Train Stopping Control via Quantized Braking and Feasible Region
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional Train Automatic Stopping Control (TASC) systems face challenges in accurately stopping trains at predetermined positions due to uncertainties in train parameters and environmental conditions, leading to potential overrun or underrun, as they rely on infinite velocity profiles that are difficult to generate and prone to errors.
Innovation Solution
The system employs a finite set of braking system actions to maintain the train within a feasible region defined by stopping constraints, using mixed integer optimization and convex optimization to select control actions that ensure precise stopping, even with variations in train parameters and quantization errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional TASC systems use infinite velocity profiles for train stopping control, then the train can theoretically achieve smooth deceleration, but the system becomes computationally complex and prone to errors due to parameter uncertainties
Solution Approach 1:
The patent transforms the continuous velocity profile parameters into a finite set of discrete control actions. Instead of computing continuous velocity functions, the system uses a finite set of braking force levels (e.g., 0%, 25%, 50%, 75%, 100% of maximum braking force) to control train deceleration. This discretization simplifies computation and improves reliability by eliminating the complexity of generating and tracking infinite velocity profiles while maintaining effective stopping control.
2Adaptability or versatility
If the system pre-generates multiple reference velocity profiles based on different assumptions, then it can prepare for various conditions, but the selection process becomes time-consuming and resource-intensive
Solution Approach 1:
The patent implements dynamic reevaluation of the optimal control sequence at each time step based on current train state (position, velocity) and environmental conditions (rail friction, grade). Instead of statically selecting from pre-generated profiles, the system continuously computes the optimal finite control sequence using model predictive control, adapting to changing conditions in real-time. This eliminates the time-consuming profile selection process while maintaining versatility across different rail conditions.
3Reliability
If the system uses continuous braking force control, then smooth train deceleration can be achieved, but the braking system experiences chatter and wear due to frequent adjustments
Solution Approach 1:
The patent segments the continuous braking force range into a finite set of discrete braking levels. The braking system operates at these discrete levels rather than continuously adjusting force, which eliminates chatter caused by frequent small adjustments. The finite control sequence planning ensures smooth transitions between discrete levels while maintaining stopping precision through optimized control sequences that account for train dynamics and external disturbances.
4Measurement precision
If the system accounts for uncertainty in train parameters and environmental conditions, then more accurate stopping can be achieved, but the computational burden increases significantly
Solution Approach 1:
The patent performs preliminary computation of the finite control sequence using model predictive control, which optimizes the entire sequence of braking actions in advance based on current state estimates and predicted disturbances. By planning the complete control sequence beforehand and then executing it with minimal real-time adjustments, the system achieves accurate stopping despite parameter uncertainties without requiring complex continuous computation during execution. This preliminary planning approach reduces real-time computational burden while maintaining precision.
Data Source
AI summary
Methods and systems for controlling a train movement to a stop at a stopping position between a first position and a second position. Determining constraints of a velocity of the train with respect to a train position forming a feasible region (FR) for a state of the train during the movement, such that a lower curve bounding the FR has a zero velocity only at the first position, and an upper curve bounding the FR has a zero velocity only at the second position. Determining a control invariant subset (CIS) of the FR, wherein for each state within the CIS there is at least one control action having a value selected from a finite set of values that maintains the state of the train within the CIS. Controlling train movement subject to constraints by selecting a control action maintaining the state of the train within the CIS of the FR.