Effective detection algorithm of electronic information and signal processing based on multi-sensor data fusion

Appropriate vibration diagnostics when cutting a variety of products is critical for cutting quality and scrap and consumables reduction. Collecting and combining data from various sensors allows for realtime data collection during processing for system monitoring. The purpose of this study is to develop a method for effectively diagnosing milling machine vibrations using multi-sensor data fusion. It is proposed that sound and frequency signals be preprocessed to isolate specific vibration features, and then classified and aggregated for a complete vibration diagnostics system. After preliminary wavelet signal conversion, a deep belief network has been developed for each sensor to classify vibration features. When combining classification results, the fuzzy Choquet integral algorithm is used. Thus, a comprehensive judgment is obtained. It has been demonstrated that for neural networks, 100 epochs are sufficient for learning to take place. It has been discovered that the accuracy of classification and inference of the correct solution increases from 75% to 98% as the number of vibration feature combinations increases. When compared to the Dempster-Schafer evidence theory, the fuzzy integral fusion algorithm provides a high level of vibration detection accuracy of up to 98.7%. The resulting diagnostic procedure is simple and effective. It can be used to control the cutting process on machine tools in industrial settings. Improvements to the scheme are suggested, including the use of intelligent technologies to automate the process.