Multivariate Data Fusion and Health Analysis of Coal Processing Facilities

Abstract

This paper presents a comprehensive health diagnosis and predictive maintenance management system based on multi - source data fusion. Firstly, the research constructs a data acquisition framework covering multi - source sensors such as vibration, temperature, and pressure. Wavelet transform and empirical mode decomposition are employed for signal processing, and multi - dimensional features in the time domain, frequency domain, and time - frequency domain are extracted. Subsequently, the local preserving projection algorithm is introduced to achieve effective dimensionality reduction and visualization of high - dimensional features. At the diagnostic level, a fuzzy clustering - fuzzy integral fusion diagnostic algorithm based on fuzzy measure theory is innovatively proposed to address the challenge of composite fault identification. At the predictive level, by constructing a degradation feature evaluation system encompassing correlation, monotonicity, and robustness, the self - organizing map network is used to fuse multiple features into a comprehensive health index, and the long - short - term memory network model is utilized to accurately predict the evolution trend of the equipment's health status. Finally, based on the health status assessment and remaining useful life prediction, a full - life - cycle predictive maintenance strategy is formulated, and an intelligent maintenance decision - support system is designed.