As a preparation method of solid preparation,dry granulation technology has the characteristics of high efficiency,pure physical processing and can be recycled.It has been widely used in the preparation of moisture sensitive and heat sensitive granules.In view of the current dry granulation process is mostly through artificial experience to adjust,lack of theoretical data analysis,equipment is mainly based on imitation,there are not high efficiency,card,particle forming rate is not high,serious powder leakage and other problems.In this paper,a small dry granulator LGS5 as a platform,through the experiment and process modeling and simulation methods,analysis of different process parameters on the plate and granulation,dry granulation forming technology for a preliminary study.At the same time,based on machine learning method,the prediction model considering powder properties was built,and the model accuracy was proved by experimental data set.The research will be meaningful to improve process comprehending,enhancing granulating efficiency and quality.The main research contents are as follows:(1)The equipment and technological process of dry granulation are introduced.In view of the intermediate process,the basic conditions of roller forming are analyzed theoretically.The FEM method were introduced.Based on the parameters of dry granulation equipment of the enterprise,a simplified model of tablet pressing was established by Abaqus software.The influences of conveying velocity,roller speed,gap and friction factor on the density of the tablet were studied.The rationality of simulation was verified by experiments.(2)Experimental study on dry granulation.Three commonly used excipients were selected and their properties were characterized.Based on the FEM simulation,the forming rule of the midbody under different ratio of screw speed and roller speed was studied experimentally.Selected the dry granule in the process of the screw and the whole grain rotational speed,and roller speed,fluid power,based on the Box-Behnken experiment scheme,study the significant difference of the process parameters on the particle forming rate.Powder properties and process parameters data got by experiment.A prediction model was built for SH-CG1,and the error of the model was proved to be 4.8%,and the optimal process parameters in the horizontal range were confirmed.(3)The prediction of the particle forming rate of dry granulation.Based on the experimental data,an index factor was introduced as the material properties.Including the process parameters,a prediction model between the process parameters of dry granulation and the particle forming rate was established.By comparing the model with the nonlinear fitting vector machine model,the particle swarm optimization algorithm is brought in to ameliorate the basic model,the average prediction error is3.4%,which could direct the selection of the early stage process. |