Process Simulation And Prediction Of Polyester Fiber Melt Conveying

As the largest synthetic fiber variety,polyester fiber has become more and more widely used,and the demand of its quality has been increased.The production process of polyester fiber is divided into three parts: polymerization,melt transportation and spinning.Among them,the melt transporting link has the function of connecting the upper and lower,mainly to transport the products of the polymerization link safely,stably and evenly to the spinning section for spinning production.The quality of the melt conveyed directly determines the quality of the final spinning product.In order to ensure the quality of the melt transported in the melt conveying section,a simple process simulation was carried out for this part,and the performance index of the melt was predicted by using relevant models.The quality of the melt in the production process can be predicted in advance so that various preventions can be done in time.In the industrial background of intelligent manufacturing,based on the application of intelligent control and optimization technology to the production process of polyester fiber,this paper mainly completes the following work:(1)Establish a process simulation of the production line of the melt conveying link to ensure that the data flow of the melt-related indicators can operate normally.Mainly according to the connection nodes of each component of the melt conveying link,and the corresponding mechanism model is established for each segment,and the established mechanism models are connected in order according to the order of component connection in the actual production process to realize the points to segmentation prediction.Then,the program is written in Python language,and the prediction results of all the segments are visually displayed,and the whole melt conveyance process is simulated and predicted.(2)In the melt transport segmentation model established in the first part,due to the complexity of the melt transport link,each model will inevitably have errors.And the tandem mode between multiple models causes the cumulative transfer of errors between the models,making the final prediction results inaccurate.In order to avoid this,two solutions have been proposed.One is to change the multi-model connection method,set the breakpoint in the appropriate place,and introduce a time series prediction model,and use its output as the input of the subsequent model,thus preventing the continued transmission of the error.The general idea of the second scheme is to insert an error correction model at the critical point where the cumulative error can be ignored.This model can correct the data with errors so that there is no error or the error is small.There may be no first solution high,but it is simple to operate and does not interrupt the continuity of the data stream.Finally,the second solution is selected to correct the error of the series model system of the whole melt conveying link.(3)A hybrid integration model is proposed for the problem that the prediction accuracy of the mechanism model in the melt conveying process is not high.This model combines the mechanism model and the data-driven model,and uses the data-driven model to compensate the error of the mechanism model,so that the prediction accuracy of the whole model can be improved.Different from the conventional hybrid model,the data-driven model part of the hybrid integration model proposed in this paper is combined with the GRU algorithm and the SVR algorithm.The combination method is to weight the average of the two by a set of weight vectors to increase the whole model`s fault tolerance rate.Finally,this thesis summarizes all the work content,and analyzes its shortcomings in terms of relevant content and current results.These shortcomings also point the way for future research.