Plunge Milling Force And Modeling Research

Plunge milling is mainly used for roughing and semi-finishing and is a kind of machiningmethod which can achieve high metal removal rate. It widely used in the field of aerospaceand mould industry. This article selects the cold work die steel Cr12as test materials, mainlystudying milling process of milling force and establishing model of plunge milling force.The establishment of the milling force model for the selection of machining parameters,machining process planning and selection and design of tool has important guiding role.Milling force formation is a complex nonlinear dynamics process, involving the blankmaterials, the processing environment, machine performance and other factors. BP (BackPropagation) neural network as one of the most widely used neural network model at present,since it is composed of a large number of connected neurons, it has the very good learningability and predict performance. BP neural network can approximate arbitrary linear andnonlinear function that can be used to establish the milling force prediction model. But the BPnetwork belongs to the gradient descent algorithm, it has shortcomings of slow convergencespeed and long training time. Genetic algorithm is an efficient parallel global search algorithm,has good robustness, and can avoid local minimum problem of the traditional BP algorithm.In this paper, the research content mainly includes:(1) According to the characteristics of the plunge milling, based on cutting graphics, thetheoretical analysis of the milling force was carried out.(2) Design the single factor test and multifactor orthogonal experiment, and on CNCmachining center the milling force is measured through milling force measuring device.(3) Principle of BP neural network and theory of genetic algorithm were summarized.The testing data are normalized and the BP neural network model of milling force isestablished.(4) Geting the optimal weight and threshold of BP neural network model by usinggenetic algorithm, the genetic neural network model is established by combining the geneticalgorithm and neural network.(5) Data validation and error analysis was carried on the network model by using thenetwork training data and test data and compare the network prediction precision before andafter the optimized using genetic algorithm.