Speed Tracking Control Using An ANFIS Model For High-speed Electric Multiple Unit

The high-speed electric multiple unit (EMU) is a nonlinear dynamical system composedof complicated equipment. It operates in complicated environments with various time-spacedistribution features. The traditional approach to operating the high-speed EMU is based uponactual operation conditions and a given V-S curve (velocity versus position curve). It cannotguarantee the high standards of operational performance requirements.To improve the performance of high-speed EMU, an ANFIS model is developed usingEMU running data and traction/braking characteristics. A model predictive controller isdesigned based on the model for speed tracking control of the high-speed EMU. Specificstudies are as follows:1. The study takes CRH380AL as the research objective. Based on the actual operationconditions and running curve of CRH380AL, a data-driven adaptive neurofuzzy inferencesystem (ANFIS) is exploited to model the running process. We first determine the optimalnumber of fuzzy rules by using the subtractive clustering algorithm. Second, we employ aleast square method and back propagation learning algorithm to optimize the parameters toobtain the optimal ANFIS model of EMU running process. The experiment results show thehigh precision of proposed ANFIS model.2. Based on the ANFIS model, we propose a generalized predictive control algorithm toensure the safe, punctual, comfortable and efficient operation of high-speed EMU. For thecase of normal, emergency brake and parameters variation, we present a comparativeexperiment with the PID controller and the adaptive robust controller using the data obtainedfrom the CRH380AL. The experiment results show that the proposed speed trackingcontroller has good robustness, tracking property and control effect.3. A high-speed EMU is a train consisting of multiple trailer and motor cars. To capturethe motion characteristics of the EMU, this study proposes a multi-ANFIS model using theactual running data and traction/braking characteristics. The experiment results on the actualCRH380A show the accuracy and precision of the presented multi-ANFIS model.4. A model predictive controller is designed based on the multi-ANFIS model forsynchronous tracking control of each car body for harmonic running. Experiments have beencarried out on the real CRH380A running data showing significant improvements forhigh-speed EMU operations.