Performance testing method for train air compressor, train and storage medium

Abstract

The present invention provides a performance detection method for a train air compressor, a train, and a storage medium. The method includes: establishing a first network model between historical data on the status of air-consuming equipment and a first total air pressure change during a decrease in total air pressure when each air compressor is turned off; establishing a second network model between historical data on the status of each air compressor, an output value of the first network model, and a second total air pressure change during a rise in total air pressure when each air compressor is turned on and working normally; inputting the acquired real-time data on the status of air-consuming equipment into the first network model to obtain a first output value; inputting the real-time data on the status of each air compressor and the first output value into the second network model to obtain a second output value; and determining the performance of each air compressor based on the second output value and the actual total air pressure change during the train operation. The present invention can realize early warning and preventive maintenance of train air compressor performance, thereby improving the operational quality, efficiency, and safety of the vehicle.

Description

Technical Field

[0001] The present invention relates to the technical field of train braking, and in particular to a performance detection method for a train air compressor, a train and a storage medium. Background Art

[0002] As more and more EMUs enter operation, China's high-speed rail will face significant challenges in safe operation and routine inspection and maintenance. Safe operation and reduced maintenance costs are pressing challenges for China's high-speed train development. Air brakes are a crucial component of the train's braking system. Air compressors, also known as air compressors, provide the air supply that powers the air brakes. Their performance significantly impacts the safe operation of trains, necessitating diagnostics and early warning of compressor anomalies.

[0003] As the train's ventilation system, the air compressor continuously supplies compressed air to maintain a total air pressure of 800-950 kPa. Deterioration in air compressor performance can lead to abnormal total air pressure, impacting the safe operation of the train. Currently, by measuring the air compressor's air flow time on EMUs, a standard for abnormal air compressor air flow time has been established. If the main air compressor air flow time exceeds the preset standard during EMU operation, an early warning will be issued.

[0004] However, air leakage from other air-consuming equipment in both faulty and non-faulty states can also cause abnormal total air pressure, thereby affecting the air compressor performance judgment result, making it impossible to determine whether the abnormality is caused by the degradation of air compressor performance or air leakage from other air-consuming equipment. Summary of the Invention

[0005] The embodiments of the present invention provide a performance detection method for a train air compressor, a train, and a storage medium to solve the problem in the prior art that the performance of the air compressor cannot be accurately determined.

[0006] In a first aspect, an embodiment of the present invention provides a method for detecting the performance of a train air compressor, comprising:

[0007] Establishing a first network model between historical data of air-consuming equipment status and corresponding first total air pressure change during a process of total air pressure drop when each air compressor in the train is shut down;

[0008] Establishing a second network model between the historical status data of each air compressor during the process of total air pressure rising when each air compressor in the train is turned on and working normally, the output value of the first network model, and the second total air pressure change corresponding to the historical status data of each air compressor;

[0009] Acquire real-time status data of air-consuming equipment and each air compressor when each air compressor is turned on and working normally during train operation;

[0010] Inputting the real-time status data of the air-consuming equipment into the first network model to obtain a first output value; inputting the real-time status data of each air compressor and the first output value into the second network model to obtain a second output value;

[0011] The performance of each air compressor is determined according to the second output value and the actual total wind pressure change during the train operation.

[0012] In one possible implementation, establishing a first network model between historical data on the status of air-consuming equipment and corresponding first total air pressure changes during a decrease in total air pressure when each air compressor in the train is shut down includes:

[0013] Obtain historical data on the status of air-consuming equipment and the corresponding first total air pressure change during the process of total air pressure reduction when each air compressor in the train is turned off;

[0014] A first network model is obtained by training using the historical data of the status of the air-consuming equipment as an input value and the first total wind pressure change as an output value.

[0015] In a possible implementation, the training of the first network model using the historical data of the status of the air-consuming equipment as an input value and the first total wind pressure change as an output value includes:

[0016] Taking the historical data of the status of the air-consuming equipment and the corresponding first total wind pressure change as a sample set, and dividing the sample set into a training sample set and a verification sample set according to a preset ratio;

[0017] Randomly generate initial weight values ​​and initial bias values, and build N preset neural network models, where N is a positive integer;

[0018] Setting different numbers of hidden neurons in the N preset neural network models, and using the training sample sets to train the N preset neural network models respectively;

[0019] Inputting the verification sample sets into the N first neural network models obtained after training for verification, and determining the target number of hidden neurons when the verification results meet the preset conditions;

[0020] Constructing M preset neural network models with the target number of hidden neurons, where M is a positive integer;

[0021] Randomly generate M groups of variables, each group of variables includes a set of weight values ​​and a set of bias values, and use the training sample set to train the M preset neural network models respectively;

[0022] Inputting the verification samples into the trained M second neural network models for verification respectively, and determining the target weight value and the target deviation value when the verification result meets the second preset condition;

[0023] A preset neural network model obtained by training is constructed using the target number of hidden neurons, the target initial weight value, and the target initial deviation value to obtain a first network model.

[0024] In a possible implementation, the verification sample sets are respectively input into the N first neural network models obtained after training for verification, and when the verification results meet preset conditions, the target number of hidden neurons is determined, including:

[0025] Inputting the verification sample set into the N first neural network models obtained after training respectively, and calculating the corresponding N first mean absolute errors;

[0026] When the corresponding number of hidden neurons in the two first neural network models increases, the corresponding first mean absolute error decreases and the decrease value is within a preset range, determining the hidden neuron data with the least number in the two first neural networks as the target number of hidden neurons;

[0027] The verification samples are respectively input into the trained M second neural network models for verification. When the verification results meet the second preset condition, a target weight value and a target deviation value are determined, including:

[0028] Inputting the verification samples into the trained M second neural network models respectively, and calculating M second mean absolute error values;

[0029] Each group of variables is used as input and the mean absolute error value is used as the evaluation value. A genetic algorithm is performed based on each mean absolute error value and each corresponding group of variables to determine the optimal group of variables, and the optimal group of variables is used as the target weight value and the target deviation value.

[0030] In one possible implementation, establishing a second network model between historical status data of each air compressor during a total air pressure increase process when each air compressor in the train is turned on and working normally, an output value of the first network model, and a second total air pressure change corresponding to the historical status data of each air compressor includes:

[0031] Obtaining historical status data of each air compressor and corresponding second total air pressure change during the total air pressure increase process when each air compressor in the train is turned on and working normally;

[0032] The second network model is obtained by training using the historical data of the status of each air compressor and the output value of the first network model as input values ​​and the second total wind pressure change as output value.

[0033] In a possible implementation, after establishing the first network model between the historical data of the status of air-consuming equipment and the corresponding first total air pressure change during the process of total air pressure decreasing when each air compressor in the train is shut down, the method further includes:

[0034] Acquire real-time data on the status of air-consuming equipment and the real-time change in total air pressure during train operation when the total air pressure drops and all air compressors are not turned on;

[0035] updating the first network model using the real-time data of the status of the air-consuming equipment as an input value and the real-time change in the total wind pressure as an output value;

[0036] The step of inputting the real-time data of the status of the air-consuming equipment into the first network model to obtain a first output value includes:

[0037] The real-time data of the status of the air-consuming equipment is input into the updated first network model to obtain a first output value.

[0038] In a possible implementation, determining the performance of each air compressor according to the second output value and the actual total air pressure change during the train operation includes:

[0039] Obtaining t second output values ​​output by the second network model within a preset t sampling periods, where t is a positive integer;

[0040] Obtain t actual total air pressure changes corresponding to each air compressor being turned on and working normally during the train operation within the preset t sampling periods;

[0041] Calculating mean square errors between the t second output values ​​and the t actual total wind pressure changes;

[0042] The performance of each air compressor is determined according to the change of the mean square error value.

[0043] In a possible implementation, determining the performance of each air compressor based on the change in the mean square error value includes:

[0044] Calculating an evaluation index based on the mean square error value and the mean square error standard value;

[0045] The performance of each air compressor is determined according to the evaluation index.

[0046] In a second aspect, an embodiment of the present invention provides a train, comprising an electronic device, the electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the method for detecting the performance of a train air compressor as described in the first aspect or any possible implementation of the first aspect are implemented.

[0047] In a third aspect, an embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the performance detection method of the train air compressor as described in the first aspect or any possible implementation method of the first aspect.

[0048] An embodiment of the present invention provides a performance detection method for a train air compressor, a train and a storage medium. By using historical data of the air compressor in different states to establish a first network model and a second network model, the data collected in real time during the operation of the train is input into the two network models to obtain the performance of the air compressor. This is more accurate than the existing method of determining the performance of the air compressor only by counting the air compressor's blowing time, and realizes performance warning and preventive maintenance of the train air compressor, thereby improving the vehicle's operating quality, operating efficiency and operating safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0050] Figure 1 This is a flow chart of a method for detecting the performance of a train air compressor provided by an embodiment of the present invention;

[0051] Figure 2 is a flowchart of establishing a first network model according to an embodiment of the present invention;

[0052] Figure 3 1 is a schematic diagram of the relationship between a first total wind pressure change value and time provided by an embodiment of the present invention;

[0053] Figure 4 Schematic diagram of the relationship between the output value of the first network model and the actual total wind pressure change value provided by an embodiment of the present invention;

[0054] Figure 5 1 is a schematic diagram of the relationship between the second total wind pressure change value and time provided by an embodiment of the present invention;

[0055] Figure 6 Schematic diagram of the relationship between the output value of the second network model and the actual total wind pressure change value provided by an embodiment of the present invention;

[0056] Figure 7 1 is a schematic structural diagram of a performance detection device for a train air compressor provided by an embodiment of the present invention;

[0057] Figure 8 is a schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0058] In the following description, specific details such as particular system structures and techniques are provided for purposes of illustration, not limitation, to facilitate a thorough understanding of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.

[0059] In order to make the purpose, technical solutions and advantages of the present invention more clear, specific embodiments will be described below with reference to the accompanying drawings.

[0060] Figure 1 The following is a flowchart of a method for detecting the performance of a train air compressor provided by an embodiment of the present invention:

[0061] Step 101 : establishing a first network model between historical data of status of air-consuming equipment and corresponding first total air pressure variation during a process of total air pressure reduction when each air compressor in the train is shut down.

[0062] In order to study the characteristics of air compressor performance degradation and eliminate the influence of other air-consuming devices, it is necessary to first establish an air consumption model of each air-consuming device. Therefore, the first network model is established in step 101.

[0063] In one embodiment, see Figure 2 , establishing a first network model between the historical data of the status of air-consuming equipment and the corresponding first total air pressure change during the process of total air pressure decreasing when each air compressor in the train is turned off, may include:

[0064] Step 201 : obtaining historical data on the status of air-consuming equipment and corresponding first total air pressure changes during a process of total air pressure reduction when each air compressor in the train is shut down.

[0065] See also Figure 3 The data marked in the bold box is the data of the total air pressure drop when the air compressor is turned off in the historical data. The status of each air-consuming device and the first total air pressure change at the next moment are used as a sample to form a first data sample set. The structure of each sample in the first sample set is:

[0066]

[0067]

[0068] Among them, s i is a data sample, i is the sample number, xi The input part of the sample is the state of each air-consuming device [k1, k2, ... n ], n represents the number of air-consuming equipment, y i The output part of the sample is the total wind pressure change ΔP at the next moment z , m represents the number of samples, S represents m data samples, X represents the input part of m samples, and Y represents the output part of m samples.

[0069] It should be noted that the air-consuming equipment on the EMU may include pantographs, doors, fans, air intake and exhaust devices, air ducts, air purification equipment, air pressure fluctuation control devices, etc.

[0070] In step 202 , a first network model is obtained by training using the historical data of the status of the air-consuming equipment as input and the first total air pressure change as output.

[0071] When training the first network model, the sample set is divided into a training sample set and a validation sample set. The training sample set is used to perform BP neural network training on the relationship between the status of the air-consuming equipment and the change in the first total air pressure to obtain a neural network model. The validation sample set is used to verify the neural network model, and the mean absolute error (MAE) is calculated to determine the final target network model, i.e., the first network model.

[0072] Optionally, the historical data of the status of the air-consuming equipment and the corresponding first total air pressure change are used as a sample set, and the sample set is divided into a training sample set and a verification sample set according to a preset ratio;

[0073] Randomly generate initial weight values ​​and initial bias values, and build N preset neural network models, where N is a positive integer;

[0074] The numbers of hidden neurons in N preset neural network models are set to be different, and the N preset neural network models are trained respectively using the training sample sets;

[0075] The validation sample sets are input into the N first neural network models obtained after training for validation. When the validation results meet the preset conditions, the target number of hidden neurons is determined.

[0076] Construct M preset neural network models with the target number of hidden neurons, where M is a positive integer;

[0077] Randomly generate M groups of variables, each group of variables includes a set of weight values ​​and a set of bias values, and use the training sample set to train M preset neural network models respectively;

[0078] Inputting the verification samples into the trained M second neural network models for verification, and determining the target weight value and the target deviation value when the verification result meets the second preset condition;

[0079] A neural network model obtained by training is constructed by selecting a target number of hidden neurons, a target initial weight value, and a target initial bias value to obtain a first network model.

[0080] In one embodiment, the validation sample sets are respectively input into the N first neural network models obtained after training for validation. When the validation results meet preset conditions, the target number of hidden neurons is determined, including:

[0081] Input the validation sample set into the N first neural network models obtained after training, and calculate the corresponding N first mean absolute errors;

[0082] When the corresponding number of hidden neurons in the two first neural networks increases, the corresponding first mean absolute error decreases and the decrease value is within a preset range, the hidden neuron data with the least number in the two first neural networks is determined as the target number of hidden neurons.

[0083] For example, N is 6, that is, the corresponding numbers of hidden neurons in the 6 first neural network models can be 1, 2, 3...6 respectively. If the number of hidden neurons in the two first neural network models changes from 4 to 5, the MAE corresponding to the first neural network with 4 hidden neurons is reduced by 10% from the MAE corresponding to the first neural network with 5 hidden neurons, and the MAE corresponding to the increase of hidden neurons from 5 to 6 is reduced by 1%, then the number of hidden neurons is selected to be 5.

[0084] The verification samples are respectively input into the M second neural network models after training for verification. When the verification results meet the second preset condition, the target weight value and the target deviation value are determined, including:

[0085] Input the validation samples into the trained M second neural network models respectively, and calculate M second mean absolute error values;

[0086] Each group of variables is used as input and the mean absolute error value is used as the evaluation value. The genetic algorithm is calculated based on each mean absolute error value and the corresponding group of variables to determine the optimal group of variables. The optimal group of variables is used as the target weight value and target deviation value.

[0087] For example, using MAE as the evaluation function, randomly generated weights and biases are encoded using real numbers. Genetic algorithms such as crossover and mutation are then applied to the real-time encodings to generate a new generation of weights and biases. After multiple iterations, the minimum set of weight and bias real-number encodings is obtained. The weight and bias values ​​corresponding to the minimum set of weight and bias real-number encodings are then determined as the target weight and bias values.

[0088] The following example illustrates the training process of the first network model;

[0089] Using historical operating data when the air compressor was not turned on, a sample set was constructed with the status of air-consuming devices (such as the two pantographs, vehicle door status, parking brake status, each vehicle's air brake force setting, and each vehicle's air spring pressure) as input, and the total air pressure change as output. The sample set was divided into a training sample set and a validation sample set. A BP neural network was trained on the relationship between the status of air-consuming devices and the total air pressure change using the training sample set, resulting in neural network model 1. The validation sample set was used to validate the neural network model, and the mean absolute error (MAE) was calculated.

[0090] The number of input layer neurons is 28, and the number of input parameters is 28, including 2 pantograph states, door states, parking brake states, air brake force setting values ​​of each vehicle, and air spring pressure values ​​of each vehicle. The number of output layer neurons is 1, and the output is the total wind pressure change. By selecting different numbers of hidden layer neurons to calculate MAE, it is found that when the number of hidden neurons is 9, the MAE does not decrease significantly compared with the number of hidden neurons is 8, and the MAE does not decrease significantly with the increase of the number of hidden neurons. Therefore, the number of hidden layer neurons is selected as 9.

[0091] The initial number of weights is n iw =n i n h +n h n o =28*8+8*1=232;

[0092] The number of deviations is n ib =n h +n o =8+1=9.

[0093] Real number coding is used for n iw +n ib = 241 genes are processed to form an individual, and the genetic algorithm is optimized with MAE as the evaluation function to obtain the optimal weight value and bias value, which are input into the neural network model 1 to obtain the final first network model. The output of the first network model is consistent with the actual total wind pressure change as shown in Figure 4 As shown in the figure, the dotted curve is the model output, and the solid curve is the actual total wind pressure change.

[0094] Step 102: Establish a second network model between the historical status data of each air compressor, the output value of the first network model, and the second total air pressure change corresponding to the historical status data of each air compressor during the process of increasing total air pressure when each air compressor in the train is turned on and working normally.

[0095] Based on the first network model, a relationship model between the air compressor status, the output of the first network model, and the second total air pressure change was established. Using historical operating data of air compressors operating normally and in the on state, the second network model was trained with the air compressor status and the output of the first network model as input and the second total air pressure change as output.

[0096] A second network model is established between the historical status data of each air compressor, the output value of the first network model, and the second total air pressure change corresponding to the historical status data of each air compressor during the process of total air pressure rising when each air compressor in the train is turned on and operates normally, including:

[0097] Obtain the historical data of each air compressor status and the corresponding second total air pressure change during the total air pressure increase process when each air compressor in the train is turned on and working normally; see Figure 5 The data marked in the bold box is the second total air pressure change corresponding to the total air pressure rising process when each air compressor is turned on and working normally.

[0098] The state of each air compressor, the output value of the first network model and the second total wind pressure change are taken as a sample to form a second data sample set. The structure of each sample contained in the second data sample set is

[0099]

[0100]

[0101] Among them, s i is a data sample, i is the sample number, x i The input part of the sample is the state of each air-consuming device [k1, k2, ... n ] and the output value V0 of the first network model, n represents the number of wind-consuming devices, y i The output part of the sample is the total wind pressure change ΔP at the next moment z , m represents the number of samples, S represents m data samples, X represents the input part of m samples, and Y represents the output part of m samples.

[0102] The second network model is trained by taking the historical data of the status of each air compressor and the output value of the first network model as input values ​​and the second total wind pressure change as output value.

[0103] The second data sample set is divided into a training sample set and a verification sample set. The training sample set is used to perform BP neural network training on the relationship between the status of each air compressor, the output value of the first network model and the change in the second total wind pressure to obtain a second neural network model. The second neural network model is verified using the verification sample set, and the mean absolute error (MAE) is calculated. The number of hidden neurons, the optimal initial weight value and the deviation are determined using the same method as in the training of the first network model. The second neural network model is trained to obtain the final second network model, and the mean square error of the verification sample set is recorded as a benchmark for evaluating the performance of the air compressor during train operation. It should be noted that the method of training the second network model is the same as that of training the first network model, and will not be repeated here.

[0104] For example:

[0105] Using historical operating data from air compressors operating normally and in the on state, a sample set was constructed with the two air compressor states and the output of the first network model as input. The sample set was divided into a training sample set and a validation sample set. The training sample set was used to train a BP neural network on the relationship between the two air compressor states, the output of the first network model, and the change in the second total air pressure, resulting in Neural Network Model 2. The validation sample set was used to validate the neural network model, and the mean absolute error (MAE) was calculated.

[0106] The number of neurons in the input layer is 3, which is the number of input parameters and includes the states of two air compressors and the output value of the first network model. The number of neurons in the output layer is 1, and the output is the change in total wind pressure. By selecting different numbers of hidden neurons in the hidden layer to calculate the MAE, it is found that when the number of hidden neurons in the hidden layer is 4, the MAE does not decrease significantly compared with when the number of hidden neurons in the hidden layer is 3, and the MAE does not decrease significantly with the increase of the number of hidden layer neurons. Therefore, the number of hidden layer neurons is selected as 3.

[0107] Initial weight number n iw =n i n h +n h n o =3*3+3*1=12;

[0108] The number of initial deviations n ib =n h +n o =3+1=4;

[0109] Real number coding is used for n iw +n ib= 16 genes are processed to form an individual, and the genetic algorithm is optimized with MAE as the evaluation function to obtain the optimal initial weight value and deviation, which are input into the neural network model 2 to obtain the final second network model. The output of the second network model is consistent with the actual total wind pressure change as shown in Figure 6 As shown in the figure, the dotted curve is the model output, and the solid curve is the actual total wind pressure change.

[0110] Step 103 : Acquire the real-time status data of the air-consuming equipment and the real-time status data of each air compressor when each air compressor is turned on and operates normally during the operation of the train.

[0111] After the first network model and the second network model are trained, the first network model and the second network model can be used to detect data from actual train operation to determine the performance of the air compressor.

[0112] Step 104: input the real-time status data of the air-consuming equipment into the first network model to obtain a first output value; input the real-time status data of each air compressor and the first output value into the second network model to obtain a second output value.

[0113] During the operation of the train, since the conditions of each air-consuming device may change or malfunction, such as blockage or air leakage of the air-consuming device, the first network model between the historical data of the status of each air-consuming device and the first total air pressure change will change. Therefore, during the operation of the train, a new sample set is formed using the data when the air compressor is not turned on, and the first network model is updated and trained in real time to ensure that the first network model remains accurate when the condition of the air-consuming device changes or malfunctions.

[0114] In one embodiment, after establishing a first network model between the historical data of the status of air-consuming equipment and the corresponding first total air pressure change during the process of total air pressure reduction when each air compressor in the train is turned off, the method further includes:

[0115] Acquire real-time data on the status of air-consuming equipment and the real-time change in total air pressure during train operation when the total air pressure drops and all air compressors are not turned on;

[0116] The first network model is updated using the real-time data of the status of the air-consuming equipment as input and the real-time change of the total air pressure as output;

[0117] Optionally, inputting the real-time data of the status of the air-consuming equipment into the first network model to obtain a first output value includes:

[0118] The real-time data of the status of the air-consuming equipment is input into the updated first network model to obtain a first output value.

[0119] Step 105 : determining the performance of each air compressor according to the second output value and the actual total air pressure change during the train operation.

[0120] Optionally, determining the performance of each air compressor based on the second output value and the actual total air pressure change during the train operation includes:

[0121] Obtain t second output values ​​output by the second network model within a preset t sampling period, where t is a positive integer; for example, the t second output values ​​output by the second network model may be [V 2o1 ,V 2o2 ,…V 2ot ];

[0122] Obtain t actual total air pressure changes corresponding to each air compressor being turned on and working normally during the train operation within the preset t sampling periods; for example, t actual total air pressure changes can be [ΔP1, ΔP2, … ΔP t ];

[0123] Calculate the mean square error between t second output values ​​and t actual total wind pressure changes; optionally, use Calculate the mean square error between the t second output values ​​and the t actual total wind pressure changes, where S represents the calculated mean square error.

[0124] The performance of each air compressor is determined based on the change in the mean square error value.

[0125] Optionally, the performance of each air compressor is determined based on the change in the mean square error value, including:

[0126] Calculate the evaluation index based on the mean square error value and the mean square error standard value;

[0127] Determine the performance of each air compressor based on the evaluation indicators.

[0128] The evaluation index is calculated based on the mean square error value Here, S0 represents the standard mean square error (SME). As the air compressor performance deteriorates during train operation, the relationship between the actual train total air pressure change, the air compressor status, and the output value of the first network model changes, and the deviation from the second network model increases, causing the SME to gradually increase. Therefore, in theory, Q = 1 indicates the best air compressor performance. Therefore, when Q ≤ 1, the closer Q is to 1, the better the air compressor performance; the smaller Q is, the worse the air compressor performance. Therefore, when Q is less than the preset value, for example, when Q is less than 1 / 10 to 1 / 20, the SME is 10-20 times the original value, indicating a significant decline in air compressor performance, and an early warning of air compressor performance is issued.

[0129] The embodiment of the present invention establishes a first network model between the historical status data of air-consuming equipment and the corresponding first total air pressure change during the process of total air pressure decreasing when each air compressor in the train is turned off; establishes a second network model between the historical status data of each air compressor, the output value of the first network model and the second total air pressure change corresponding to the historical status data of each air compressor during the process of total air pressure increasing when each air compressor in the train is turned on and working normally; obtains the real-time status data of air-consuming equipment and the real-time status data of each air compressor when each air compressor is turned on and working normally during the operation of the train; inputs the real-time status data of air-consuming equipment into the first network model to obtain a first output value; inputs the real-time status data of each air compressor and the first output value into the second network model to obtain a second output value; and determines the performance of each air compressor based on the second output value and the actual total air pressure change during the operation of the train. The embodiment of the present invention uses historical data of the air compressor under different states to establish a first network model and a second network model, so that the data collected in real time during the operation of the train are input into the two network models to obtain the performance of the air compressor. This is more accurate than the existing method of determining the performance of the air compressor by only counting the air compressor's blowing time, and realizes early warning and preventive maintenance of the train air compressor performance, thereby improving the operational quality, operational efficiency and operational safety of the vehicle. In addition, after the first network model is established, the first network model is updated according to the real-time data collected during the operation of the train, so that when the air compressor is in a faulty or non-faulty state, the influence of the failure of each air-consuming equipment during operation on the change value of the total air pressure can be more accurately reflected, thereby obtaining a more accurate air compressor performance judgment result.

[0130] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0131] The following are device embodiments of the present invention. For details not fully described therein, reference may be made to the corresponding method embodiments described above.

[0132] Figure 7 A schematic diagram of the structure of a performance detection device for a train air compressor provided by an embodiment of the present invention is shown. For ease of explanation, only the parts related to the embodiment of the present invention are shown, which are described in detail as follows:

[0133] like Figure 7 As shown, a performance detection device 7 for a train air compressor includes: a model building module 71 , an acquisition module 72 and a calculation module 73 .

[0134] A model building module 71 is used to build a first network model between the historical data of the status of air-consuming equipment and the corresponding first total air pressure change during the process of total air pressure reduction when each air compressor in the train is turned off;

[0135] The model building module 71 is further configured to build a second network model based on the historical status data of each air compressor, the output value of the first network model, and the second total air pressure change corresponding to the historical status data of each air compressor during the process of total air pressure rising when each air compressor in the train is turned on and working normally;

[0136] An acquisition module 72 is used to acquire real-time data on the status of air-consuming equipment and each air compressor when each air compressor is turned on and working normally during train operation;

[0137] The calculation module 73 is used to input the real-time status data of the air-consuming equipment into the first network model to obtain a first output value; and input the real-time status data of each air compressor and the first output value into the second network model to obtain a second output value;

[0138] The calculation module 73 is further configured to determine the performance of each air compressor according to the second output value and the actual total air pressure change during the train operation.

[0139] In one possible implementation, when the model building module 71 builds a first network model between the historical data of the status of air-consuming equipment and the corresponding first total air pressure change during the process of total air pressure reduction when each air compressor in the train is shut down, it is used to:

[0140] Obtain historical data on the status of air-consuming equipment and the corresponding first total air pressure change during the process of total air pressure reduction when each air compressor in the train is turned off;

[0141] The first network model is obtained by training using the historical data of the status of the air-consuming equipment as input and the first total air pressure change as output.

[0142] In one possible implementation, the model building module 71 uses the historical data of the status of the wind-consuming equipment as input and the first total wind pressure change as output to train the first network model, and is used to:

[0143] Taking the historical data of the status of the air-consuming equipment and the corresponding first total air pressure change as a sample set, the sample set is divided into a training sample set and a verification sample set according to a preset ratio;

[0144] Randomly generate initial weight values ​​and initial bias values, and build N preset neural network models, where N is a positive integer;

[0145] The numbers of hidden neurons in N preset neural network models are set to be different, and the N preset neural network models are trained respectively using the training sample sets;

[0146] The validation sample sets are input into the N first neural network models obtained after training for validation. When the validation results meet the preset conditions, the target number of hidden neurons is determined.

[0147] Construct M preset neural network models with the target number of hidden neurons, where M is a positive integer;

[0148] Randomly generate M groups of variables, each group of variables includes a set of weight values ​​and a set of bias values, and use the training sample set to train M preset neural network models respectively;

[0149] Inputting the verification samples into the trained M second neural network models for verification, and determining the target weight value and the target deviation value when the verification result meets the second preset condition;

[0150] A neural network model obtained by training is constructed by selecting a target number of hidden neurons, a target initial weight value, and a target initial bias value to obtain a first network model.

[0151] In one possible implementation, the model building module 71 inputs the verification sample sets into the N first neural network models obtained after training for verification. When the verification results meet the preset conditions, the target number of hidden neurons is determined by:

[0152] Input the validation sample set into the N first neural network models obtained after training, and calculate the corresponding N first mean absolute errors;

[0153] When the corresponding number of hidden neurons in the two first neural network models increases, the corresponding first mean absolute error decreases and the decrease value is within a preset range, determining the hidden neuron data with the least number in the two first neural networks as the target number of hidden neurons;

[0154] The model building module 71 inputs the verification samples into the trained M second neural network models for verification. When the verification results meet the second preset conditions, the target weight value and the target deviation value are determined to:

[0155] Input the validation samples into the trained M second neural network models respectively, and calculate M second mean absolute error values;

[0156] Each group of variables is used as input and the mean absolute error value is used as the evaluation value. The genetic algorithm is calculated based on each mean absolute error value and the corresponding group of variables to determine the optimal group of variables. The optimal group of variables is used as the target weight value and target deviation value.

[0157] In one possible implementation, when the model building module 71 builds a second network model between the historical status data of each air compressor, the output value of the first network model, and the second total air pressure change corresponding to the historical status data of each air compressor during the process of increasing total air pressure when each air compressor in the train is turned on and working normally, it is used to:

[0158] Obtaining historical status data of each air compressor and corresponding second total air pressure change during the total air pressure increase process when each air compressor in the train is turned on and working normally;

[0159] The second network model is trained by taking the historical data of the status of each air compressor and the output value of the first network model as input values ​​and the second total wind pressure change as output value.

[0160] In one possible implementation, after establishing the first network model between the historical data of the status of air-consuming equipment and the corresponding first total air pressure change during the process of total air pressure reduction when each air compressor in the train is shut down, the model building module 71 is further configured to:

[0161] Acquire real-time data on the status of air-consuming equipment and the real-time change in total air pressure during train operation when the total air pressure drops and all air compressors are not turned on;

[0162] The first network model is updated using the real-time data of the status of the air-consuming equipment as input and the real-time change of the total air pressure as output;

[0163] Inputting the real-time data of the status of the air-consuming equipment into the first network model to obtain a first output value includes:

[0164] The real-time data of the status of the air-consuming equipment is input into the updated first network model to obtain a first output value.

[0165] In one possible implementation, when the calculation module 73 determines the performance of each air compressor based on the second output value and the actual total air pressure change during the train operation, it is configured to:

[0166] Obtaining t second output values ​​output by the second network model within a preset t sampling periods, where t is a positive integer;

[0167] Obtain t actual total air pressure changes corresponding to each air compressor being turned on and working normally during the train operation within the preset t sampling periods;

[0168] Calculating mean square error values ​​between t second output values ​​and t actual total wind pressure changes;

[0169] The performance of each air compressor is determined based on the change in the mean square error value.

[0170] In one possible implementation, when the calculation module 73 determines the performance of each air compressor based on the change in the mean square error value, it is configured to:

[0171] Calculate the evaluation index based on the mean square error value and the mean square error standard value;

[0172] Determine the performance of each air compressor based on the evaluation indicators.

[0173] The above-mentioned train air compressor performance detection device establishes a first network model between the historical data of the status of air-consuming equipment and the corresponding first total air pressure change during the decrease of total air pressure when each air compressor in the train is turned off through a model establishment module, and establishes a second network model between the historical data of the status of each air compressor, the output value of the first network model and the second total air pressure change corresponding to the historical data of the status of each air compressor during the increase of total air pressure when each air compressor in the train is turned on and working normally; the acquisition module acquires the real-time status data of air-consuming equipment and the real-time status data of each air compressor when each air compressor is turned on and working normally during the operation of the train; the calculation module inputs the real-time status data of air-consuming equipment into the first network model to obtain a first output value, and inputs the real-time status data of each air compressor and the first output value into the second network model to obtain a second output value, and then determines the performance of each air compressor based on the second output value and the actual total air pressure change during the operation of the train. The embodiment of the present invention uses historical data of the air compressor under different states to establish a first network model and a second network model, so that the data collected in real time during the operation of the train are input into the two network models to obtain the performance of the air compressor. This is more accurate than the existing method of determining the performance of the air compressor by only counting the air compressor's blowing time, and realizes early warning and preventive maintenance of the train air compressor performance, thereby improving the operational quality, operational efficiency and operational safety of the vehicle. In addition, after the first network model is established, the first network model is updated according to the real-time data collected during the operation of the train, so that when the air compressor is in a faulty or non-faulty state, the influence of the failure of each air-consuming equipment during operation on the change value of the total air pressure can be more accurately reflected, thereby obtaining a more accurate air compressor performance judgment result.

[0174] An embodiment of the present invention provides a train, including electronic equipment, such as Figure 8 As shown, the electronic device 8 of this embodiment includes: a processor 80, a memory 81, and a computer program 82 stored in the memory 81 and executable on the processor 80. When the processor 80 executes the computer program 82, the steps in the above-mentioned embodiments of the performance detection method for the train air compressor are implemented, such as Figure 1 Alternatively, when the processor 80 executes the computer program 82, the functions of the modules / units in the above-mentioned device embodiments are realized, for example Figure 7 Functions of the modules / units 71 to 73 shown.

[0175] For example, the computer program 82 may be divided into one or more modules / units, one or more modules / units being stored in the memory 81 and executed by the processor 80 to implement the present invention. One or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, and the instruction segments are used to describe the execution process of the computer program 82 in the electronic device 8. For example, the computer program 82 may be divided into Figure 7 Modules / units 71 to 73 are shown.

[0176] The electronic device 8 may include, but is not limited to, a processor 80 and a memory 81. Those skilled in the art will understand that Figure 8 It is only an example of the electronic device 8 and does not constitute a limitation of the electronic device 8. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0177] The processor 80 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor.

[0178] The memory 81 may be an internal storage unit of the electronic device 8, such as a hard disk or memory of the electronic device 8. The memory 81 may also be an external storage device of the electronic device 8, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 8. Furthermore, the memory 81 may include both an internal storage unit of the electronic device 8 and an external storage device. The memory 81 is used to store the computer program and other programs and data required by the electronic device. The memory 81 may also be used to temporarily store data that has been output or is about to be output.

[0179] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0180] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0181] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0182] In the embodiments provided by the present invention, it should be understood that the disclosed devices / electronic devices and methods can be implemented in other ways. For example, the device / electronic device embodiments described above are merely schematic. For example, the division of the modules or units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0183] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0184] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0185] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, it can implement the steps of the above-mentioned performance detection method embodiments of each train air compressor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form, etc. The computer-readable medium may include: any entity or device that can carry the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0186] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A performance testing method for a train air compressor, characterized in that: include: Establishing a first network model between historical data of air-consuming equipment status and corresponding first total air pressure change during a process of total air pressure drop when each air compressor in the train is shut down; Establishing a second network model between the historical status data of each air compressor during the process of total air pressure rising when each air compressor in the train is turned on and working normally, the output value of the first network model, and the second total air pressure change corresponding to the historical status data of each air compressor; Acquire real-time status data of air-consuming equipment and each air compressor when each air compressor is turned on and working normally during train operation; Inputting the real-time data of the status of the wind-consuming equipment into the first network model to obtain a first output value; Inputting the real-time status data of each air compressor and the first output value into the second network model to obtain a second output value; The performance of each air compressor is determined according to the second output value and the actual total wind pressure change during the train operation.

2. The performance detection method for a train air compressor according to claim 1, characterized in that: The first network model is established between the historical data of the status of air-consuming equipment and the corresponding first total air pressure change during the process of total air pressure decreasing when each air compressor in the train is turned off, including: Obtain historical data on the status of air-consuming equipment and the corresponding first total air pressure change during the process of total air pressure reduction when each air compressor in the train is turned off; A first network model is obtained by training using the historical data of the status of the air-consuming equipment as an input value and the first total wind pressure change as an output value.

3. The performance detection method for a train air compressor according to claim 2, characterized in that: The training of the first network model using the historical data of the status of the air-consuming equipment as an input value and the first total wind pressure change as an output value includes: Taking the historical data of the status of the air-consuming equipment and the corresponding first total wind pressure change as a sample set, and dividing the sample set into a training sample set and a verification sample set according to a preset ratio; Randomly generate initial weight values ​​and initial bias values, and build N preset neural network models, where N is a positive integer; Setting different numbers of hidden neurons in the N preset neural network models, and using the training sample sets to train the N preset neural network models respectively; Inputting the verification sample sets into the N first neural network models obtained after training for verification, and determining the target number of hidden neurons when the verification results meet the preset conditions; Constructing M preset neural network models with the target number of hidden neurons, where M is a positive integer; Randomly generate M groups of variables, each group of variables including a set of weight values ​​and a set of bias values, and use the training sample set to train the M preset neural network models respectively; Inputting the verification samples into the trained M second neural network models for verification respectively, and determining the target weight value and the target deviation value when the verification result meets the second preset condition; A neural network model obtained by training is constructed using the target number of hidden neurons, the target weight value, and the target deviation value to obtain a first network model.

4. The performance detection method for a train air compressor according to claim 3, characterized in that: The verification sample sets are respectively input into the N first neural network models obtained after training for verification, and when the verification results meet the preset conditions, the target number of hidden neurons is determined, including: Inputting the verification sample set into the N first neural network models obtained after training respectively, and calculating the corresponding N first mean absolute errors; When the corresponding number of hidden neurons in the two first neural network models increases, the corresponding first mean absolute error decreases and the decrease value is within a preset range, determining the number of hidden neurons with the least number in the two first neural networks as the target number of hidden neurons; The verification samples are respectively input into the trained M second neural network models for verification. When the verification results meet the second preset condition, a target weight value and a target deviation value are determined, including: Inputting the verification samples into the trained M second neural network models respectively, and calculating M second mean absolute error values; Each group of variables is used as input and the second mean absolute error value is used as the evaluation value. A genetic algorithm is performed based on each second mean absolute error value and each corresponding group of variables to determine the optimal group of variables, and the optimal group of variables is used as the target weight value and the target deviation value.

5. The performance detection method for a train air compressor according to claim 1, characterized in that: The second network model is established between the historical status data of each air compressor during the process of total air pressure rising when each air compressor in the train is turned on and working normally, the output value of the first network model, and the second total air pressure change corresponding to the historical status data of each air compressor, including: Obtaining historical status data of each air compressor and corresponding second total air pressure change during the total air pressure increase process when each air compressor in the train is turned on and working normally; The second network model is obtained by training using the historical data of the status of each air compressor and the output value of the first network model as input values ​​and the second total wind pressure change as output value.

6. The performance testing method for a train air compressor according to claim 1, characterized in that: After establishing the first network model between the historical data of the status of air-consuming equipment and the corresponding first total air pressure change during the process of decreasing the total air pressure when each air compressor in the train is turned off, the method further includes: Acquire real-time data on the status of air-consuming equipment and the real-time change in total air pressure during train operation when the total air pressure drops and all air compressors are not turned on; updating the first network model using the real-time data of the status of the air-consuming equipment as an input value and the real-time change in the total wind pressure as an output value; The step of inputting the real-time data of the status of the air-consuming equipment into the first network model to obtain a first output value includes: The real-time data of the status of the air-consuming equipment is input into the updated first network model to obtain a first output value.

7. The performance detection method for a train air compressor according to any one of claims 1 to 6, characterized in that: The determining the performance of each air compressor according to the second output value and the actual total air pressure change during the train operation includes: Obtaining t second output values ​​output by the second network model within a preset t sampling periods, where t is a positive integer; Obtain t actual total air pressure changes corresponding to each air compressor being turned on and working normally during the train operation within the preset t sampling periods; Calculating mean square errors between the t second output values ​​and the t actual total wind pressure changes; The performance of each air compressor is determined according to the change of the mean square error value.

8. The performance testing method for a train air compressor according to claim 7, characterized in that: Determining the performance of each air compressor according to the change of the mean square error value includes: Calculating an evaluation index based on the mean square error value and the mean square error standard value; The performance of each air compressor is determined according to the evaluation index.

9. A train comprising an electronic device, the electronic device comprising a memory and a processor, the memory being used to store a computer program, the processor being used to call and run the computer program stored in the memory, characterized in that: When the processor executes the computer program, the steps of the method for detecting the performance of a train air compressor as described in any one of claims 1 to 8 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for detecting the performance of a train air compressor as described in any one of claims 1 to 8 are implemented.

Images