A process model-based monitoring and fault diagnosis methodology for free-form surface machining process

Multi-axis milling of free-form surfaces such as dies, molds, and various aerospace components is gaining wider attention. Because of the continuously varying cutter engagement conditions in these applications, the cutting force may change dramatically along the tool path, which makes the selection of process parameters and the monitoring of process faults difficult. Traditionally, process engineers rely on empirical data to deal with process planning and process monitoring. This suffers from the drawbacks that the experimental procedure is usually time-consuming and new tests have to be run whenever there is a change in cutting conditions, tool path and workpiece material.; In this work, a mechanistic force model is developed for ball end milling of free-form surfaces. The workpiece surface is represented by discretized point vectors. The modeling approach employs the cutting edge profile in either analytical or measured form. The engaged cut geometry is determined by classification of the elemental cutting point positions with respect to the workpiece surface. The chip load model determines the undeformed chip thickness distribution along the cutting edge with consideration of various process faults. Given a 5-axis tool path, shape-driving profiles are generated and piecewise ruled surfaces are used to construct the tool swept envelope. The tool swept envelope is then used to update the workpiece surface geometry employing a Z-map method. A series of 3-axis and 5-axis surface machining tests on Ti6Al4V were conducted to validate the model. The model shows good computational efficiency, and the force predictions are found in good agreement with the measured data.; Based on the process model, a fault detection and fault diagnosis methodology is proposed. The method has the capability of not only detecting the presence but also identifying the magnitudes of faults, which include flute chipping, breakage and spindle/cutter axes runout. A threshold-based fault detection method is developed based on the analysis of harmonic power distribution in the cutting force signal. A genetic algorithm approach is used to search and determine the fault pattern and magnitudes. The results obtained from these methods are validated through both steady state and free-form surface machining tests on 1018 steel.