In machining processes utilizing machine tools, anomalies such as wear, breakage, missing parts, and chipping of the cutting tools can occur. Continuing to machine with these unnoticed tool anomalies can lead to the production of defective items. Consequently, there is a demand in manufacturing settings for technologies capable of detecting these tool anomalies in real-time during production.
Chipping, in particular, is a phenomenon where a part of the cutting tool's edge breaks off minutely, making it one of the anomalies difficult to detect visually. This leads to unintended changes in the machined surface's shape and a decrease in the dimensional accuracy of the product, thus becoming a cause for defects.
When tool anomalies occur, the load on the cutting tool exhibits a pattern of fluctuation different from the norm. By capturing this fluctuation pattern, it is possible to detect anomalies from the electric current flowing through the motors of the machine tools, thus allowing for the detection of tool anomalies without direct observation.
Utilizing this technology, we conducted experiments to detect minute chippings through the analysis of electric current flowing in the motors.
When repeatedly machining products of the same shape, the pattern of load fluctuation on the cutting tool tends to be similar for each cycle. Consequently, it is expected that the electric current data for each cycle will exhibit a similar pattern of fluctuation. By applying this principle and detecting variations in the electric current data that are abnormal, it becomes possible to determine the occurrence of tool anomalies.
Utilizing this method, we installed clamp-style current sensors on the motors of the machine tools and performed the cutting operation of the same product 60 times, collecting the electric current data for each instance. Subsequently, to estimate the fluctuation pattern of the load on the cutting tool, we overlaid and conducted a comparative analysis of the electric current data for each machining cycle.
The data obtained from the experiment is shown in Figure 3.
In the figure, the blue and red lines represent the electric current data during normal operation and at the time of chipping occurrence, respectively.
Figure 4 illustrates an expanded view of the electric current data during the time when chipping occurred. From this, it can be observed that the electric current data during chipping shows a slightly different fluctuation pattern. This pattern of fluctuation cannot be detected by traditional threshold-based detection methods for electric current values. Therefore, the development of a new anomaly detection algorithm was necessary.
The developed anomaly detection algorithm extracts multiple fluctuation patterns that appear during chipping from the electric current data as features and makes determinations based on their statistical characteristics. This approach enables more accurate detection of chipping compared to traditional methods.
The results of chipping detection using this anomaly detection algorithm are presented in Figure 5. In the figure, the blue and red dots represent the degrees of anomaly during normal operation and chipping occurrence, respectively, while the red line indicates the statistically calculated decision threshold.
The degree of anomaly calculated by the anomaly detection algorithm from the electric current data shows similar values for cycles 1 to 59 under normal conditions, but a sharp increase is observed in the 60th cycle, at the time of chipping occurrence. Thus, the subtle differences in fluctuation patterns, barely noticeable in the electric current data, become significantly apparent when using the developed anomaly detection algorithm, allowing for precise detection of chipping occurrences.
This technology is equipped with the capability to detect the occurrence of minute chippings, which are difficult to discern visually. This innovative approach enables the early detection of machining defects, effectively preventing the continuous production of defective items during the manufacturing process. MAZIN will utilize this technology to aim for enhanced efficiency and improved quality in production environments and will continue to drive forward with advanced experiments in the future.
