Tool life prognostics in CNC turning of AISI 4140 steel using neural network based on computer vision

Abstract One of the essential requirements for intelligent manufacturing is low cost and reliable predictions tool life during machining. It crucial to monitor condition cutting achieve cost-effective high-quality Tool conditioning monitoring (TCM) determining remaining useful assure uninterrupted machining manufacturing. The same can be done by direct indirect wear measurement prediction techniques. In methods, data acquired from sensors resulting in some ambiguity, such as noise, reliability, complexity. However, available images significantly less chances ambiguity with proper acquisition system. which provide higher accuracy than involve collecting worn tools at different stages process predict life. this context, a novel system proposed examine progressive utilizing artificial neural network (ANN). Experiments were performed on AISI 4140 steel material under dry conditions carbide inserts. speed, feed, depth cut, white pixel counts are considered input parameters model, flank along predicted output. captured using an industrial camera turning operation regular intervals. ANN training set predicts life, especially sigmoid function rectified linear unit (ReLU) activation ANN. showed 86.5%, ReLU resulted 93.3% predicting model’s maximum minimum root mean square error (RMSE) 1.437 0.871 min. outcomes showcased ability image processing modeling potential approach developing low-cost that measure operations.