模型参数不确定的连续系统自学习定性控制

Self Learning Qualitative Control for Continuous System with Uncertain Model Parameters

对模型中含有不确定性因素的非线性系统给出了一种定性控制方法，避免了模糊控制器设计中存在的主观性和模糊规则难以获取等缺点．先将状态进行定性划分，再对每个定性状态根据期望约束利用不精确模型求出控制的参考范围及初值，最后对控制量在线自学习调整．此方法具有模糊控制的控制效果．

A qualitative control method is proposed for a nonlinear system with uncertain factor. The approach circumvents the disadvantages in the design of fuzzy controller, such as subjectivity, difficulty to obtain fuzzy rules,etc. Qualitative partitions for system states are first made. Reference range and initial values of control are then obtained to satisfy the expected constraint conditions for each qualitative state by approximate models. The control values can thus be self adapted and adjusted on line. This is equivalent to fuzzy control.