Research On The Detection Rate Of Multi-parameter Optimization Foreign Fiber Sorter Based On Neural Network

Due to the influence of foreign fibers on yarn quality,foreign fiber sorter is commonly designed by image processing at home and abroad to detect and reject foreign fibers.However,the unsatisfactory rejection effect causes many enterprises to carry out secondary manual detection.This paper takes the multi-parameter optimization as the research means to study the image detection parameters,fan frequency and camera white balance value of foreign fiber sorter with different neural networks for improving the detection rate of foreign fibers.Analyzing the process of foreign fiber detection and rejection,referring to the complexity and precision of the model,KPCA(Kernel Principal Component Analysis)algorithm is used to screen the influencing factors of foreign fiber detection rate and extract the main elements.The influencing factors of foreign fiber detection rate are further analyzed,and the RBF(Radical Basis Function)neural network is used to fit the main elements and detection rate to deduce the prediction model of foreign fiber detection rate.As the influence of image detection parameters on detection rate is analyzed,GABP neural network algorithm is improved,and the optimal parameter prediction model for image detection is established.According to the classification of different fibers,the image detection parameters are optimized and verified by experiments.Based on the optimal parameters of image detection and RBF neural network prediction model of foreign fiber detection rate as the objective function,the MIGA(Multi-islands Genetic Algorithms)is used to analyze the multi-parameter optimization process to determine the optimal parameter combination value of image detection parameters,fan frequency and camera white balance value,and the effectiveness of the combination value is verified by experiments.The detection effect of the optimal parameter combination value in actual operation is analyzed,and a multi-parameter optimal control system for foreign fiber detection rate based on neural network is proposed to adjust the optimal parameters real-timely.The RBF neural network prediction model is mainly used to preliminarily predict the foreign fiber detection rate,and the PSO-BP neural network controller is used to optimize the multi-parameters and design an online correction system to ensure the accuracy of the data and the stability of the neural network,and the simulation experiment is carried out to verify it.The research on multi-parameter optimization of foreign fiber sorter detection rate based on neural network can improve the foreign fiber detection rate in foreign fiber cleaning equipment and improve the dependence on manual selection in actual foreign fiber sorting operation,which has higher theoretical and application value.