Predicting kerf quality characteristics in laser cutting of basalt fibers reinforced polymer composites using neural network and chimp optimization

In this study, an optimized artificial intelligence model is developed to predict the kerf quality characteristics in laser cutting of basalt fibers reinforced polymer composites. The model is composed of Long Short-Term Memory (LSTM) and Chimp Optimization Algorithm (CHOA). The latter is used as an internal optimizer to obtain the optimal parameters of the network model. The developed model was compared with three other models, namely standalone LSTM, LSTM optimized using Heap-Based Optimizer (HBO), and LSTM optimized using Manta Ray Foraging Optimization (MRFO). All models were trained and tested using experimental data considering five process control factors (cutting speed, air pressure, pulse frequency, pulse width and lamp current) and three process response (kerf width, kerf taper and kerf deviation). Response surface methodology was used to design the experimental plan. The accuracy of the models was evaluated and compared to each other using different statistical measures. LSTM-CHOA succeeded to predict kerf quality characteristics of the cut composites despite of their heterogeneous and anisotropic structure and it outperformed the three other models. The root mean squared error of the predicted kerf width, kerf deviation and kerf taper using LSTM-CHOA decreased by about 27.43%, 60% and 56.6%, respectively, compared with that of standalone LSTM.