Intelligent car transfer and garage transfer management method based on machine learning

Through the intelligent relocation and relocation management method based on machine learning, the train scheduling model is established using the LSTM algorithm, which solves the problem of both operating requirements and overall maintenance time in train maintenance, and realizes orderly maintenance and time reduction of the train system, which reduces the Maintenance staff burden.

CN119090485AInactive Publication Date: 2024-12-06湖南润伟智能机器有限公司 +1

Patent Information

Application Number
CN202411145282.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-20
Publication Date
2024-12-06
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing technology cannot take into account the train's operating needs and overall maintenance time during train maintenance, resulting in extended maintenance tasks, backlog of tasks and delays.

Method used

Based on machine learning, intelligent vehicle migration and relocation management method is used to establish a train scheduling model using the LSTM algorithm. By acquiring and preprocessing the maintenance scheduling data of train entry and exit, the maintenance level and priority are divided into training sets, verification sets and test sets. Optimize the model to minimize the completion time of maintenance scheduling tasks and meet the priority time requirements.

Benefits of technology

The orderly maintenance of the train system has been achieved, the overall maintenance time has been shortened, the burden on railway maintenance staff has been reduced, and the efficiency of railway operation has been improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119090485A_ABST
    Figure CN119090485A_ABST
Patent Text Reader

Abstract

The invention provides a machine learning-based intelligent train transfer and garage transfer management method, which comprises the following steps of 1, acquiring recent train in-out garage maintenance scheduling data, and preprocessing the data; 2, marking the priority of each piece of scheduling information in the data; 3, dividing the data into training set data, verification set data and test set data; 4, establishing a train scheduling model based on a deep learning algorithm; 5, training and optimizing the train dispatching model by using the training set and verification set data based on a model optimization algorithm; 6, evaluating and verifying the optimization result of the train dispatching model by using the test set data; and step 7, receiving train in-out dispatching information today, and obtaining a dispatching strategy through the train dispatching model. According to the method, the train dispatching model is established based on the LSTM algorithm, ordered maintenance of a train system is guaranteed, the total maintenance time is shortened, and the burden of railway maintenance workers is relieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of intelligent train dispatching, and in particular to an intelligent train and warehouse relocation management method based on machine learning. Background Art

[0002] During the use of railway trains, parts will gradually wear, corrode and be damaged. In order to keep the vehicles in good quality, ensure driving safety and extend the service life of the vehicles, various inspections and repairs must be carried out on the railway vehicles. The vehicle planned preventive maintenance system implemented by China Railway is divided into two categories according to the repair content: regular inspections and daily maintenance. It is necessary to gradually realize condition-based repairs, that is, repair the vehicle after it breaks down, which can reduce costs.

[0003] In the routine maintenance and repair process of trains, due to the large number of railway vehicles and the small number of maintenance and inspection depots, it is necessary to sort the maintenance of railway trains. In the prior art, maintenance is generally carried out in sequence according to the time sequence of the maintenance and inspection needs of the trains, or in sequence according to the time of the next running of the trains. However, the maintenance needs of trains are different, and the degree and time of the required inspections are different. Sorting only according to the maintenance demand time or the interval time of the next running of the train cannot take into account the running needs of the trains and the overall maintenance time; when the running needs of the trains are not met, the trains will not be able to complete the maintenance and depart on time, causing delays; and if the overall maintenance time of the trains is too long, subsequent maintenance tasks will continue to pile up, which may cause a large number of tasks to be unable to be completed on time.

[0004] Application Contents

[0005] This application is made in view of the above-mentioned issues, and its purpose is to provide an intelligent vehicle relocation and depot management method based on machine learning, establish a train scheduling model based on the LSTM algorithm, minimize the completion time of maintenance scheduling tasks and the time requirements that meet the priorities as the optimization goal of the model, take into account the time requirements of the train and the overall maintenance time, ensure the orderly maintenance of the train system, shorten the overall maintenance time, and reduce the burden on railway maintenance personnel.

[0006] Specifically, the first aspect of the present application provides an intelligent vehicle relocation and warehouse relocation management method based on machine learning, comprising the following steps:

[0007] Step 1: Obtain recent train in-and-out maintenance scheduling data and pre-process the data;

[0008] By obtaining the recent train in-and-out maintenance scheduling data, the maintenance level and maintenance time of each train are obtained, and the average maintenance time of each maintenance level is used as the basis for the subsequent maintenance sorting time. The maintenance level of the railway train is divided into five levels, among which the first and second levels of maintenance are daily maintenance and operation maintenance, and the third level and above are advanced maintenance. Since the fifth-level maintenance of the train requires the EMU to be driven back to the main engine manufacturer for comprehensive maintenance, which the railway bureau cannot complete, in this embodiment, only the first to fourth level maintenance is recorded and scheduled;

[0009] Step 2: Mark the priority of each scheduling information in the data;

[0010] The priority of each dispatching information is divided according to the maintenance level of the train dispatching maintenance and the time interval for the next departure, so as to serve as one of the bases for subsequent maintenance sorting;

[0011] Step 3: Divide the data into training set, validation set and test set data;

[0012] The training set is used in the model training process to provide input features and targets so that the model can learn the mapping relationship between features and targets. The training set is the basis for model learning and parameter optimization. Through multiple iterative training in the training set, the model gradually improves its model capabilities.

[0013] The validation set is a set of samples set aside during the model training process. It is usually separated from the training set and does not participate in the model training. The validation set is used to adjust the model's hyperparameters, evaluate the model's generalization ability on unseen data, and prevent overfitting. By evaluating the model on the validation set, you can ensure that the model does not overfit to the training data and ignore the characteristics of the new data;

[0014] The test set is a dataset that is completely independent of the training set and validation set and is used to evaluate the performance of the model in the final application. The test set is used to evaluate the final performance of the model, especially its generalization ability on unknown data. The results of the test set can be used as the final indicator of the success of the model. The data of the test set should be the data that the model may encounter in actual applications, so it should not be used to adjust model parameters or select features.

[0015] Step 4: Establish a train dispatching model based on deep learning algorithm;

[0016] Step 5: Use the training set and validation set data to train and optimize the train scheduling model based on the model optimization algorithm according to the data priority;

[0017] The core function of the model optimization algorithm is to adjust the model parameters in an iterative manner to minimize the loss function, thereby finding the best model parameters and accelerating the model training process.

[0018] Step 6: Use the test set data to evaluate and verify the optimization results of the train scheduling model;

[0019] The test set is used to evaluate the final performance of the model, especially its generalization ability on unknown data. The results of the test set can be used as the final indicator of the success of the model. The results obtained from the test set data are compared with the time spent on previous scheduling results to evaluate the effectiveness of the model;

[0020] Step 7: Collect relevant data information within the station, receive today's train in and out scheduling information, and obtain the scheduling strategy through the train scheduling model.

[0021] After the model training is completed, the current scheduling task data is input into the model to obtain the optimal scheduling strategy for train maintenance scheduling;

[0022] Furthermore, the preprocessing includes data cleaning and data normalization.

[0023] Furthermore, the priority of the dispatching information is divided according to the five-level maintenance level that the train needs to undergo and the time difference between the next departure of this train.

[0024] Furthermore, the data is divided into a training set, a validation set and a test set data, and K-fold cross validation is adopted.

[0025] Furthermore, establishing a train scheduling model based on a deep learning algorithm includes establishing a train scheduling model based on an LSTM algorithm.

[0026] Furthermore, the optimization goal of the train scheduling model is to minimize the completion time of the maintenance scheduling task and the time requirement that meets the priority.

[0027] Furthermore, the use of the training set and validation set data is based on a model optimization algorithm to train and optimize the train scheduling model according to data priority, specifically including:

[0028] Step 1: Use the training set data to train the train scheduling model;

[0029] Use the training set data to forward propagate the model and calculate the gradient of the loss function with respect to each parameter;

[0030] Step 2: Based on the model optimization algorithm, update the parameters of the train scheduling model according to the data of the training process;

[0031] According to the Adam algorithm, each parameter is updated using the current gradient, the momentum of the previous gradient, and an estimate of the variance;

[0032] Step 3: During the training process, use the validation set data to evaluate the generalization ability of the model at regular intervals.

[0033] During the training process, you need to use the validation set regularly to evaluate the generalization ability of the model, which helps prevent overfitting and adjust hyperparameters, such as the learning rate. The Adam algorithm automatically adjusts the learning rate of each parameter. If you find that the performance of the model on the validation set begins to decline, you may need to readjust the model or its hyperparameters;

[0034] Step 4: When the model training performance on the validation set no longer improves, stop training the train scheduling model.

[0035] When the performance on the validation set no longer improves, stop training. This is a common technique to prevent overfitting and can help us find a good balance between model complexity.

[0036] Furthermore, the model optimization algorithm is the Adam optimization algorithm.

[0037] Furthermore, the training and optimization of the train scheduling model according to the data priority specifically includes: selecting the appropriate number of hidden layers, the number of neurons in each layer, and the learning rate of the neural network model while satisfying the scheduling data priority.

[0038] Furthermore, the use of the test set data to evaluate and verify the optimization results of the train scheduling model specifically includes:

[0039] Step 1: Use the data in the test set as the input of the train scheduling model;

[0040] Step 2: Evaluate whether the output of the train scheduling model meets the time requirements in the priority level;

[0041] Step 3: Use past train scheduling results as a comparison to verify the time-shortening effect of the train scheduling model.

[0042] In the second aspect, the present application also provides a computing device, which has the function of implementing the method described in the first aspect above. The beneficial effects can be found in the description of the first aspect, which will not be repeated here. The function can be implemented by hardware, or by executing the corresponding software implementation through hardware. The hardware or software includes one or more modules corresponding to the above functions. In one possible design, the structure of the device includes an acquisition module, a training module, and optionally, a construction module. These modules can implement the function of the training node in the method example of the first aspect above. Please refer to the detailed description in the method example for details, which will not be repeated here.

[0043] In the third aspect, the present application also provides a computing device, which is used to implement the functions of the method described in the first aspect above. The beneficial effects can be found in the description of the first aspect and will not be repeated here. The structure of the computing device includes a processor and a memory, and the memory is used to store instructions and / or data. The memory is coupled to the processor, and when the processor executes the program instructions stored in the memory, the function of the training node in the example of the first aspect above can be implemented. The structure of the computing device also includes a communication interface for communicating with other devices.

[0044] In a fourth aspect, the present application also provides a computer-readable storage medium, in which instructions are stored, and when the computer-readable storage medium is run on a computer, the computer executes the method in the above-mentioned first aspect and various possible designs of the first aspect.

[0045] In a fifth aspect, the present application also provides a computer program product comprising instructions, which, when executed on a computer, enables the computer to execute the method in the above-mentioned first aspect and various possible designs of the first aspect.

[0046] In a sixth aspect, the present application also provides a computing chip, which is connected to a memory, and the chip is used to read and execute a software program stored in the memory, and to execute the methods in the above-mentioned first aspect and various possible implementation methods of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the embodiments of the present drawings or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present drawings. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

[0048] Figure 1 This is a flow chart of the intelligent vehicle and warehouse relocation management method based on machine learning;

[0049] Figure 2 Flowchart for training and optimizing the train scheduling model.

[0050] The purpose, features and advantages of this figure will be further described in conjunction with the embodiments with reference to the accompanying drawings. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is described and illustrated 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 intended to limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0052] Obviously, the drawings described below are only some examples or embodiments of the present application. For ordinary technicians in this field, the present application can also be applied to other similar scenarios based on these drawings without creative work. In addition, it can also be understood that although the efforts made in this development process may be complicated and lengthy, for ordinary technicians in this field related to the content disclosed in this application, some changes in design, manufacturing or production based on the technical content disclosed in this application are just conventional technical means, and should not be understood as insufficient content disclosed in this application.

[0053] If not otherwise specified, all embodiments and optional embodiments of the present application can be combined with each other to form a new technical solution.

[0054] Unless otherwise specified, all technical features and optional technical features of this application can be combined with each other to form a new technical solution.

[0055] If there is no special explanation, all steps of the present application can be performed sequentially or randomly, preferably sequentially. For example, the method includes steps (a) and (b), which means that the method may include steps (a) and (b) performed sequentially, or may include steps (b) and (a) performed sequentially. For example, the method may further include step (c), which means that step (c) may be added to the method in any order. For example, the method may include steps (a), (b) and (c), or may include steps (a), (c) and (b), or may include steps (c), (a) and (b), etc.

[0056] If there is no special explanation, the "include" and "comprising" mentioned in this application are open-ended or closed-ended. For example, the "include" and "comprising" may mean that other components not listed may also be included or only the listed components may be included or only the listed components may be included.

[0057] If not specifically stated, in this application, the term "or" is inclusive. For example, the phrase "A or B" means "A, B, or both A and B". More specifically, any of the following conditions satisfies the condition "A or B": A is true (or exists) and B is false (or does not exist); A is false (or does not exist) and B is true (or exists); or both A and B are true (or exist).

[0058] In order to better understand the solutions of the embodiments of the present application, some relevant terms and concepts that may be involved in the embodiments of the present application are first introduced below.

[0059] (1) Artificial intelligence (AI), also known as intelligent machinery or machine intelligence, refers to machines that are made by humans and can display intelligence. Generally speaking, AI refers to the technology that displays human intelligence through ordinary computer programs.

[0060] (2) Machine learning (ML). Machine learning is the core of artificial intelligence. The theory of machine learning mainly designs and analyzes some algorithms that allow computers to learn automatically. Machine learning algorithms are a type of algorithm that automatically analyzes data to obtain patterns and uses the patterns to predict unknown data. Therefore, the core of machine learning is data, algorithms (models), and computing power (computer computing power). The application areas of machine learning are very wide, including data mining, data classification, computer vision, natural language processing (NLP), biometric recognition, search engines, medical diagnosis, detection of credit card fraud, securities market analysis, DNA sequencing, speech and handwriting recognition, strategic games and robot applications. Machine learning is to design an algorithm model to process data and output the results that users want. Users can continuously tune the algorithm model to form more accurate data processing capabilities.

[0061] (3) Deep learning (DL). Deep learning is a method of machine learning. Its concept originates from the study of artificial neural networks. Multilayer perceptrons with multiple hidden layers are a type of deep learning structure, so deep learning is often referred to as deep neural networks. Compared with general machine learning, deep learning can automatically extract features, that is, automatically combine simple features into more complex features, and use these combinations to perform multi-layer weight learning to solve problems. The motivation for studying deep learning is to establish a neural network that simulates the human brain for analysis and learning. It imitates the mechanism of the human brain to interpret data, such as images, sounds, and text. Deep learning first emerged in image recognition, but in just a few years, deep learning has been extended to various fields of machine learning and has performed well. It has been applied in various fields such as image recognition, speech recognition, audio processing, natural language recognition, robot bioinformatics processing, search engines, human-computer games, online advertising targeting, medical automatic diagnosis, and finance.

[0062] (4) Neural networks (NN) are mathematical models that imitate the behavioral characteristics of animal neural networks and perform distributed parallel information processing. The purpose of information processing can be achieved by adjusting the interconnected relationships between a large number of nodes within the neural network. Neural networks have the ability to self-learn and adapt. Neural networks are usually used in artificial intelligence model training and data derivation processing.

[0063] Specifically, a neural network may usually include multiple layers connected end to end, such as convolutional layers, fully connected layers (FC), activation layers or pooling layers.

[0064] (5) Deep neural network (DNN), also known as multi-layer neural network, can be understood as a neural network with multiple hidden layers. According to the position of different layers, DNN can be divided into three categories: input layer, hidden layer, and output layer. Generally speaking, the first layer is the input layer, the last layer is the output layer, and the layers in between are all hidden layers. The layers are fully connected, that is, any neuron in the i-th layer must be connected to any neuron in the i+1-th layer. Although DNN looks complicated, the work of each layer is not complicated. Simply put, it is the following linear relationship expression: where is the input vector, is the output vector, is the offset vector, W is the weight matrix (also called coefficient), and α() is the activation function. Each layer simply performs such a simple operation on the input vector to obtain the output vector.

[0065] (6) Loss function. In the process of training a deep neural network, we hope that the output of the deep neural network is as close as possible to the value we really want to predict. Therefore, we can compare the predicted value of the current network with the target value we really want, and then update the weight vector of each layer of the neural network according to the difference between the two (of course, there is usually an initialization process before the first update, that is, pre-configuring parameters for each layer in the deep neural network). For example, if the predicted value of the network is high, adjust the weight vector to make it predict a lower value, and continue to adjust until the deep neural network can predict the target value we really want or a value very close to the target value we really want. Therefore, we need to pre-define "how to compare the difference between the predicted value and the target value", which is the loss function or objective function. They are important equations used to measure the difference between the predicted value and the target value. Taking the loss function as an example, the higher the output value (loss) of the loss function, the greater the difference. Then the training of the deep neural network becomes a process of minimizing this loss as much as possible.

[0066] (7) Long Short-Term Memory (LSTM) is a time-recurrent neural network designed to solve the long-term dependency problem of general RNNs (recurrent neural networks). All RNNs have a chain form of repeated neural network modules. In standard RNNs, this repeated structural module has only a very simple structure, such as a tanh layer.

[0067] (8) Adam (Adaptive Moment Estimation) is a deep learning optimization algorithm proposed by Diederik P. Kingma and Jimmy Ba in 2014. It combines the ideas of momentum and RMSProp, and aims to adjust the learning rate of each parameter by calculating the first-order moment estimate and second-order moment estimate of the gradient, thereby achieving more efficient network training.

[0068] (9) K-fold cross validation is a model evaluation method. Its principle is to randomly divide the original data set into K subsets, one of which is used as the validation set and the remaining K-1 subsets as the training set, and perform K times of model training and validation. Finally, the mean of the K times of validation results is used as the model evaluation result, thereby obtaining a more reliable model effect evaluation indicator.

[0069] like Figure 1-2 As shown, in this embodiment, a smart vehicle relocation and warehouse relocation management method based on machine learning is provided, comprising the following steps:

[0070] Step 1: Obtain recent train in-and-out maintenance scheduling data and pre-process the data;

[0071] By obtaining the recent train in-and-out maintenance scheduling data, the maintenance level and maintenance time of each train are obtained, and the average maintenance time of each maintenance level is used as the basis for the subsequent maintenance sorting time. The maintenance level of the railway train is divided into five levels, among which the first and second levels of maintenance are daily maintenance and operation maintenance, and the third level and above are advanced maintenance. Since the fifth-level maintenance of the train requires the EMU to be driven back to the main engine manufacturer for comprehensive maintenance, which the railway bureau cannot complete, in this embodiment, only the first to fourth level maintenance is recorded and scheduled;

[0072] Step 2: Mark the priority of each scheduling information in the data;

[0073] The priority of each dispatching information is divided according to the maintenance level of the train dispatching maintenance and the time interval for the next departure, so as to serve as one of the bases for subsequent maintenance sorting;

[0074] Step 3: Divide the data into training set, validation set and test set data;

[0075] The training set is used in the model training process to provide input features and targets so that the model can learn the mapping relationship between features and targets. The training set is the basis for model learning and parameter optimization. Through multiple iterative training in the training set, the model gradually improves its model capabilities.

[0076] The validation set is a set of samples set aside during the model training process. It is usually separated from the training set and does not participate in the model training. The validation set is used to adjust the model's hyperparameters, evaluate the model's generalization ability on unseen data, and prevent overfitting. By evaluating the model on the validation set, you can ensure that the model does not overfit to the training data and ignore the characteristics of the new data;

[0077] The test set is a dataset that is completely independent of the training set and validation set and is used to evaluate the performance of the model in the final application. The test set is used to evaluate the final performance of the model, especially its generalization ability on unknown data. The results of the test set can be used as the final indicator of the success of the model. The data of the test set should be the data that the model may encounter in actual applications, so it should not be used to adjust model parameters or select features.

[0078] The validation set is used in the training process. Generally, during training, the validation set is run once after several epochs to see the effect, because verifying the model too frequently will affect the training speed. The first advantage of doing this is that problems with the model or parameters can be discovered in a timely manner, such as when the model diverges on the validation set, abnormal results occur, or the mAP does not increase or increases very slowly. At this time, the training can be terminated in time, and the parameters can be readjusted or the model can be adjusted without waiting until the end of training. Another benefit is to verify the generalization ability of the model. If the effect on the validation set is much worse than that on the training set, it is time to consider whether the model is overfitting. At the same time, different models can also be compared through the validation set. In general neural networks, we use validation data sets to find the optimal network depth, or to determine the stopping point of the backpropagation algorithm or to select the number of hidden layer neurons in the neural network;

[0079] Step 4: Establish a train dispatching model based on deep learning algorithm;

[0080] Step 5: Use the training set and validation set data to train and optimize the train scheduling model based on the model optimization algorithm according to the data priority;

[0081] The core function of the model optimization algorithm is to adjust the model parameters in an iterative manner to minimize the loss function, thereby finding the best model parameters and accelerating the model training process.

[0082] Step 6: Use the test set data to evaluate and verify the optimization results of the train scheduling model;

[0083] The test set is used to evaluate the final performance of the model, especially its generalization ability on unknown data. The results of the test set can be used as the final indicator of the success of the model. The results obtained from the test set data are compared with the time spent on previous scheduling results to evaluate the effectiveness of the model;

[0084] Step 7: Collect relevant data information within the station, receive today's train in and out scheduling information, and obtain the scheduling strategy through the train scheduling model.

[0085] After the model training is completed, the current scheduling task data is input into the model to obtain the optimal scheduling strategy for train maintenance scheduling;

[0086] Furthermore, preprocessing includes data cleaning and data normalization.

[0087] In this embodiment, data cleaning is to remove inconsistent or unreasonable data in the display format of time, date, and train number in the acquired train dispatching data;

[0088] Data normalization is to make the data dimensionless so that data of different specifications can be converted to the same specification.

[0089] Furthermore, the priority of the dispatching information is divided according to the five-level maintenance level that the train needs to undergo and the time difference between the next departure of this train.

[0090] In this embodiment, the fifth-level maintenance cannot be carried out at the railway bureau station, so it is not considered, and the setting basis is the time difference between the four maintenance levels and the next departure of this train;

[0091] The maintenance level is 1, and the time difference between the next departure of this train is within 6 hours, so the priority is 1;

[0092] If the maintenance level is 1 or 2, and the time difference between the next departure of this train is 6 to 12 hours, the priority is 2;

[0093] If the maintenance level is 1 or 2, and the time difference between the next departure of this train is 12 to 24 hours, the priority is 3;

[0094] If the maintenance level is within 3 and the time difference between the next departure of this train is 24 to 48 hours, the priority is 4;

[0095] If the maintenance level is 4 and the required maintenance time is too long, the priority level will be 5 and the train operation will be suspended;

[0096] The above priority levels are divided according to actual conditions. When the maintenance level is 1, it is daily maintenance, which takes about four hours and needs to be arranged as soon as possible for daily departure maintenance. When the maintenance level increases, the time required for maintenance will increase significantly, so the railway control system will arrange the departure time for a longer time, and there will be no situation where the maintenance level is very high and the train is in a hurry to depart.

[0097] Furthermore, the data is divided into training set, validation set and test set data, and K-fold cross validation is adopted.

[0098] In this embodiment, K-fold cross validation randomly divides the original data set into 20 subsets, one of which is used as a validation set, and the remaining 19 subsets are used as training sets, and 19 model training and validation are performed;

[0099] Furthermore, a train scheduling model is established based on a deep learning algorithm, including establishing a train scheduling model based on an LSTM algorithm.

[0100] Furthermore, the optimization goal of the train scheduling model is to minimize the completion time of the maintenance scheduling task and the time requirement that meets the priority.

[0101] Furthermore, the train scheduling model is trained and optimized based on the model optimization algorithm using the training set and validation set data according to the data priority, including:

[0102] Step 1: Use the training set data to train the train scheduling model;

[0103] Use the training set data to forward propagate the model and calculate the gradient of the loss function with respect to each parameter;

[0104] Step 2: Based on the model optimization algorithm, update the parameters of the train scheduling model according to the data of the training process;

[0105] According to the Adam algorithm, each parameter is updated using the current gradient, the momentum of the previous gradient, and an estimate of the variance;

[0106] Step 3: During the training process, use the validation set data to evaluate the generalization ability of the model at regular intervals.

[0107] During the training process, you need to use the validation set regularly to evaluate the generalization ability of the model, which helps prevent overfitting and adjust hyperparameters, such as the learning rate. The Adam algorithm automatically adjusts the learning rate of each parameter. If you find that the performance of the model on the validation set begins to decline, you may need to readjust the model or its hyperparameters;

[0108] Step 4: When the model training performance on the validation set no longer improves, stop training the train scheduling model.

[0109] When the performance on the validation set no longer improves, stop training. This is a common technique to prevent overfitting and can help us find a good balance between model complexity.

[0110] Furthermore, the model optimization algorithm is the Adam optimization algorithm.

[0111] Furthermore, the train scheduling model is trained and optimized according to the data priority, specifically including: selecting the appropriate number of hidden layers, the number of neurons in each layer, and the learning rate of the neural network model while satisfying the scheduling data priority.

[0112] Furthermore, the test set data is used to evaluate and verify the optimization results of the train scheduling model, including:

[0113] Step 1: Use the data in the test set as the input of the train scheduling model;

[0114] Step 2: Evaluate whether the output of the train scheduling model meets the time requirements in the priority level;

[0115] That is, check whether the departure time of the centralized dispatching data meets the requirements and whether the train can be dispatched within the specified time;

[0116] Step 3: Use past train scheduling results as a comparison to verify the time-shortening effect of the train scheduling model.

[0117] In this embodiment, the total time spent in the train scheduling model using the test set data is compared with the scheduling time results of the same data in past train scheduling, saving 21% of time, significantly reducing the scheduling time during train maintenance and improving railway operating efficiency.

[0118] It should be noted that the present application is not limited to the above-mentioned embodiments. The above-mentioned embodiments are only examples, and the embodiments having the same structure as the technical idea and exerting the same effect within the scope of the technical solution of the present application are all included in the technical scope of the present application. In addition, without departing from the scope of the main purpose of the present application, various modifications that can be thought of by those skilled in the art to the embodiments and other methods of combining some of the constituent elements in the embodiments are also included in the scope of the present application.

Claims

1. An intelligent vehicle and warehouse relocation management method based on machine learning, characterized in that: The following steps are involved: Step 1: Obtain recent train in-and-out maintenance scheduling data and pre-process the data; Step 2: Mark the priority of each scheduling information in the data; Step 3: Divide the data into training set, validation set and test set data; Step 4: Establish a train dispatching model based on deep learning algorithm; Step 5: Use the training set and validation set data to train and optimize the train scheduling model based on the model optimization algorithm according to the data priority; Step 6: Use the test set data to evaluate and verify the optimization results of the train scheduling model; Step 7: Collect relevant data information within the station, receive today's train in and out scheduling information, and obtain the scheduling strategy through the train scheduling model.

2. The intelligent vehicle and warehouse relocation management method based on machine learning according to claim 1 is characterized in that: The preprocessing includes data cleaning and data normalization.

3. The intelligent vehicle and warehouse relocation management method based on machine learning according to claim 1 is characterized in that: The priority of the dispatching information is divided according to the five-level maintenance level that the train needs to undergo and the time difference between the next departure of this train.

4. The intelligent vehicle and warehouse relocation management method based on machine learning according to claim 1 is characterized in that: The data is divided into a training set, a validation set and a test set, and K-fold cross validation is used.

5. The intelligent vehicle and warehouse relocation management method based on machine learning according to claim 1 is characterized in that: The establishing of a train scheduling model based on a deep learning algorithm includes establishing a train scheduling model based on an LSTM algorithm.

6. The intelligent vehicle and warehouse relocation management method based on machine learning according to claim 5 is characterized in that: The optimization goal of the train scheduling model is to minimize the completion time of the maintenance scheduling task and the time requirement that meets the priority.

7. The intelligent vehicle and warehouse relocation management method based on machine learning according to claim 1 is characterized in that: The train scheduling model is trained and optimized based on the model optimization algorithm using the training set and the validation set data according to the data priority, specifically including: Step 1: Use the training set data to train the train scheduling model; Step 2: Based on the model optimization algorithm, update the parameters of the train scheduling model according to the data of the training process; Step 3: During the training process, use the validation set data to evaluate the generalization ability of the model at regular intervals. Step 4: When the model training performance on the validation set no longer improves, stop training the train scheduling model.

8. The intelligent vehicle and warehouse relocation management method based on machine learning according to claim 7 is characterized in that: The model optimization algorithm is the Adam optimization algorithm.

9. The intelligent vehicle and warehouse relocation management method based on machine learning according to claim 1 is characterized in that: The training and optimization of the train scheduling model according to the data priority specifically includes: selecting the appropriate number of hidden layers, the number of neurons in each layer, and the learning rate of the neural network model while satisfying the scheduling data priority.

10. The intelligent vehicle and warehouse relocation management method based on machine learning according to claim 1 is characterized in that: The use of test set data to evaluate and verify the optimization results of the train scheduling model specifically includes: Step 1: Use the data in the test set as the input of the train scheduling model; Step 2: Evaluate whether the output of the train scheduling model meets the time requirements in the priority level; Step 3: Use past train scheduling results as a comparison to verify the time-shortening effect of the train scheduling model.

Citation Information

Patent Citations

  • Urban rail train inspection and transportation integrated plan optimization method considering operation calendar

    CN118036933A

  • Low earth orbit satellite parallel simulation optimization method and system based on LSTM and Bayesian optimization

    CN118157797A

Cited By

  • Scheduling and control system and method for freight train

    CN119705561A