Railway vehicle wheel polygonal fault diagnosis method and system

Abstract

The application discloses a rail vehicle wheel polygon fault diagnosis method and system, comprising: obtaining the stiffness and displacement of a first level suspension steel spring, the stiffness, displacement and pressure of a second level suspension air spring, a wheel diameter, a wheel set mass, a bogie mass, a vehicle body mass and vehicle operation parameter data; calculating the peak-to-peak value, average value, standard deviation, kurtosis, waveform factor, pulse factor, margin factor and peak factor of the above data respectively, and constructing vehicle parameter characteristics, suspension system parameter characteristics and characteristics of the fusion of vehicle parameters and suspension system parameters; inputting the constructed characteristics into a trained wheel polygon fault diagnosis model to obtain a wheel polygon fault diagnosis result. The application adopts a method of fusing vehicle operation parameters and vehicle suspension system parameters for modeling, constructs various characteristics capable of accurately reflecting a wheel polygon fault state, and realizes real-time diagnosis of a wheel fault without affecting vehicle operation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wheel fault detection, and in particular to a rail vehicle wheel polygon fault diagnosis method and system. BACKGROUND

[0002] The statements in this section merely provide background information related to the present application and do not necessarily constitute the prior art.

[0003] The wheel is a bearing and rolling unit of the rail vehicle. With the increase of the running mileage of the vehicle, irregular wear is prone to occur in the wear process of the wheel, and the original circular profile of the wheel is changed into a polygon. This phenomenon is called wheel polygon. The order of the wheel polygon represents the number of sides of the wheel. For example, if the wheel polygon is 17 orders, it means that the wheel has been worn into a 17-sided polygon. The wheel polygon will cause the vehicle vibration to intensify, reduce the passenger ride comfort, and reduce the service life of the vehicle related parts. Therefore, the degree of wheel polygon needs to be diagnosed in a timely manner, and the wheel needs to be refinished in severe cases. The fault degree of the wheel polygon represents the wear degree of the wheel, which is usually represented by the distance between the circular profile of the wheel and the polygon side after wear, and the unit is generally mm. The greater the distance, the more serious the fault degree.

[0004] Currently, the detection of the wheel polygon is mainly divided into static detection and dynamic detection. The static detection refers to accurately measuring the wheel profile after the vehicle is parked in the garage by professional measuring equipment. This method has high accuracy, but it affects the operation of the vehicle and cannot achieve real-time online detection. The dynamic detection method includes two types: trackside equipment-based detection method and vehicle-mounted equipment-based detection method. The trackside equipment-based detection method refers to installing stress sensors near the rail to measure the wheel-rail vertical force, but this method has requirements for the arrangement of the sensors and cannot achieve real-time online detection. The vehicle-mounted equipment-based detection method currently mainly installs vibration sensors at the axle box to diagnose the wheel polygon by analyzing the time domain and frequency domain characteristics of the vibration at the axle box. This method has good real-time performance and relatively high accuracy, but it needs to collect high-frequency signals from the vibration sensor in real time, and then perform filtering, Fourier transform, Hilbert transform and other types of frequency spectrum analysis, which requires a large amount of computing resources. Therefore, the system composed of vibration sensors, data acquisition devices and data analysis processing devices is expensive and large in size, which is not suitable for large-scale installation on rail vehicles. SUMMARY

[0005] In order to solve the above problems, the present application proposes a rail vehicle wheel polygon fault diagnosis method and system, which adopts a vehicle running parameter and vehicle primary and secondary suspension system parameter fusion modeling method to construct various features that can accurately reflect the wheel polygon fault state, and realizes real-time diagnosis of the wheel fault without affecting the operation of the vehicle.

[0006] According to a first aspect of an embodiment of the present application, a rail vehicle wheel polygon fault diagnosis method is provided, comprising:

[0007] Obtaining the stiffness and displacement of the primary suspension steel spring of the vehicle, the stiffness, displacement and pressure of the secondary suspension air spring, the wheel diameter, the mass of the wheelset, the mass of the bogie, the mass of the vehicle body and the vehicle operation parameter data;

[0008] Respectively calculating the peak-to-peak value, the average value, the standard deviation, the kurtosis, the waveform factor, the pulse factor, the margin factor and the peak factor of the above data, and respectively constructing the vehicle parameter feature, the suspension system parameter feature and the feature of the fusion of the vehicle parameter and the suspension system parameter;

[0009] Inputting the constructed feature into a trained wheel polygon fault diagnosis model to obtain the fault diagnosis result of the wheel polygon.

[0010] As an optional solution, the wheel polygon fault diagnosis model specifically comprises:

[0011] An input layer for respectively inputting the vehicle parameter feature, the suspension system parameter feature and the feature of the fusion of the vehicle parameter and the suspension system parameter;

[0012] An intermediate layer, the vehicle parameter feature and the suspension system parameter feature enter an automatic learning layer after passing through multiple convolution layers and pooling layers, the automatic learning layer obtains the optimal connection relationship between the upper layer neurons and the lower layer neurons through training, and the output of the automatic learning layer enters a full connection layer; the feature of the fusion of the vehicle parameter and the suspension system parameter also enters the full connection layer after passing through multiple convolution and pooling layers;

[0013] The full connection layer classifies the received data, and the classification result is sent to an output layer; meanwhile, the output of the full connection layer is also fed back to the automatic learning layer;

[0014] The output layer receives the data input by the full connection layer, and outputs the wheel polygon fault diagnosis result.

[0015] As an optional solution, the vehicle parameter feature is constructed, specifically comprising:

[0016] The vehicle speed, the peak-to-peak value within the speed setting time, the average value within the speed setting time, the standard deviation within the speed setting time, the total mass of the vehicle, the mileage change within the setting time and the mode of the traction and braking level within the setting time.

[0017] The suspension system parameter feature is constructed, specifically comprising:

[0018] primary suspension steel spring displacement, primary suspension steel spring displacement kurtosis, primary suspension steel spring displacement waveform factor, primary suspension steel spring displacement pulse factor, primary suspension steel spring displacement margin factor, primary suspension steel spring displacement peak factor, secondary suspension air spring displacement, secondary suspension air spring displacement kurtosis, secondary suspension air spring displacement waveform factor, secondary suspension air spring displacement pulse factor, secondary suspension air spring displacement margin factor, and secondary suspension air spring displacement peak factor.

[0019] The features of the vehicle parameters and the suspension system parameters are fused, and specifically include:

[0020] average value of the primary suspension steel spring displacement when the vehicle speed is greater than a certain set value, average value of the secondary suspension air spring displacement when the vehicle speed is greater than a certain set value, peak-to-peak value of the primary suspension steel spring displacement when the vehicle speed is greater than a certain set value, peak-to-peak value of the secondary suspension air spring displacement when the vehicle speed is greater than a certain set value, weighted average of the average value and the standard deviation of the primary suspension steel spring displacement within a set time, weighted average of the average value and the standard deviation of the secondary suspension air spring displacement within a set time, weighted average of the average value and the kurtosis of the primary suspension steel spring displacement within a set time, weighted average of the average value and the kurtosis of the secondary suspension air spring displacement within a set time, weighted average of the average value and the waveform factor of the primary suspension steel spring displacement within a set time, weighted average of the average value and the waveform factor of the secondary suspension air spring displacement within a set time, weighted average of the average value and the pulse factor of the primary suspension steel spring displacement within a set time, weighted average of the average value and the pulse factor of the secondary suspension air spring displacement within a set time, weighted average of the average value and the margin factor of the primary suspension steel spring displacement within a set time, weighted average of the average value and the margin factor of the secondary suspension air spring displacement within a set time, weighted average of the average value and the peak factor of the primary suspension steel spring displacement within a set time, and weighted average of the average value and the peak factor of the secondary suspension air spring displacement within a set time.

[0021] According to a second aspect of the embodiment of the present application, there is provided a rail vehicle wheel polygonal fault diagnosis system, comprising:

[0022] a data acquisition module configured to acquire the stiffness and displacement of the primary suspension steel spring, the stiffness, displacement and pressure of the secondary suspension air spring, the wheel diameter, the mass of the wheelset, the mass of the bogie, the mass of the vehicle body, and the vehicle operation parameter data;

[0023] a feature extraction module configured to calculate the peak-to-peak value, average value, standard deviation, kurtosis, waveform factor, pulse factor, margin factor and peak factor of the above data respectively, and construct the vehicle parameter features, the suspension system parameter features, and the features of the vehicle parameters and the suspension system parameters fused.

[0024] The fault diagnosis module is configured to input the constructed features into the trained wheel polygon fault diagnosis model to obtain a wheel polygon fault diagnosis result.

[0025] According to a third aspect of the embodiments of the present application, a terminal device is provided, which comprises a processor and a memory, the processor is configured to implement instructions, and the memory is configured to store a plurality of instructions, which are adapted to be loaded and executed by the processor to implement the wheel polygon fault diagnosis method of the rail vehicle.

[0026] According to a fourth aspect of the embodiments of the present application, a computer readable storage medium is provided, which stores a plurality of instructions, which are adapted to be loaded and executed by a processor of a terminal device to implement the wheel polygon fault diagnosis method of the rail vehicle.

[0027] Compared with the prior art, the present application has the following beneficial effects:

[0028] (1) The present application respectively acquires vehicle operating parameter data and suspension system parameter data, and respectively constructs vehicle parameter features, suspension system parameter features, and features of the fusion of vehicle parameters and suspension system parameters, which are all features capable of fully reflecting the wheel polygon fault state. A vibration system is formed by a wheel set, a primary suspension steel spring, a frame, a secondary suspension air spring, and a vehicle body, and the inherent characteristics (stiffness, mass, etc.) of the system determine that the system resonance excited under certain specific vehicle working conditions (speed, traction brake level, etc.) is the strongest, so the fusion of vehicle parameters and suspension system parameters can fully reflect the suspension system response characteristics under specific vehicle working conditions and fully capture the wheel polygon fault features.

[0029] (2) The present application adopts a method of fusing vehicle operating parameters and vehicle primary and secondary suspension system parameters to model, and constructs various features capable of accurately reflecting the wheel polygon fault state, thereby realizing real-time diagnosis of wheel faults without affecting vehicle operation.

[0030] (3) The present application constructs a feedback neural network combining artificial fusion features and automatic learning features, automatically learns vehicle parameter features and suspension system parameter features, can fully learn the correlation features between the two types of parameters, reduces the number of neural network parameters and the amount of calculation, reduces the redundancy information of the neural network, and improves the calculation speed; combines the automatically learned correlation features with the artificially constructed features of the fusion of vehicle parameters and suspension system parameters, can fully combine machine automatic learning and expert experience, and improves the accuracy of the neural network model. The full connection layer is fed back to the input of the automatic learning layer to form a ring-shaped network, which can feed back the learning of the automatic learning layer and realize automatic optimization of the network structure of the automatic learning layer.

[0031] Advantages of the additional aspects of the application will become apparent in the following description, become apparent from the following description, or be learned by the practice of the application. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 Flow chart of rail vehicle wheel polygon fault diagnosis method in the embodiment of the application;

[0033] Figure 2 Schematic diagram of vehicle suspension system;

[0034] Figure 3 Schematic diagram of wheel polygon fault diagnosis model structure in the embodiment of the application;

[0035] Figure 4 Schematic diagram of automatic learning layer learning process in the embodiment of the application;

[0036] Wherein, 101. vehicle body, 102. secondary suspension air spring, 103. frame, 104. primary suspension steel spring, 105. wheel set, 106. wheel, 107. first displacement sensor, 108. pressure sensor, 109. second displacement sensor. DETAILED DESCRIPTION

[0037] It should be noted that the following detailed description is merely exemplary in nature and is intended to provide further description of the application. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0038] It is to be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of example embodiments consistent with the present application. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0039] Embodiment one

[0040] In one or more embodiments, a rail vehicle wheel polygon fault diagnosis method is disclosed, in combination Figure 1 , specifically comprising the following process:

[0041] (1)Obtain the stiffness and displacement of the primary suspension steel spring of the vehicle, the stiffness, displacement and pressure of the secondary suspension air spring, the wheel diameter, the wheelset mass, the bolster mass, the car body mass and the vehicle operation parameter data;

[0042] In this embodiment, first, a vehicle and suspension system dynamics model is constructed, as shown in Figure 2 Analysis shows that the transmission path of the vibration impact caused by the wheel polygon is: from the wheel to the wheelset, through the primary suspension steel spring to the bolster, and then through the secondary suspension air spring to the car body.

[0043] Therefore, the parameters of the related components on the transmission path are highly related to the wheel polygon fault, and the following suspension system parameters are screened out as the data to be collected for fault diagnosis: the wheel (106) diameter D, the wheelset (105) mass m1, the bolster (103) mass m2, the car body (101) mass m3, the primary suspension steel spring (104) stiffness k1, the secondary suspension air spring (102) stiffness k2, the primary suspension steel spring displacement h1 and the secondary suspension air spring displacement h2 and the secondary suspension air spring pressure P2.

[0044] In combination with Figure 2 A first displacement sensor (107) is installed at the primary suspension steel spring (104) of the vehicle to measure the displacement h1 of the primary suspension steel spring; a second displacement sensor (109) is installed at the secondary suspension air spring (102) of the vehicle to measure the displacement h2 of the secondary suspension air spring; and a pressure sensor (108) is installed at the secondary suspension air spring (102) of the vehicle to measure the pressure P2 of the secondary suspension air spring. The remaining data are inherent parameters of the railway vehicle suspension system and can be directly obtained.

[0045] Secondly, the vehicle operation parameters are also closely related to the fault response of the wheel polygon, and in this embodiment, the vehicle network is used to collect the vehicle operation information, including the vehicle speed v, the mileage s and the traction and braking level P / B.

[0046] After obtaining the above data, a data preprocessing process is still needed, mainly including the following two aspects: abnormal data management and data normalization.

[0047] ①Abnormal data management: delete abnormal values and interpolate missing values according to certain rules.

[0048] ②Data normalization: uniformly convert data of different dimensions and different orders of magnitude to data between 0 and 1 according to certain rules.

[0049] (2) Calculate the peak-to-peak value, average value, standard deviation, kurtosis, waveform factor, pulse factor, margin factor and peak factor of the above data respectively, and select the characteristics that are obviously different between the wheel fault and the normal state as the feature data based on the calculation results.

[0050] Specifically:

[0051] ① Peak-to-peak value: refers to the difference between the maximum value x max and the minimum value x min in a certain period of time, and the calculation expression is:

[0052] pp=x max -x min .

[0053] ② Average value: refers to the sum of all data divided by the number of data in a group of data, which refers to the average of all amplitudes. The calculation expression is:

[0054]

[0055] where N is the number of data in a group of data, and x i is the i-th data.

[0056] ③ Standard deviation: refers to the dispersion degree of a group of data, reflecting the amplitude of vibration. The calculation expression is:

[0057]

[0058] where N is the number of data in a group of data, x i is the i-th data, and μ is the variance.

[0059] ④ Kurtosis: kurtosis is the fourth power of the amplitude, and after the fourth power relationship of a pulse signal is changed, the high amplitude is highlighted, and the low amplitude is suppressed. The calculation expression is:

[0060]

[0061] where N is the number of data in a group of data, x i is the i-th data, and μ is the average value.

[0062] ⑤ Waveform factor: the root mean square value divided by the absolute average value, which is very sensitive to initial non-periodic impact. The calculation expression is:

[0063]

[0064] ⑥ Pulse factor: the peak value divided by the absolute average value. Like the peak value index, it is used to detect whether there is impact vibration in the signal. The calculation expression is:

[0065]

[0066] ⑦ Margin factor: Generally used to detect the wear condition of mechanical equipment. The calculation expression is:

[0067]

[0068] in,

[0069] ⑧ Peak factor: This is the peak value divided by the root mean square value. The calculation expression is:

[0070]

[0071] In this embodiment, vehicle parameter features, suspension system parameter features, and features that combine vehicle parameters and suspension system parameters are constructed based on the calculation results.

[0072] The vehicle parameter characteristics include: vehicle speed, peak-to-peak value within the vehicle speed setting time, average value within the vehicle speed setting time, standard deviation within the vehicle speed setting time, total vehicle mass, mileage change within the setting time, and mode of traction and braking levels within the setting time.

[0073] The characteristics of the suspension system include: primary suspension steel spring displacement, primary suspension steel spring displacement kurtosis, primary suspension steel spring displacement waveform factor, primary suspension steel spring displacement pulse factor, primary suspension steel spring displacement margin factor, primary suspension steel spring displacement peak factor, secondary suspension air spring displacement, secondary suspension air spring displacement kurtosis, secondary suspension air spring displacement waveform factor, secondary suspension air spring displacement pulse factor, secondary suspension air spring displacement margin factor, and secondary suspension air spring displacement peak factor.

[0074] The characteristics of the fusion of vehicle parameters and suspension system parameters include: the average displacement of the primary suspension steel springs when the vehicle speed is greater than a certain set value; the average displacement of the secondary suspension air springs when the vehicle speed is greater than a certain set value; the peak-to-peak value of the primary suspension steel spring displacement when the vehicle speed is greater than a certain set value; the peak-to-peak value of the secondary suspension air spring displacement when the vehicle speed is greater than a certain set value; the weighted average of the average value over the vehicle speed setting time and the standard deviation of the primary suspension steel spring displacement; the weighted average of the average value over the vehicle speed setting time and the standard deviation of the secondary suspension air spring displacement; the weighted average of the average value over the vehicle speed setting time and the kurtosis of the primary suspension steel spring displacement; the weighted average of the average value over the vehicle speed setting time and the kurtosis of the secondary suspension air spring displacement; and the average value over the vehicle speed setting time. The weighted average of the mean and the displacement waveform factor of the primary suspension steel spring, the weighted average of the mean and the displacement waveform factor of the secondary suspension air spring within the vehicle speed setting time, the weighted average of the mean and the displacement pulse factor of the primary suspension steel spring within the vehicle speed setting time, the weighted average of the mean and the displacement pulse factor of the secondary suspension air spring within the vehicle speed setting time, the weighted average of the mean and the displacement margin factor of the primary suspension steel spring within the vehicle speed setting time, the weighted average of the mean and the displacement margin factor of the secondary suspension air spring within the vehicle speed setting time, the weighted average of the mean and the displacement peak factor of the primary suspension steel spring within the vehicle speed setting time, and the weighted average of the mean and the displacement peak factor of the secondary suspension air spring within the vehicle speed setting time.

[0075] (3) Input the constructed features into the trained wheel polygon fault diagnosis model to obtain the fault diagnosis results of the wheel polygon.

[0076] In this embodiment, the wheel polygon fault diagnosis model employs a feedback neural network that combines manually fused features with automatically learned features, such as... Figure 3 As shown, each circle represents a neuron, and each arrow indicates a relationship between upper and lower layer neurons, including a weight value. The upper layer neuron is calculated as the sum of the products of all lower layer neurons connected to it and their corresponding weight values. The output of each lower layer neuron (the calculated value of that layer neuron) is used as the input to the upper layer neuron for computation.

[0077] The main structure of the feedback neural network in this embodiment specifically includes:

[0078] The input layer is used to input vehicle parameter features, suspension system parameter features, and features fused from vehicle parameters and suspension system parameters, respectively.

[0079] In the intermediate layer, vehicle parameter features and suspension system parameter features pass through multiple convolutional and pooling layers before entering the automatic learning layer. The output of the automatic learning layer (the calculated value of the neurons in this layer) enters the fully connected layer. The features that fuse vehicle parameters and suspension system parameters also enter the fully connected layer after passing through multiple convolutional and pooling layers.

[0080] like Figure 4 As shown, the initial state of the automatic learning layer is: neurons in this layer are connected to all neurons in the lower layer. After each round of learning and training, the neural network calculates the connection weights between neurons in this layer and neurons in the lower layer. At this time, the connection weight with the smallest weight in the connection between each upper layer neuron and the lower layer neuron is forcibly reset to 0, the weight is fixed, and learning and training are performed again. This learning and training process is repeated until the number of connections between each upper layer neuron and the lower layer neuron equals the set value. Figure 4 When the value is set to 2, training ends. At this point, each upper-layer neuron is connected to only 2 lower-layer neurons (other connection weights are 0). The number of connections between upper-layer and lower-layer neurons is relatively small, which can significantly reduce the computational load of the neural network and reduce redundant information.

[0081] The fully connected layer classifies the received data, and the classification results are sent to the output layer. At the same time, the output of the fully connected layer is also fed back to the automatic learning layer as input to the automatic learning layer, forming a ring network structure.

[0082] The output layer receives data from the fully connected layer and outputs the wheel polygon fault diagnosis results.

[0083] In this embodiment, the training process for the wheel polygon fault diagnosis model is as follows:

[0084] Fault simulation tests were conducted using a test bench. First, polygonal fault samples of the wheels were manually fabricated, and the order and severity of the fault were recorded. Then, wheels with known polygon orders and fault severity were mounted on the test bench for testing. Sensors were placed at appropriate locations to obtain polygonal data for wheels with different orders and fault severity. Alternatively, relevant data from actual vehicle operation could be downloaded, and the wheels disassembled to measure the polygon order and fault severity. The experimental data and actual vehicle data together constituted the training dataset for the fault diagnosis model.

[0085] The training dataset is fed into the neural network to train the model. The model structure is adjusted based on the training results, including pruning or growing, to ultimately obtain a fault diagnosis model that meets the required accuracy and recall. The built model can undergo periodic incremental learning to optimize and update its version, improving its generalization ability. As one implementation method, model training can be performed on a ground server.

[0086] The relevant parameter features obtained in step (2) are input into the trained wheel polygon fault diagnosis model to obtain the order and fault degree of the wheel polygon; finally, the wheel polygon fault diagnosis result is combined with the maintenance strategy, and according to the wheel polygon order and fault degree, the requirements for the wheel diameter difference of the same wheelset, the same bogie and the same railway vehicle are considered, and a reasonable wheel turning repair proposal is given.

[0087] Embodiment two

[0088] In one or more embodiments, a railway vehicle wheel polygon fault diagnosis system is disclosed, comprising:

[0089] A data acquisition module is configured to acquire the stiffness and displacement of the primary suspension steel spring of the vehicle, the stiffness, displacement and pressure of the secondary suspension air spring, the wheel diameter, the wheelset mass, the bogie mass, the vehicle body mass and the vehicle operation parameter data;

[0090] A feature extraction module is configured to calculate the peak-to-peak value, the average value, the standard deviation, the kurtosis, the waveform factor, the pulse factor, the margin factor and the peak factor of the above data respectively, and to construct the vehicle parameter features, the suspension system parameter features and the features of the fusion of the vehicle parameters and the suspension system parameters respectively;

[0091] A fault diagnosis module is configured to input the constructed features into the trained wheel polygon fault diagnosis model to obtain the fault diagnosis result of the wheel polygon.

[0092] It should be noted that the specific implementation of each module has been described in detail in Embodiment one, and will not be described here in detail.

[0093] Embodiment three

[0094] In one or more embodiments, a terminal device is disclosed, comprising a server, the server comprising a memory, a processor and a computer program stored on the memory and executable on the processor, and the processor implements the railway vehicle wheel polygon fault diagnosis method in Embodiment one when executing the program. For brevity, this will not be described here.

[0095] It should be understood that in the present embodiment, the processor can be a central processing unit CPU, and the processor can also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, ready-to-program gate arrays FPGA or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc.

[0096] The memory can include read-only memory and random access memory, and provide instructions and data to the processor, a portion of the memory can also include non-volatile random access memory. For example, the memory can also store device type information.

[0097] In the implementation process, each step of the above method can be completed by integrated logic circuit of hardware in the processor or instruction in the form of software.

[0098] Embodiment four

[0099] In one or more embodiments, a computer readable storage medium is disclosed, wherein a plurality of instructions are stored, the instructions are suitable for being loaded and executed by a processor of a terminal device to implement the rail vehicle wheel polygon fault diagnosis method described in embodiment one.

[0100] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit it, although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that: the specific embodiments of the present application can still be modified or replaced by the equivalent, without departing from the spirit and scope of the present application, any modification or equivalent replacement, which should be covered in the protection scope of the claims of the present application.

Claims

1. A method of diagnosing a polygonal fault of a wheel of a railway vehicle, characterized in that, The application relates to a vehicle wheel polygon fault diagnosis method and device. The vehicle wheel polygon fault diagnosis method comprises the following steps: acquiring the rigidity and displacement of a primary suspension steel spring of a vehicle, the rigidity, displacement and pressure of a secondary suspension air spring, the wheel diameter, the wheel pair mass, the frame mass, the vehicle body mass and vehicle operation parameter data; calculating the peak-to-peak value, the average value, the standard deviation, the kurtosis, the waveform factor, the pulse factor, the margin factor and the peak value factor of the above data respectively, and constructing vehicle parameter features, suspension system parameter features and features of the fusion of the vehicle parameters and the suspension system parameters; inputting the constructed features into a trained vehicle wheel polygon fault diagnosis model to obtain the fault diagnosis result of the vehicle wheel polygon; the vehicle wheel polygon fault diagnosis model comprises: an input layer for inputting the vehicle parameter features, the suspension system parameter features and the features of the fusion of the vehicle parameters and the suspension system parameters respectively; an automatic learning layer into which the vehicle parameter features and the suspension system parameter features are inputted after passing through a plurality of convolution layers and pooling layers, the automatic learning layer obtaining the optimal connection relationship between the upper-layer neurons and the lower-layer neurons through training, and the output of the automatic learning layer being inputted into a full connection layer; the features of the fusion of the vehicle parameters and the suspension system parameters also enter the full connection layer after passing through a plurality of convolution layers and pooling layers; the full connection layer classifies the received data, and the classification result is sent to an output layer; meanwhile, the output of the full connection layer is also fed back to the automatic learning layer; an output layer receiving the data inputted by the full connection layer and outputting the vehicle wheel polygon fault diagnosis result; the features of the fusion of the vehicle parameters and the suspension system parameters comprise:

2. A rail vehicle wheel polygonal fault diagnosis method according to claim 1, characterized in that, the average value of the displacement of the primary suspension steel spring when the vehicle speed is greater than a certain set value, the average value of the displacement of the secondary suspension air spring when the vehicle speed is greater than a certain set value, the peak-to-peak value of the displacement of the primary suspension steel spring when the vehicle speed is greater than a certain set value, the peak-to-peak value of the displacement of the secondary suspension air spring when the vehicle speed is greater than a certain set value, the weighted average of the average value within a set vehicle speed time and the standard deviation of the displacement of the primary suspension steel spring, the weighted average of the average value within a set vehicle speed time and the standard deviation of the displacement of the secondary suspension air spring, the weighted average of the average value within a set vehicle speed time and the kurtosis of the displacement of the primary suspension steel spring, the weighted average of the average value within a set vehicle speed time and the kurtosis of the displacement of the secondary suspension air spring, the weighted average of the average value within a set vehicle speed time and the waveform factor of the displacement of the primary suspension steel spring, the weighted average of the average value within a set vehicle speed time and the waveform factor of the displacement of the secondary suspension air spring, the weighted average of the average value within a set vehicle speed time and the pulse factor of the displacement of the primary suspension steel spring, the weighted average of the average value within a set vehicle speed time and the pulse factor of the displacement of the secondary suspension air spring, the weighted average of the average value within a set vehicle speed time and the margin factor of the displacement of the primary suspension steel spring, the weighted average of the average value within a set vehicle speed time and the margin factor of the displacement of the secondary suspension air spring, the weighted average of the average value within a set vehicle speed time and the peak value factor of the displacement of the primary suspension steel spring and the weighted average of the average value within a set vehicle speed time and the peak value factor of the displacement of the secondary suspension air spring.

3. A rail vehicle wheel polygonal fault diagnostic method according to claim 1 or 2, characterized in that, The vehicle operation parameter data comprises the vehicle speed, the mileage and the traction brake level. the vehicle parameter features comprise: The vehicle speed, the peak-to-peak value in the speed setting time, the average value in the speed setting time, the standard deviation in the speed setting time, the total mass of the vehicle, the mileage change in the setting time, and the mode of the traction brake level in the setting time.

4. A rail vehicle wheel polygonal fault diagnosis method according to claim 1, characterized in that, The suspension system parameter features are constructed, and specifically include: The first level suspension steel spring displacement, the first level suspension steel spring displacement kurtosis, the first level suspension steel spring displacement waveform factor, the first level suspension steel spring displacement pulse factor, the first level suspension steel spring displacement margin factor, the first level suspension steel spring displacement peak factor, the second level suspension air spring displacement, the second level suspension air spring displacement kurtosis, the second level suspension air spring displacement waveform factor, the second level suspension air spring displacement pulse factor, the second level suspension air spring displacement margin factor, and the second level suspension air spring displacement peak factor.

5. A rail vehicle wheel polygonal fault diagnosis method according to claim 1, characterized in that, The wheel polygon fault diagnosis result includes the order and the fault degree of the wheel polygon.

6. A rail vehicle wheel polygonal fault diagnosis system using the rail vehicle wheel polygonal fault diagnosis method according to claim 1, characterized by, The method includes: The data acquisition module is configured to acquire the stiffness and displacement of the first level suspension steel spring, the stiffness, displacement, and pressure of the second level suspension air spring, the wheel diameter, the wheelset mass, the bolster mass, the carbody mass, and the vehicle operation parameter data; The feature extraction module is configured to calculate the peak-to-peak value, the average value, the standard deviation, the kurtosis, the waveform factor, the pulse factor, the margin factor, and the peak factor of the above data, respectively, to construct the vehicle parameter features, the suspension system parameter features, and the features of the fusion of the vehicle parameters and the suspension system parameters; The fault diagnosis module is configured to input the constructed features into the trained wheel polygon fault diagnosis model to obtain the wheel polygon fault diagnosis result.

7. A terminal device comprising a processor and a memory, the processor being configured to implement instructions; the memory being configured to store a plurality of instructions, wherein the terminal device is configured to perform the method according to any one of claims 1-6. The instructions are adapted to be loaded and executed by the processor to perform the rail vehicle wheel polygon fault diagnosis method of any one of claims 1-5.

8. A computer-readable storage medium having stored therein a plurality of instructions, wherein the instructions, when executed by a processor, cause the processor to perform operations comprising: The instructions are adapted to be loaded and executed by the processor of the terminal device to perform the rail vehicle wheel polygon fault diagnosis method of any one of claims 1-5.

Images