RNN Neural Network Intelligent Control Of Chaotic Motion Of A Class Of Impact Vibration System With Clearance

Vibro-impact systems with clearance are ubiquitous in engineering systems,and the erratic motion produced by such systems will reduce its operation accuracy,cause fatigue damage to parts and components,and directly impact its service life.Dynamic analysis and chaos control on a vibro-impact system with gaps to mitigate its negative effects on engineering systems is therefore of great research importance and engineering application value in light of actual engineering requirements.The chaotic motion of a rigid vibro-impact system with gaps is controlled in this paper using an intelligent optimization strategy based on ridgelet neural network(abbreviated as RNN).The primary task is the following:First,a dynamic model of a rigid vibro-impact system with two gaps and a single degree of freedom is developed.By establishing Poincaré mapping and calculating the system’s Lyapunov exponent,the dynamic behaviors of the system,such as bifurcation,chaotic motion,and periodic motion,are studied qualitatively.Using the maximal Lyapunov exponent of the controlled system,the energy consumed by the control system,and the control rate as weighted performance indexes,the respective performance index functions for two distinct control objectives are determined.Second,On the basis of an RNN neural network with three layers and five ridges,a chaotic controller is built,the basic parameters of the controller are optimized using the IFSAPSO algorithm,and an IFSAPSO-RNN neural network chaotic control strategy is proposed.To address the issue that the standard particle swarm optimization algorithm(abbreviated as PSO)is prone to local optimum,an improved fractional-order simulated annealing particle swarm optimization algorithm(abbreviated as IFSAPSO)is developed by combining hyperbolic tangent nonlinear inertia weight,linear gradual learning factor,and fractional simulated annealing mechanism.The algorithm’s performance is evaluated using six selected benchmark test functions and the Lorenz chaotic system identifiability function.Finally,the IFSAPSO-RNN neural network chaotic controller is then used to control the chaotic motion of a single-degree-of-freedom rigid impact vibration system with intervals in a fixed period and fixed state,and the energy consumption and control rate of the controller are evaluated quantitatively.The simulation results demonstrate that under the control objectives of fixed period and fixed state,the control strategy can realize chaotic control of such systems seamlessly,rapidly,and with reduced energy consumption.