|
Abstract: The characteristics of extension synthetic judgment and
extension trouble diagnosis are introduced. Then a new weights
assignment algorithm is proposed to ravel out the limitation that most
of exiting weights assignment algorithms used in extension synthetic
judgment cannot work efficiently under the changing environments. The
algorithm can solve online weights assignment problem through alternant
influence between individuals and environment. A method is recommended
to define the function of fitness of individual in genetic algorithm
(GA) under the alternant manner according to the character of genetic
learning. At the same time adaptive crossover and mutation strategy for
GA is studied, and parameters for GA used in weights assignment
algorithm are presented. The algorithm has been used in tool storage
extenion trouble diagnosis and showed upper efficiency and precision.
Key words: Weights assignment Genetic algorithm Machine
learning Extension judgment
CLC No: O212
国家自然科学基金(50175103)和国家“863”计划(2002AA411110)资助项目. Received 20020828,
received in revised form 20021018
|