TECHNIQUES FOR EVALUATING AND IMPROVING THE OPERATIONAL EFFICIENCY OF POSITIVE TRAIN CONTROL BRAKING ENFORCEMENT ALGORITHMS

摘要

Software algorithms are used in Positive Train Control (PTC) systems to predict train stopping distance and to enforce a penalty brake application. These algorithms have been shown to be overly conservative, leading to operational inefficiencies by interfering with normal train operations. A braking enforcement algorithm that can safely stop trains to prevent authority and speed limit violations without impacting existing railroad operations is critical to successful widespread implementation of PTC. Due to operational issues observed with early PTC braking enforcement algorithms, a number of techniques are proposed and evaluated to improve the operational efficiency of these algorithms, with emphasis on applicability to PTC systems currently being implemented. Transportation Technology Center, Inc. (TTCI) is employing a new methodology for evaluation of braking algorithms that uses Monte Carlo simulation techniques to statistically evaluate the performance of the algorithm, with limited need for field testing to verify the simulation results. In the Monte Carlo process, computer simulations are run repeatedly using randomly selected input values to predict the resulting probability distribution of stopping locations. The method provides a higher level of confidence in algorithm performance with reduced time and cost compared to traditional methods, which rely heavily on field testing. For freight trains, the method utilizes a detailed train dynamics simulation model previously developed and validated by the Association of American Railroads (AAR). For passenger trains, TTCI is developing and validating a new model capable of simulating brake systems and components specific to passenger and commuter equipment. New methods for addressing operational efficiency of braking algorithms focus on improving the accuracy of stopping distance prediction and reducing the potential variation from the prediction. Techniques investigated by TTCI include adaptive functions, which measure train braking performance en route and adapt the algorithm to these characteristics; emergency brake backup, which uses feedback following a penalty application to determine if additional emergency braking is required to stop the train short of the target; an improved target offset function, which relies on statistical multi-variable regression of thousands of stopping distance simulations; and including information about dynamic braking effort in the stopping distance prediction. Results from TTCI's investigations show potential to reduce the operational impact, by demonstrating the probability of stopping excessively short of the target is significantly less than that of previous algorithms. The techniques are already being adopted by PTC onboard suppliers for the largest North American railroads, and many are applicable to railways worldwide.