Research On Operating Condition Analysis And Intelligent Fault Diagnosis Of Pumped Storage Units

With the continuous advancement of energy structure transformation,the pumped storage unit(PSU)plays a crucial role of peak load and frequency regulation,electric energy consumption and accident backup in modern power system owing to its characteristic of flexible start-up,fast response speed and wide operating range.For the purpose of improving the power supply quality and safeguaranteeing the grid,the construction of smart pumped storage power station has become a new development direction in hydropower industry.In this research,the information of basic station equipment and operation data were fully utilized to establish the mechanism model of PSU,process status signals and analyze the system operating condition quantificationally.Further more,signal feature extraction and intelligent fault diagnosis were carried out.Based on which,the visualization networking evaluation system were designed and developed under the advanced system integration theory and software technology to satisfy the need of practical engineering application.The primary contents and innovative achievements are as follows:Compared with hydroelectric generating units,the operating modes of PSU is more complicated and switches more frequently.The wide hydraulic fluctuations in operating transition process cause a serious threat to stable operation of equipments such as pipelines and generating units.Based on the hydraulic propagation characteristics of pressure pipeline,surge shaft,globe valve and pumped turbine,the characteristic line equations were introduced to model the pumped turbine regulating system mechanically.The GB-BP neural network based method was proposed to fit the pump turbine characteristic curve so as to enhance the interpolation accuracy of generating unit parameters.The established model's precision was verified to be high by comparing the simulation results with actual data under various operation modes.Preprocessing and analyzing the hydro-mechanical-electrical status signals are significant ways to evaluate the operating stability of PSU.Considering the characteristic of pressure fluctuation signal and the shortcomings of original EWT,an improvement measure in spectrum segmentation based on multi scale-space during empirical wavelet transform and a trend extraction approach to guarantee the completeness of Hilbert spectrum were proposed.On the basis of theory research,the methodology was applied to the analysis of actual pressure fluctuation signal under pump outage condition.It turned out that the intrinsic modes including primary frequencies were effectively extracted.Combined with the generating unit parameter simulation curves obtained by mechanism model,PSU's work condition under that operating mode was analyzed.To achieve unsupervised fault diagnosis and operating mode recongnizition of PSU,it is a key scientific issue to solve the problem that tradition FCM clustering algorithm fail to select optimum cluster number automatically,thereby being ineffective to identify unknow fault patterns.An improved hybrid backtracking search algorithm which optimizes the cluster validity index by evolving the cluster number and cluster centers coded separately in binery and real number simultaneously was proposed.The final subset of signal features is acquired by empirical wavelet transform,mixed feature extraction and principal component analysis successively.The methodology achieved high classification accuracy in the bearing fault diagnosis experiment and the PSU's operating modes recongnizition experiment based on pressure fluctuation signals.In oder to meet the repuirement of operation analysis in practical project,overcome the shortcomings that front-end and back-end highly coupled and unclear responsibilities in traditional hydropower station control optimization and condition evaluation system.A front-end and back-end separated architecture which adopted Vue.js as front-end framework and spring boot framework as back-end was designed.The pumped storage unit operating condition evaluation system including the functions of PSU simulation,signal and index analysis,etc.was developed and applied in pratical project.