| 摘要 |
<P>PROBLEM TO BE SOLVED: To provide an elevator system for creating a neural network which can execute optimum car assigning operation without manually creating an ideal fuzzy rule. <P>SOLUTION: A group managing control device 1 for the elevator system comprises a neural net car assignment part 12 for executing car assigning operation, an empirical assignment elevator simulation part 14 for creating one or more learning sample data in accordance with traffic information data and a preset car assigning rule, and an optimizing processing part 15. The optimizing processing part 15 uses one or more learning sample data for learning one or more neural networks corresponding to the construction of an elevator group to create one neural network, and then it searches an optimum value for the weighting factor of the neural network in accordance with evolutionary calculation and optimizes the weighting factor of the car assignment part 12 in accordance with its searching result. <P>COPYRIGHT: (C)2005,JPO&NCIPI |
