Optimization of Parameters and Prediction of Response Values Using Regression and ANN Model in Resistance Spot Welding of 17-4 Precipitation Hardened Stainless Steel

17-4[Formula: see text]PH steel, also known as UNS 17400 steel or called SAE type 630 stainless steel, is a very important category of steel which has tremendous and extraordinary properties. The superior properties include high corrosion resistance, high hardness and high strength due to the conversion or phase change of the austenite to martensite by cooling the material to room temperature after heating to a temperature around 1030[Formula: see text]C. Normally, this procedure is called as precipitation hardening and hence the name. Due to its extensive properties, the 17-4[Formula: see text]PH steel finds its applications in a variety of industries including pump shafts, oil path, mechanical seals, and within the aerospace industry, parts of the aircrafts, chemical industries and petroleum industries. When a material is used in any kind of applications, fastening is the major process involved. Hence, there should be a standard welding procedure involved in generating a permanent fastening. In this work, resistant spot welding is considered as the welding process and the major parameters are considered which have a crucial effect on the whole process. The responses are considered to be the nugget diameter and tensile strength which denote the weld quality majorly. The process parameters with the help of literature are considered and they are electrode force, voltage and weld current. Taguchi method is used to design the experiments along with the NN tool to generate and predict the new response values. The results show that the major affecting factor is electrode force followed by current and then the voltage. Comparison is done to choose a better model for predicting the optimum responses with the given values of the input parameters. The results are pretty much accurate for both but still the regression model yields better results and almost similar values to the experimental values. Therefore, better results can also be obtained by ANN model by continuous training of the model.