Rail transit operation scheme analysis method and system

By deploying sensors and machine learning algorithms in the rail transit system, collecting and analyzing data in real time and automatically optimizing operation solutions, the problem of inefficiency of traditional methods is solved, and the intelligence and automation of rail transit operations are realized, and operational efficiency and passengers are improved. Satisfaction.

CN119005505AInactive Publication Date: 2024-11-22NANJING ACTOR TRANSPORTATION TECHNOLOGY CO LTD
View PDF 0 Cites 4 Cited by

Patent Information

Application Number
CN202411036607.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-31
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing rail transit operation plan analysis methods rely on traditional methods, are inefficient, inaccurate decision-making, and cannot meet actual needs.

Method used

By deploying sensors on rail transit vehicles and sites, collecting operation data in real time, performing data preprocessing and big data analysis, building an operation plan analysis model based on machine learning algorithms, and automatically selecting and optimizing operation plans using intelligent decision-making algorithms.

Benefits of technology

It has realized the intelligence and automation of rail transit operations, improved operational efficiency and passenger satisfaction, reduced operating time and energy consumption, and enhanced operational flexibility and adaptability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119005505A_ABST
    Figure CN119005505A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of rail transit operation, and discloses a rail transit operation scheme analysis method and system, and the method comprises the following steps: deploying sensor equipment on a rail transit vehicle and a station, and collecting operation data information in real time; performing cleaning, denoising and normalization processing on the collected data, performing statistics and analysis on the preprocessed data by using a big data analysis technology, constructing a rail transit operation scheme analysis model based on a machine learning algorithm, and performing evaluation and comparison on different operation schemes according to real-time data and the trained model; and an intelligent decision algorithm is utilized to automatically select an optimal operation scheme, and real-time adjustment and optimization are carried out. According to the rail transit operation scheme analysis method and system, by introducing the artificial intelligence technology, big data analysis and machine learning, comprehensive analysis and optimization of the rail transit operation scheme are achieved, the flexibility and adaptability of operation are improved, and the method and system have wide application prospects.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of rail transit operation technology, and in particular to a rail transit operation plan analysis method and a system thereof. Background Art

[0002] Rail transit systems play a vital role in urban mobility, providing efficient and sustainable transportation for millions of people. However, the operation of rail transit networks is complex, and optimizing their efficiency and performance poses significant challenges.

[0003] The existing rail transit operation plan analysis method and its system rely on traditional methods. The formulation and optimization of operation plans mainly rely on manual experience and rules, which are inefficient, inaccurate in decision-making, still have limitations, and cannot meet actual needs. Therefore, it is necessary to design corresponding technical solutions to solve them. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the deficiencies in the prior art, the present invention provides a rail transit operation plan analysis method and a system thereof, which solve the technical problems that the rail transit operation plan analysis method and the system thereof rely on traditional methods, the formulation and optimization of operation plans mainly rely on manual experience and rules, are inefficient, make inaccurate decisions, have limitations, and cannot meet actual needs.

[0006] (II) Technical solution

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a rail transit operation plan analysis method and system thereof, the method steps include the following:

[0008] S1. Data collection: Deploy sensor equipment on rail transit vehicles and stations to collect operational data information in real time;

[0009] S2. Data preprocessing: cleaning, denoising and normalizing the collected data to ensure the accuracy and consistency of the data;

[0010] S3. Data analysis: Use big data analysis technology to count and analyze the pre-processed data to extract useful information and features;

[0011] S4. Model training: Based on machine learning algorithms, build a rail transit operation plan analysis model and use historical data for training and optimization;

[0012] S5. Solution evaluation: Evaluate and compare different operation solutions based on real-time data and trained models;

[0013] S6. Intelligent decision-making: Based on the evaluation results, use intelligent decision-making algorithms to automatically select the best operation plan, and adjust and optimize it in real time.

[0014] Preferably, in step S1, the collected operational data information includes: vehicle position, vehicle speed, number of passengers, vehicle status, vehicle temperature, door switch status, train running time, train stop time, compartment internal temperature and humidity, and vehicle energy consumption;

[0015] Vehicle location: Real-time acquisition of vehicle location information on the track to monitor vehicle running trajectory and calculate travel distance;

[0016] Vehicle speed: record the real-time speed of the vehicle to evaluate the vehicle's driving conditions and operating efficiency;

[0017] Passenger count: The number of passengers is counted in real time through sensors in the car or counters on the doors to assess the crowding level and passenger load of the vehicle;

[0018] Vehicle status: monitor the operating status of the vehicle, including whether the vehicle is operating normally or has any faults;

[0019] Vehicle temperature: measures the temperature inside the vehicle to regulate the operation of the air conditioner;

[0020] Door open / close status: records the open / close status of the doors to ensure the safety of passengers and a smooth ride;

[0021] Train running time: records the running time of the train from the starting station to the terminal station, which is used to evaluate the running speed and punctuality of the train;

[0022] Train stop time: records the time the train stays at each station, which is used to evaluate the train transfer efficiency and station congestion;

[0023] Temperature and humidity inside the car: monitor the temperature and humidity inside the car to provide passengers with a comfortable riding environment;

[0024] Vehicle energy consumption: It is used to evaluate the vehicle's energy efficiency and energy utilization by measuring the vehicle's energy consumption.

[0025] Preferably, in step S2, the specific steps of data preprocessing include the following:

[0026] S201, data cleaning: clean the collected data to remove duplicate, missing or erroneous data to ensure the integrity and accuracy of the data;

[0027] S202, data denoising: removing noise from data through filtering technology to improve data quality and stability;

[0028] S203, data normalization: normalizing the data to unify data of different ranges to the same scale;

[0029] S204, data conversion: converting or transforming the data, converting the time format into a standard format, and converting the text data into a numeric type;

[0030] S205, data integration: integrating data from different sensors or devices to form a complete data set;

[0031] S206, data sampling: sampling the data to reduce the data volume or balance the data distribution;

[0032] S207, data feature extraction: extracting useful features from the original data, including average speed, travel distance and congestion level;

[0033] S208, data standardization: standardize the data to meet statistical requirements.

[0034] Preferably, in step S3, the pre-processed data is statistically analyzed using big data analysis technology, which specifically includes the following steps:

[0035] S301, descriptive statistical analysis: by calculating the statistical indicators of the mean, variance, maximum value and minimum value of the data, the overall characteristics of the data are described and analyzed to assist in understanding the distribution, trend and outliers of the data;

[0036] S302, Correlation Analysis: By analyzing the correlation between data, find the correlation and mutual influence between data, including correlation coefficient analysis, covariance analysis and regression analysis, which are used to reveal the intrinsic relationship between data;

[0037] S303, cluster analysis: divide the data into several similar groups, so that the data within the same group has high similarity, while the data between different groups has low similarity, which is used to assist in identifying potential patterns and groups in the data, thereby providing a basis for subsequent decision-making;

[0038] S304, Time Series Analysis: For time series data, by analyzing the trend, periodicity and seasonality of the data, predict future development trends and change patterns. Time series analysis methods include smoothing method, exponential smoothing method and ARIMA model;

[0039] S305, Visual Analysis: Display data in an intuitive and understandable way through charts, graphs and visualization techniques to assist in discovering patterns and trends in the data.

[0040] Preferably, in step S4, the specific steps of constructing a rail transit operation plan analysis model based on a machine learning algorithm are as follows:

[0041] S401, data preparation: organize historical rail transit operation data into a format suitable for machine learning algorithm processing, including input features and output labels;

[0042] S402, Feature Engineering: Process and transform input features to extract useful information.

[0043] Feature engineering methods include feature selection, feature scaling, or feature combination;

[0044] S403, model selection: select a suitable machine learning algorithm based on the nature of the problem and the characteristics of the data, including linear regression;

[0045] y=w0+w1x1+w2x2+...+wn*xn;

[0046] Among them, y is the output label, x1, x2, ..., xn are the input features, and w0, w1, w2, ..., wn are the parameters of the model;

[0047] S404, model training: using historical data to train the model, and optimizing the parameters of the model by minimizing the loss function, so as to enable the model to better fit the data;

[0048] S405, model evaluation: use evaluation indicators to evaluate the performance of the model, including mean square error or accuracy, and use cross-validation method to evaluate the model;

[0049] S406, model tuning: tuning the model according to the evaluation results, including adjusting the model's hyperparameters and adding regularization terms;

[0050] S407, model application: use the trained model to predict and analyze new data to obtain relevant results of the rail transit operation plan.

[0051] Preferably, in step S5, the method steps for evaluating and comparing different operation plans include the following:

[0052] S501, determine evaluation indicators: according to the objectives and needs of rail transit operations, determine evaluation indicators, including operating time, passenger satisfaction or energy consumption, to quantify the effects and performance of the operation plan and provide a basis for evaluation and comparison;

[0053] S502, data collection and processing: according to the evaluation indicators, relevant data is collected in real time, and pre-processed and normalized to ensure the accuracy and comparability of the data;

[0054] S503, model application: using the trained model, predict and analyze the real-time data to obtain the evaluation results of each operation plan, and evaluate the performance of each plan on different indicators based on the difference between the model's predicted value and the actual data;

[0055] S504, indicator evaluation and comparison: conduct a comprehensive evaluation and comparison of the evaluation results of each operation plan, use statistical methods or visualization methods to analyze the evaluation results, compare the differences and advantages and disadvantages of each plan in different indicators, and find the optimal operation plan.

[0056] Preferably, in step S6, the intelligent decision-making algorithm uses a combination of a particle swarm algorithm and a deep reinforcement learning algorithm to automatically select the optimal operation plan and perform real-time adjustment and optimization. The specific method steps include the following:

[0057] S601, collect and prepare historical operation data of rail transit, construct state space and action space, and map the selection and adjustment of operation plans into a combination of states and actions;

[0058] S602, setting an initial particle group of the PSO algorithm, and randomly assigning an initial position and speed to each particle, and representing the position of each particle as a set of parameters of an operation plan;

[0059] S603, using deep reinforcement learning algorithm to build an intelligent decision-making model;

[0060] During the training process, the model learns the decision-making strategy for selecting and adjusting operation plans by interacting with the environment;

[0061] Using historical data and the particle swarm generated by PSO, the model continuously optimizes the decision-making strategy through feedback signals;

[0062] S604, in each iteration, PSO calculates the fitness function according to the current position and velocity of the particle;

[0063] According to the fitness function, the speed and position of the particles are updated, and the parameter combination of the new operation plan is searched;

[0064] PSO will use the evaluation results of the DRL model to guide the search of the particle swarm and gradually approach the optimal operation plan;

[0065] S605, after each iteration, evaluating the performance index of the optimal operation plan according to the parameter combination of the optimal operation plan obtained by the PSO search;

[0066] Based on the real-time collected data and current status, the DRL model is used to make real-time decisions and adjust the parameters of the operation plan;

[0067] Adjust and optimize operation plans in real time to adapt to changing passenger demands and actual traffic conditions.

[0068] Preferably, in the PSO algorithm, the fitness function is used to evaluate the performance of each particle, and the evaluation indicators related to the DRL model include:

[0069] Adaptability = passenger satisfaction score - operating cost;

[0070] Fitness = α*passenger satisfaction score + β*operation efficiency - γ*operation cost;

[0071] Among them, α, β, and γ are weight coefficients;

[0072] In the DRL model, a reward function is used to evaluate the quality of each state and action, guiding the model to learn the optimal decision-making strategy, including:

[0073] Reward = passenger satisfaction score - operating cost;

[0074] Reward = α*passenger satisfaction score + β*operation efficiency - γ*operation cost;

[0075] Similarly, α, β, and γ are weight coefficients.

[0076] Preferably, in deep reinforcement learning, the Q-value function represents the expected cumulative reward of performing an action in a given state, and a neural network is used to approximate the Q-value function, i.e., the Q network.

[0077] The specific Q-value function is defined according to the problem, including:

[0078] Q(s,a)=R(s,a)+γ*max(Q(s',a'))

[0079] Among them, s and a represent state and action respectively, R(s,a) represents the immediate reward for performing action a in state s, s' represents the next state, a' represents the optimal action in the next state, and γ is the discount factor.

[0080] (III) Beneficial effects

[0081] Compared with the prior art, the beneficial effects of the present invention are as follows: by introducing artificial intelligence technology, through big data analysis and machine learning, comprehensive analysis and optimization of rail transit operation plans can be achieved to improve operational efficiency and passenger satisfaction; it can automatically process and analyze a large amount of operational data, and make intelligent decisions based on real-time conditions, thereby realizing the intelligence and automation of rail transit operations; compared with traditional rule-based and experience-based methods, it can more accurately predict and evaluate the effects of different plans, which can reduce operating time, improve operational stability, and reduce energy consumption. At the same time, it can timely adjust and optimize operation plans, improve operational flexibility and adaptability, and has broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] Figure 1 It is a schematic diagram of the overall method steps of the present invention;

[0083] Figure 2 This is a schematic diagram of the collected operational data information of the present invention;

[0084] Figure 3 It is a schematic diagram of the specific steps of data preprocessing of the present invention;

[0085] Figure 4 This is a schematic diagram of the specific steps of using big data analysis technology to perform statistics and analysis on pre-processed data in the present invention;

[0086] Figure 5 A schematic diagram of the specific steps of constructing a rail transit operation plan analysis model based on a machine learning algorithm of the present invention;

[0087] Figure 6 A schematic diagram of the method steps for evaluating and comparing different operation plans of the present invention;

[0088] Figure 7 The intelligent decision-making algorithm of the present invention adopts a combination of a particle swarm algorithm and a deep reinforcement learning algorithm to automatically select the optimal operation plan, and perform real-time adjustment and optimization of the specific method steps. DETAILED DESCRIPTION

[0089] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0090] See also Figure 1-Figure 7 The embodiment of the present invention provides a technical solution: a rail transit operation plan analysis method and a system thereof, the method steps include the following:

[0091] S1. Data collection: Deploy sensor equipment on rail transit vehicles and stations to collect operational data information in real time;

[0092] S2. Data preprocessing: cleaning, denoising and normalizing the collected data to ensure the accuracy and consistency of the data;

[0093] S3. Data analysis: Use big data analysis technology to count and analyze the pre-processed data to extract useful information and features;

[0094] S4. Model training: Based on machine learning algorithms, build a rail transit operation plan analysis model and use historical data for training and optimization;

[0095] S5. Solution evaluation: Evaluate and compare different operation solutions based on real-time data and trained models;

[0096] S6. Intelligent decision-making: Based on the evaluation results, use intelligent decision-making algorithms to automatically select the best operation plan, and adjust and optimize it in real time.

[0097] Further improved, in step S1, the collected operational data information includes: vehicle position, vehicle speed, number of passengers, vehicle status, vehicle temperature, door switch status, train running time, train stop time, compartment internal temperature and humidity, and vehicle energy consumption;

[0098] Vehicle location: Real-time acquisition of vehicle location information on the track to monitor vehicle running trajectory and calculate travel distance;

[0099] Vehicle speed: record the real-time speed of the vehicle to evaluate the vehicle's driving conditions and operating efficiency;

[0100] Passenger count: The number of passengers is counted in real time through sensors in the car or counters on the doors to assess the crowding level and passenger load of the vehicle;

[0101] Vehicle status: monitor the operating status of the vehicle, including whether the vehicle is operating normally or has any faults;

[0102] Vehicle temperature: measures the temperature inside the vehicle to regulate the operation of the air conditioner;

[0103] Door open / close status: records the open / close status of the doors to ensure the safety of passengers and a smooth ride;

[0104] Train running time: records the running time of the train from the starting station to the terminal station, which is used to evaluate the running speed and punctuality of the train;

[0105] Train stop time: records the time the train stays at each station, which is used to evaluate the train transfer efficiency and station congestion;

[0106] Temperature and humidity inside the car: monitor the temperature and humidity inside the car to provide passengers with a comfortable riding environment;

[0107] Vehicle energy consumption: It is used to evaluate the vehicle's energy efficiency and energy utilization by measuring the vehicle's energy consumption.

[0108] Further improved, in step S2, the specific steps of data preprocessing include the following:

[0109] S201, data cleaning: clean the collected data to remove duplicate, missing or erroneous data to ensure the integrity and accuracy of the data and avoid erroneous results in subsequent analysis;

[0110] S202, data denoising: removing noise from data through filtering technology to improve data quality and stability and reduce the interference of noise on analysis results;

[0111] The Kalman filter algorithm is used to estimate the state of the signal by comprehensively utilizing the system model and observation data. It is suitable for linear systems and Gaussian noise and can effectively handle the noise of dynamic systems.

[0112] S203, data normalization: normalizing the data to unify data of different ranges to the same scale;

[0113] S204, data conversion: converting or transforming the data, converting the time format into a standard format, and converting the text data into a numeric type;

[0114] S205, data integration: integrating data from different sensors or devices to form a complete data set;

[0115] S206, data sampling: sampling data to reduce the amount of data or balance data distribution, and improve the efficiency and accuracy of analysis and modeling;

[0116] S207, data feature extraction: extracting useful features from the original data, including average speed, travel distance and congestion level;

[0117] S208, data standardization: standardize the data to meet statistical requirements, including a mean of 0 and a variance of 1, for training and optimization of machine learning algorithms.

[0118] The goal of data preprocessing is to improve the quality and availability of data to ensure the accuracy and reliability of subsequent analysis and modeling, improve the quality and consistency of data, reduce the impact of errors and noise, and make the data more in line with the requirements of analysis and modeling, thereby obtaining more accurate and reliable results.

[0119] Further improved, in step S3, the pre-processed data is counted and analyzed using big data analysis technology, specifically including the following steps:

[0120] S301, descriptive statistical analysis: by calculating the statistical indicators of the mean, variance, maximum value and minimum value of the data, the overall characteristics of the data are described and analyzed to assist in understanding the distribution, trend and outliers of the data;

[0121] S302, Correlation Analysis: By analyzing the correlation between data, find the correlation and mutual influence between data, including correlation coefficient analysis, covariance analysis and regression analysis, which are used to reveal the intrinsic relationship between data;

[0122] S303, cluster analysis: divide the data into several similar groups, so that the data within the same group has high similarity, while the data between different groups has low similarity, which is used to assist in identifying potential patterns and groups in the data, thereby providing a basis for subsequent decision-making;

[0123] S304, Time Series Analysis: For time series data, by analyzing the trend, periodicity and seasonality of the data, predict future development trends and change patterns. Time series analysis methods include smoothing method, exponential smoothing method and ARIMA model;

[0124] S305, Visual Analysis: Display data in an intuitive and understandable way through charts, graphs and visualization techniques to assist in discovering patterns and trends in the data, deepen understanding of the data, and help decision makers better interpret and use the data.

[0125] Select appropriate methods for statistics and analysis based on specific needs and data characteristics to extract useful information and features, providing support for subsequent model training and solution evaluation.

[0126] Further improved, in step S4, the specific steps of constructing a rail transit operation plan analysis model based on a machine learning algorithm are as follows:

[0127] S401, data preparation: organize historical rail transit operation data into a format suitable for machine learning algorithm processing, including input features and output labels;

[0128] S402, Feature Engineering: Process and transform input features to extract useful information.

[0129] Feature engineering methods include feature selection, feature scaling, or feature combination;

[0130] S403, model selection: select a suitable machine learning algorithm based on the nature of the problem and the characteristics of the data, including linear regression;

[0131] y=w0+w1x1+w2x2+...+wn*xn;

[0132] Among them, y is the output label, x1, x2, ..., xn are the input features, and w0, w1, w2, ..., wn are the parameters of the model;

[0133] S404, model training: using historical data to train the model, and optimizing the parameters of the model by minimizing the loss function, so as to enable the model to better fit the data;

[0134] S405, model evaluation: use evaluation indicators to evaluate the performance of the model, including mean square error or accuracy, and use cross-validation method to evaluate the model;

[0135] S406, model tuning: tuning the model according to the evaluation results, including adjusting the model's hyperparameters and adding regularization terms;

[0136] S407, model application: use the trained model to predict and analyze new data to obtain relevant results of the rail transit operation plan.

[0137] As a further improvement, in step S5, the method steps for evaluating and comparing different operation plans include the following:

[0138] S501, determine evaluation indicators: according to the objectives and needs of rail transit operations, determine evaluation indicators, including operating time, passenger satisfaction or energy consumption, to quantify the effects and performance of the operation plan and provide a basis for evaluation and comparison;

[0139] S502, data collection and processing: according to the evaluation indicators, relevant data is collected in real time, and pre-processed and normalized to ensure the accuracy and comparability of the data;

[0140] S503, model application: using the trained model, predict and analyze the real-time data to obtain the evaluation results of each operation plan, and evaluate the performance of each plan on different indicators based on the difference between the model's predicted value and the actual data;

[0141] S504, indicator evaluation and comparison: conduct a comprehensive evaluation and comparison of the evaluation results of each operation plan, use statistical methods or visualization methods to analyze the evaluation results, compare the differences and advantages and disadvantages of each plan in different indicators, and find the optimal operation plan.

[0142] By evaluating and comparing real-time data and combining it with pre-set evaluation indicators, the effects and performance of different operating plans can be objectively evaluated and compared, thus providing a scientific basis for decision-making and adjusting and optimizing the operating plan in real time.

[0143] Further improved, in step S6, the intelligent decision-making algorithm uses a combination of a particle swarm algorithm and a deep reinforcement learning algorithm to automatically select the optimal operation plan and perform real-time adjustment and optimization. The specific method steps include the following:

[0144] S601, collect and prepare historical operation data of rail transit, construct state space and action space, and map the selection and adjustment of operation plans into a combination of states and actions;

[0145] S602, setting an initial particle group of the PSO algorithm, and randomly assigning an initial position and speed to each particle, and representing the position of each particle as a set of parameters of an operation plan;

[0146] S603, using deep reinforcement learning algorithm to build an intelligent decision-making model;

[0147] During the training process, the model learns the decision-making strategy for selecting and adjusting operation plans by interacting with the environment;

[0148] Using historical data and the particle swarm generated by PSO, the model continuously optimizes the decision-making strategy through feedback signals (reward or value function);

[0149] S604, in each iteration, the PSO calculates the fitness function (according to the evaluation index of the DRL model) according to the current position and speed of the particle;

[0150] According to the fitness function, the speed and position of the particles are updated, and the parameter combination of the new operation plan is searched;

[0151] PSO will use the evaluation results of the DRL model to guide the search of the particle swarm and gradually approach the optimal operation plan;

[0152] S605, after each iteration, evaluating the performance index of the optimal operation plan according to the parameter combination of the optimal operation plan obtained by the PSO search;

[0153] Based on the real-time collected data and current status, the DRL model is used to make real-time decisions and adjust the parameters of the operation plan;

[0154] Adjust and optimize operation plans in real time to adapt to changing passenger demands and actual traffic conditions.

[0155] Through continuous iteration and optimization, the method combining particle swarm algorithm and deep reinforcement learning algorithm can automatically select the optimal operation plan, and adjust and optimize in real time to improve the efficiency and service quality of rail transit operation. The specific parameter settings, model architecture and optimization strategies will be adjusted and optimized according to actual conditions.

[0156] DRL is the abbreviation of Deep Reinforcement Learning. Deep reinforcement learning is a method that combines deep learning and reinforcement learning to solve decision-making problems with complex state space and action space. In deep reinforcement learning, the agent learns the optimal behavior strategy through interaction with the environment to maximize the cumulative reward.

[0157] The core of deep reinforcement learning is reinforcement learning, which is a method of making decisions through interactive learning between the agent and the environment. In reinforcement learning, the agent chooses to perform different actions by observing the state of the environment, and evaluates the quality of the action through the rewards or penalties given by the environment. Through continuous interaction with the environment and trial and error, the agent gradually learns the strategy of choosing the optimal action under different states.

[0158] Deep reinforcement learning introduces deep learning techniques, especially neural networks, to approximate the representation of value functions or policy functions. Deep neural networks can model complex state spaces and output corresponding actions or value functions. By using deep learning techniques, deep reinforcement learning can handle problems with high-dimensional state spaces and action spaces.

[0159] Deep reinforcement learning has achieved remarkable results in many fields. By combining deep learning and reinforcement learning methods, deep reinforcement learning can learn and formulate efficient decision-making strategies in complex environments, and has broad application prospects.

[0160] Further improved, in the PSO algorithm, the fitness function is used to evaluate the performance of each particle, and the evaluation indicators related to the DRL model include:

[0161] Adaptability = passenger satisfaction score - operating cost;

[0162] Fitness = α*passenger satisfaction score + β*operation efficiency - γ*operation cost;

[0163] Among them, α, β, and γ are weight coefficients;

[0164] In the DRL model, a reward function is used to evaluate the quality of each state and action, guiding the model to learn the optimal decision-making strategy, including:

[0165] Reward = passenger satisfaction score - operating cost;

[0166] Reward = α*passenger satisfaction score + β*operation efficiency - γ*operation cost;

[0167] Similarly, α, β, and γ are weight coefficients.

[0168] Specifically, in deep reinforcement learning, the Q-value function represents the expected cumulative reward of performing an action in a given state, and a neural network is used to approximate the Q-value function, namely the Q network.

[0169] The specific Q-value function is defined according to the problem, including:

[0170] Q(s,a)=R(s,a)+γ*max(Q(s',a'))

[0171] Among them, s and a represent state and action respectively, R(s,a) represents the immediate reward for performing action a in state s, s' represents the next state, a' represents the optimal action in the next state, and γ is the discount factor.

[0172] In summary, by introducing artificial intelligence technology, through big data analysis and machine learning, comprehensive analysis and optimization of rail transit operation plans can be achieved to improve operational efficiency and passenger satisfaction; it can automatically process and analyze large amounts of operational data, and make intelligent decisions based on real-time conditions, thereby realizing the intelligence and automation of rail transit operations; compared with traditional rule-based and experience-based methods, it can more accurately predict and evaluate the effects of different plans, which can reduce operating time, improve operational stability, and reduce energy consumption. At the same time, it can timely adjust and optimize operation plans, improve operational flexibility and adaptability, and has broad application prospects.

[0173] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the attached claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any figure mark in the claims should not be regarded as limiting the claims involved.

[0174] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A rail transit operation plan analysis method, characterized in that: The method steps include the following: S1. Data collection: Deploy sensor equipment on rail transit vehicles and stations to collect operational data information in real time; S2. Data preprocessing: cleaning, denoising and normalizing the collected data to ensure the accuracy and consistency of the data; S3. Data analysis: Use big data analysis technology to count and analyze the pre-processed data to extract useful information and features; S4. Model training: Based on machine learning algorithms, build a rail transit operation plan analysis model and use historical data for training and optimization; S5. Solution evaluation: Evaluate and compare different operation solutions based on real-time data and trained models; S6. Intelligent decision-making: Based on the evaluation results, the intelligent decision-making algorithm is used to automatically select the optimal operation plan, and adjust and optimize it in real time; The intelligent decision-making algorithm combines the particle swarm algorithm and the deep reinforcement learning algorithm to automatically select the optimal operation plan and make real-time adjustments and optimizations. The specific methods and steps are as follows: S601, collect and prepare historical operation data of rail transit, construct state space and action space, and map the selection and adjustment of operation plans into a combination of states and actions; S602, setting an initial particle group of the PSO algorithm, and randomly assigning an initial position and speed to each particle, and representing the position of each particle as a set of parameters of an operation plan; S603, using deep reinforcement learning algorithm, builds an intelligent decision-making model; during the training process, the model learns to select and adjust the decision-making strategy of the operation plan by interacting with the environment; using historical data and the particle swarm generated by PSO, the model continuously optimizes the decision-making strategy through feedback signals; S604, in each iteration, PSO calculates the fitness function according to the current position and speed of the particle; according to the fitness function, updates the speed and position of the particle, and searches for the parameter combination of the new operation plan; PSO will use the evaluation results of the DRL model to guide the search of the particle swarm and gradually approach the optimal operation plan; S605, after each iteration, evaluate the performance index of the optimal operation plan parameter combination obtained by PSO search; make real-time decisions and adjust the parameters of the operation plan using the DRL model according to the real-time collected data and current status; adjust and optimize the operation plan in real time to adapt to the changing passenger demand and actual traffic conditions; In the PSO algorithm, the fitness function is used to evaluate the performance of each particle, and the evaluation indicators related to the DRL model include: Adaptability = passenger satisfaction score - operating cost; Fitness = α*passenger satisfaction score + β*operation efficiency - γ*operation cost; Among them, α, β, and γ are weight coefficients; In the DRL model, a reward function is used to evaluate the quality of each state and action, guiding the model to learn the optimal decision-making strategy, including: Reward = passenger satisfaction score - operating cost; Reward = α*passenger satisfaction score + β*operation efficiency - γ*operation cost; Similarly, α, β, and γ are weight coefficients; In deep reinforcement learning algorithms, the Q-value function is used to represent the expected cumulative reward of performing an action in a given state, and a neural network is used to approximate the Q-value function, namely the Q network. The specific Q-value function is defined according to the problem, including: Q(s,a)=R(s,a)+γ*max(Q(s',a')) Among them, s and a represent state and action respectively, R(s,a) represents the immediate reward for performing action a in state s, s' represents the next state, a' represents the optimal action in the next state, and γ is the discount factor.

2. The rail transit operation plan analysis method according to claim 1, characterized in that: In step S1, the collected operational data information includes: vehicle position, vehicle speed, number of passengers, vehicle status, vehicle temperature, door switch status, train running time, train stop time, compartment internal temperature and humidity, and vehicle energy consumption; Vehicle location: Real-time acquisition of vehicle location information on the track to monitor vehicle trajectory and calculate travel distance; Vehicle speed: record the real-time speed of the vehicle to evaluate the vehicle's driving conditions and operating efficiency; Passenger count: The number of passengers is counted in real time through sensors in the car or counters on the doors to assess the crowding level and passenger load of the vehicle; Vehicle status: monitor the operating status of the vehicle, including whether the vehicle is operating normally or has any faults; Vehicle temperature: measures the temperature inside the vehicle to regulate the operation of the air conditioner; Door open / close status: records the open / close status of the doors to ensure the safety of passengers and a smooth ride; Train running time: records the running time of the train from the starting station to the terminal station, which is used to evaluate the running speed and punctuality of the train; Train stop time: records the time the train stays at each station, which is used to evaluate the train transfer efficiency and station congestion; Temperature and humidity inside the car: monitor the temperature and humidity inside the car to provide passengers with a comfortable riding environment; Vehicle energy consumption: It is used to evaluate the vehicle's energy efficiency and energy utilization by measuring the vehicle's energy consumption.

3. The rail transit operation plan analysis method according to claim 1, characterized in that: In step S2, the specific steps of data preprocessing include the following: S201, data cleaning: clean the collected data to remove duplicate, missing or erroneous data to ensure the integrity and accuracy of the data; S202, data denoising: removing noise from data through filtering technology to improve data quality and stability; S203, data normalization: normalizing the data to unify data of different ranges to the same scale; S204, data conversion: converting or transforming the data, converting the time format into a standard format, and converting the text data into a numeric type; S205, data integration: integrating data from different sensors or devices to form a complete data set; S206, data sampling: sampling the data to reduce the data volume or balance the data distribution; S207, data feature extraction: extracting useful features from the original data, including average speed, travel distance and congestion level; S208, data standardization: standardize the data to meet statistical requirements.

4. The rail transit operation plan analysis method according to claim 1, characterized in that: In step S3, the pre-processed data is statistically analyzed using big data analysis technology, which specifically includes the following steps: S301, descriptive statistical analysis: by calculating the statistical indicators of the mean, variance, maximum value and minimum value of the data, the overall characteristics of the data are described and analyzed to assist in understanding the distribution, trend and outliers of the data; S302, Correlation Analysis: By analyzing the correlation between data, find the correlation and mutual influence between data, including correlation coefficient analysis, covariance analysis and regression analysis, which are used to reveal the intrinsic relationship between data; S303, cluster analysis: divide the data into several similar groups, so that the data within the same group has high similarity, while the data between different groups has low similarity, which is used to assist in identifying potential patterns and groups in the data, thereby providing a basis for subsequent decision-making; S304, Time Series Analysis: For time series data, by analyzing the trend, periodicity and seasonality of the data, predict future development trends and change patterns. Time series analysis methods include smoothing method, exponential smoothing method and ARIMA model; S305, Visual Analysis: Display data in an intuitive and understandable way through charts, graphs and visualization techniques to assist in discovering patterns and trends in the data.

5. The rail transit operation plan analysis method according to claim 1, characterized in that: In step S4, the specific steps of constructing a rail transit operation plan analysis model based on a machine learning algorithm are as follows: S401, data preparation: organize historical rail transit operation data into a format suitable for machine learning algorithm processing, including input features and output labels; S402, feature engineering: processing and transforming input features to extract useful information. Feature engineering methods include feature selection, feature scaling or feature combination; S403, model selection: select a suitable machine learning algorithm based on the nature of the problem and the characteristics of the data, including linear regression; y=w0+w1x1+w2x2+...+wn*xn; Among them, y is the output label, x1, x2, ..., xn are the input features, and w0, w1, w2, ..., wn are the parameters of the model; S404, model training: using historical data to train the model, and optimizing the parameters of the model by minimizing the loss function, so as to enable the model to better fit the data; S405, model evaluation: use evaluation indicators to evaluate the performance of the model, including mean square error or accuracy, and use cross-validation method to evaluate the model; S406, model tuning: tuning the model according to the evaluation results, including adjusting the model's hyperparameters and adding regularization terms; S407, model application: use the trained model to predict and analyze new data to obtain relevant results of the rail transit operation plan.

6. The rail transit operation plan analysis method according to claim 1, characterized in that: In step S5, the method steps for evaluating and comparing different operation plans include the following: S501, determine evaluation indicators: according to the objectives and needs of rail transit operations, determine evaluation indicators, including operating time, passenger satisfaction or energy consumption, to quantify the effects and performance of the operation plan and provide a basis for evaluation and comparison; S502, data collection and processing: according to the evaluation indicators, relevant data is collected in real time, and pre-processed and normalized to ensure the accuracy and comparability of the data; S503, model application: using the trained model, predict and analyze the real-time data to obtain the evaluation results of each operation plan, and evaluate the performance of each plan on different indicators based on the difference between the model's predicted value and the actual data; S504, indicator evaluation and comparison: conduct a comprehensive evaluation and comparison of the evaluation results of each operation plan, use statistical methods or visualization methods to analyze the evaluation results, compare the differences and advantages and disadvantages of each plan in different indicators, and find the optimal operation plan.

7. Rail transit operation plan analysis system, characterized by: It includes data acquisition module, data preprocessing module, data analysis module, model training module, solution evaluation module and intelligent decision-making module; The data acquisition module is used to collect real-time operational data information monitored by sensor equipment deployed on rail transit vehicles and stations; The data preprocessing module is used to clean, denoise and normalize the collected data; The data analysis module is used to perform statistics and analysis on the preprocessed data and extract useful information and features; The model training module is used to build a rail transit operation plan analysis model and use historical data for training and optimization; The solution evaluation module is used to evaluate and compare different operation solutions based on real-time data and trained models; The intelligent decision-making module is used to automatically select the optimal operation plan based on the evaluation results and use the intelligent decision-making algorithm to adjust and optimize in real time.

Citation Information

Cited By

  • Intelligent rail transit signal system and automatic scheduling method thereof

    CN119659715A

  • An intelligent rail transit signal system and its automatic dispatching method

    CN119659715B

  • Linkage scheme prediction method, device and equipment for rail transit and storage medium

    CN120373578A

  • Rail transit linkage scheme prediction method, device and equipment and storage medium

    CN120373578B