Study On System Of Household Photovoltaic Microgrids

Because the grid-connected power generation of the household photovoltaic(PV) microgrid is in the grid terminals, and load can use up the electric energy generated from photovoltaic cells, the electric energy loss in the process of transmission can be greatly reduced, which makes a full use of the advantages of distributed renewable energy power generation being close to the local load center and of no pollution, and overcomes the disadvantages of geographical dispersion of solar energy and other renewable energies. Compared with PV power station household PV microgrid has greater advantages both in flexibility and efficiency. Therefore, based on the connection of the PV power generation and energy storage system with the grid, this paper analyzes the establishment of household PV microgrid and studies the island detection method and household PV microgrid island operation. In addition, it also studies how to make the inverter output voltage, frequency and waveform distortion rate meet the requirements of power quality in the household PV microgrid island operation, and with the analysis of genetic algorithm it formulates a scheduling plan for the household to use PV micro grid system.In this paper, the main work and achievements are as follows:1. The design of a 3 kW single-phase PV grid-connected inverter, which can be in parallel operation and off-grid operation, the output voltage 220 V, 50 Hz sine wave, Rated power of 3 k W, waveform distortion rate less than 5%, the power factor of 1,conforming to the design requirements.2. The design of a set of household PV power generation microgrid experiment device. It sets up a household PV grid experimental apparatus on the analysis of PV grid inverter, which contains 3 kW photovoltaic panels, two groups of 12 V, 200 Ah battery, PV grid inverter 3 kW, 3 kW bi-directional inverter and local ac load, etc., and with access to household PV microgrid experiment data it verifies the master-slave control strategy of the PV microgrid smooth switching.Theoretical analyses, simulation and experimental data show that the household PV microgrid topology structure can be both in parallel operation and off-grid operation, and the proposed master-slave control strategy can absorb photovoltaic energy easily and efficiently, which can efficiently provide electricity to the local load and ensure the stable and reliable operation of the system.3. In order to meet the need of smart grid development, this paper takes an example of a household load in Tai’an, Shandong province. With energy storage level control and genetic algorithm, it makes a rational electricity plan of residential PV microgrid. The results are as follows:(1)From the perspective of usersThe household can get the income of ￥ 2.76 yuan by selling electricity before the optimization of PV microgrid, while the income rises to ￥4.925 yuan after the optimization since the established model can shift the load time through the reasonable arrangement of energy storage level, which can maximize the household income and prove that energy utilization of photovoltaic power generation is the largest.(2)From the perspective of gridAfter optimizing the time-shifting load scheduling, and with the cooperation of energy storage and time-shifting load it can achieve the function of peak shifting, which is of great importance to increase power supply capacity and improve power supply reliability.4. Passive islanding detection methods have a lot of non-detection, while active islanding detection methods, though non-detection zones are relatively small, have harmonic pollution to power quality. In order to improve the power quality, this paper proposes a new passive islanding detection method for the grid inverter, which is the islanding detection method based on wavelet-neural network identification. This method utilizes public coupling point voltage signal acquisition which is based on various scales of the wavelet transform to extract the characteristic coefficient of energy. The extracted feature vector not only has translation invariant feature, but also reflects the signal time-frequency local characteristics, which improve the real-time feature of island detection. Using the model with strong ability to recognize the three layers BP neural networks can effectively identify the island and nonisland state. With Matlab simulation results the island detection method shows that its discriminant accuracy of identifying island and non-island state can reach 98.45%, whose blind area is much smaller than the traditional passive detection method. Since there is no adding of disturbance quantity to the control signal, there is no harmonic pollution, nor bad impact on power quality, which makes up for the inadequacy of active test. Moreover, it will not cause voltage amplitude, frequency limit or system damage, nor bring active power fluctuations, which is advantageous to the photovoltaic and other renewable energy applications.