Study On Modeling And Optimizing Control Of Soft Capsule Dropping Pills Pharmaceutical Process

This article takes soft capsule dropping pills pharmaceutical production process as the research object.In order to improve the yield of soft capsule dropping pills and its stability of product quality,the modeling and optimizing control are studied based on intelligent control,data-driven modeling and intelligent optimization algorithms. The main work in this dissertation is presented as follows.Firstly,the soft capsule dropping pills pharmaceutical production process is analyzed systematically.The soft capsule dropping pills process contains a number of subsystems,each subsystem with its own characteristics,such as gelatin solution generation subsystem is a very complex industrial processe,which includes a number of links,there are complex physical and chemical changes in the process,with obvious non-linear and many intermediate control variables.The gelatin temperature subsystem,paraffin oil level subsystem and pulse pressure subsystem are also very complexed with the characteristics of non-linear,time-varying parameters and time delay and so on.On this basis,aiming at the problems of the process and the different characteristics of the subsystems,design the appropriate control strategy to achieve better control results.Then,the gelatin solution production process subsystem was studied.This paper investigates multistage inverse modeling for a class of serially connected industrial large-scale systems.To control the quality of the production,inverse modeling is to obtain the model of the required process conditions and the control variable set points of a process system by backward reasoning with the specified product qualities as a starting point.The existing methods of inverse modeling establish the inverse model for the whole process which is generally difficult and rough.To reduce the difficulty and to improve the accuracy of the model,multistage inverse modeling method is proposed in this paper.The inverse models are established by using least squares support vector machine(LSSVM)and BP neural network (BP-NN)respectively.As an illustration of the effectiveness of the proposed method, we consider the production quality control problem of the gelatin solution production process.The simulation results indicate the model based on the proposed method has smaller error and higher hit rates.And then,the modeling and control of gelatin temperature control subsystem, paraffin oil level subsystem and pulse pressure subsystem are studied respectively. The mathematical models of the subsystems are established respectively.Aiming at the problems of nonlinear,time-variant and coupling existing in soft capsule dropping pills system,three kinds of novel intelligent control strategies are proposed based on fuzzy logic controller,adaptive control,PID controller and intelligent control based on pattern recognition method.They are namely hybrid fuzzy PID(Fuzzy-PID), model reference fuzzy adaptive PID(MRFA-PID),and adaptive control based on pattern recognition(PR-PID).The proposed methods can improve the dynamic response,regulation precision and robustness of the closed-loop system as well as guarantees the basic requirement on stability and product quality.They are used in the above three subsystems respectively.The ultimate goal of this study is to improve the soft capsule dropping pills pass rate and stability of product quality.The key process parameters of soft capsule dropping pills,including settings of the subsystems need an an optimization design. Therefore,on the basis of the completion of various sub-systems modeling and optimization control,the process of soft capsule dropping pills product quality modeling and optimization were studied.Soft capsule dropping pills product quality model was established based on data-driven modeling method.Soft capsule dropping pills product quality control system is a multi-input and multi-output complex system. First of all,the process parameters and a two-level hierarchy index system of soft capsule pills product quality were proposed based on the analysis to the production process.In the soft capsule dropping pills production process,there are so many variables in product quality-oriented modeling and the relationship among them are quite complex,which make physical-driven and statistical modeling very difficult, even if not impossible.So,data-driven modeling method is used to establish a higher precision product quality model of the soft capsule dropping pills.The model was established based on least squares support vector machine(LSSVM),whose inputs are the process parameters,namely gelatin solution viscosity,gelatin solution temperature,paraffin oil temperature,and pulse pressure,and outputs are the secondary quality indexes,namely spherical degree,tailed degree,breaking pill rate and collodion silk degree.On this basis,Analysis hierarchy process(AHP)was used to determine the weights of the secondary quality indexes.And then,particle swarm optimization(PSO) algorithm was used to optimize the process parameters in order to improve the yield of soft capsule pills,which is a multi-objective optimization problem.The nominal values of the process parameters corresponding to the highest yield can be obtained. The yield increases by 2.7 percent when the optimizing parameters are used to the soft capsule dropping pills process via off-line instruction,which indicating that the method of using LSSVM and PSO in product quality modeling and optimization of soft capsule dropping pills is reasonable.It provides a new way for soft capsules process optimization.