摘要 |
PURPOSE:To perform the optimum allocation control by forming multiple relational models between evaluation index weighted values and group management control response results, relating the relational models with the traffic demand, and obtaining the demand state peculiar to a building based on the actual control response result via on-line learning. CONSTITUTION:The inference result evaluation section 24 of a learning controller 1-1 calculates the traffic demand for every fixed period in each time zone and outputs it to a inference section 21, and the weighted value for a partial system model is calculated. A combination of control parameters changed with control parameters alpha of evaluation indexes in the preset range are inputted to a partial model section 22, and group management control response inference results for the combination of control parameters are outputted based on the above weighted values and the associative function by a synthetic section 23. An evaluation section 24 evaluates the inference results based on the evaluation criteria judged to be optimum for each building where an elevator system is installed, extracts the optimum inference result, and selects the evaluation index weighted value corresponding to it as the optimum control parameter. |