Modeling And Control Of Output Fiber Morphology Distribution PDF For High Concentration Refiner Based On Stochastic Distribution Control

The high concentration refining system is an extremely important part of the chemical-mechanical pulping process.Its operating performance and quality of output fiber morphology distribution PDF(probability density function)not only affect the quality of production directly,but also affects power consumption of the subsequent papermaking process.However,the output fiber morphology of the refining process exhibits random distribution characteristics,which does not meet the Gaussian distribution.This is mainly because a series of complex mechanical action of lateral and vertical extrusion to the fiber during the process of gradually dissociating into single fiber and physical properties itself.It's difficult to establish effective pulping process model with the conventional mechanism mathematical method so as to control the fiber morphology.It's essential and urgent to propose methods of modeling and control to systems with characteristics of stochastic distribution in order to realize control of the output fiber morphology distribution PDF and solve problems of energy consumption and operation optimization of chemical mechanical pulping process.This paper did a series of modeling and control work to high concentration refining system with stochastic distribution characteristics based on the stochastic distribution control theory.Stochastic distribution control theory is proposed for bounded non-gaussian random distribution dynamic system,the core of which lies in the control to the shape of the probability density function of the system output variables.It's able to solve the problems of modelling and control to the fiber morphological distribution PDF by applying the theory.this article launched a series of modeling and control study revolves around the high consistency pulping system,the main work is as follows:(1)First of all,this paper introduces the production technology of chemical mechanical pulping process and made analysis to the process variable and dynamic characteristic of the high concentration refining system.Then it introduced the method of PDF approximation and weights decoupling the output fiber morphology distribution PDF based on B-spline neural network and the principle of the approximation of the Bounded stochastic dynamic system output probability density function with B-spline functions.(2)This paper established a probability density function(PDF)model of the high concentration refining system output fiber morphology distribution based on the theory of stochastic distribution and wavelet neural network(WNN)with WNN's powerful nonlinear function approximation and adaptive fault tolerance ability.The simulation experiment shows the effectiveness of the proposed modeling method.Then,this paper establish the state space model between the main input variables and the output fiber morphology distribution PDF of the high concentration refining system,which provides the model basis for the the subsequent control to the fiber morphology distribution PDF.(3)In the aspect of control,this paper designed a iterative learning control algorithm of output PDF of stochastic dynamic system based on norm optimization and apply it to the model of high concentration refining stochastic dynamic system.Adaptive iterative learning rate is introduced to accelerate the reflection.It correct the system input gradually by minimizing the defined performance indicators.Finally the effectiveness of the algorithm is verified by simulation experiment.(4)In order to widen the range of application of the proposed algorithm,nonlinear term,State disturbance and output disturbance are added to the liner high concentration refining system model got in the the third chapte.This paper proposed a robust iterative learning control algorithm based on norm optimization for output PDF of the uncertain nonlinear high concentration refining system and analyzes the control strategy of the algorithm when used in the model of nonlinear high density mill stochastic dynamic systems.The tracking of the output fiber distribution PDF is simulated for constant and dynamic value of two cases respectively.The simulation experiments show the accuracy and effectiveness of the algorithm.