CN-121990025-A - Intelligent train operation control system and construction method thereof

CN121990025ACN 121990025 ACN121990025 ACN 121990025ACN-121990025-A

Abstract

The invention discloses an intelligent train operation control system which comprises four subsystems, a driver monitoring subsystem, a man-machine co-driving subsystem, an adaptive protection subsystem and an automatic driving subsystem, wherein the four subsystems are in communication connection through the same control network and are used for monitoring the state of a driver and implementing grading safety response, the man-machine co-driving subsystem is used for providing driving assistance and converting driving experience into system parameters, the adaptive protection subsystem is used for dynamically optimizing operation strategies and protection parameters based on reinforcement learning, the automatic driving subsystem is used for realizing autonomous perception, decision and control of the environment, and the four subsystems are integrated in an existing train operation control framework. The invention realizes the dynamic adjustment of the protection parameters based on reinforcement learning, and greatly improves the operation efficiency on the premise of ensuring the safety.

Inventors

XUE FENG
JIAO WEI
YANG WEN
Ou Guoen
AN HONGFEI
DENG HAO

Assignees

卡斯柯信号有限公司

Dates

Publication Date: 20260508
Application Date: 20260330

Claims (12)

1. An intelligent train operation control system is characterized by comprising the following four subsystems which are in communication connection through the same control network: A driver monitoring subsystem configured to monitor driver status and implement a graduated safety response; The man-machine co-driving subsystem is configured to provide driving assistance and convert driving experience into system parameters; The self-adaptive protection subsystem is configured for dynamically optimizing an operation strategy and protection parameters based on reinforcement learning; The automatic driving subsystem is configured for realizing autonomous sensing, decision and control of the environment; Wherein the four subsystems are integrated in an existing train operation control architecture.
2. The intelligent train operation control system of claim 1, further comprising a layered architecture for data interaction and instruction transmission over a standardized interface, comprising: the sensing layer comprises a driver state sensor, a train state sensor, an environment sensor and a track state sensor and is used for collecting data of a driver, a train, an environment and a track; The data layer is used for cleaning, fusing and preprocessing the data; And the execution layer is used for executing the control decision.
3. The intelligent train operation control system of claim 2 wherein said driver monitoring subsystem is deployed at said decision layer and said execution layer, comprising: a state evaluation module for evaluating driver state based on the physiological and behavioral data; and the response module is used for triggering the grading response action according to the evaluation result.
4. The intelligent train operation control system according to claim 3, wherein the man-machine co-driving subsystem is deployed at the decision-making layer and the execution layer, comprising: the experience extraction module is used for converting the operation experience of the driver into an algorithm model; And the task allocation module is used for allocating the repetitive tasks to the machine and the decision-making tasks to the driver.
5. The intelligent train operation control system of claim 2, wherein the adaptive protection subsystem is deployed at the decision layer, comprising: The data fusion module is used for integrating the perception data from the state of a driver, the state of train equipment, the state of the environment and the state of the track; The strategy optimization module is used for optimizing an operation control strategy by adopting a reinforcement learning algorithm and dynamically adjusting the protection parameters according to the real-time state of the train.
6. The intelligent train operation control system according to claim 5, further comprising a digital twin module connected to the adaptive protection subsystem for constructing a train operation digital twin model, and performing full-scene simulation verification on the optimized strategy output by the reinforcement learning optimization module.
7. The intelligent train operation control system according to claim 2, wherein the automatic driving subsystem is deployed in the decision layer and the execution layer, and the automatic driving subsystem adopts a multiple safety redundancy mechanism, and comprises an environment modeling module for fusing radar, vision and GNSS data to construct a high-precision model of a train operation environment, and an autonomous decision module for realizing dynamic adjustment of a driving plan and autonomous coping of an emergency event based on a deep learning and reinforcement learning algorithm.
8. The intelligent train operation control system of claim 1, wherein the co-operating man-machine subsystem comprises an intelligent interactive interface configured to present line data, equipment status, and optimal operating advice in real time.
9. The construction method of the intelligent train operation control system is characterized by comprising the following steps: S1, constructing a driver monitoring subsystem for monitoring the state of a driver and configuring a grading response; s2, constructing a man-machine co-driving subsystem, which is used for providing driving assistance and converting driving experience into system parameters; s3, constructing a self-adaptive protection subsystem for expanding the perception data and utilizing a reinforcement learning dynamic optimization strategy; S4, constructing an automatic driving subsystem for realizing full-autonomous sensing, decision making and control; The four subsystems are integrated in the existing train operation control framework and are in communication connection through the same control network.
10. The method of claim 9, wherein S1 comprises: Monitoring the driver with visual and physiological sensors; performing state evaluation based on multi-mode data fusion; A driver status database is established for generating personalized driving advice.
11. The method of claim 9, wherein in S3, the reinforcement learning optimization strategy is used to design a multi-objective rewards function targeting security, efficiency and economy, and to perform simulation verification in a digital twin system.
12. The method of claim 11, further comprising the iterative optimization step of collecting system operational data and feeding back to the corresponding algorithm model, continuously training and optimizing the model based on the feedback data, and forming a closed loop iteration after digital twin verification.

Description

Intelligent train operation control system and construction method thereof Technical Field The invention relates to the field of rail transit, in particular to an intelligent train operation control system and a construction method thereof. Background The China train operation control system (CTCS) is used as core technical equipment for guaranteeing safe and efficient operation of high-speed railways in China, and along with the increase of the density of rail transit networks, the complexity of operation scenes and the rapid development of artificial intelligence technology, the limitation of the traditional train control system architecture based on fixed block or quasi-mobile block and depending on preset rule protection in the aspects of intellectualization, self-adaption and man-machine cooperation modes is gradually revealed, so that the development requirements of future railways are difficult to be completely met. Specifically, the prior art mainly has the following prominent problems: At present, high-speed railways with speed per hour of 250km/h and above in the global range commonly adopt a driver manual driving mode under the protection of ATP (automatic train protection). The potential safety hazards caused by human factors are difficult to completely eliminate through the existing regular protection system. The existing CTCS does not fully consider the braking performance difference, equipment abrasion and real-time transition of the line environment (such as the influence of wind, rain, snow and fog on the braking distance) of different vehicle types. The conservative fixed parameter setting causes that an excessive safe redundancy distance has to be reserved in actual operation, so that hidden loss of line passing capacity and reduction of energy efficiency are caused, and fine capacity scheduling based on a real-time state cannot be realized. Traditional train control systems rely primarily on pre-written logic rules and offline data. The method lacks of deep mining and autonomous learning capability for massive operation data, and cannot dynamically optimize a control strategy according to historical operation data and real-time working conditions. It is difficult to achieve maximization of the operation efficiency with ensuring safety. Although the industry has increasingly stringent requirements for full-automatic train driving, the existing manned driving mode is directly spanned to high-level full-automatic driving, and the technology complexity is exponentially increased, the security authentication difficulty is extremely high, and the compatibility with the existing huge stock equipment is poor. In summary, a novel train operation control method or system which can integrate artificial intelligence technology, has self-learning self-adaptive capability and supports progressive upgrade is developed, so as to solve the problems of high artificial dependence, poor adaptability, insufficient intellectualization, difficult upgrade and the like of the existing system, and become urgent demands for promoting the high-quality development of rail transit. Disclosure of Invention The invention aims to solve the problems of insufficient intellectualization, difficult upgrading and the like of the conventional train operation control system. In order to achieve the above object, the present invention provides an intelligent train operation control system, comprising the following four subsystems communicatively connected through the same control network: A driver monitoring subsystem configured to monitor driver status and implement a graduated safety response; The man-machine co-driving subsystem is configured to provide driving assistance and convert driving experience into system parameters; The self-adaptive protection subsystem is configured for dynamically optimizing an operation strategy and protection parameters based on reinforcement learning; The automatic driving subsystem is configured for realizing autonomous sensing, decision and control of the environment; Wherein the four subsystems are integrated in an existing train operation control architecture. Optionally, the system further comprises a layered architecture for data interaction and instruction transmission through a standardized interface, and the layered architecture comprises: the sensing layer comprises a driver state sensor, a train state sensor, an environment sensor and a track state sensor and is used for collecting data of a driver, a train, an environment and a track; The data layer is used for cleaning, fusing and preprocessing the data; And the execution layer is used for executing the control decision. Optionally, the driver monitoring subsystem is disposed at the decision layer and the execution layer, including: a state evaluation module for evaluating driver state based on the physiological and behavioral data; and the response module is used for triggering the grading response action according to the evalua