Chunlin Guo

The scarcity of vibration fault sample data of large oil-immersed transformers has become a bottleneck in the development of transformer condition monitoring and fault diagnosis technology. This paper first studied the basic mathematical model of transformer vibration, then carried out vibration and electrical signal measurement tests in substations. After that, the simulation of transformer body and oil tank vibration modeling is realized by using COMSOL multi-physics field simulation software. The model built in this paper takes into account the effects of winding layered structures, winding pads, insulating oil, and fasteners on vibration transmission. This paper realizes the coupled simulation of the electromagnetic-structural-acoustic field of an oil-immersed transformer by dissecting and solving the finite element model mesh and verifies the model’s reliability in both theory and test.

At present, the number of fault samples is insufficient in the field of transformer fault diagnosis based on vibration, and the experimental data of existing research results mostly come from laboratory conditions, which is difficult to popularize in a large scale. Therefore, combining with the idea of fuzzy mathematics, this paper integrates the prior knowledge into the neural network, and extracts the function of characteristic quantity and fault probability from the trained neural network model by numerical test, which is the basis of fault diagnosis. Then, the effectiveness of this method is verified on the test sample set. The results can be used as a means to incorporate prior knowledge into intelligent algorithms, and help to study the relationship between different feature quantities. In general, this article provides a way of applying prior knowledge to vibration diagnosis algorithms, which has broad development prospects.

Game theoretic methods and simulations based on reinforcement learning (RL) are often used to analyze electricity market equilibrium. However, the former is limited to a simple market environment with complete information, and difficult to visually reflect the tacit collusion; while the conventional RL algorithm is limited to low-dimensional discrete state and action spaces, and the convergence is unstable. To address the aforementioned problems, this paper adopts deep deterministic policy gradient (DDPG) algorithm to model the bidding strategies of generation companies (GenCos). Simulation experiments, including different settings of GenCo, load and network, demonstrate that the proposed method is more accurate than conventional RL algorithms, and can converge to the Nash equilibrium of complete information even in the incomplete information environment. Moreover, the proposed method can intuitively reflect the different tacit collusion level by quantitatively adjusting GenCos' patience parameter, which can be an effective means to analyze market strategies.

The continuous growth in the number of electric vehicles (EVs) places higher demands on public charging station (PCS) planning. This study proposes a PCS locating and sizing method based on the discrete distribution of EV charging demand. First, the distribution of EV charging demand is predicted by the distribution of gas station sales, functional areas, and population density. Next, a method of determining the PCS capacity based on the distribution of the charging demand is given. Then, the PCS locating and sizing model with the goal of minimising the total social cost is established, and the particle swarm optimisation algorithm is used to solve the problem. Finally, the method is verified by a case in Beijing. The results show that after obtaining several convenient parameters and using a simple algorithm calculation, this method can effectively reduce the construction and operation costs of PCSs and give full play to the social benefits of PCSs.

Motivated by the development of electric vehicles (EVs), this paper addresses the planning of EV charging infrastructures (EVCIs). Considering that the optimal locations for EVCIs in most cities with a tight land supply are difficult to obtain, it is significant to study the planning of EVCIs in such cities. This paper proposes a planning method and model of EVCIs under the condition that alternative locations of centralized charging infrastructures (CCIs) are known. Firstly, based on the principle of energy equivalence, the sales volume of the fuel of gas stations in the planned area is converted into the equivalent electric quantity (EEQ) and the EEQ is divided according to the demands of different EVs for charging. Then, an equal product of the load and distance (EPLD) criterion is used to distribute the EEQ to CCIs and distributed charging infrastructures (DCIs) located in areas for load forecasting. The final plan is given out after checking the constraints and planning rationality. In addition, the net present value (NPV), the average charging distance, and the total harmonic distortion (THD) rate are used to evaluate the planning scheme. Finally, the feasibility and practicability of the proposed method are verified by a case study in Beijing.