A Study Of The Application Of Neural Network In Elevator Group Control

The number of buildings is increasing,we lift the performance requirements for more and more high,the single elevator can no longer meet the passengers on the traffic demand,multi elevator appeared,solve the elevator between how connected effectively become the most important issue.The neural network has the good self-learning function,to continuously improve the performance of the system itself in the system through the data,data analysis and prediction is widely used,the optimization of the elevator control system plays an extremely important role.However,the neural network is easy to fall into the maximum value of the system,resulting in a large output error.In recent years,particle swarm optimization(PSO),which is gradually applied to the field of elevator control,has been used to optimize the neural network algorithm,which can avoid the disadvantage of neural network easily fall into the maximum value,and improve the performance of the system.First,paper analyzes the main components of the elevator,with a 16 storey commercial building design background,the building elevator group control system is designed,and the PLC,inverter and other core electrical equipment system design,software simulation analysis of the control algorithm,elevator control system PLC program design.Secondly,design the evaluation system of the elevator function according to the actual situation of elevator,using MATLAB neural network modeling,training and testing of elevator evaluation index,neural network has the disadvantages of slow training speed,easy to fall into local maxima and weak global search capability,using neural network self-learning function are not always available evaluation parameter.Then,the dissertation analyzed the basic principle of particle swarm optimization algorithm,particle swarm algorithm of neural network initial weights and thresholds,through MATLAB simulation analysis,neural network and particle swarm optimization can reduce the number of iterations after training and optimization of learning time,easy to find the global optimal solution and improve the shortcomings of neural network the train speed is slow,easy to fall into local maxima and weak global search capability,convergence performance advantages such as faster convergence speed and better,to achieve the expected goal.Finally,through the process of traffic flow model of the elevator,in a typical traffic mode,the application of Labwindos/CVI software to simulate the elevator group control system,to verify the feasibility of the neural network of particle swarm optimization algorithm in elevator group control and superiority,shows that the particle swarm optimization algorithm of neural network in elevator group control research significance.