A numerical control machining tool path step error prediction method based on BP neural network

Area of Science:

• Manufacturing Engineering
• Computational Science
• Artificial Intelligence

Background:

• Accurate step error calculation is crucial for high-quality numerical control (NC) machining tool paths.
• Current iterative methods for step error calculation are time-consuming and accuracy-limited.
• Neural networks offer potential for faster and more accurate calculations via parallel processing and continuous learning.

Purpose of the Study:

• To develop and validate a novel BP neural network model for predicting step errors in NC machining tool paths.
• To improve the efficiency and accuracy of step error calculation compared to traditional geometric methods.
• To demonstrate the effectiveness of the proposed neural network in real-world machining scenarios.

Main Methods:

• A BP neural network model was constructed using core parameters for step error calculation.
• Z-score normalization was applied to standardize data and mitigate the impact of singular parameters.
• Dropout technique and Stochastic Gradient Descent with Momentum (SGDM) optimizer were employed during neural network training to enhance stability and prevent overfitting.

Main Results:

• The neural network model successfully predicted step errors for samples from three different surface models.
• Prediction error decreased with increased sample training, indicating model convergence.
• After training on 15% of samples, over 99% of predicted step errors had an absolute error less than 1 μm.
• The computation time was reduced to one-third compared to the traditional geometric method.

Conclusions:

• The proposed BP neural network model provides an effective and efficient solution for calculating NC machining tool path step errors.
• The method achieves high accuracy and significantly reduces computation time, making it suitable for practical applications.
• This approach demonstrates the potential of neural networks in optimizing complex manufacturing processes.