steering motor 260, which provides a reaction force corresponding to the operation of the steering wheel 210 to the driver; and a wheel actuating motor 280, which is used to move the wheel W. The steering motor 260 provides a steering feeling to the driver. In addition, the wheel actuating motor 280 is used to move the rack 290 connected to the wheel W. The rack 290 is moved by the drive of the wheel actuating motor 280, so that the steering of the vehicle can be achieved.

[0058] In addition, the sensing information of the steering sensor 230 is transmitted to the electronic control unit 240. Based on the sensing information of the steering sensor 230, the electronic control unit 240 may send control signals to the steering motor controller 250 for controlling the steering motor 260 and the wheel actuation motor controller 270 for controlling the wheel actuation motor 280.

[0059] Figure 2 The diagram shows the detailed configuration and operation of a steering motor controller of a steer-by-wire system of a vehicle.

[0060] Reference Figure 2 The steering motor controller 250 may include a command generator 251 , a steering torque controller 252 , a steering motor torque controller 253 and a steering motor current controller 254 .

[0061] The command generator 251 generates a command steering torque T corresponding to the input value. s，d , the command steering torque T s，d is the torque that needs to be transmitted to the steering wheel 210. At this time, the input value may be the propulsion force of the rack 290.

[0062] The steering torque controller 252 receives the command steering torque T s，d and the steering torque T measured by the steering wheel 210 s，m The difference between the two generates the command steering motor torque T d , the command steering motor torque T d is the torque that the steering motor 260 needs to generate. For example, the steering torque T s，m The measurement may be performed by the steering sensor 230 .

[0063] Command steering torque T s，d and measuring the steering torque T s，m The difference is the difference between the command and the measured value for the steering torque, which can be defined as the first residual. In other words, the steering torque controller 252 can generate the command steering motor torque T by receiving the first residual. d .

[0064] The steering motor torque controller 253 receives the command steering motor torque T d The measured steering motor torque T m The difference between the two generates the command current I d , the command current I d is the current that needs to be supplied to the steering motor 260 .

[0065] Command steering motor torque T d and measure the steering motor torque T m The difference between the command and the measured value of the motor torque can be defined as the second residual. In other words, the steering motor torque controller 253 can generate the command current I by receiving the second residual. d .

[0066] The steering motor current controller 254 receives the command current I d and the measured current I of the motor m The difference between the current supplied to the steering motor 260 is thereby controlled. Based on this, the voltage V input to the steering motor 260 can be determined. d In addition, the measured current I m The angle θ of the steering motor 260 can be m (position) can be calculated. In addition, the measured current I of the steering motor 260 can be used to calculate the position of the steering motor 260. m Estimate the measured steering motor torque T m .

[0067] Figure 3 : is a diagram showing the configuration of a steering motor abnormality detection device according to one embodiment of the present invention. Figure 4 : is a diagram showing a model for acquiring an estimated value or for detecting abnormality in a steering motor abnormality detection device according to an embodiment of the present invention.

[0068] A steering motor abnormality detection device 100 according to an embodiment of the present invention is used to detect an abnormality of a steering motor 260, which is provided in a steer-by-wire system 200 of a vehicle and is used to provide a reaction force to an operation of a steering wheel 210. More specifically, the steering motor abnormality detection device 100 can detect a decrease in output before the steering motor 260 completely fails.

[0069] According to an embodiment of the present invention, the steering motor abnormality detection device 100 can predict the performance degradation of the steering motor 260 before the steering motor 260 fails by using an artificial intelligence-based digital twin algorithm. The digital twin algorithm can virtually construct a vehicle's steer-by-wire system.

[0070] Reference Figure 3 According to an embodiment of the present invention, a steering motor abnormality detection device 100 may include a memory 110 and a processor 120 .

[0071] The memory 110 stores one or more instructions that can be executed by the processor 120 .

[0072] The memory 110 may include a hardware device configured to store and execute program instructions. The memory 110 may include storage media such as a read-only memory (ROM), a random access memory (RAM), a flash memory, etc. In addition, the memory 110 may also include magnetic media such as floppy disks and magnetic tapes, optical recording media such as compact disk read only memory (CD-ROM), digital video disks (DVD), magneto-optical media such as floppy disks, etc.

[0073] The processor 120 executes the one or more instructions. For example, the processor 120 may be a hardware unit that performs calculations and controls in a computer. The processor 120 may include one or more arithmetic logic units (ALUs) and registers.

[0074] The processor 120 inputs the input value associated with the driving of the rack 290 into the artificial neural network model 20 by executing the one or more instructions, and obtains one or more estimated values ​​associated with the steering output by the artificial neural network model 20, wherein the rack 290 receives the driving force from the wheel actuator driven corresponding to the operation of the steering wheel 210, thereby moving the wheel W of the vehicle. The wheel actuator may be the wheel actuating motor 280 described above.

[0075] The input value may be the propulsion force of the rack 290. The propulsion force of the rack 290 may be obtained from a Controller Area Network (CAN) 300 of the vehicle. For example, the wheel actuating motor controller 270 may receive feedback of the propulsion force of the rack 290 and send the propulsion force information of the rack through the CAN 300.

[0076] As described above, the input value may be acquired through the vehicle's CAN 300. Thus, according to the present invention, there is no need to use an additional sensor in connection with abnormality detection of the steering motor 260.

[0077] In this regard, the memory 110 may be directly or indirectly connected to the controller area network 300 of the vehicle. The memory 110 may receive the input value from the controller area network 300. Therefore, the processor 120 may obtain the input value from the memory 110.

[0078] The one or more estimated values ​​may include a first estimated value for a first residual, the first residual being a difference between a command and a measured value for a steering torque, the steering torque being a steering torque transmitted to the steering wheel 210 corresponding to the input value. In addition, the one or more estimated values ​​may further include a second estimated value for a second residual, the second residual being a difference between a command and a measured value for a motor torque, the motor torque being a motor torque transmitted to the steering motor 260 corresponding to the input value.

[0079] The artificial neural network model 20 receives the input value and outputs one or more estimated values. The artificial neural network model 20 receives the input value and can output a first estimated value and a second estimated value.

[0080] The one or more estimated values ​​are obtained for comparison with one or more measured values ​​related to steering. At this time, the one or more measured values ​​may include: a first measured value as a measured value of the first residual; and a second measured value as a measured value of the second residual.

[0081] In one embodiment of the present invention, the artificial neural network model 20 may be composed of a generative adversarial network (GAN).

[0082] In one embodiment of the present invention, the artificial neural network model 20 may be a deep learning based neural twin model for the vehicle's steer-by-wire system 200. In other words, the artificial neural network model 20 may function as a virtual twin model of the vehicle's steer-by-wire system.

[0083] Figure 5 is a diagram showing a detailed configuration of an artificial neural network model.

[0084] Reference Figure 5 , the artificial neural network model 20 may include a generator 21 and a discriminator 22.

[0085] The artificial neural network model 20 can be constructed by alternately performing learning with the generator 21 and the discriminator 22, and the data used in performing the learning can be data obtained when the vehicle and the steer-by-wire system are in a normal state. Thus, one or more of the estimated values ​​can follow the measured values ​​of the first residual and the measured values ​​of the second residual obtained when the vehicle and the steer-by-wire system are in a normal state.

[0086] The generator 21 receives the input value and generates one or more estimated values. The generator 21 may include a neural network. The generator 21 may be composed of a multivariate transformer. That is, the artificial neural network model 20 may be a generative adversarial network (GAN) based on a multivariate transformer.

[0087] In one embodiment of the present invention, the generator 21 may receive an input value associated with the driving of a rack, the rack receiving a driving force from a wheel actuator driven in response to the operation of a steering wheel, thereby moving the wheels of the vehicle. As described above, the input value may be a propulsion force of the rack.

[0088] The generator 21 may calculate the input value and output more than one estimated value. More specifically, the generator 21 may output an estimated value of the first residual and an estimated value of the second residual.

[0089] When the steer-by-wire system 200 of the vehicle is in a normal state, the estimated value estimated by the generator 21 follows the measured value of the first residual and the measured value of the second residual outputted corresponding to the input value.

[0090] Figure 6 is a diagram showing a detailed configuration of a generator of an artificial neural network model.

[0091] Reference Figure 6 , the generator 21 may include an encoder 21a and a decoder 21b.

[0092] The encoder 21a and the decoder 21b constitute a network for estimating one or more estimated values ​​associated with the steering of the vehicle. In other words, the generator 21 can be composed of a network of an encoder 21a-decoder 21b structure trained in an end-to-end manner to estimate one or more of the estimated values.

[0093] The encoder 21a may be designed to extract high-level features from input values ​​using a transformer block and linear operations. At this time, the transformer block may include an embedding layer, a self-attention layer, and a feed-forward layer.

[0094] First, the embedding layer in the Transformer block can use a trainable embedding matrix to encode spatiotemporal information into the input sequences.

[0095] Next, deep features can be extracted through a self-attention layer, and refinement can be performed using a feed-forward layer.

[0096] At this time, the self-attention layer can use a scaled dot-product attention mechanism. The self-attention layer can be a major component in the Transformer-based Deep Neural Network (DNN) architecture, which enables the model to process and understand dependencies in the input sequence.

[0097] The scaled inner product attention first applies layer normalization to the input sequence and processes the result into three intrinsic vectors of query (Q) vector, key (K) vector and value (V) vector through linear operation.

[0098] Next, an attention score AS for each position in the input sequence may be calculated based on the derived vector. The calculation formula of the attention score AS may be as shown in Formula 1 below.

[0099] [Formula 1]

[0100]

[0101] Among them, σ(·) represents the softmax operation, ∥x input ∥ dim represents the feature dimension of the input, and (·) represents the inner product.

[0102] The attention score quantifies the relative importance of each position in the input sequence with respect to a given query. That is, higher scores are assigned to more appropriate positions.

[0103] To calculate the query vector, key vector and initial input value of decoder 21b The output of the transformer block can be provided to two different linear operations.

[0104] The decoder 21b may use the acquired values ​​to estimate one or more of the estimated values ​​in an auto-regressive manner similar to Sequence-to-Sequence Recurrent Neural Networks.

[0105] Similar to the encoder 21a, the decoder 21b may include a transformer block and a linear operation. In addition, the decoder 21b may also include a cross-attention layer within the transformer block.

[0106] The cross-attention layer can use the query vector and key vector of encoder 21a to perform scaled inner product attention to learn the dynamic correlation between the input signal and the inferred signal.

[0107] The discriminator 22 receives the input value and measured data including one or more measured values, and outputs a discriminator value for the measured data.

[0108] The discriminator 22 may determine a conditional probability. The discriminator 22 determines the probability that the corresponding data is true or false based on the input features. The discriminator 22's discriminator value may be a classification label. For example, the discriminator 22's discriminator value may be 0 or 1.

[0109] On the other hand, the discriminator 22 may also additionally receive estimation data including the input value and one or more estimation values, and output a discrimination value for the estimation data. Such a process may be performed in advance when the generator 21 and the discriminator 22 perform learning.

[0110] In this way, the discriminator 22 can output a discriminant value for the measured data during the operation for detecting abnormalities, and at the same time, can output a discriminant value for the estimated data during the operation for learning.

[0111] The discriminator 22 may be formed of a discriminator network used in adversarial training of a generative adversarial network (GAN). In other words, the generator 21 and the discriminator 22 may be trained by adversarial training to improve estimation performance and set a generative adversarial network (GAN) model.

[0112] In more detail, the discriminator 22 may perform optimization alternately with the generator 21 to solve a Wasserstein min-max problem such as Equation 2.

[0113] [Formula 2]

[0114]

[0115] Among them, x D，Real represents the actual data set consisting of input values ​​and steering-related measured values, x D，Fake represents a fictitious data sample consisting of input values ​​and steering-related estimates.

[0116] In the artificial neural network model 20, the generator 21 is trained to deceive the D，Real and x D，Fake The generator 21 can thus generate an estimation value for following the steer-by-wire system in a normal state.

[0117] In association with this, the loss function for generative adversarial network (GAN) training can be defined as the following formula 3.

[0118] [Formula 3]

[0119]

[0120] Where N is the batch size, L G is the loss function of the generator, L D is the loss function of the discriminator.

[0121] The first term of the loss of generator 21 corresponds to the mean absolute error loss L for supervised training MAE The remaining terms of the loss of the generator 21 together with the loss of the discriminator 22 constitute the generative adversarial network (GAN) loss function L GAN .

[0122] The processor 120 detects whether an abnormality occurs in the steering motor by comparing one or more estimated values ​​with one or more measured values ​​associated with the steering. As described above, the one or more estimated values ​​may include a first estimated value for the first residual and a second estimated value for the second residual. In addition, the one or more measured values ​​may include: a first measured value as a measured value of the first residual; and a second measured value as a measured value of the second residual.

[0123] The processor 120 can input the error data related to the difference between the one or more estimated values ​​and the one or more measured values ​​into the abnormality detection model 30 by executing one or more of the instructions, and thereby determine whether the steering motor is abnormal. For example, the abnormality detection model 30 can use a one-class support vector machine (OCSVM) algorithm.

[0124] There may be a plurality of data sets, the data sets including the input value, one or more estimated values, and one or more measured values. The error data may include: the mean value and standard deviation of the error between the measured value and the estimated value obtained from each of the plurality of data sets; the maximum absolute error between the measured value and the estimated value obtained from each of the plurality of data sets; and the discriminator value for the measured data obtained from each of the plurality of data sets.

[0125] In one embodiment of the present invention, the error data may include: the average value of the error between the first estimated value and the first measured value; the standard deviation between the first estimated value and the first measured value; the maximum absolute error between the first estimated value and the first measured value; the average value of the error between the second estimated value and the second measured value; the standard deviation between the second estimated value and the second measured value; the maximum absolute error between the second estimated value and the second measured value; and an identification value of an identifier for the measured data including the input value, the first measured value, and the second measured value.

[0126] A plurality of data sets including one or more estimated values ​​and one or more measured values ​​may be considered as a batch. For example, 100 to 150 data sets (as a specific example, 128 data sets) may be considered as a batch, and the error data of each batch may be obtained.

[0127] The error data acquired for one lot may be input to the abnormality detection model 30. In addition, the abnormality level of the steering motor (degradation in performance of the steering motor) may be detected based on the output (feature) acquired from the abnormality detection model 30 that received the error data.

[0128] Regarding the anomaly detection metric, we can consider the F1 score given by the following formula 4.

[0129] [Formula 4]

[0130]

[0131] Among them, TP is true positive, FP is false positive, FN is false negative, and positive indicates that the performance of the brake motor is reduced.

[0132] On the other hand, as described above, the discriminator 22 can calculate not only the discrimination value for the measured data included in the data set, but also the discrimination value for the estimated data included in the data set. The discrimination value of the estimated data of the data set by the discriminator 22 can be fed back to the generator 21 of the artificial neural network model 20, thereby being used in training.

[0133] So far, the structure of the steering motor abnormality detection device 100 according to one embodiment of the present invention has been described in detail. Next, the operation of the steering motor abnormality detection device 100 will be described in detail.

[0134] Figure 7 1 is a diagram showing the operation of the steering motor abnormality detection device according to one embodiment of the present invention.

[0135] Reference Figure 7 , the steering motor abnormality detection device 100 according to an embodiment of the present invention can perform the following operations.

[0136] First, the memory 110 stores the input value At this time, the one or more measured values ​​may include the first measured value R1 of the measured first residual. m and the second measured value R2 of the measured second residual m .

[0137] As mentioned above, the input value It can be the propulsion force of the rack. In addition, multiple input values ​​can be The first measured value R1 m And the second measured value R2 m That is, in one embodiment of the present invention, the batch of data can be used as a unit to perform data processing and calculation.

[0138] Then, the processor 120 can execute one or more instructions to convert the input value The processor 120 can input the input value to the artificial neural network model 20, and obtain one or more estimated values ​​output by the artificial neural network model 20. More specifically, the processor 120 can The first estimated value R1 outputted by the generator 21 as the estimated value of the first residual is obtained. eand the second estimate R2 which is an estimate of the second residual e .

[0139] In addition, processor 120 may execute more than one instruction, thereby converting the input value The first measured value R1 m And the second measured value R2 m is input to the artificial neural network model 20, and the discrimination value DS is obtained. More specifically, the processor 120 converts the input value Including the first measured value R1 m The first measured data, and the second measured value R2 m The second measured data is input to the discriminator 22, and the discrimination value DS is obtained.

[0140] Next, the processor 120 outputs the error data D Error . Error data D Error Can be output in different data batches.

[0141] More specifically, there may be a plurality of data sets, each of which includes the input value, one or more estimated values, and one or more measured values.

[0142] At this time, the error data D Error May include: first estimated value R1 e With the first measured value R1 m The average value of the error between E(R1) ; The first estimated value R1 e With the first measured value R1 m The standard deviation between E(R1) ; The first estimated value R1 e With the first measured value R1 m The maximum absolute error between E(R1) ; The second estimated value R2 e The second measured value R2 m The average value of the error between E(R2) ; The second estimated value R2 e The second measured value R2 m The standard deviation between E(R2) ; The second estimated value R2 e The second measured value R2 m The maximum absolute error between e(R2) ; For input value Including the first measured value R1 m The first measured data and the second measured value R2 m The discrimination value DS of the discriminator of the second measured data.

[0143] Finally, the processor 120 executes one or more instructions, thereby converting the error data D Error The anomaly detection model 30 is inputted, and the output of the anomaly detection model 30 is obtained. As described above, the anomaly detection model 30 may include a one-class support vector machine (OCSVM) algorithm.

[0144] Based on the received error data D Error Based on the output (features) obtained by the abnormality detection model 30, the processor 120 can detect the abnormality level (performance degradation of the steering motor) of the steering motor of the wire control system.

[0145] The above has described in detail the steering motor abnormality detection device 100 according to one embodiment of the present invention. Next, a steering motor abnormality detection method will be described.

[0146] Figure 8 is a sequence diagram of a steering motor abnormality detection method according to an embodiment of the present invention.

[0147] According to a steering motor abnormality detection method (step S100 ) according to an embodiment of the present invention, an abnormality of a steering motor is detected, wherein the steering motor is configured in a steer-by-wire system of a vehicle and provides a reaction force to an operation of a steering wheel.

[0148] Reference Figure 8 , the steering motor abnormality detection method (step S100) according to an embodiment of the present invention can be performed as follows.

[0149] First, the processor 120 inputs an input value associated with the driving of a rack into the artificial neural network model 20, and obtains one or more estimated values ​​associated with steering output by the artificial neural network model 20 (step S110), wherein the rack receives a driving force from a wheel actuator driven corresponding to the operation of the steering wheel and moves the wheels of the vehicle.

[0150] The input value may be a propulsion force of the rack. For example, the propulsion force of the rack may be obtained from the controller area network 300 of the vehicle.

[0151] The one or more estimated values ​​may include a first estimated value for a first residual, the first residual being a difference between a command and a measured value for a steering torque, the steering torque being a steering torque transmitted to the steering wheel 210 corresponding to the input value. In addition, the one or more estimated values ​​may further include a second estimated value for a second residual, the second residual being a difference between a command and a measured value for a motor torque, the motor torque being a motor torque transmitted to the steering motor 260 corresponding to the input value.

[0152] The one or more estimated values ​​are obtained for comparison with one or more measured values ​​related to steering. At this time, the one or more measured values ​​may include: a first measured value as a measured value of the first residual; and a second measured value as a measured value of the second residual.

[0153] In one embodiment of the present invention, the artificial neural network model 20 may include a generative adversarial network (GAN). In other words, the artificial neural network model 20 may be a deep learning-based neural twin model for a vehicle's steer-by-wire system.

[0154] The artificial neural network model 20 may include a generator 21 and a discriminator 22. The generator 21 receives the input value and generates the estimated value. The generator 21 may be composed of a multivariate transformer. In addition, the discriminator 22 may receive measured data including the input value and one or more measured values, and output a discriminant value for the measured data.

[0155] In the step of acquiring the one or more estimated values ​​(step S110 ), the processor 120 may output the input value to the generator 21 , and acquire the one or more estimated values ​​generated by the generator 21 .

[0156] On the other hand, the artificial neural network model 20 is constructed by performing learning by the generator 21 and the discriminator 22, and the input value and the one or more measured values ​​used in performing the learning can be obtained when the vehicle and the steer-by-wire system are in a normal state. Thus, the one or more estimated values ​​can follow the steering-related measured values ​​obtained when the steer-by-wire system is in a normal state.

[0157] Next, the processor 120 detects whether an abnormality occurs in the steering motor by comparing the one or more estimated values ​​with one or more measured values ​​related to the steering (step S120 ).

[0158] The processor 120 can input error data related to the difference between the estimated value and the measured value into the abnormality detection model 30 by executing one or more instructions, and determine whether the steering motor is abnormal. For example, the abnormality detection model 30 can use a one-class support vector machine (OCSVM) algorithm.

[0159] Fig. 9 is a detailed sequence diagram of steps for detecting whether an abnormality occurs in a steering motor in a steering motor abnormality detection method according to an embodiment of the present invention.

[0160] Reference Fig. 9 The step of detecting whether an abnormality occurs in the steering motor (step S120) can be performed as follows.

[0161] First, the processor 120 inputs the measured data including the input value and one or more measured values ​​into the discriminator 22 , and obtains the discrimination value for the measured data generated by the discriminator 22 (step S121 ).

[0162] Next, the processor 120 inputs error data including a discrimination value for the measured data and a numerical value related to a difference between one or more estimated values ​​and one or more measured values ​​into the abnormality detection model 30 and obtains an output of the abnormality detection model 30 (step S122).

[0163] There may be multiple data sets, each of which includes the input value, one or more estimated values, and one or more measured values. The error data may include: the average value and standard deviation of the error between the measured value and the estimated value obtained from each of the multiple data sets; the maximum absolute error between the measured value and the estimated value obtained from each of the multiple data sets; and the identification value of the identifier for the measured data obtained from each of the multiple data sets.

[0164] In one embodiment of the present invention, the error data may include: the average value of the error between the first estimated value and the first measured value; the standard deviation between the first estimated value and the first measured value; the maximum absolute error between the first estimated value and the first measured value; the average value of the error between the second estimated value and the second measured value; the standard deviation between the second estimated value and the second measured value; the maximum absolute error between the second estimated value and the second measured value; and an identification value of an identifier for the input value, the first measured data including the first measured value, and the second measured data including the second measured value.

[0165] A plurality of data sets including one or more estimated values ​​and one or more measured values ​​may be considered as a batch. 100 to 150 data sets (as a specific example, 128 data sets) may be considered as a batch, and the error data may be acquired for each batch.

[0166] The error data acquired for one lot may be input to the abnormality detection model 30. In addition, the degree of abnormality (degradation in performance of the steering motor) of the steering motor may be detected based on the output (feature) acquired from the abnormality detection model 30 receiving the error data.

[0167] In addition, the present invention also provides a non-transitory computer-readable storage medium storing a program for executing the steering motor abnormality detection method. Specifically, the present invention can provide a non-transitory computer-readable storage medium storing a program including at least one instruction for executing the steering motor abnormality detection method.

[0168] At this time, the instructions may include not only the machine code generated by the compiler, but also the high-level language code executable by the computer.

[0169] The recording medium may also include magnetic media (Magnetic Media) such as hard disks, floppy disks and magnetic tapes, optical recording media (Optical Media) such as compact disk read only memory (CD-ROM) and digital video disk (DVD), magneto-optical media (Magneto-Optical Media) such as floppy disks (Floptical Disk), hardware devices such as read-only memory (ROM), random access memory (RAM), flash memory, etc., which are constructed in a manner of storing and executing program instructions.

[0170] Regarding the anomaly detection metric, the F1 score as described above may be considered.

[0171] As a result of the simulation, according to the present invention, it is possible to predict the performance of an F1 score of 0.85 or more until the maximum output of the brake motor is reduced to 10%. In other words, according to the present invention, the degree of deterioration of the steering motor, which is provided in the steer-by-wire system and provides a reaction force to the operation of the steering wheel, can be accurately predicted before the steering motor, which is configured in the steer-by-wire system and provides a reaction force to the operation of the steering wheel, completely fails.

[0172] According to the present invention, it is possible to predict before a steering motor of a steer-by-wire system fails, thereby enabling preventive measures to be taken before the steering motor fails, thereby preventing the failure of the steering motor.

[0173] Thus, the present invention provides an estimate of the future behavior and remaining useful life of the system and predictive diagnostics suitable for predictive maintenance repair applications, thereby effectively prompting preventive measures and maintenance before a steering motor failure occurs in a steer-by-wire system.

[0174] Although the embodiments of the present invention have been described, the technical concept of the present invention is not limited to the embodiments presented in this specification, and those skilled in the art who understand the technical concept of the present invention can easily propose other embodiments within the scope of the same technical concept by adding, changing, deleting, adding, etc. to the constituent elements. However, this will also be considered to fall within the scope of the present invention.

Claims

1. A steering motor abnormality detection device, which detects abnormality of a steering motor, wherein the steering motor is arranged in a steer-by-wire system of a vehicle and provides a reaction force to an operation of a steering wheel, characterized in that: include: memory for storing one or more instructions, and A processor, executing the one or more instructions; The processor inputs input values ​​related to the driving of the rack into the artificial neural network model by executing the one or more instructions, and obtains one or more estimated values ​​related to steering output by the artificial neural network model, and detects whether the steering motor has an abnormality by comparing the one or more estimated values ​​with one or more measured values ​​related to steering. The rack receives driving force from a wheel actuator driven in response to the operation of the steering wheel to move the wheels of the vehicle.

2. The steering motor abnormality detection device according to claim 1, wherein: The input value is the propulsion force of the rack.

3. The steering motor abnormality detection device according to claim 1, wherein: The one or more estimates include a first estimate for a first residual, the first residual being a difference between a command and a measurement for a steering torque, the steering torque being a steering torque delivered to a steering wheel of the vehicle in response to the input value, The one or more measured values ​​include a first measured value that is a measured value of the first residual.

4. The steering motor abnormality detection device according to claim 3, wherein: The one or more estimated values ​​also include a second estimated value for a second residual, the second residual being a difference between a command and a measurement for a motor torque, the motor torque being a motor torque delivered to the steering motor in response to the input value, The one or more measured values ​​further include a second measured value that is a measured value of the second residual.

5. The steering motor abnormality detection device according to claim 1, wherein: The input value is obtained through a controller area network of the vehicle.

6. The steering motor abnormality detection device according to claim 1, wherein: The artificial neural network model is composed of a generative adversarial network. The generative adversarial network comprises: a generator that receives the input value and generates the one or more estimated values; and The discriminator receives the input value and measured data including the one or more measured values ​​and outputs a discriminant value for the measured data.

7. The steering motor abnormality detection device according to claim 6, wherein: The processor inputs error data into an abnormality detection model by executing the one or more instructions and determines whether an abnormality occurs in the steering motor, wherein the error data is related to a difference between the one or more estimated values ​​and the one or more measured values.

8. The steering motor abnormality detection device according to claim 7, wherein: There are a plurality of data sets, each of which includes the input value, the one or more estimated values, and the one or more measured values. The error data includes: a mean and a standard deviation of the errors between measured values ​​and estimated values ​​obtained from each of a plurality of said data sets; the maximum absolute error between the measured value and the estimated value obtained from each of the plurality of said data sets; and A discriminator value for the measured data obtained from each of the plurality of data sets.

9. A method for detecting an abnormality of a steering motor, wherein the method detects an abnormality of a steering motor, wherein the steering motor is arranged in a steer-by-wire system of a vehicle and provides a reaction force to an operation of a steering wheel, wherein: include: The processor inputs an input value related to driving of a rack into an artificial neural network model, and obtains one or more estimated values ​​related to steering output by the artificial neural network model, wherein the rack receives a driving force from a wheel actuator driven in response to an operation of the steering wheel and moves the wheels of the vehicle; as well as The processor detects whether an abnormality occurs in the steering motor by comparing the one or more estimated values ​​with one or more measured values ​​related to steering.

10. The method for detecting abnormality of a steering motor according to claim 9, wherein: The input value is the propulsion force of the rack.

11. The method for detecting abnormality of a steering motor according to claim 9, wherein: The one or more estimates include a first estimate for a first residual, the first residual being a difference between a command and a measurement for a steering torque, the steering torque being a steering torque delivered to a steering wheel of the vehicle in response to the input value, The one or more measured values ​​include a first measured value that is a measured value of the first residual.

12. The method for detecting abnormality of a steering motor according to claim 11, wherein: The one or more estimated values ​​also include a second estimated value for a second residual, the second residual being a difference between a command and a measurement for a motor torque, the motor torque being a motor torque delivered to the steering motor in response to the input value, The one or more measured values ​​further include a second measured value that is a measured value of the second residual.

13. The method for detecting abnormality of a steering motor according to claim 9, wherein: The artificial neural network model is composed of a generative adversarial network. The generative adversarial network comprises: a generator that receives the input value and generates the one or more estimated values; and The discriminator receives the input value and measured data including the one or more measured values ​​and outputs a discriminant value for the measured data.

14. The method for detecting abnormality of a steering motor according to claim 13, wherein: The step of detecting whether the steering motor is abnormal comprises: The processor inputs the measured data into the discriminator, and obtains a discriminator value for the measured data generated by the discriminator; and The processor inputs a discrimination value for the measured data and error data including numerical values ​​related to differences between the one or more estimated values ​​and the one or more measured values ​​into an abnormality detection model and obtains an output of the abnormality detection model.

15. A computer-readable storage medium storing a program, wherein: The program includes at least one instruction for executing the steering motor abnormality detection method according to claim 9.

Images