Elevator safety assessment based on IGA-BPNN

GAO Zong-shuai1, XI Tao1, XU Wei-xiong1, WANG Li-jing2
(1.School of Mechanical Engineering, Tiangong University, Tianjin 300387, China; 2.School of Control and
Mechanical Engineering, Tianjin Chengjian University, Tianjin 300384, China)

Abstract: Aiming at the safety problems of construction elevators, a health evaluation index system was established based on expert investigation method, ReliefF and Pearson. The weight of each health indicator was calculated by analytic hierarchy process, and the health grade of construction elevators was divided based on triangular fuzzy number. The analytical calculations of 5 test functions with Particle Swarm Optimization (PSO), Wolf Pack Algorithm (WPA) and Improved Genetic Algorithm (IGA) were compared. It was found that IGA has higher accuracy and convergence speed. A health assessment model for construction elevators was presented to solve those problems based on improved genetic algorithm-back propagation neural network (IGA-BPNN). An adaptive cross probability and mutation probability calculation strategy was proposed by IGA, which improved the ability of GA to find the global optimal solution. IGA was used to optimize the initial weights and thresholds of the BP neural network to form the IGA-BPNN model. GA-BPNN and IGA-BPNN were used to predict and judge the health level of construction elevators. The results indicate that the IGA-BPNN algorithm has higher accuracy and precision in predicting the health level of the elevator.
Key words: construction elevators; triangular fuzzy number; health assessment index system; improved genetic algorithm（IGA）; BP neural network