Casing Process Parameter Prediction Model of Wind Turbine Bearing Pedestal Based on BP Neural Network
Author of the article:CHEN Deping;CHEN Ying;LUO Jianshe;XIE Jian;CHEN Liming
Author's Workplace：School of Materials Engineering, Chengdu Technological College, Chengdu 611730, China
Key Words：BP neural network orthogonal test casting process parameters shrinkage cavity and porosity numerical simulation
Abstract：
A prediction model for the casting quality of large wind turbine bearing pedestal based on the concept of KBE and BP neural network was established combining the orthogonal test design method and casting simulation. The pouring temperature, pouring time and the initial temperature of the mold were taken as the input values of BP network training samples, and the shrinkage cavity defect area of the bearing seat, the solidification time of the bearing seat and the maximum temperature difference of the casting after the solidification of the bearing seat obtained by the simulation software Procast were taken as the model target values. The results show that using this model can predict the result value of random combinations of process parameters of castings, through simulation test and the comparison of predictive value,the results obtained are consistent with in two ways, to shorten the development cycle, large casting reduces the manufacture cost, and can give the best process parameter combination and guidelines for the actual production can be fast and efficient.
