Control Traffic Flow Of Modern Elevator Group Based On Genetic Algorithm

With the development of modern city, more and more skyscraper is built in the city today. It is indispensable to transport passengers with elevator group composed of several lifts. Accordingly, the controlling algorithm of elevator group has already become the focus of research at home and abroad. For a complicated elevator group control system (EGCS), because of many kinds of demands, in this paper many integrated intelligent allocation elevator algorithms are adopted to analyze traffic flow and proceed to pattern-recognition optimally. Consequently, a variety of algorithms of allocating elevator are adopted according to different patterns. To many disadvantages of current methods, this paper presents an algorithm of allocating elevator based on GA. The approach increases running efficiency of elevator group, and reduces energy-consuming. Simulation experiment validates the performance of the method.Firstly, we reviewed the development and current status of EGCS in domestic and international, and introduced advantages and shortcomings of a variety of EGCS. We also established evaluation function of multi-object for EGCS, which is based on evaluating four performance indexes and analyzing uncertain factors in the course of allocating elevator.Secondly, we proposed two ways of traffic flow pattern-recognition. â‘ Normal fuzzy neural network is applied to recognize the traffic pattern and genetic algorithm which is used to improve parameters of membership functions. â‘¡ New method based on immune evolutionary k-means clustering algorithm is presented to identify traffic pattern. Experimental results illustrate that the two methods are capable of recognizing traffic patterns. However, the second method is superior to the first one.Thirdly, a control algorithm of elevator group system with inter-floor traffic pattern based on hybridized genetic algorithm is proposed in this paper. In this algorithm, we search an optimal allocating method based on the goal function of the minimal average waiting time of the passengers. We established the adaptive function according to the running state of elevators and the hall-call signal among layers. At the early stage of the evolution, we use standard genetic algorithm to search globalsolution in solution space. When the population is convergent around the optimal result, we can put the adaptive orthogonal local search into the algorithm to improve the local search ability. Simulation results demonstrated that the method conquers preferably the disadvantage of the standard genetic algorithm that has slow convergence speed at a later time, thereby it meets the real-time control of EGCS.Finally, a virtual environment is established to simulate actual EGCS and validate performance indexes of the methods presented in this paper.