Penetration Detection for Small-Hole Drilling EDM Based on LSTM

To address the problem of complex discharge state changes during perforation in EDM small-hole drilling, which result in untimely detection and low accuracy, this paper proposes a state monitoring method that integrates Long Short-Term Memory (LSTM) networks with machining parameters. First, the processing depth, speed, current, and voltage signals are normalized to achieve feature dimensionality reduction. Then, a state prediction model is constructed using a Long Short-Term Memory (LSTM) network, enabling the prediction of the current-stage data’s state value based on historical machining data. Finally, the perforation state of small-hole machining is determined by comparing the measured and predicted values. The input of the model includes machining depth, speed and machining current, and the output is whether penetration occurs. A penetration detection platform was established to validate the proposed method. The measured data indicate that the penetration recognition rate for 200 holes reaches 96%, and the minimum penetration distance is maintained below 0.5mm. Therefore, this method can achieve effective penetration monitoring in small-hole EDM machining, and it exhibits good model stability and accuracy.