摘要 |
PURPOSE: To provide an elevator group control system which can prevent impairing of allotment performance even if a traffic condition greatly fluctuates after elevator operation. CONSTITUTION: When a car call is issued, allotment is carried out by a first allotment means 12 using a neural network and by a second allotment means 13 using a conventional method, and a comparison means 15 compares the allotment results to see whether they agree. When the rate of agreement is not higher than a predetermined value, the neural network is made to relearn by a learning means 14, during which the method of allotment is changed by an allotment method changing means 17 and allotment is carried out by the second allotment means 13 using conventional method. Thus, even if the traffic condition in a building greatly fluctuates and the allotment performance of the neural network cannot respond temporarily, allotment is carried out by the conventional method, so the allotment performance can be prevented from worsening steeply. |