Detection of cutting tool breakage and wear in CNC machi̇ne tools with appropriate signal processing methods

In industries such as machining, automotive, and defense, which require high precision and continuity, the ability to maintain uninterrupted production processes is directly related to the early detection of faults such as wear and breakage in cutting tools used in CNC machine tools. These failure conditions not only reduce production quality but also lead to increased downtime and significant increases in operating costs. While various methods have been developed in the literature to monitor cutting tool condition, many of these methods are challenging to integrate into real-time systems and operate with limited accuracy rates. In this context, this study proposes an innovative hybrid analysis approach that combines the Undecimated Wavelet Transform (UWT) and Root Mean Square (RMS) methods to effectively detect cutting tool wear and fracture tendency. Experimental studies were conducted in a real factory environment on a CNC sliding automatic lathe, and vibration data was obtained using a Dytran 3055D2 vibration sensor and a Sinus Apollo data acquisition card at a sampling frequency of 40 kHz. The UWT method was used to separate the temporal-frequency components of the signal by leveling them; then, RMS values were calculated based on the detail coefficients to create a hybrid score for each part. Threshold value analysis was applied to these scores, and parts exceeding the threshold value of 0.8 indicated that the tool had reached a critical point in the wear process. When the system detected 2-3 parts consecutively exceeding this threshold, it automatically generated a tool change signal, enabling intervention before tool breakage. The analyses revealed that the proposed UWT+RMS hybrid method, particularly with two-level UWT decomposition, yielded the most consistent and distinct results. The method successfully monitored the early stages of tool wear, enabling both production continuity and the reuse of the tool after sharpening. Additionally, the method can be easily integrated into real-time monitoring systems. In this regard, the proposed method is both applicable in the production field and provides an efficient monitoring solution in terms of hardware. With this study, UWT and RMS analysis techniques, which are typically used separately in the literature, were evaluated together for the first time in a hybrid structure and successfully tested on a system operating on a production line. The method not only detects cutting tool wear and breakage processes but also enables timely intervention in these processes, offering concrete benefits such as extending tool life, reducing maintenance costs, and lowering scrap rates. Despite its simple structure, the developed system stands out as a solution that operates with high accuracy, has a low processing load, and is applicable in real production environments. Future studies aim to test the system with different tool types and workpiece materials, integrate multiple sensors to achieve a more comprehensive structure, and enhance predictive maintenance functions using pattern recognition-based methods.