Search

Aiming at the classification problem of power users' behavior under the background of big data, a classification method of power users' behavior based on ant colony optimize extreme learning machine (ACO-ELM) was proposed. First, the problem of optimizing the initial input weight and hidden layer neuron threshold of the extreme learning machine is transformed into the problem of ants seeking the best path in the ant colony algorithm. Through the updating, mutation and heredity of ant colony algorithm, the extreme learning machine is trained, and the input weight and threshold with the minimum error are selected to improve its prediction accuracy. ACO-ELM was used to classify power users according to power load data. Experimental simulation verifies that ACO-ELM has better prediction accuracy and accurate classification of power users.

A power network planning method with coal-to-electricity considering technical and economical adaptability is proposed in this paper to effectively solve the problem that traditional planning method does not consider the influence of coal-to-electricity. First, a technical and economical analysis model considering is set up, which breaks through the modeling difficulties of coal-to-electricity. The scale and specific household distribution of coal-to-electricity is quantitatively determined. Meanwhile, the investment composition, added electricity price and investment payback period are calculated, which provide useful advice for implementing coal-to-electricity and formulating subsidy policies. Besides, an engineering solution method based on multi-factors adaptability analysis is proposed to universally and practically solve the above model. Eventually, an actual application is shown to further illustrate the proposed model.

To address the issue that existing studies cannot take the impact of the existing service area and dynamic changes in traffic on planning of highway charging stations(CS) into consideration, a new practical planning method is proposed for the charging station of electric vehicle(EV) on the highway. First, taking into account the impact of the existing service areas, a multi-objective planning model was established, with the aiming of minimizing annual construction and maintenance costs and the daily charge waiting time of EVs. Then, a dynamic traffic simulation method based on link transmission model (LTM) is proposed. Further, the model was solved based on multi-objective particle swarm optimization and VIKOR method to obtain the optimal planning scheme. Finally, the validity of the proposed method are verified by the circular highway network.

In view of the fact that the configuration of some equipment for EV battery swapping station was not taken into account and the analysis of the charging behavior during the operation process was relatively coarse, a comprehensive optimal allocation method of EV battery swapping station was proposed based on the ordered charging strategy. Firstly, a comprehensive optimal allocation model of spare battery pack, charging equipment and swapping equipment was constructed. Secondly, based on the TOU power price mechanism, an ordered charging strategy of electric vehicles was formulated under a certain configuration scheme. Thirdly, a method of calculating the users' waiting time based on the queuing theory was proposed. Furthermore, an improved genetic algorithm based on elitism preservation strategy was proposed. Finally, the validity and practicability of this method were verified by example analysis.

Nowadays, the scale of distributed photovoltaic generation integrated into the distributed network and its penetration rate are growing continuously. The increase of installed capacity of distributed PV generation, sensitive to the market and national policies, is highly uncertain. Besides, the output of PV generation is highly random and clustered. This has posed great challenges to the medium and long term load forecasting of power system. In this paper, a neural network expert system model is proposed to predict the installed capacity of distributed PV on the target year. A method is put forward to predict the effective clipping capacity of the distributed PV generation during peak hours, providing new ideas for long term load forecasting with the distributed PV power integrated into the grid.

With the continuous development of information technology and the improvement of distribution automation level. Especially, the amount of on-line monitoring and statistical data is increasing, and large data is used data distribution system, describes the technology to collect, data analysis and data processing of the data distribution system. The artificial neural network mining algorithm and the large data are researched in the fault diagnosis and prediction of the distribution network.