Edge-based embedded algorithms for diagnosing bearing and machining issues in machining center spindles

Maintaining spindle health is vital in the metal-cutting industry. Without proper monitoring, spindle operation can lead to significant damage that worsens with continued use. As a result, implementing an effective spindle maintenance system is essential for any metal-cutting facility seeking to improve spindle performance and reduce operating costs. The most vulnerable component of the spindle is the bearing, which is prone to failure due to factors such as inadequate lubrication, over-lubrication, contamination, overloading, excessive temperatures, or misalignment. These issues can result in specific faults, including inner race faults (BPFI), outer race faults (BPFO), ball defects (BSF), or cage failures (FTF). Spindle failure can also occur due to a machining phenomenon known as chatter, which arises from incorrect settings of the radial depth of cut and spindle speed. One method to mitigate this issue is to create a stability lobe diagram (SLD) before machining. However, generating an SLD for every spindle is often impractical due to the lack of frequency response function (FRF) measurements for individual spindles. Therefore, detecting chatter in real-time during the machining process offers a practical approach to preventing spindle damage and minimizing maintenance costs. In this article, we present algorithms for diagnosing bearing faults such as BPFI, BPFO, BSF, and FTF using methods like Kurtosis, Hilbert Transform, and Wavelet Transform. Additionally, we propose a chatter detection algorithm designed for embedding into edge devices through the use of a redesigned filter. The results demonstrate that the developed algorithms are both accurate and computationally efficient, meeting the performance requirements of edge devices.