Research On Classification And Identification Of Pressure Fluctuation Characteristics In Draft Tube Of Francis Turbine

Since the 21 st century,all countries in the world have been paying attention to sustainable development.As a distributed generation technology,renewable energy is gradually becoming the development trend of all countries in the world.Hydropower has become one of the most promising renewable energy power generation projects,and it is gradually replacing the traditional thermal power generation.Hydraulic turbine is the core equipment of hydropower station,which is used to drive the generator to generate electricity.Francis turbine is widely used because of its small size,high efficiency,and wide application range of head.The pressure fluctuation of draft tube is an important factor affecting the stability of Francis turbine.According to the pressure fluctuation characteristics of draft tube,it can be classified or identified to determine the working condition of turbine.The formation mechanism of pressure fluctuation and the extraction method of characteristic signal are systematically analyzed in this thesis,and a characteristic classification method and a characteristic identification method of pressure fluctuation signal of draft tube are proposed.In the characteristic classification method,firstly,wavelet packet is used to decompose the pressure fluctuation signal in three layers to obtain eight sub waveforms,and the sample entropy and permutation entropy are used to calculate the characteristic vector of the obtained sub waveforms.Secondly,the fuzzy C-means clustering algorithm is used to cluster the extracted characteristic vectors,and the characteristics with high similarity are classified into one class,which realizes the automatic classification of vibration signal characteristics according to their own characteristics.Finally,because the classification accuracy of SVM is affected by its kernel function and penalty factor,the equilibrium optimizer algorithm is used to optimize these two parameters.The EO-SVM of draft tube pressure fluctuation characteristic classification model is constructed,and the effectiveness and accuracy of the model for the characteristic classification of draft tube are verified.In the characteristic identification method,firstly,in view of the shortcoming that the manta ray foraging optimization algorithm is easy to fall into local optimum,the algorithm is improved in four aspects:(1)elite opposition-based learning algorithm was used to optimize the initial population;(2)the first 50% of the initialized population was selected as a new population to ensure high-quality population;(3)in the chain foraging,the adaptive t distribution was used to replace the chain factor to optimize the updating strategy of individuals at the chain foraging;(4)in order to ensure the stability of the algorithm,some expressions of multiplication r are deleted in chain foraging and spiral foraging.Then,the pressure fluctuation signal is decomposed and reconstructed by discrete wavelet transform,the low frequency coefficient and high frequency coefficient of one-dimensional discrete wavelet transform,as well as the vertical and diagonal components of low frequency coefficient,high frequency coefficient reconstructed by two-dimensional discrete wavelet transform are extracted as characteristic vectors;secondly,fuzzy C-means clustering algorithm is used to classify the extracted characteristics into four categories;finally,based on the improved manta ray foraging optimization algorithm,the smoothing factor of the probabilistic neural network is optimized,and the ITMRFO-PNN identification model is constructed to identify the signal characteristics,and the validity and accuracy of the model are verified.