A Gear-box Fault Diagnosis Method Based On Shuffled Frog Leaping Algorithm To Optimize The BP Neural Network

With development of Chinese equipment manufacturing industry,the manufacturing industry will enter an intelligent era by "Made in China 2025",such as “China Eye”,“The tip of the crown”,“The Jiaolong” and “Shenzhou” spacecraft.At the same time,the development and progression of mechanical fault diagnosis technology is also undergoing profound changes.Gearbox is an important part of mechanical equipment.Its operating conditions directly affect the safety and stability of the equipment system,involving in economic losses even casualties once in disorder.Accordingly,it is of great importance to monitor the condition of gearbox for the safe operation.The hybrid frog leaping algorithm(SFLA)is a population-based collaborative search optimization method simulating natural biological behavior.The individuals of the frogs are sorted and divided into several mold groups,and the evolution of the memes is carried out.The global information exchange is realized by the mixing of the mold groups.Therefore,SFLA combines the advantages of the algorithm with the local search function and the particle swarm optimization algorithm simulating the behavior of the birds,resulting in easy implementation,high precision,fast convergence,large coverage area and favorable global selection.The purpose of this study is to explore the diagnosis of gear-box fault with JZQ250 transfer-box.The model of shuffled frog leaping algorithm(SFLA)combined with back propagation(BP)neural network was created in the study,based on the efficient global optimization efficacy of SFLA on optimizing the structure of BP neural network.Compared to the BP neural network,this model has more accurate diagnostic capability and higher convergence rate with shorter training time,higher training accuracy and less local optimum.Through a series of training and testing,the results demonstrate that this model could improve the stability and accuracy of diagnosis.