Study On Finite State Feedback Robust Control Of Railway Steel Bridge Based On PSO-BP Algorithm

The structure of railway steel bridge is an important facility in the railway transportation,and it plays an irreplaceable role in relieving the increasingly severe transportation situation.However,under various adverse excitations,the vibration and fatigue life of railway steel bridges have become more and more serious,and thus affect their safety and durability.Therefore,it is necessary to adopt the active control of mature theory and good damping effect to restrain the response of vibration,yet most active control algorithms are not ideal for the railway steel bridge structure due to the uncertainty caused by the crassitude structural members and multi degree of freedom of the railway steel bridge.In contrast,robust control algorithm can effectively deal with the above uncertainties.However,general robust active control algorithm needs full state feedback,and it will be costly and unrealistic to arrange sensors on full degree of freedom of steel bridge structures.Therefore,it is necessary to find an effective method to apply robust control algorithm to the vibration damping of railway steel bridge.Aiming at the two problems above,this paper proposes two intelligent algorithms based on BP neural network and particle swarm optimization to predict the full state response of structure,then adopts the robust H_∞algorithm to control vibration,and applies it to a railway steel bridge structure.The main contents of this study include the following aspects:(1)Aiming at vibration problem of nominal system of railway steel bridge,a linear state H_∞feedback control method with robust stability is introduced.Based on the algorithm,a robust H_∞control method is analyzed and deduced after presenting specific expression of uncertain parameters.(2)Setting up the finite element model of the railway steel bridge structure,analyzing its modal characteristics and dynamic response under train load and bidirectional seismic excitation to determine the dangerous point of vertical vibration.Illustrating the necessity of setting up AMD system in dangerous point and adjacent area,and giving its layout and mechanical model.(3)Analyzing the vibration damping effect of the railway steel bridge nominal system using the linear state H_∞feedback control to verify its suitability as the basis of the robust H_∞control theory.Discussing vibration attenuation effect of the railway steel bridges with single and composite uncertain mass,stiffness and control force using the robust H_∞algorithm under seismic excitation to demonstrate the effectiveness of the algorithm.Verifing robustness of the proposed method by node vibration control results of the railway steel bridge with more adverse uncertainty using the robust H_∞under train load.(4)Aiming at the demand for structure full state feedback of robust control algorithm,establishing an identifier by utilizing PSO optimized BP neural network with global optimal initial weights and thresholds to predict full freedom acceleration based on the acceleration response of partial structure freedom.Then the structural acceleration response predicted by the PSO-BP network is transformed into velocity and displacement through the time domain integration and the wavelet transform baseline drift removal method.Thus,the full state response of the railway steel bridge can be applied to the prediction of the robust H_∞control force.In summary,this paper combines vibration control of the railway steel bridge with intelligent algorithm effectively,establishes the theoretical method of robust H_∞control based on finite state observer using PSO-BP neural network which has practical significance to vibration control and life prolongation of the uncertain railway steel bridge.