Heavy-load train energy consumption model construction method and device, electronic equipment and storage medium

Abstract

The invention provides a heavy-load train energy consumption model construction method and device, electronic equipment and a storage medium. The method comprises the steps that vehicle parameters of each time of heavy haul train, the position and duration of each time of air braking of the heavy haul train in a plurality of long and large downhill sections, the maximum deviation value of the running speed in a constant-speed section, running data and energy consumption data are collected; dividing the collected data into a plurality of sample result data according to the trip times of the heavy haul train; and inputting the sample training data and the sample result data into a pre-constructed neural network model for training to obtain a heavy load train energy consumption model. And the neural network model is trained by taking the vehicle parameters, the air braking position and duration, the speed maximum deviation value and the driving data as sample training data, so that the energy consumption prediction of the trained heavy load train energy consumption model can be more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of rail transit energy consumption analysis, and in particular to a method, device, electronic equipment and storage medium for constructing a heavy-load train energy consumption model. Background Art

[0002] Since heavy-load trains consume a lot of energy, energy consumption analysis of heavy-load trains is particularly important.

[0003] At present, the energy consumption analysis of heavy-load trains is mainly based on the actual operation data of heavy-load trains. The operation data of heavy-load trains is used as training samples to train the neural network, and the trained neural network model is used to predict and analyze the energy consumption.

[0004] However, traditional neural networks have poor prediction effects on the energy consumption of heavy-load trains and are unable to accurately analyze the energy consumption of heavy-load trains. Summary of the invention

[0005] Based on this, it is necessary to provide a method, device, electronic device and storage medium for constructing a heavy-load train energy consumption model to address the above-mentioned technical problems.

[0006] A method for constructing a heavy-load train energy consumption model, comprising:

[0007] Collect vehicle parameters of each heavy-load train trip;

[0008] Collecting the position and duration of each air brake of the heavy-loaded train in multiple long downhill sections for each trip, and collecting the maximum speed deviation value of the heavy-loaded train running in the constant speed section;

[0009] Collecting driving data of the heavy-load train for each trip;

[0010] Collecting energy consumption data of the heavy-load train for each trip;

[0011] According to the number of trips of the heavy-load train, the collected vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the driving data are divided into a plurality of sample training data, and the energy consumption data are divided into a plurality of sample result data;

[0012] The sample training data and the sample result data are input into a pre-built neural network model for training. The neural network model is used to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data to obtain a heavy-load train energy consumption model.

[0013] In one embodiment, the step of inputting the sample training data and the sample result data into a pre-built neural network model for training, using the neural network model to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data to obtain the heavy-load train energy consumption model includes:

[0014] The sample training data and the sample result data are input into a pre-constructed neural network model for training. The neural network model is used to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between driving data and the energy consumption data. During the training of the neural network model, when the changes in three adjacent loss function values ​​are all less than a first preset threshold, the training is completed and the heavy-load train energy consumption model is obtained.

[0015] In one embodiment, it also includes:

[0016] During the training process of the heavy-load train energy consumption model, when it is impossible to achieve that three adjacent loss function value changes are all less than the first preset threshold within the preset time, the number of neurons in each layer of the neural network is increased until three adjacent loss function value changes are all less than the first preset threshold, and the heavy-load train energy consumption model is obtained.

[0017] In one embodiment, the step of inputting the sample training data and the sample result data into a pre-built neural network model for training, using the neural network model to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data to obtain the heavy-load train energy consumption model includes:

[0018] The sample training data and the sample result data are input into a pre-built neural network model for training, and the neural network model is used to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data, so as to obtain a preliminary training heavy-load train energy consumption preliminary analysis model;

[0019] Get sample test data;

[0020] Inputting the sample test data into the preliminary analysis model of energy consumption of the heavy-load train to obtain the test predicted energy consumption;

[0021] The actual energy consumption in the sample test data is compared with the test predicted energy consumption to obtain a comparison result, and the preliminary analysis model of the heavy-load train energy consumption is optimized according to the comparison result to obtain the heavy-load train energy consumption model.

[0022] In one embodiment, the comparing the actual energy consumption in the sample test data with the predicted energy consumption of the test to obtain a comparison result, optimizing the preliminary analysis model of the heavy-load train energy consumption according to the comparison result, and obtaining the heavy-load train energy consumption model includes:

[0023] Calculating the mean square error between the actual energy consumption in the sample test data and the test predicted energy consumption;

[0024] When the mean square error between the actual energy consumption in the sample test data and the test predicted energy consumption is less than a second preset threshold, the training is completed and the heavy-load train energy consumption model is obtained;

[0025] When the mean square error between the actual energy consumption in the sample test data and the test predicted energy consumption is greater than or equal to the second preset threshold, the number of neurons in each layer of the neural network is changed, and the process returns to the step of inputting the sample training data and the sample result data into a pre-built neural network model for training.

[0026] In one embodiment, it also includes:

[0027] Acquiring multiple types of measured data of the heavy-load train, wherein the multiple types of measured data include vehicle parameters, position and duration of air brakes, maximum speed deviation values ​​in each constant speed section, and driving data;

[0028] Each time prediction is made, the value of one of the multiple types of measured data is changed, and the values ​​of the other measured data are kept unchanged, and the multiple types of measured data are respectively input into the heavy-load train energy consumption model for multiple predictions to obtain multiple predicted energy consumptions related to the values ​​of the measured data;

[0029] Based on the values ​​of each of the measured data and each of the predicted energy consumptions, respectively establish the relationship between the measured data of each type and the predicted energy consumption, analyze the relationship between the measured data of each type and the predicted energy consumption, and obtain the importance of the measured data of each type to the energy consumption;

[0030] Based on the importance of various types of measured data to energy consumption, an operation optimization plan for the heavy-load train is generated.

[0031] An electronic device comprises a memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the steps of the track assembly shape and position detection method described in any of the above embodiments when executing the computer program.

[0032] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the track assembly shape and position detection method described in any of the above embodiments.

[0033] A computer program, when executed by a processor, implements the steps of the track assembly shape and position detection method described in any of the above embodiments.

[0034] The above-mentioned heavy-load train energy consumption model construction method, device, electronic device and storage medium use vehicle parameters, the position and duration of air braking, the maximum speed deviation value in each constant speed section and driving data as sample training data to train the neural network model, so that the trained heavy-load train energy consumption model can fully learn the influence of the formation length, vehicle weight data, the position and duration of air braking, and the maximum speed deviation value in the constant speed section on energy consumption, thereby making the energy consumption prediction more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A schematic diagram of a process for constructing a heavy-load train energy consumption model in one embodiment;

[0036] Figure 2 is a structural block diagram of an electronic device in one embodiment;

[0037] Figure 3 A logical diagram of forward propagation of prediction results of a hidden layer of a neural network model in one embodiment;

[0038] Figure 4 A curve diagram of an activation function in one embodiment;

[0039] Figure 5 A logical diagram of reverse feedback of a hidden layer of a neural network model in one embodiment;

[0040] Figure 6 A schematic flow chart of a method for constructing a heavy-load train energy consumption model in another embodiment;

[0041] Figure 7 It is a flowchart of a method for constructing an energy consumption model of a heavy-load train in another embodiment. DETAILED DESCRIPTION

[0042] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0043] Embodiment 1

[0044] In this embodiment, Figure 1 As shown, a method for constructing a heavy-load train energy consumption model is provided, which includes:

[0045] Step 110, collecting vehicle parameters of each heavy-load train.

[0046] Specifically, the vehicle parameters can also be called vehicle conditions, for example, the vehicle parameters include locomotive power, train length, and vehicle weight data, wherein the locomotive power is the power of the traction device of the heavy-load train, and the locomotive power is proportional to the energy consumption, the train length is the length of the train unit towed by the locomotive, and the vehicle weight data is the total weight of the train. In one embodiment, the vehicle parameters also include train resistance data, which includes the air resistance, mechanical resistance, and rolling resistance of the train.

[0047] It is worth mentioning that the basic resistance data of the train mainly includes air resistance, mechanical resistance and rolling resistance. At high speeds, air resistance dominates. As the length of the train increases, the rear vehicle can use the air flow field formed by the front vehicle to reduce air resistance. This effect is particularly significant when driving at high speeds. The front vehicle can "divert" the air resistance, significantly reducing the air resistance that the vehicle behind the vehicle needs to overcome. This results in a reduction in the actual resistance of the train as a whole. Therefore, the air resistance of each train formation is related to the length of the formation. The air resistance of each train formation can be calculated through the formation length and basic air resistance.

[0048] Step 120, collecting the position and duration of each air brake of the heavy-loaded train in multiple long downhill sections for each trip, and collecting the maximum speed deviation value of the heavy-loaded train running in the constant speed section.

[0049] In this embodiment, the entire line section that the heavy-load train passes through is divided into a plurality of long downhill sections, potential constant speed sections, and start-stop sections according to the line slopes, curves, and speed limit conditions of the route that the heavy-load train passes through. Among them, the maximum speed deviation value of each heavy-load train in the constant speed section is collected, and the maximum speed deviation value in the constant speed section is the difference between the highest speed and the lowest speed in the constant speed section. In addition, the position of the heavy-load train when air braking is adopted in the constant speed section and the duration of air braking are collected.

[0050] It should be understood that heavy-load trains need to use air brakes to ensure that the train does not have an accident when passing through long downhill sections. Under the premise of ensuring safety, reasonable air brake position and duration can allow the train to pass safely and quickly. Unreasonable air brake position and duration will greatly increase the train's running time and reduce the train's operating efficiency, and even cause abnormal stops between stations. Therefore, collecting the position and duration of air brakes at each time in multiple long downhill sections can accurately obtain the energy consumption relationship in the long downhill stage.

[0051] Step 130, collecting the driving data of the heavy-load train for each trip.

[0052] In this embodiment, the driving data includes the line length of the heavy-loaded train's route, the running time excluding stops, the normal stop number data and the abnormal stop number data, wherein the line length is the total length of the line passed by the heavy-loaded train, the running time excluding stops is the running time of the heavy-loaded train on the line minus the stop time, the normal stop number data refers to the number of planned stops at the station, and the abnormal stop number data is the number of unplanned stops.

[0053] The train stop time will first affect the total running time of the train. Freight trains do not have a stop plan at many stations along the way. Compared with not stopping, stopping will increase the additional energy consumption of braking and starting, and will generate additional starting and stopping time. While increasing the running energy consumption, it will also extend the running time of the train. Abnormal stops of trains often occur in long downhill sections due to unreasonable application of air brakes, resulting in failure of the braking system, which will greatly increase the total running time of the train.

[0054] Step 140, collecting energy consumption data of the heavy-load train for each trip.

[0055] In this embodiment, data on the energy consumed by each heavy-load train is collected.

[0056] Step 150, according to the number of trips of the heavy-load train, the collected vehicle parameters, the position and duration of air braking, the maximum speed deviation value in each constant speed section, and the driving data are divided into multiple sample training data, and the energy consumption data are divided into multiple sample result data.

[0057] In this embodiment, according to the trips of the heavy-load train, the various data collected above are divided into sample training data of multiple trips, and the sample training data of each trip corresponds to the sample result data of one trip. In this embodiment, the sample training data and the sample result data are both sample data, which are used to train the neural network model.

[0058] The data is divided into multiple different sample training data according to the number of heavy-load trains, so that different data can be matched with different trains, making the samples more targeted and the sample training data more orderly, avoiding the clutter of sample data, effectively improving the training efficiency, and making the trained model more accurate.

[0059] Step 160, input the sample training data and the sample result data into a pre-constructed neural network model for training, and use the neural network model to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data to obtain a heavy-load train energy consumption model.

[0060] In this embodiment, the sample training data and the sample result data are input into the neural network model for training, and the neural network model learns the relationship between different sample training data and sample result data, so as to train and obtain the energy consumption model of the heavy-load train.

[0061] In the above embodiment, vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section and driving data are used as sample training data to train the neural network model, so that the trained heavy-load train energy consumption model can fully learn the influence of the formation length, vehicle weight data, the position and duration of air brakes, and the maximum speed deviation value in the constant speed section on energy consumption, thereby making the energy consumption prediction more accurate.

[0062] Embodiment 2

[0063] In this embodiment, based on the above embodiment, the steps of inputting the sample training data and the sample result data into a pre-built neural network model for training, using the neural network model to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data, and obtaining the heavy-load train energy consumption model include:

[0064] The sample training data and the sample result data are input into a pre-constructed neural network model for training. The neural network model is used to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between driving data and the energy consumption data. During the training of the neural network model, when the changes in three adjacent loss function values ​​are all less than a first preset threshold, the training is completed and the heavy-load train energy consumption model is obtained.

[0065] In this embodiment, the neural network model includes four hidden layers, and the number of neurons in each hidden layer decreases layer by layer according to the direction in which the results are passed forward. For example, the number of neurons in each hidden layer is 20, 14, 8, and 4, respectively. Each hidden layer passes forward layer by layer to calculate the predicted energy consumption, and the deviation is fed back in reverse according to the difference between the predicted energy consumption and the actual energy consumption, and multiple iterations are performed to correct the model. After each feedback, the loss function of each feedback is calculated. When the changes in the loss function values ​​for three consecutive times are less than the first preset value, it is determined that the neural network model training is completed, and the heavy-load train energy consumption model is obtained.

[0066] Specifically, the calculation of each adjacent layer during the forward pass is as follows (taking the first two layers as an example):

[0067] z [1] =W [1] x+b [1]

[0068] a [1] =σ(z [1] )

[0069] z [2] =W [1] a [1] +b [2]

[0070] a [2] =σ(z [2] )

[0071] Where the superscript is the number of layers; z is the value before the activation function of each layer; a is the value after the activation function, W and b are the biases of each layer, and x defaults to a [0] , σ is the activation function.

[0072] The forward pass process is as follows Figure 3 As shown, the activation function alternately uses the leakyRELU function and the sigmoid function, such as Figure 4 As shown, the neural network has nonlinear properties.

[0073] LeakyRELU function

[0074] Sigmoid function:

[0075] The reverse feedback process is as follows: Figure 5 As shown in Figure 2, the reverse feedback bias correction model is:

[0077] energy consumption, m is the number of samples, n is the number of neurons in this layer, and L is the regularization coefficient to prevent the model from overfitting.

[0078] The reverse feedback deviation correction calculation model is:

[0079]

[0080] Where W' and b' are updated parameters,

[0081] and Before the update, its 馈 Partial derivative, α is the learning rate.

[0082] Model training is complete:

[0083]

[0084] In this embodiment, when J 馈When the difference between the values ​​obtained in three consecutive calculations is less than the first preset threshold, it indicates that the model has stabilized. Therefore, when the changes in the loss function values ​​of three consecutive times are all less than the first preset threshold, the heavy-load train energy consumption training is completed and the heavy-load train energy consumption model is obtained.

[0085] Embodiment 3

[0086] In this embodiment, based on the above embodiment, during the training process of the heavy-load train energy consumption model, when it is impossible to achieve three adjacent loss function value changes that are all less than the first preset threshold within the preset time, the number of neurons in each layer of the neural network is increased until three adjacent loss function value changes are all less than the first preset threshold, and the heavy-load train energy consumption model is obtained.

[0087] In this embodiment, if no results are obtained after a long period of training (preset time), the number of neurons in each hidden layer of the neural network is increased, and the iterative calculation is continued until the changes in the loss function values ​​of three consecutive times are all less than the first preset threshold. Specifically, within the preset time, in the reverse feedback deviation correction calculation model, if J 馈 If J 馈 If it keeps decreasing and changes slowly, increase the learning rate α. If a stable model cannot be obtained after the preset time, increase the number of neurons in each hidden layer of the neural network.

[0088] Embodiment 4

[0089] In this embodiment, based on the above embodiment, the steps of inputting the sample training data and the sample result data into a pre-built neural network model for training, using the neural network model to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data, and obtaining the heavy-load train energy consumption model include:

[0090] The sample training data and the sample result data are input into a pre-built neural network model for training, and the neural network model is used to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data, so as to obtain a preliminary training heavy-load train energy consumption preliminary analysis model;

[0091] Get sample test data;

[0092] Inputting the sample test data into the preliminary analysis model of energy consumption of the heavy-load train to obtain the test predicted energy consumption;

[0093] The actual energy consumption in the sample test data is compared with the test predicted energy consumption to obtain a comparison result, and the preliminary analysis model of the heavy-load train energy consumption is optimized according to the comparison result to obtain the heavy-load train energy consumption model.

[0094] In this embodiment, the sample test data is input into the preliminary analysis model of heavy-load train energy consumption for testing to obtain the test predicted energy consumption, and the predicted test predicted energy consumption is compared with the actual energy consumption in the sample test data. According to the comparison results, the preliminary analysis model of heavy-load train energy consumption can be corrected to obtain the heavy-load train energy consumption model.

[0095] Embodiment 5

[0096] In this embodiment, based on the above embodiment, the actual energy consumption in the sample test data is compared with the test predicted energy consumption to obtain a comparison result, and the heavy-load train energy consumption preliminary analysis model is optimized according to the comparison result to obtain the heavy-load train energy consumption model, which includes:

[0097] Calculating the mean square error between the actual energy consumption in the sample test data and the test predicted energy consumption;

[0098] When the mean square error between the actual energy consumption in the sample test data and the test predicted energy consumption is less than a second preset threshold, the training is completed and the heavy-load train energy consumption model is obtained;

[0099] When the mean square error between the actual energy consumption in the sample test data and the test predicted energy consumption is greater than or equal to the second preset threshold, the number of neurons in each layer of the neural network is changed, and the process returns to the step of inputting the sample training data and the sample result data into a pre-built neural network model for training.

[0100] In this embodiment, the sample training data and the sample result data are input into a pre-built neural network model for training. During the training of the neural network model, when the changes in the three adjacent loss function values ​​are all less than the first preset threshold, the training is completed, and a preliminary training heavy-load train energy consumption preliminary analysis model is obtained; then, the preliminary obtained heavy-load train energy consumption preliminary analysis model is tested and verified. The specific process is to obtain sample test data, and use the sample test data to test and verify the heavy-load train energy consumption preliminary analysis model.

[0101] It is worth mentioning that the sample test data is also sample data. For example, sample data is collected and randomly divided into a training set and a test set at a ratio of ten to one, wherein the training set includes sample training data and sample result data, and the test set is sample test data, including sample data and test result data.

[0102] The sample test data is input into the preliminary analysis model of heavy-duty train energy consumption to obtain the test predicted energy consumption, and the variance of the test predicted energy consumption and the actual energy consumption in the test result data is calculated. The calculation process is as follows:

[0103]

[0104] Among them, y i is the actual energy consumption of the ith sample, is the estimated energy consumption of the ith sample, and D is the variance. When the value of D is less than the second preset threshold, the model training is completed and a stable heavy-load train energy consumption model is obtained.

[0105] When the value of D is greater than or equal to the second preset threshold, it indicates that the model is unstable and the prediction result is inaccurate, then the number of neurons in each layer of the neural network is increased, and the pre-built neural network model is trained using the training set until the variance D is less than the second preset threshold.

[0106] Embodiment 6

[0107] In this embodiment, based on the above embodiment, the method further includes: obtaining multiple types of measured data of the heavy-load train, wherein the multiple types of measured data include vehicle parameters, position and duration of air brakes, maximum speed deviation values ​​in each constant speed section, and driving data;

[0108] Each time prediction is made, the value of one of the multiple types of measured data is changed, and the values ​​of the other measured data are kept unchanged, and the multiple types of measured data are respectively input into the heavy-load train energy consumption model for multiple predictions to obtain multiple predicted energy consumptions related to the values ​​of the measured data;

[0109] Based on the values ​​of each of the measured data and each of the predicted energy consumptions, respectively establish the relationship between the measured data of each type and the predicted energy consumption, analyze the relationship between the measured data of each type and the predicted energy consumption, and obtain the importance of the measured data of each type to the energy consumption;

[0110] Based on the importance of various types of measured data to energy consumption, an operation optimization plan for the heavy-load train is generated.

[0111] In this embodiment, in each round of calculation, each measured data is input into the heavy-load train energy consumption model for prediction; the specific process is: in the first round, each measured data is input into the heavy-load train energy consumption model for a prediction, then, the value of one of the measured data is changed, the values ​​of other measured data are kept unchanged, and the data is input into the heavy-load train energy consumption model again for a prediction, then, the value of the measured data is changed again, the values ​​of other measured data are kept unchanged, and the data is input into the heavy-load train energy consumption model again for a prediction... In this way, the relationship between the changed measured data and the predicted energy consumption is obtained, for example, with the changed measured data as the horizontal axis and the predicted energy consumption as the vertical axis, a curve representing the relationship between the changed measured data and the predicted energy consumption is obtained; in the next round, each measured data is input into the heavy-load train energy consumption model for a prediction, then, the value of another measured data is changed, the values ​​of other measured data are kept unchanged, and the above process is followed again. After multiple tests, another curve of the relationship between the measured data and the predicted energy consumption is obtained. In this way, multiple rounds of calculations are performed until the relationship between various measured data and the predicted energy consumption is obtained. According to the relationship between each measured data and the predicted energy consumption, the importance of various types of measured data to energy consumption is analyzed. According to the importance of various measured data, an operation optimization plan can be obtained. The operation optimization plan includes an operation plan, a power plan, and a transportation capacity plan. Specifically, the operation plan is used to guide the driver to operate the heavy-loaded train. The power plan is used to provide suggestions for the operating power of the heavy-loaded locomotive. It can provide suggestions based on the locomotive power and the parameters that affect the transportation capacity of the heavy-loaded train. It can provide leading suggestions for the configuration of various types of trains on the line. The transportation capacity plan includes suggestions on the length of the formation, the weight of the whole vehicle, and the running time. By adjusting the parameters such as the length of the formation, the weight of the whole vehicle, and the running time, a single train formation method that meets the transportation volume requirements and saves energy can be obtained.

[0112] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but the steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of the steps, and the steps can be executed in other orders. For example, steps 110 to 140 can be performed simultaneously, or steps 140 to 110 can be executed. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0113] Embodiment 7

[0114] In this embodiment, a specific embodiment is provided.

[0115] In this embodiment, Figure 6 and Figure 7 As shown, the method for constructing the energy consumption model of a heavy-load train includes:

[0116] S1. Collect the actual operation data of heavy-load trains, extract the locomotive power, marshaling length, vehicle weight, running time, maximum speed deviation value of each constant speed section, air brake position and duration of each long downhill section, normal stop number, abnormal stop number, and corresponding energy consumption of each heavy-load train. And randomly divide the data set into training set and test set.

[0117] S2. Build a heavy-duty train energy consumption analysis model, use a neural network model combined with gradient descent, and use the data in the training set to train the model. When the change in the loss function value for three consecutive times is less than the design value, the heavy-duty train energy consumption analysis model is initially trained. If no results are obtained after a long period of training, increase the number of neurons in each layer.

[0118] S3. Use the test set to verify the trained heavy-load train energy consumption analysis model. When the mean square error is less than the design value, the heavy-load train energy consumption model training is completed. Otherwise, it means that the model is overfitting. Return to S2 to change the number of model neurons and retrain the model.

[0119] S4. According to actual needs, the trained heavy-load train energy consumption analysis model is input with the locomotive power, train length, line length, vehicle weight, running time excluding stops, maximum speed deviation value in each constant speed section, position and duration of each air brake in long downhill sections, number of normal stops, number of abnormal stops of the heavy-load train that has not been tested on the actual vehicle, which needs to be predicted, to obtain the estimated energy consumption of the heavy-load train that has not been tested on the actual vehicle.

[0120] S5. Keep some variables in S4 unchanged, and change some of them to get the relationship between train energy consumption and them. And provide suggestions for driver operation optimization, train formation and operation diagram optimization of the line.

[0121] In one embodiment, the step S1 specifically includes the following sub-steps:

[0122] S11. Collect vehicle conditions of each heavy-load train, including locomotive power, train length, and vehicle weight data.

[0123] S12. Divide the entire line section into a plurality of long downhill sections, potential constant speed sections and start-stop sections according to line ramps, curves and speed limit conditions.

[0124] S13, collecting the position and duration of each air brake in multiple long downhill sections; extracting the difference between the highest speed and the lowest speed in each interval in the potential constant speed section.

[0125] S14. Collect driving data, including the length of the heavy-load train line, the running time excluding stops, the number of normal stops, and the number of abnormal stops.

[0126] S15. Divide the collected heavy-load train data into multiple samples according to the trip number, and randomly divide them into training set and test set in a ratio of ten to one.

[0127] Specifically, the main factors affecting the basic resistance of a train are air resistance, mechanical resistance, and rolling resistance. At high speeds, air resistance is dominant, and as the length of the train increases, the rear vehicle can use the air flow field formed by the front vehicle to reduce air resistance. This effect is particularly significant when driving at high speeds, as the front vehicle can "divide" the air resistance, significantly reducing the air resistance that the vehicle behind the vehicle needs to overcome. This results in a reduction in the actual resistance of the train as a whole.

[0128] After simulation verification, it is found that running a train at a constant speed in a flat section is more energy-efficient than running a train at a variable speed, so the maximum speed deviation value of each constant speed section is increased.

[0129] Heavy-loaded trains need to use air brakes when passing through long downhill sections to ensure that the train does not have accidents. Under the premise of ensuring safety, reasonable air brake position and duration can allow the train to pass safely and quickly. Unreasonable air brake position and duration will greatly increase the train's running time, reduce the train's operating efficiency, and even cause abnormal stops between stations.

[0130] The train stop time will first affect the total running time of the train. Freight trains do not have a stop plan at many stations along the way. Compared with not stopping, stopping will increase the additional energy consumption of braking and starting, and will generate additional starting and stopping time. While increasing the running energy consumption, it will also extend the running time of the train. Abnormal stops of trains often occur in long downhill sections due to unreasonable application of air brakes, resulting in failure of the braking system, which will greatly increase the total running time of the train.

[0131] In one embodiment, step S2 includes the following sub-steps:

[0132] S21. Import the training set in step S1 into the constructed heavy-load train energy consumption analysis model.

[0133] S22. The model initially has four hidden layers, and the number of neurons in each hidden layer is 20, 14, 8, and 4 respectively. Each layer forward passes the calculation to predict the energy consumption, and the deviation correction model is fed back according to the difference between the predicted energy consumption and the actual energy consumption.

[0134] S23, the calculation of each adjacent layer in the forward transmission process is as follows (taking the first two layers as an example):

[0135] z [1] =W [1] x+b [1]

[0136] a [1] =σ(z [1] )

[0137] z [2] =W [1] a [1] +b [2]

[0138] a [2] =σ(z [2] )

[0139] Where the superscript is the number of layers; z is the value before the activation function of each layer; a is the value after the activation function, W and b are the biases of each layer, and x defaults to a [0] , σ is the activation function.

[0140] The forward transmission diagram is as follows Figure 3 As shown:

[0141] S24, the activation function alternately uses the leakyRELU function and the sigmoid function, such as Figure 4 As shown, the neural network has nonlinear properties.

[0142] LeakyRELU function

[0143] Sigmoid function:

[0144] S25, the process of reverse feedback is as follows Figure 5 As shown in Figure 2, the reverse feedback bias correction model is:

[0145]

[0146] Take the last layer as an example, where y i is the actual energy consumption of the ith sample, is the estimated energy consumption of the ith sample, m is the number of samples, n is the number of neurons in this layer, and L is the regularization coefficient to prevent overfitting of the model.

[0147] S26, reverse feedback deviation correction calculation:

[0148]

[0149] Where W' and b' are updated parameters,

[0150] and Before the update, the 馈 Partial derivative, α is the learning rate.

[0151] S27, model training ends:

[0152]

[0153] When the changes in the loss function values ​​for three consecutive times are all less than the design value, the initial training of the heavy-load train energy consumption is completed. 馈 If J 馈 If it keeps decreasing and changes slowly, increase the learning rate α. Return to S22 and adjust the number of neurons in each layer appropriately.

[0154] In one embodiment, step S3 includes the following sub-steps:

[0155] S31, import the test set data separated by S1 into the heavy-load train energy consumption analysis model trained by S2, only calculate the predicted energy consumption, and do not modify the model.

[0156] S32. Calculate the mean square error between the predicted energy consumption and the actual energy consumption.

[0157]

[0158] When D meets the requirements, the model training is completed, otherwise return to S22.

[0159] In one embodiment, step S4 includes the following sub-steps:

[0160] S41. Input the data of locomotive power, train length, vehicle weight, running time, maximum speed deviation value in each constant speed section, position and duration of each air brake in the long downhill section, normal stop times, and abnormal stop times of the heavy-load train for predicting energy consumption into the energy consumption analysis model.

[0161] S42. Calculate the estimated energy consumption for the trip through model calculation.

[0162] S43. Keeping other variables unchanged and changing one of the variables, a curve showing the variation of the train energy consumption with the variable under the current setting can be obtained, and thereby the importance of each factor on the traction energy consumption can be obtained.

[0163] In one embodiment, step S5 includes the following sub-steps:

[0164] The data of maximum speed deviation in each constant speed section, air brake position and duration in long downhill sections, normal stop times and abnormal stop times in S51 and S4 are mainly used to evaluate the driver's control level. In actual vehicles, the energy consumption used by excellent drivers of the same vehicle is significantly lower than the average energy consumption. By adjusting these data, suggestions for optimizing the driver's control level can be obtained.

[0165] The locomotive power in S52 and S4 affects the transportation capacity of heavy-load trains, and can provide leading suggestions for the configuration of various types of trains on this line.

[0166] The train length, vehicle weight and running time in S53 and S4 are mainly used to evaluate the transport volume. By adjusting these parameters, a single train train formation that meets the transport volume requirements and saves energy can be obtained.

[0167] Embodiment 8

[0168] In this embodiment, a heavy-load train energy consumption model construction device is provided, comprising:

[0169] The first collection module is used to collect vehicle parameters of each heavy-load train;

[0170] The second collection module is used to collect the position and duration of each air brake of the heavy-loaded train in multiple long downhill sections for each trip, and collect the maximum speed deviation value of the heavy-loaded train running in the constant speed section;

[0171] A third collection module is used to collect the driving data of the heavy-load train for each trip;

[0172] A fourth collection module is used to collect energy consumption data of the heavy-load train for each trip;

[0173] A data division module is used to divide the collected vehicle parameters, the position and duration of air brakes, the maximum deviation value of speed in each constant speed section, and driving data into a plurality of sample training data, and divide the energy consumption data into a plurality of sample result data according to the number of trips of the heavy-load train;

[0174] A training module is used to input the sample training data and the sample result data into a pre-built neural network model for training, and use the neural network model to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data to obtain a heavy-load train energy consumption model.

[0175] In one embodiment, the training module is used to input the sample training data and the sample result data into a pre-constructed neural network model for training, and use the neural network model to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between driving data and the energy consumption data. During the training of the neural network model, when the changes in three adjacent loss function values ​​are all less than a first preset threshold, the training is completed and the heavy-load train energy consumption model is obtained.

[0176] In one embodiment, the training module is also used to increase the number of neurons in each layer of the neural network during the training process of the heavy-load train energy consumption model, when it is impossible to achieve three adjacent loss function value changes less than a first preset threshold within a preset time, until the three adjacent loss function value changes are less than the first preset threshold, to obtain the heavy-load train energy consumption model.

[0177] In one embodiment, the training module includes:

[0178] A preliminary model training unit is used to input the sample training data and the sample result data into a pre-built neural network model for training, and use the neural network model to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data, so as to obtain a preliminary analysis model of heavy-duty train energy consumption that has been preliminarily trained;

[0179] A sample test data acquisition unit, used to acquire sample test data;

[0180] A test prediction energy consumption acquisition unit, used for inputting the sample test data into the heavy-load train energy consumption preliminary analysis model to obtain the test prediction energy consumption;

[0181] A comparison unit is used to compare the actual energy consumption in the sample test data with the test predicted energy consumption to obtain a comparison result, and optimize the preliminary analysis model of the heavy-load train energy consumption according to the comparison result to obtain the heavy-load train energy consumption model.

[0182] In one embodiment, the comparison unit is also used to calculate the mean square error between the actual energy consumption in the sample test data and the test predicted energy consumption; when the mean square error between the actual energy consumption in the sample test data and the test predicted energy consumption is less than a second preset threshold, the training is completed and the heavy-load train energy consumption model is obtained; when the mean square error between the actual energy consumption in the sample test data and the test predicted energy consumption is greater than or equal to the second preset threshold, the number of neurons in each layer of the neural network is changed, and the step of inputting the sample training data and the sample result data into a pre-built neural network model for training is returned.

[0183] In one embodiment, the apparatus further comprises:

[0184] A measured data acquisition module, used to acquire multiple types of measured data of the heavy-load train, wherein the multiple types of measured data include vehicle parameters, position and duration of air brakes, maximum speed deviation values ​​in each constant speed section, and driving data;

[0185] A numerical value change input module, used for changing the value of one of the multiple types of measured data each time a prediction is made, keeping the values ​​of the other measured data unchanged, and inputting the multiple types of measured data into the heavy-load train energy consumption model for multiple predictions to obtain multiple predicted energy consumptions related to the values ​​of the measured data;

[0186] An importance calculation module, for establishing, based on the values ​​of each of the measured data and each of the predicted energy consumptions, the relationship between the measured data of each type and the predicted energy consumption, analyzing the relationship between the measured data of each type and the predicted energy consumption, and obtaining the importance of the measured data of each type to the energy consumption;

[0187] The optimization scheme generating module is used to generate the operation optimization scheme of the heavy-load train based on the importance of various types of measured data to energy consumption.

[0188] For the specific limitations of the heavy-load train energy consumption model construction device, please refer to the limitations of the heavy-load train energy consumption model construction method above, which will not be repeated here. Each unit in the above-mentioned heavy-load train energy consumption model construction device can be implemented in whole or in part through software, hardware and their combination. The above-mentioned units can be embedded in or independent of the processor in the electronic device in the form of hardware, or can be stored in the memory of the electronic device in the form of software, so that the processor can call and execute the operations corresponding to the above-mentioned units.

[0189] Embodiment 9

[0190] In this embodiment, an electronic device is provided. Its internal structure diagram can be shown as follows: Figure 2As shown. The electronic device includes a processor, a memory, a network interface, a display screen and an input device connected through a system bus. Among them, the processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program, and the non-volatile storage medium is deployed with a database, which is used to store train travel data, including coordinate data, acceleration data, attitude angle data, etc. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the electronic device is used to communicate with other electronic devices that deploy application software. When the computer program is executed by the processor, a method for constructing a heavy-load train energy consumption model is implemented. The display screen of the electronic device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the electronic device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the housing of the electronic device, or an external keyboard, touchpad or mouse, etc.

[0191] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a partial structure related to the scheme of the present application, and does not constitute a limitation on the electronic device to which the scheme of the present application is applied. The specific electronic device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0192] In one embodiment, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method for constructing a heavy-load train energy consumption model described in any of the above embodiments when executing the computer program.

[0193] Embodiment 10

[0194] In this embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for constructing a heavy-load train energy consumption model described in any of the above embodiments are implemented.

[0195] Embodiment 11

[0196] In this embodiment, a computer program is provided, and when the computer program is executed by a processor, the steps of the method for constructing a heavy-load train energy consumption model described in any of the above embodiments are implemented.

[0197] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0198] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0199] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A method for constructing a heavy-load train energy consumption model, characterized in that: include: Collect vehicle parameters of each heavy-load train trip; Collecting the position and duration of each air brake of the heavy-loaded train in multiple long downhill sections for each trip, and collecting the maximum speed deviation value of the heavy-loaded train running in the constant speed section; Collecting driving data of the heavy-load train for each trip; Collecting energy consumption data of the heavy-load train for each trip; According to the number of trips of the heavy-load train, the collected vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the driving data are divided into a plurality of sample training data, and the energy consumption data are divided into a plurality of sample result data; The sample training data and the sample result data are input into a pre-built neural network model for training. The neural network model is used to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data to obtain a heavy-load train energy consumption model.

2. The method according to claim 1, characterized in that The step of inputting the sample training data and the sample result data into a pre-built neural network model for training, using the neural network model to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data to obtain the heavy-load train energy consumption model includes: The sample training data and the sample result data are input into a pre-constructed neural network model for training. The neural network model is used to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between driving data and the energy consumption data. During the training of the neural network model, when the changes in three adjacent loss function values ​​are all less than a first preset threshold, the training is completed and the heavy-load train energy consumption model is obtained.

3. The method according to claim 2, characterized in that Also includes: During the training process of the heavy-load train energy consumption model, when it is impossible to achieve that three adjacent loss function value changes are all less than the first preset threshold within the preset time, the number of neurons in each layer of the neural network is increased until three adjacent loss function value changes are all less than the first preset threshold, and the heavy-load train energy consumption model is obtained.

4. The method according to claim 1, characterized in that: The step of inputting the sample training data and the sample result data into a pre-built neural network model for training, using the neural network model to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data to obtain the heavy-load train energy consumption model includes: The sample training data and the sample result data are input into a pre-built neural network model for training, and the neural network model is used to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data, so as to obtain a preliminary training heavy-load train energy consumption preliminary analysis model; Get sample test data; Inputting the sample test data into the preliminary analysis model of energy consumption of the heavy-load train to obtain the test predicted energy consumption; The actual energy consumption in the sample test data is compared with the test predicted energy consumption to obtain a comparison result, and the preliminary analysis model of the heavy-load train energy consumption is optimized according to the comparison result to obtain the heavy-load train energy consumption model.

5. The method according to claim 4, characterized in that The comparing the actual energy consumption in the sample test data with the predicted energy consumption of the test to obtain a comparison result, optimizing the preliminary analysis model of the heavy-load train energy consumption according to the comparison result, and obtaining the heavy-load train energy consumption model comprises: Calculating the mean square error between the actual energy consumption in the sample test data and the test predicted energy consumption; When the mean square error between the actual energy consumption in the sample test data and the test predicted energy consumption is less than a second preset threshold, the training is completed and the heavy-load train energy consumption model is obtained; When the mean square error between the actual energy consumption in the sample test data and the test predicted energy consumption is greater than or equal to the second preset threshold, the number of neurons in each layer of the neural network is changed, and the process returns to the step of inputting the sample training data and the sample result data into a pre-built neural network model for training.

6. The method according to any one of claims 1 to 5, characterized in that: Also includes: Acquiring multiple types of measured data of the heavy-load train, wherein the multiple types of measured data include vehicle parameters, position and duration of air brakes, maximum speed deviation values ​​in each constant speed section, and driving data; Each time prediction is made, the value of one of the multiple types of measured data is changed, and the values ​​of the other measured data are kept unchanged, and the multiple types of measured data are respectively input into the heavy-load train energy consumption model for multiple predictions to obtain multiple predicted energy consumptions related to the values ​​of the measured data; Based on the values ​​of each of the measured data and each of the predicted energy consumptions, respectively establish the relationship between the measured data of each type and the predicted energy consumption, analyze the relationship between the measured data of each type and the predicted energy consumption, and obtain the importance of the measured data of each type to the energy consumption; Based on the importance of various types of measured data to energy consumption, an operation optimization plan for the heavy-load train is generated.

7. A device for constructing a heavy-load train energy consumption model, characterized in that: include: The first collection module is used to collect vehicle parameters of each heavy-load train; The second collection module is used to collect the position and duration of each air brake of the heavy-loaded train in multiple long downhill sections for each trip, and collect the maximum speed deviation value of the heavy-loaded train running in the constant speed section; A third collection module is used to collect the driving data of the heavy-load train for each trip; A fourth collection module is used to collect energy consumption data of the heavy-load train for each trip; A data division module is used to divide the collected vehicle parameters, the position and duration of air brakes, the maximum deviation value of speed in each constant speed section, and driving data into a plurality of sample training data, and divide the energy consumption data into a plurality of sample result data according to the number of trips of the heavy-load train; A training module is used to input the sample training data and the sample result data into a pre-built neural network model for training, and use the neural network model to learn vehicle parameters, the position and duration of air brakes, the maximum speed deviation value in each constant speed section, and the relationship between the driving data and the energy consumption data to obtain a heavy-load train energy consumption model.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images