摘要 |
PROBLEM TO BE SOLVED: To provide a method and a device for controlling a physical system that grows gradually as the size of an operation range and complication of functions to be learned are increased. SOLUTION: This method contains a step 100 for defining the initial structure having a functional approximation, a step 102 for receiving the condition data expressing the real condition of a physical system, a step 104 for renewing the functional approximation by renewing a learning parameter of each of local models to separately minimize the predetermined erroneous index per each of the local models on the basis of a tracking error and an approximation error, a step 106 for computing a control variable on the basis of the control regulation of a control system by using the renewed functional approximation, a step 108 for outputting the computed control variable to the physical system, and a step for repeating the step 102 for receiving, the step 104 for renewing, the step 106 for computing and the step 108 for outputting. COPYRIGHT: (C)2004,JPO
|