Fault Diagnosis Of Power Transformer Winding Based On Variational Mode Decomposition Energy Entropy And Support Vector Machine

Power transformers are the core hub energy conversion equipment of power systems.The faults of power transformers are often accompanied by serious consequences such as large-scale power outages.Therefore,the safe and stable operation of power transformers is critical to the reliability of power systems.The traditional transformer winding fault detection method may require the transformer to exit the operation,or the hysteresis for the detection of the latent fault,and it is difficult to find the fault early in the occurrence of the latent fault.Aiming at these shortcomings,this paper proposes a fault diagnosis method for power transformer windings based on vibration spectrum analysis,in order to realize the accurate diagnosis of early latent faults in power transformer windings.For this reason,the variational mode decomposition is introduced into the fault diagnosis of power transformer windings.The detection and fault feature extraction of the surface vibration of transformer tank based on the energy entropy of variational mode decomposition is discussed in detail.A support vector machine fault diagnosis model based on Pareto particle swarm optimization.Intelligent diagnosis of power transformer winding faults.In view of the incomplete problem of vibration analysis of power transformer vibration spectrum analysis,this paper introduces the basic structure of power transformer and four core vibration sources,and how the vibration signals of multiple vibration sources are finally coupled to each other through different transmission paths.The surface of the transformer tank is analyzed in detail.The vibration mechanism of the core and winding of the two core sources of the transformer is analyzed in detail.The quantitative expression of the vibration acceleration of the winding and the core is given.Considering that the transformer vibration acceleration signal obtained on the surface of the transformer tank has nonlinear non-stationary characteristics,the theory of variational mode decomposition(VMD)decomposition is introduced.The simulation analysis shows that the VMD is decomposed in the case of discontinuous signals,frequency close signals,and pulse signals.Compared with EMD,it has obvious superiority,effectively avoiding two types of modal aliasing and over-decomposition,and accurately reflecting the source signal characteristics.For the problem that the VMD model parameters are difficult to determine,the two core parameters of the VMD are determined using the wave method.The VMD energy entropy is used to make a preliminary assessment of the health of the transformer winding,and the specific criteria are given.The VMD energy entropy is not affected by the magnitude of the load current and has strong stability.Then,for the problem that the VMD energy entropy can not be quasi-deterministic fault type,the VMD-SVM joint fault diagnosis model is constructed to realize the accurate diagnosis of the winding pad detachment,the winding radial stretching and the winding torsion.Aiming at the problem that the two core parameters of the support vector machine are difficult to determine,the Pareto particle swarm optimization method is used to obtain the optimal parameters by multi-objective parallel optimization of the two core parameters of the support vector machine.Finally,the VMD-SVM fault diagnosis model is tested by using the transformer instance fault data.For comparison,the signal decomposition processing method is replaced by EMD to obtain the vibration feature vector,and the EMD-SVM diagnostic model is constructed.The parameter optimization algorithm of the support vector machine is replaced.For the genetic algorithm,the GA-SVM diagnostic model is constructed,and the three models are tested using the actual transformer fault and normal data.The example test results show that the proposed VMD-SVM has the highest diagnostic accuracy and the diagnostic accuracy rate reaches 98.75.%.Accurate diagnosis of latent faults in power transformer windings is achieved.