Study On Wood Drying Modeling And Internal Model Control Based On LS-SVM Strategy

Wood is one of the most important assets that forests provide to human beings.Compared with the forest resources of other countries and regions in the world,the coverage rate of our country’s forest is low,the distribution is uneven,and the forest resources that can be cut are few.With the continuous emergence of environmental protection and ecological problems,and the increasingly intensified contradiction between wood supply and demand,the rational use of existing forest resources and the improvement of comprehensive utilization of wood have become one of the hot issues that people pay attention to.Wood drying is the main method and important link to ensure and improve the quality of wood,reduce the loss of wood and improve the utilization rate of wood.However,wood drying is a nonlinear process with the characteristics of large lag,strong coupling and time-varying,and the drying process is susceptible to uncertain factors such as external interference.It is difficult to establish an ideal mathematical model and the control effect of traditional control methods is not good.Aiming at the above problems,the thesis proposes to use the Least Squares Support Vector Machine(LSSVM)which has the characteristics of arbitrary approximation ability,good generalization ability and small sample learning ability to nonlinear systems to build wood drying model and its inverse model,combined with Internal Model Control(IMC)strategy,new methods suitable for wood drying process control is studied.Aiming at the problem that it is difficult to establish an ideal mathematical model for wood drying which is a complex nonlinear process,the actual drying data of ash and birch are taken as samples,the wood drying model(the temperature and humidity control model and the drying schedule model)are identified positively and inversely by LS-SVM,respectively.The control model and control inverse model provide the necessary conditions for design of control system;the drying schedule model(wood moisture content prediction model)provides a strong basis for implementation and optimization of moisture content drying schedule;the drying schedule inverse model makes the drying medium state parameters in the drying kiln achieve continuous adjustment.The research results show that it is feasible to establish the wood drying model and inverse model by LS-SVM,and has certain prediction accuracy and generalization ability.The performance of LS-SVM model based on RBF kernel function is affected by the penalty factor C and the kernel function parameter σ,and the model parameters selected by experience cannot make its performance reach the best.Therefore,the thesis proposes a Modified Ant Colony Algorithm(MACA)to optimize the model parameters combination automatically to improve the calculation speed,modeling accuracy and generalization ability.At the same time,five standard test functions are used to test the optimization ability of MACA,and compared with the algorithms proposed in related literatures and the traditional Ant Colony Optimization(ACO)algorithm.The research results show that the optimization ability of MACA is better than the comparison algorithm.The RMSE,RE,RMSRE and MAE are used as performance evaluation indexes,and the performance of three wood drying schedule models,LS-SVM,ACO-LSSVM and MACA-LSSVM,is compared.The research results show that the prediction speed based on MACA-LSSVM is better than LS-SVM and comparable to ACOLSSVM.The modeling accuracy and model stability are better than LS-SVM and ACOLSSVM.Aiming at the problems of multiple adjustment parameters,low temperature and humidity control accuracy and seriously affected by external uncertainties in traditional control method of wood drying process,the thesis adopts the IMC algorithm with the advantages of simple structure,intuitive parameter setting and strong robustness,and the control model based on LS-SVM is used as the internal model of IMC system,the control inverse model is used as the internal model controller,and the first-order low-pass filter is used as the feedback filter to enhance the robustness and stability of the IMC system.The results show that l SSVM-IMC has good tracking ability and disturbance suppression ability,and the system can still show certain robustness even with model errors.In order to remove the coupling relationship between temperature and humidity in drying process,the inverse system method is introduced.In this method,the α-order inverse system based on LS-SVM is connected in series before the original system to form a pseudo-linear system,and an input filter and a feedback filter are introduced to improve the tracking ability and anti-disturbance ability of the system,and the IMC strategy is combined for control.The research results show that the IMC based on LS-SVM α-order inverse system has good tracking ability,which is better than traditional PID control and LSSVM-IMC method;the LSSVM α-order inverse system IMC is less affected by the model error,and its ability to suppress disturbance is better than that of LSSVM-IMC,and it has a certain decoupling effect.The temperature and humidity control accuracy meets the requirements of wood drying process.The establishment of wood drying model and its inverse model,the optimization of model parameters,and the realization of wood drying IMC method based on LS-SVM strategy have important research significance to optimize the drying process,improve the control level of wood drying process,ensure the wood drying quality,and shorten the drying cycle,etc.