Histogram as features for fault detection of multi point cutting tool – A data driven approach

Abstract

In the milling process, the condition of the tool is the important factor that determines the quality of the machined product. Hence, monitoring the condition of the tool becomes essential. There should be a mechanism to find out whether the tool is in good condition or affected by some faults. If so, the fault in the tool should be known to the user in order to prevent unnecessary breakdown. Face milling tool condition monitoring is taken up for the study. Tools with various conditions were considered viz., Tool in good condition, tool in broken condition, tool with the flaking condition, and tool with flank wear. The machining parameters were varied with three different values such as spindle speed (250 rpm, 300 rpm, and 350 rpm), feed rate (14 mm/min 20 mm/min and 28 mm/min), and depth of cut (0.5 mm, 0.75 mm, and 1 mm) and the percentage variation of the considered machining parameters with the average is found as 17% for spindle speed, 33% and 38% for feed rate, and 33% for depth of cut. Vibration signals have been acquired with the help of an accelerometer for various conditions of the tool under varying machining parameters. The raw signals were then processed and were extracted thirty-five histogram features. From the pool of features, only the useful features were selected by the decision tree algorithm. A set of rules have been generated from the useful features and fed as an input to the fuzzy classifier. The same set of features was used to classify the faults with the Artificial Neural Network (ANN) algorithm and their fault classification capacities were analyzed. The results show that both the algorithms exhibit promising results and ensure the robustness of the signals. Also, the best classifier is proposed for the fault classification of the milling tool.