Power Demand Modeling And Adaptive Control Strategy For Tracked Vehicles

The driver's throttle and brake pedal commands are always interpreted as the power demand satisfied by the vehicular powertrain. The power demand is a stochastic variable in real time, and thus it can be modeled by a discrete-time stochastic dynamic process named Markov chain. After investigating the development of Markov chain in vehicles' application, research on Markov chain for tracked vehicles is rare. The main content of this thesis is shown as below.(1) Multi-dimensional Markov chain-based driver model for tracked vehicles.Because the steering principle in tracked vehicle is different from that in wheeled vehicles, it is necessary to consider the heading power and steering power simultaneously in tracked vehicles. A stochastic driver model incorporating the heading and steering motion is established in this paper. The three-dimensional Markov chain-based driver model consists of the angular velocity, average heading velocity, and power demand.(2) Online updating algorithm for the transition probability matrix and state prediction algorithm for velocity and power demand.An online updating algorithm based on the nearest-neighborhood method and offline estimation process is derived for renewing the transition probability matrix. This algorithm has the flexibility to weigh the past and current information to update the transition probability matrix through the forgetting factor and the effective memory depth.Fuzzy encoding is proposed to extend the online updating algorithm on the basis of membership function and fuzzy set theory. Then, the state prediction algorithm using the updated transition probability matrix from the fuzzy encoding method and nearest-neighborhood method is deduced in detail. Finally, the simulation results indicate that the prediction accuracy of fuzzy encoding method is better than that of the nearest-neighborhood method.(3) Adaptive control strategy for a hybrid electric tracked vehicle through stochastic dynamic programming.Based on the proposed Markov chain-based driver model and online updating algorithm, an adaptive control strategy is developed for a series hybrid electric tracked vehicle through stochastic dynamic programming. Simulation results illuminate that better fuel economy is achieved with the help of the updated Markov chain-based driver model. Subsequently the adaptability of the online updating algorithm is validated for different driving schedules.