A transformer is an important part of power transmission and transformation equipment. Once a fault occurs, it may cause a large-scale power outage. The safety of the transformer is related to the safe and stable operation of the power system. Aiming at the problem that the diagnosis result of transformer fault diagnosis method is not ideal and the model is unstable, a transformer fault diagnosis model based on improved particle swarm optimization online sequence extreme learning machine (IPSO-OS-ELM) algorithm is proposed. The improved particle swarmoptimization algorithm is applied to the transformer fault diagnosis model based on the OS-ELM, and the problems of randomly selecting parameters in the hidden layer of the OS-ELM and its network output not stable enough, are solved by optimization. Finally, the effectiveness of the improved fault diagnosis model in improving the accuracy is verified by simulation experiments.

The smart grid concept is predicated upon the pervasive With the construction and development of distribution automation, distributed power supply needs to be comprehensively considered in reactive power optimization as a supplement to reactive power. The traditional reactive power optimization of a distribution network cannot meet the requirements of an active distribution network (ADN), so the Improved Grey Wolf Optimizer (IGWO) is proposed to solve the reactive power optimization problem of the ADN, which can improve the convergence speed of the conventional GWO by changing the level of exploration and development. In addition, a weighted distance strategy is employed in the proposed IGWO to overcome the shortcomings of the conventional GWO. Aiming at the problem that reactive power optimization of an ADN is non-linear and non-convex optimization, a convex model of reactive power optimization of the ADN is proposed, and tested on IEEE33 nodes and IEEE69 nodes, which verifies the effectiveness of the proposed model. Finally, the experimental results verify that the proposed IGWO runs faster and converges more accurately than the GWO.

Against the background of increasing installed capacity of wind power in the power generation system, high-precision ultra-short-term wind power prediction is significant for safe and reliable operation of the power generation system. We present a method for ultra-short-term wind power prediction based on a copula function, bivariate empirical mode decomposition (BEMD) algorithm and gated recurrent unit (GRU) neural network. First we use the copula function to analyze the nonlinear correlation between wind power and external factors to extract the key factors influencing wind power generation. Then the joint data composed of the key factors and wind power are decomposed into a series of stationary subsequence data by a BEMD algorithm which can decompose the bivariate data jointly. Finally, the prediction model based on a GRU network uses the decomposed data as the input to predict the power output in the next four hours. The experimental results show that the proposed method can effectively improve the accuracy of ultra-short-term wind power prediction.

In order to solve the problem of harmonic waves caused by battery energy storage (BES) and distributed generation (DG) inverters in an active distribution network, an intelligent optimal dispatching method based on a modified flower pollination algorithm (MFPA) is proposed. Firstly, the active distribution network dispatching model considering the power quality (PQ) problem caused by BES and DG is proposed. In this model, the objective function considers the additional network loss caused by a harmonic wave, as well as the constraints of the harmonic wave and voltage unbalance. Then, the MFPA is an improvement of a flower pollination algorithm (FPA). Because the MFPA has the characteristics of higher solution accuracy and better convergence than the FPA and it is not easy to fall into local optimal, the MFPA is used to solve the proposed model. Finally, simulation experiments are carried out on IEEE 37 bus and IEEE 123 bus systems, respectively. The experimental results show that this method can achieve satisfactory power quality while optimizing the total active power loss of the branch. The comparative experimental results show that the developed algorithm has better convergence than the FPA.