CAJ | 학술논문

为有效控制离散非线性系统，使系统控制策略能够应对状态域的所有初始状态，在近似动态规划方法的基础上，提出一个未固定初始状态的带ε误差限的离散非线性系统优化控制算法。研究初始状态对离散系统控制策略的影响，确定在初始状态域边界上寻找最优初始点的方法。所求控制策略使初始状态域的所有性能指标函数在最大迭代步数内收敛，使性能指标与最优性能指标保持在精度ε内。为了易于实现算法，使用神经网络来近似性能指标函数和最优控制策略。结合实例，对该算法进行仿真分析，分析结果表明了算法的有效性。
위유효공제리산비선성계통，사계통공제책략능구응대상태역적소유초시상태，재근사동태규화방법적기출상，제출일개미고정초시상태적대ε오차한적리산비선성계통우화공제산법。연구초시상태대리산계통공제책략적영향，학정재초시상태역변계상심조최우초시점적방법。소구공제책략사초시상태역적소유성능지표함수재최대질대보수내수렴，사성능지표여최우성능지표보지재정도ε내。위료역우실현산법，사용신경망락래근사성능지표함수화최우공제책략。결합실례，대해산법진행방진분석，분석결과표명료산법적유효성。
To control of discrete nonlinear systems effectively ,and make the system control strategy cope with all the initial states in the state domain ,a newε-optimal control algorithm with unfixed initial state based on the adaptive dynamic programming ap-proach was proposed .The study of the initial state’s effects on the discrete system control strategy was done to find the optimal initial point at the initial state domain boundaries .This algorithm made performance index function in the initial state domain iteratively converge within an errorεin finite time .For facilitating the implementation of theεoptimal control algorithm ,neural networks were used to approximate the performance index function and compute the optimal control policy respectively .Simula-tion analysis demonstrates the effectiveness of the algorithm .