Compressibility Prediction of Reduced Water Atomized Iron Powder Using Adaptive Neuro-Fuzzy Model

摘要

At present, reduced iron powders are the main grades of powder in production and consumption in powder metallurgy industry. One of the most important processing properties which determine the ultimate properties of part is compressibility of metal powders. Compressibility is a function of particle shape, density, hardness and size distribution, which is defined as a dependence of the compact green density on compacting pressure. Various particle size distributions have different apparent densities and lead to changes in compressibility of powders. In this study an adaptive neuro-fuzzy model is introduced to establish the relationship between the compressibility of WPL200 iron powder as a function of particle size distributions and apparent density. In an effort to construct the model, particle size distributions, apparent density and compaction pressure are employed as input variables while green density is the only output argument. To verify the accuracy of model 10% of experimental data is used as testing data. Results show that there is a satisfactory agreement between experimental data and predicted values and the average percentage of error is less than 6%, which demonstrates the high prediction capability of the model.