Automated Modulated Parameter Implementation Using Neural Network for Enhancement of Paint Spray

Artificial neural networks (ANNs) were introduced to be implemented for the paint spray process. ANNs are biologically inspired computational networks. It is a technology based on studies of the brain and nervous system. ANN models simulate the electrical activities of the brain and nervous system. ANN models include three layers (an input layer, hidden layer and output layer). ANN can be simulated in Matlab to predict the optimum conditions for maximum production. Surrounding conditions are the inputs for simulation. Data are collected and tabulated as a database for the model to generate the result. The result will be evaluated, analysed in the end. The prediction was trained using 70% of the data, and in the validation process, 30% of data was used. The optimum number of neurons is determined by training the network using a different number of neurons, and the performance of each network is compared. The statistical performance measures consisting of root mean square error (RMSE) and the square root of the coefficient of determination (R). A high predictive accuracy ANN model will possess the value of R close to one and low RMSE. The results of the training set were not high, R = 68.466% validation set with R = 63.173%.