Acoustic Sensing Approach for Real-Time Monitoring of Ultrasonic Bone Cutting States toward Predictive Maintenance of Cutting Tools
Jingyu Wang, Yu Liu, Linwei Wang, Jinguang Li, Yumeng Sun, Anxing Zhang, Shiwei Wang, Boming HuangAbstract
Ultrasonic surgical instruments are widely used in orthopedic procedures for their precision and efficiency. However, maintaining optimal cutting conditions at the blade–bone interface is crucial for both surgical quality and equipment longevity. Cutting speed is a key factor influencing both efficiency and tool health—low speeds reduce efficiency and increase wear, whereas excessive speeds can cause thermal damage to bone and accelerate blade degradation. Real-time monitoring of cutting speed is therefore essential not only to ensure surgical precision but also to enable predictive maintenance of the cutting equipment. To address this, we propose an acoustic-based cutting state monitoring method built on an enhanced deformable bidirectional long short-term memory–attention-enhanced exponentially weighted moving average (DBiLSTM-AEWMA) architecture. A deformable convolution (DC) module with a bottleneck residual structure is first used as a front-end encoder to adaptively capture localized, speed-related waveform distortions while keeping the model lightweight. The resulting feature sequence is then processed by a bidirectional long short-term memory (BiLSTM) network to model temporal dependencies, and a multihead attention mechanism emphasizes informative frames. An attention-enhanced exponentially weighted moving average (AEWMA) layer further smooths the predicted probabilities, reducing noise while maintaining responsiveness to rapid state transitions. Experimental results showed that the proposed method achieves 98.2% accuracy and reduces detection delay to 52 ms, satisfying real-time monitoring requirements. By providing real-time, acoustics-based monitoring of cutting speed, the proposed framework not only enhances surgical precision but also supports predictive maintenance strategies, enabling early detection of tool wear and performance degradation and improving proactive management of ultrasonic surgical instruments.