Rail transit intelligent operation and maintenance system

Abstract

The invention discloses a rail transit intelligent operation and maintenance system, and relates to the technical field of intelligent traffic. The system comprises a vehicle-mounted sensing unit, a central cloud service unit and a vehicle control unit, wherein the vehicle-mounted sensing unit is used for collecting vehicle driving parameters in a target area; the central cloud service unit is used for constructing a traffic prediction model and predicting vehicles in the target area according to the vehicle driving parameters and the traffic prediction model to obtain a current vehicle driving environment state; the traffic prediction model is constructed based on self-supervised learning and a Transform mechanism; the vehicle driving environment states comprise smoothness and congestion; the vehicle control unit is used for controlling the vehicle to execute a corresponding driving control method according to map navigation and the current vehicle driving environment state; the driving control method comprises straight movement control and overtaking movement control. The overtaking action control is determined according to the overtaking judgment condition. The traffic safety can be improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and in particular to an intelligent operation and maintenance system for rail transportation. Background Art

[0002] Although the current rail transit operation and maintenance system has achieved a certain level of automation and informatization, the traditional operation and maintenance methods have gradually revealed their limitations in the face of the growing traffic volume and complex and changing road environment. The traditional rail transit system mainly relies on fixed signal control systems and manual monitoring to achieve vehicle scheduling and management. This method is incapable of handling large-scale, highly dynamic traffic flows.

[0003] First, traditional rail transit systems often rely on limited and fixed sensors to obtain basic vehicle driving parameters, such as speed, position, etc. However, these parameters often cannot fully reflect the vehicle's driving status and changes in the surrounding environment. The lack of this information makes it difficult for the system to make accurate judgments and decisions.

[0004] Secondly, the existing rail transit system lacks an efficient and intelligent prediction model to analyze vehicle driving parameters in real time and predict future traffic conditions. Traditional prediction methods are often based on historical data or simple statistical models, which cannot accurately capture the dynamic characteristics and nonlinear changes of traffic flow, resulting in inaccurate prediction results and difficulty in providing strong support for vehicle control.

[0005] Finally, traditional rail transit systems usually rely on fixed signal systems and manual instructions to control the movement of vehicles, and lack the ability to dynamically adjust according to real-time traffic conditions. When faced with emergencies such as traffic congestion and accidents, the system often cannot respond effectively in time, resulting in reduced traffic efficiency and even possible safety accidents.

[0006] Therefore, in view of the shortcomings of the existing rail transit operation and maintenance system, it is particularly important to develop a rail transit intelligent operation and maintenance system that can comprehensively collect vehicle driving parameters, build an efficient prediction model, and realize intelligent vehicle control based on the prediction results. Summary of the invention

[0007] The purpose of the present invention is to provide a rail transit intelligent operation and maintenance system that can improve traffic safety.

[0008] To achieve the above object, the present invention provides the following solutions: A rail transit intelligent operation and maintenance system, comprising: The vehicle-mounted sensor unit is used to collect vehicle driving parameters in the target area; the vehicle driving parameters include vehicle speed, vehicle head angle, position information and lane conditions; the lane conditions are changes in the time distance to adjacent vehicles; The central cloud service unit is used to construct a traffic prediction model, and predict the vehicles in the target area according to the vehicle driving parameters and the traffic prediction model to obtain the current vehicle driving environment state; the traffic prediction model is constructed based on self-supervised learning and Transformer mechanism; the vehicle driving environment state includes smooth and congested; A vehicle control unit is used to control the vehicle to execute a corresponding driving control method according to map navigation and the current vehicle driving environment state; the driving control method includes straight-ahead action control and overtaking action control; the overtaking action control is determined according to overtaking judgment conditions.

[0009] Optionally, the central cloud service unit also includes a training module; The training module is used to: Acquire training data; the training data includes training parameters and corresponding state labels; Build a pre-trained model based on self-supervised learning and Transformer mechanism; The training data is input into the pre-trained model, and based on the loss function and the preset number of iterations, the loss between the model output result and the state label is minimized, and training is performed according to the gradient descent strategy, and the trained pre-trained model is determined as the traffic prediction model.

[0010] Optionally, the traffic prediction model adopts Huber loss during training.

[0011] Optionally, the vehicle control unit comprises: a straight-ahead action control submodule and an overtaking action control submodule; The straight-ahead action control submodule is used to output a straight-ahead instruction to the vehicle control system when the current vehicle driving environment state is congested; The overtaking action control submodule is used to: When the current vehicle driving environment is unobstructed, determine whether overtaking is currently required according to map navigation, and when overtaking is required, determine overtaking judgment conditions according to preset road regulations; Determining a reward function corresponding to the driving parameter according to the vehicle driving parameter, the overtaking judgment condition and a preset minimum safety distance; An optimal lane-changing and overtaking action is determined according to the vehicle driving parameters and a corresponding reward function.

[0012] Optionally, the overtaking judgment condition includes an overtaking trigger judgment condition, an overtaking lane change pre-judgment condition, an overtaking judgment condition and a return lane pre-judgment condition.

[0013] Optionally, the overtaking action control submodule further includes: an overtaking judgment condition determination submodule; The overtaking judgment condition determination submodule is used to: Extract key processes and semantic definitions from preset road regulations to obtain key conditions; Performing logic extraction on the key conditions to obtain conditional logic; the conditional logic includes atomic propositions, logical operators and temporal logic; The conditional logic is abstractly encoded to obtain a linear temporal logic formula, and the linear temporal logic formula is determined as the overtaking judgment condition.

[0014] Optionally, the traffic prediction model includes an encoder, and the output of the encoder is used as the input of a non-adaptive decoder and an adaptive decoder respectively; the non-adaptive decoder and the adaptive decoder respectively calculate the prediction loss and the adaptive loss, and back-propagate the prediction loss and the adaptive loss to the encoder to update the parameters of the encoder.

[0015] Optionally, it also includes: a remote monitoring device; the remote monitoring device is used to retrieve relevant data in real time, and display the predicted current vehicle driving environment status in the form of a seaweed ball; wherein the seaweed ball includes multiple colors, each of the colors is used to display the traffic prediction results of the corresponding interval according to the set threshold interval.

[0016] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects: The present invention discloses a rail transit intelligent operation and maintenance system, the method includes an on-board sensor unit, a central cloud service unit and a vehicle control unit; wherein the on-board sensor unit is used to collect vehicle driving parameters in a target area; the central cloud service unit is used to construct a traffic prediction model, and predict the vehicles in the target area according to the vehicle driving parameters and the traffic prediction model to obtain the current vehicle driving environment state; the traffic prediction model is constructed based on self-supervised learning and Transformer mechanism; the vehicle driving environment state includes unblocked and congested; the vehicle control unit is used to control the vehicle to execute a corresponding driving control method according to map navigation and the current vehicle driving environment state; the driving control method includes straight action control and overtaking action control; the overtaking action control is determined according to the overtaking judgment condition. The present invention can improve traffic safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0018] Figure 1It is a structural schematic diagram of the rail transit intelligent operation and maintenance system of the present invention. DETAILED DESCRIPTION

[0019] 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.

[0020] The purpose of the present invention is to provide a rail transit intelligent operation and maintenance system that can improve traffic safety.

[0021] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] like Figure 1 As shown, the present invention provides a rail transit intelligent operation and maintenance system, comprising: The vehicle-mounted sensor unit is used to collect vehicle driving parameters in the target area; the vehicle driving parameters include vehicle speed, vehicle head angle, position information and lane conditions; the lane conditions are the time distance changes with adjacent vehicles.

[0023] The central cloud service unit is used to build a traffic prediction model, and predict the vehicles in the target area according to the vehicle driving parameters and the traffic prediction model to obtain the current vehicle driving environment state; the traffic prediction model is built based on self-supervised learning and Transformer mechanism; the vehicle driving environment state includes unblocked and congested.

[0024] A vehicle control unit is used to control the vehicle to execute a corresponding driving control method according to map navigation and the current vehicle driving environment state; the driving control method includes straight-ahead action control and overtaking action control; the overtaking action control is determined according to overtaking judgment conditions.

[0025] As a specific implementation, the central cloud service unit also includes a training module; the training module is used to: Acquire training data; the training data includes training parameters and corresponding state labels; construct a pre-training model based on self-supervised learning and Transformer mechanism; input the training data into the pre-training model, based on the loss function and the preset number of iterations, with the goal of minimizing the loss between the model output result and the state label, and train according to the gradient descent strategy, and determine the trained pre-training model as the traffic prediction model. Wherein, the traffic prediction model adopts Huber loss during training.

[0026] As a specific implementation, the vehicle control unit includes: a straight-ahead action control submodule and an overtaking action control submodule.

[0027] The straight-ahead action control submodule is used to output a straight-ahead instruction to the vehicle control system when the current vehicle driving environment state is congested.

[0028] The overtaking action control submodule is used to: When the current vehicle driving environment is unobstructed, determine whether overtaking is currently required based on map navigation. When overtaking is required, determine the overtaking judgment condition based on preset road regulations; determine the reward function corresponding to the driving parameter based on the vehicle driving parameters, the overtaking judgment condition and the preset minimum safety distance; determine the optimal lane change overtaking action based on the vehicle driving parameters and the corresponding reward function. The overtaking judgment condition includes overtaking trigger judgment condition, overtaking lane change pre-judgment condition, overtaking judgment condition and return lane pre-judgment condition.

[0029] The overtaking action control submodule further includes: an overtaking judgment condition determination submodule; the overtaking judgment condition determination submodule is used to: Extract key processes and semantic definitions of preset road regulations to obtain key conditions; perform logic extraction on the key conditions to obtain conditional logic; the conditional logic includes atomic propositions, logical operators and temporal logic; abstractly encode the conditional logic to obtain a linear temporal logic formula, and determine the linear temporal logic formula as the overtaking judgment condition.

[0030] As a specific implementation, the traffic prediction model includes an encoder, and the output of the encoder is used as the input of a non-adaptive decoder and an adaptive decoder respectively; the non-adaptive decoder and the adaptive decoder respectively calculate the prediction loss and the adaptive loss, and back-propagate the prediction loss and the adaptive loss to the encoder to update the parameters of the encoder.

[0031] As a specific implementation, the system in this embodiment also includes: a remote monitoring device; the remote monitoring device is used to retrieve relevant data in real time, and display the predicted current vehicle driving environment state in the form of a seaweed ball; wherein the seaweed ball includes multiple colors, each of the colors is used to display the traffic prediction results of the corresponding interval according to the set threshold interval.

[0032] Based on the above technical solution, the following embodiments are provided.

[0033] This embodiment provides a specific embodiment of a rail transit intelligent operation and maintenance system, which integrates advanced vehicle-mounted sensor technology, cloud computing and artificial intelligence algorithms to achieve comprehensive monitoring and intelligent scheduling of the driving status of rail transit vehicles. The following is a detailed description of the various components of the system and its workflow.

[0034] 1. System Architecture The system is mainly composed of four modules: vehicle-mounted sensor unit, central cloud service unit, vehicle control unit and remote monitoring equipment.

[0035] 2. Vehicle-mounted sensor unit The on-board sensor unit is deployed on each rail transit vehicle and is responsible for real-time collection of vehicle driving parameters, including but not limited to vehicle speed, vehicle head angle, precise location information, and lane conditions (especially the time difference with adjacent vehicles). These data are transmitted to the central cloud service unit in real time via wireless network.

[0036] 3. Central Cloud Service Unit Traffic prediction model construction and training: Training module: First, extract training parameters and their corresponding state labels (such as unblocked, congested) from historical data. Then, build a pre-trained model based on self-supervised learning and Transformer mechanism. Use Huber loss function as the optimization target, and train through multiple iterations until the loss between the model output result and the state label is minimized, and finally determine the traffic prediction model.

[0037] Model structure: The traffic prediction model consists of an encoder, whose output is input to a non-adaptive decoder and an adaptive decoder, which calculate the prediction loss and the adaptive loss respectively, and back-propagate these losses to the encoder to update the encoder parameters and improve the model prediction accuracy.

[0038] Real-time prediction The central cloud service unit receives the vehicle driving parameters transmitted by the on-board sensor unit, uses the trained traffic prediction model to make real-time predictions on the vehicles in the target area, determines the current vehicle driving environment status (smooth or congested), and sends the prediction results to the vehicle control unit.

[0039] 4. Vehicle Control Unit Straight motion control submodule: When receiving the "congested" state prediction result sent by the central cloud service unit, the straight-ahead action control submodule outputs a straight-ahead instruction to the vehicle's control system to ensure that the vehicle maintains safe and stable driving in congested conditions.

[0040] Overtaking action control submodule: Determination of overtaking judgment conditions: The overtaking action control submodule first uses the overtaking judgment condition determination submodule to extract the key processes and semantic definitions of the preset road regulations and convert them into linear temporal logic formulas as overtaking judgment conditions.

[0041] Overtaking decision and execution: In the "unblocked" state, the map navigation is used to determine whether overtaking is necessary. If overtaking is necessary, the reward function is calculated based on the preset road regulations, vehicle driving parameters (such as speed and position), overtaking judgment conditions (including overtaking trigger judgment conditions, overtaking lane change judgment conditions, overtaking judgment conditions and return lane judgment conditions) and the preset minimum safety distance to determine the optimal lane change and overtaking action.

[0042] 5. Remote monitoring equipment The remote monitoring device retrieves the relevant data from the central cloud service unit in real time and intuitively displays the current vehicle driving environment status in the form of seaweed balls. The seaweed balls are composed of multiple colors, each color represents the traffic prediction results of different set threshold intervals, which makes it easier for operation and maintenance personnel to quickly identify and respond to potential changes in traffic conditions.

[0043] 6. Workflow The on-board sensor unit collects vehicle driving parameters and uploads them to the central cloud service unit.

[0044] The central cloud service unit uses the traffic prediction model to make real-time predictions and determine the vehicle driving environment status.

[0045] The vehicle control unit executes the corresponding driving control method (going straight or overtaking) based on the prediction results and map navigation information.

[0046] Remote monitoring equipment displays traffic forecast results in real time, assisting operation and maintenance personnel in monitoring and management.

[0047] Through the above embodiments, this embodiment provides an efficient and intelligent rail transit intelligent operation and maintenance system, which effectively improves the safety and operation efficiency of rail transit.

[0048] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0049] This article uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only used to help understand the core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A rail transit intelligent operation and maintenance system, characterized in that: include: The vehicle-mounted sensor unit is used to collect vehicle driving parameters in the target area; The vehicle driving parameters include vehicle speed, vehicle head angle, position information and lane conditions; the lane conditions are changes in the time distance to adjacent vehicles; The central cloud service unit is used to construct a traffic prediction model, and predict the vehicles in the target area according to the vehicle driving parameters and the traffic prediction model to obtain the current vehicle driving environment state; the traffic prediction model is constructed based on self-supervised learning and Transformer mechanism; the vehicle driving environment state includes smooth and congested; A vehicle control unit is used to control the vehicle to execute a corresponding driving control method according to map navigation and the current vehicle driving environment state; the driving control method includes straight-ahead action control and overtaking action control; the overtaking action control is determined according to overtaking judgment conditions.

2. The rail transit intelligent operation and maintenance system according to claim 1 is characterized in that: The central cloud service unit also includes a training module; The training module is used to: Acquire training data; the training data includes training parameters and corresponding state labels; Build a pre-trained model based on self-supervised learning and Transformer mechanism; The training data is input into the pre-trained model, and based on the loss function and the preset number of iterations, the loss between the model output result and the state label is minimized, and training is performed according to the gradient descent strategy, and the trained pre-trained model is determined as the traffic prediction model.

3. The rail transit intelligent operation and maintenance system according to claim 2 is characterized in that: The traffic prediction model adopts Huber loss during training.

4. The rail transit intelligent operation and maintenance system according to claim 1, characterized in that: The vehicle control unit comprises: a straight-ahead action control submodule and an overtaking action control submodule; The straight-ahead action control submodule is used to output a straight-ahead instruction to the vehicle control system when the current vehicle driving environment state is congested; The overtaking action control submodule is used to: When the current vehicle driving environment is unobstructed, determine whether overtaking is currently required according to map navigation, and when overtaking is required, determine overtaking judgment conditions according to preset road regulations; Determining a reward function corresponding to the driving parameter according to the vehicle driving parameter, the overtaking judgment condition and a preset minimum safety distance; An optimal lane-changing and overtaking action is determined according to the vehicle driving parameters and a corresponding reward function.

5. The rail transit intelligent operation and maintenance system according to claim 4 is characterized in that: The overtaking judgment conditions include overtaking trigger judgment conditions, overtaking lane change pre-judgment conditions, overtaking judgment conditions and return lane pre-judgment conditions.

6. The rail transit intelligent operation and maintenance system according to claim 4, characterized in that: The overtaking action control submodule further includes: an overtaking judgment condition determination submodule; The overtaking judgment condition determination submodule is used to: Extract key processes and semantic definitions from preset road regulations to obtain key conditions; Performing logic extraction on the key conditions to obtain conditional logic; the conditional logic includes atomic propositions, logical operators and temporal logic; The conditional logic is abstractly encoded to obtain a linear temporal logic formula, and the linear temporal logic formula is determined as the overtaking judgment condition.

7. The rail transit intelligent operation and maintenance system according to claim 1, characterized in that: The traffic prediction model includes an encoder, the output of the encoder is used as the input of a non-adaptive decoder and an adaptive decoder respectively; the non-adaptive decoder and the adaptive decoder respectively calculate the prediction loss and the adaptive loss, and back-propagate the prediction loss and the adaptive loss to the encoder to update the parameters of the encoder.

8. The rail transit intelligent operation and maintenance system according to claim 1, characterized in that: It also includes: a remote monitoring device; the remote monitoring device is used to retrieve relevant data in real time, and display the predicted current vehicle driving environment status in the form of a seaweed ball; wherein the seaweed ball includes multiple colors, each of the colors is used to display the traffic prediction results of the corresponding interval according to the set threshold interval.

Images