Based On Bp Neural Network Prediction Of Cutting Force And B (??) Grinding Performance Study

In the field of metal cutting, the predicting and controlling of cutting forces is very important because cutting force influences machining quality, working durability and wear of tools and so on. There is far-reaching significance to study the predicting of cutting force. The conventional study begins with the theory and the experiments of metal cutting. In a long time, much more theoretical investigation had been done to find out the more precise theoretical formula to calculate the cutting force. However, the large number of interrelated parameters that influence the cutting forces and the complicated cutting process hold back the study. Presently, it is usual way to study the relation between cutting force and cutting parameters through the experiential formula drawn from the orthogonal cutting experiment. However, it can't meet the requirement of predicting and control of cutting force because of its poor universal utilization.In this paper, an approach to predict cutting forces with the help of artificial neural networks is proposed. Feed-forward multi-layer neural networks, trained by the error back-propagation algorithm are used. The predicting cutting force program based on neural network by using Oriented-Object language Visual C++ has been built up. In the meantime, two methods of optimization, the conjugate gradient and Quasi-Newton, have been applied in the training process of the network in order to make calculating fast and precisely. This paper also provides the feasible means to avoid the main problems of supersaturation and local minimum. And the detailed discussion has been made about the principle of building the neural network, the training parameters and normalization of data. At last, the cutting data of two nickel-based super alloys, K24 and B 1, have been provided to train the BP neural network so that check the effect of predicting and simulating cutting force.The grindability of a new kind of nickel-based superalloy B 1 has systematically been studied through experiments. According to the comparison of grinding forces in the grinding process by using several kinds of grinding wheels (CBN 100, CBN 180 and WA80), the general rule of grinding forces of different grinding wheels has been obtained respectively. On the basis of orthogonal designed experiment, the experiential formula of grinding force has been obtained and the effects of many kinds of machining parameters can be summarized under the condition when the feed speed is constant. The research is helpful to provide the basis for applying the CBN grinding wheel to machining this kind of superalloy. All of these also provide the experimental data for the studying of predicting cutting force based on BP network.