Analysis And Prediction Of Sliding Friction Chaotic Characteristics Using Improved Firefly Algorithm

With the rapid development of electric locomotive,people are in need of a more stable pantograph catenary system,while the sliding friction is regarded as an important factor affecting and determining the stability of pantograph catenary system.Therefore,the research on sliding friction of pantograph catenary has strong theoretical and practical significance.This paper will analyze the chaotic characteristic and prediction of the sliding friction of pantograph catenary.Firstly,this paper constructs basic working conditions for chaotic characteristic analysis and prediction,and based on this,changes the contact current,sliding speed and normal bias load respectively.A total of four different working conditions are used for prediction and analysis.The friction data under four different working conditions are obtained by using the self-made sliding electric-contacting machine,and the sliding friction data under the basic working conditions are analyzed.The friction sample data are analyzed by chaos theory,and the delay time and embedding dimension of the friction time sample are calculated by autocorrelation method and G-P algorithm,so as to reconstruct the phase space of the friction time sample.The chaotic characteristics of the friction sequence are verified by PCA component analysis and calculating the maximum Lyapunov exponent,which proves that the sliding friction of pantograph catenary electric-contacting has chaotic characteristics,and chaos theory can be applied to the study of sliding friction of pantograph catenary electrical contact.Then,the basic firefly algorithm is improved by using Sinusoidal mapping to optimize the initialization population,introducing mobile search scheme for the best advantage,and introducing inertia weight.Through the comparison of experimental evaluation indexes and the analysis of convergence curve,it is verified that the performance of the improved firefly algorithm is better than the other four algorithms,so as to realize the universal popularization of the updated firefly algorithm.Finally,RBF chaotic prediction model of neural network,Volterra chaotic prediction model of series and the integration of both models are established respectively.The integrated model is optimized by the improved firefly algorithm.The weight of the prediction model corresponds to the reciprocal of the error of a single algorithm,which is obtained by the improved firefly optimization.Three chaotic prediction models are used to predict the friction time series data under four different working conditions,and effects are compared from different aspects.The results show that integrated prediction model has better prediction effects.The average value of root mean square error under four working conditions is 0.0935 and that of average absolute error is 0.0751.The prediction accuracy under four working conditions is improved by 42% on average,which is better than a single algorithm.It can be applied to the prediction of sliding friction of pantograph catenary electrical contact under different working conditions.The paper has 31 figures,9 tables and 73 references.