The performance degradation monitoring of rotating machinery equipment is an important part of active maintenance technology. Effective monitoring and evaluation of its performance can avoid sudden accidents and reduce losses to the greatest extent, and is the key to accurate fault diagnosis and life prediction. This paper proposes a fusion method based on locally linear embedding (LLE) feature, to extract the signal time domain, frequency domain and frequency domain characteristics of the equipment running status of the original feature set, the time and frequency domain feature extraction using ensemble empirical mode decomposition (EEMD) algorithm will signal is decomposed into a series of adaptive to the IMF from high to low frequency component, The energy value of each IMF component was calculated as the characteristic index in the time-frequency domain. Use LLE algorithm for feature set dimension reduction, extracting equipment all different life cycle running state in manifold features, combined with the radial basis function (RBF) neural network algorithm, which can identify different running status of equipment realize the monitoring of the performance degradation, and through the measurement of bearing the full life cycle of the signal analysis to verify the effectiveness of this method. Finally, using LabVIEW as the development platform, the functions of signal acquisition, storage, playback, time-frequency domain analysis and feature extraction are designed. The performance feature fusion and recognition algorithm proposed in this paper is integrated into the system to realize the monitoring of the running state and performance degradation state of rotating machinery equipment.