Power Quality Disturbance Detection And Classification Based On Adptive VMD And PSO-SVM

The increasing number and capacity of non-linear,shock,and volatility loads has caused the power quality problems to deteriorate.These kind of problems can damage power systems and electrical equipment,resulting in data loss,switch malfunction,production line downtime,and so on.It will pose a threat to the safe and stable operation of the power system in severe cases.In-depth study of various factors affecting the decline of power quality,accurately detecting and identifying power quality problems is the basis for ensuring the safe transportability of power systems.This thesis studies the algorithm of power quality disturbance in three aspects: power quality disturbance preprocessing,detection and classification.First,the power quality standards and development status at home and abroad is explained,and common types of power quality disturbances and hazards are summarized.The principle,advantages and disadvantages of existing power quality disturbance preprocessing,detection and classification methods are analyzed.Aiming at the problem that the actual grid signal is seriously influenced by noise,this thesis proposes a noise reduction algorithm based on the improved wavelet threshold function.Comprehensively comparing the performance of various wavelet base and threshold estimation methods,selecting the optimal wavelet base and threshold estimation method,and introducing the adjustment factor to make the wavelet threshold function has the function both soft and hard threshold.As the simulation results illustrate,compared with same type of algorithm,the proposed algorithm has good denoising effect.To cope with the problem that the current power quality detection algorithm has weak anti-noise and low positioning accuracy,the power quality disturbance detection algorithm based on variational mode decomposition is studied.Based on the study of the principle of variational mode decomposition,a power variation disturbance detection method based on adaptive variational mode decomposition is proposed.The method adaptively determines the number of decomposition layers K by using the energy difference as the evaluation parameter.And it improves the detection algorithm's anti-modal aliasing,anti-false component ability and noise robustness.The adaptive variational mode decomposition is applied to detect the simulated signal and measured grid signal.The emulation experiments results demonstrate that the adaptive variational mode decomposition can effectively separate the disturbance component and improve the disturbance signal detection positioning accuracy.The adaptive variational mode decomposition algorithm,EMD and EEMD are applied to detect the simulated signals and real grid signals respectively.The result of experiments indicates that the adaptive variational mode decomposition can accurately separate the disturbance components and effectively preserve the high frequency disturbance information.The algorithm detection and positioning effect is better than EMD and EEMD algorithms.In order to overcome the problem that it is difficult to select the parameters of the supporting vector machine,the particle swarm optimization algorithm is used to optimize the parameters of the support vector machine.On the basis of the statistical characteristics,energy distribution of each layer's eigenmode function is calculated.The energy difference eigenvector with the normal voltage signal is obtained,input it into parameter-optimized support vector machine to realize classification of the power quality disturbance signal.The simulation results illustrate that the PSO-SVM is not only better in noise resistance than other classification algorithms,but also has high recognition accuracy.The proposed algorithm is compared with PNN,ELM,and unoptimized SVM classification algorithm.The simulation results show that PSO-SVM can obtain higher classification accuracy rate in the classification of single disturbance signal and compound disturbance signalFinally,based on the algorithm proposed in this thesis,a virtualized power quality disturbance detection and recognition system based on PXI and LabVIEW platform is built.The software scheme of the whole system is elaborated,and the software design of the system data acquisition,data storage,power quality detection and power quality classification is realized.By drawing on the measured grid data,the test experiment of the detection and identification system are completed,and the engineering applicability of the method is verified.