Control of Flight Simulator with PIDNN
Flight Simulator is based on the six-degree freedom of Stewart platform and its hydraulic actuator is nonlinear. Flight Simulator is proposed for high performance with good stability robustness, stability, quick response. Although Forward Neural Network (FNN) has a good ability for mimicking any nonlinearity, the static Neural Network do not have strong calculate ability because lack of dynamic gene. Integral and differential gene can deal with dynamic information, therefore, combining Neural Network with integral and differential gene (PIDNN) has a good ability of robustness and can also determine the configuration (commonly the number of neuron in Neural Network hidden layer is hard to be determined). The weight between the layers can be initialized, which has the following advantages: (1) make the speed of net convergence faster, (2) improve the probability of dapping local small apices, and (3)assure the system stability during initialization and the operation process.
Hydraulic actuator Stewart platform Forward net work Neural Network PID Static state nonlinear
Yang Zhiyu Gao Yingjie Gu Yanpeng Li Huimin Dang Qi
Institute of Electrical Engineering, Yanshan University, Qtnhuangdao, Hebei, 066004, China
国际会议
北戴河
英文
852-854
2007-06-06（万方平台首次上网日期，不代表论文的发表时间）