Research on scraper conveyor load prediction method based on wavelet transform and BP neural network

Area of Science:

• Mining Engineering
• Artificial Intelligence
• Signal Processing

Background:

• Scraper conveyor load prediction is essential for optimizing coal mining operations and enabling cooperative speed regulation between mining machinery.
• The inherent nonlinearity and non-smoothness of scraper conveyor loads, caused by unpredictable coal fall, present significant prediction challenges.

Purpose of the Study:

• To develop an accurate prediction model for scraper conveyor load by analyzing motor current.
• To enhance the prediction capabilities for cooperative speed regulation in coal mining machinery.

Main Methods:

• A Backpropagation (BP) neural network model was developed to map motor load to current.
• Wavelet transform was employed for the decomposition and reconstruction of historical scraper conveyor current data.
• Time series prediction was performed on both original and reconstructed data samples.

Main Results:

• The BP neural network model integrated with wavelet decomposition demonstrated superior prediction accuracy compared to using original data.
• Key error metrics were significantly reduced: root mean square error by 13.26%, average absolute error by 14.19%, and percentage error by 17.43%.
• The enhanced model meets the stringent accuracy requirements for scraper conveyor load prediction.

Conclusions:

• The proposed wavelet-decomposed BP neural network model offers a robust solution for scraper conveyor load prediction.
• This improved prediction accuracy provides a crucial theoretical foundation for the cooperative speed regulation of coal mining machines and scraper conveyors.
• The findings contribute to more efficient and stable underground coal mining operations.