基于数据驱动的高速列车广义预测控制器的性能监控

Performance monitoring of generalized predictive controller for high-speed train based on data-driven

针对目前ATO成熟的广义预测控制器进行性能监控的研究,使用数据驱动的方法,从列车运行现场产生的大量输入输出数据着手,利用PCA统计方法降维,分析该控制器的性能基准,使用支持向量机对4种导致控制器性能下降的因素进行分类仿真,结果表明:分类器可以实现性能诊断,分类效果理想,误判和漏报的情况很少发生。

It is aimed to research the performance monitoring of GPC which is widely accepted in ATO system,based on data-driven method,using the input and output data from the scene.Then,using the statistical method PCA to realize the dimension reduction can obtain the performance benchmark.Furthermore,using support vector machine(SVM)for four factors led to the decrease of the controller performance to simulate,the classification results show that the classifier can achieve performance diagnosis.More importantly,classified effect is ideal,mistakes and omission rarely happen.