Feature Extraction And Vibration Prediction Of Non-stationary Signal Of Hydropower Unit

International demands countries to fulfill its obligation to reduce greenhouse gas emissions,in our domestic energy consumption at the same time,environmental protection and economic sustainable development of high-speed multiple pressures such as background,the construction and development positive hydropower renewable energy for the formation of clean,low carbon,environmental protection,new energy system,is of great significance to promote the sustainable development of economy in our country.As the core device in the process of energy conversion of hydropower station,hydropower unit is a kind of multi-source nonlinear equipment with water-mechanical-electrical coupling,and its vibration signal has strong nonlinear,so it is difficult to extract and predict vibration signals.In order to ensure the safe,reliable and stable operation of hydropower units,the feature extraction and vibration prediction of non-stationary vibration signals of hydropower units are carried out in this thesis.In this thesis,the vibration theory and mechanism of hydropower units are systematically analyzed,and the limitations of current research methods are studied.Taking multi-source nonlinear vibration signals of hydropower units as the starting point,the strategy of dynamic decreasing step is adopted for fruit fly algorithm and its odor concentration determination formula is improved,which solves the problem that the algorithm is easy to fall into local optimal solution.A method to determine the threshold of modal component of noise signal based on improved fruit fly algorithm is proposed.A method of vibration signal reconstruction and feature extraction based on improved Fruit fly algorithm,ensemble empirical mode decomposition and sample entropy was proposed to optimize the noise signal threshold and sample entropy to reconstruct and extract the vibration signals of hydropower units.Based on the improved fruit fly algorithm,the smoothness factor of the generalized regression neural network model was optimized,and the vibration prediction model of hydropower unit optimized by the improved fruit fly algorithm was constructed.