Hybrid Approach for Prediction and Modelling of Abrasive Water Jet Machining Parameter on Al-NiTi Composites

Abstract

In this work, pure aluminium and NiTi is used as matrix and reinforcement to fabricate a smart composite. To get an improved mechanical property, primarily the powder metallurgy process parameters are optimized, and the best processing parameters are used for the fabrication of composites materials and subsequently the composite is used for machiningMachining studies. Abrasive Water Jet Machining (AWJM)Abrasive Water Jet Machining (AWJM) process is used to study the machinability characteristics of the Al- NiTi smart composites. To study the effect of AWJM parameters on Al-NiTi composites, the following control variables identified are Transverse Speed (TS), Applied Pressure (AP), Standoff Distance (SoD), % Wt. of reinforcements (wt%), Abrasive Size (AS). The output indices are Surface Roughness (Ra) and Kerf Angle (Ka). The experiments are designed and conducted based on the design of experiment. Further, it describes the effectiveness of the hybrid algorithm in predicting and optimizing the Abrasive Water Jet Machining (AWJM) parametersAbrasive Water Jet Machining (AWJM). Grey Relational AnalysisGrey Relational Analysis (GRA) (GRA) is used as a feature selection and optimizing tool. The result of feature selection by GRA–EntropyEntropy, reveals that the most influencing control variables are ranked in the order as AS, AP, TS, wt% and SoD. Modelling of AWJM process is done by Support Vector Machine algorithm (SVM)Support Vector Machine (SVM), and the performance of the model is compared with SVM hybrid models. A hybrid model is developed with the concept of Differential Evolutionary algorithm (DEDifferential Evolutionary (DE)) and Entropy. Hybrid SVM–Entropy model displayed increased prediction performance by 37.8% compared to the SVMSupport Vector Machine (SVM) model. GRA–SVM–Entropy hybrid model is compared with the SVM model, it is found that the prediction performance of the GRA–SVM–Entropy hybrid model increased by 49.1%. It is found from the GRAGrey Relational Analysis (GRA)–Entropy method; the optimal conditions are A2, B1, C1, D3, and E1.