Research On Closed-loop Control Algorithm And Its Implementation Strategy For Active Beam String Structure

As a kind of rigid-flexible hybrid structure,the beam string structure(BSS)has the features of beautiful structure and definite stressing.It can give full use of both rigid and flexible materials and has the advantages of convenient manufacturing and construction,so the BSS has a good prospect of application.This thesis introduces the active control process into the BSS,making it an active BSS(ABSS)which can automatically adjust its shape according to the changes of the external environment.In order to achieve this goal,this thesis has carried out an intensive research and investigation of the improvement of optimization algorithm,the single and multi-objective optimal control of structure and the prediction of structural optimal control solution.In this thesis,a large space random search method,namely the genetic algorithm,is used to solve the optimization problem of ABSS.An improved genetic algorithm——gradient genetic algorithm is proposed to overcome the shortcomings of the traditional genetic algorithm's poor local convergence ability,a gradient descent operator is introduced into the genetic algorithm to improve its local search ability,convergence speed and search accuracy,thus the gradient genetic algorithm can achieve global optimization.Then,based on the stiffness equation of the actuator,the structural finite element equation considering the actuator coupling is deduced.Through the sensitivity analysis method,the optimization model of the structural internal force and displacement linear control is established.When taking into account the structural nonlinearity,finite element software ANSYS is used to analyze the geometric nonlinearity of the structure.The procedure of structure optimization of ABSS is compiled by using MATLAB and ANSYS.Using gradient genetic algorithm and nonlinear finite element software,the single-objective nonlinear optimization control problem of ABSS can be solved by the procedure.According to the control requirements of different objectives,the multi-objective optimization problem of ABSS considering internal forces and displacements of structures is studied.A multi-objective fuzzy optimization model based on objective priority is established,and a two-step fuzzy optimization algorithm is proposed to solve this model to realize the multi-objective control of ABSS.Finally,according to the optimal control solution under different load conditions,the stress state information which can reflect the load condition of the structure is selected,and the corresponding relationship between the stress state information and the optimal control solution is established,which is used as the training sample,then the neural network for the closed-loop control of the structure under the specified optimization target is formed.When the external environment changes,the optimal control solution corresponding to the new stress state information can be predicted through the neural network.The production mechanism of training samples is mainly studied,and the influence of the number of samples and the type of samples on the predictive performance of neural network is analyzed;The particle swarm optimization(PSO)algorithm is used to optimize the BP neural network,so that the improved neural network can jump out of the local optimal solution;Aiming at the problem that the data of the structure stress state information is missing,the k-nearest neighbor algorithm is used to supplement,then predict with the new data so that the control problem of ABSS for missing collected data can be solved.