Research On Data Mining Based Fault Diagnosis Of PV Array

The rapid development of new energy, photovoltaic industry is closely concerned. While the PV system high price is always the most important factor of restricting the development of photovoltaic industry. The PV investor’s one-time investment is too high, the cost can’t recover in short time.In this paper, on the basis of a large amounts of data information collection, analysis the nation’s PV industry development plan, and the State Grid company for the development of photovoltaic industry support policy, and JiaXing city, ZheJiang province, current situation of the PV industry. The main reason is that the photovoltaic power generation system can’t be guaranteed, PV arrays’ fault rate is high, while the hot spot phenomenon is the most common of all. The diagnosis fault can’t be processed timely and effective as soon as the phenomenon has occurred. If the fault diagnosis of PV modules can effectively, the photovoltaic power generation system’s economy would improve mostly.According to PV module fault diagnosis research is still in the initial stage. Some electric power fault diagnosis technology can be copied, but all have limitations. For PV array of the fault diagnosis technology is mainly divided into two categories:ir test and electrical test. The fault diagnosis technology of PV array is based on artificial judgment, in fact, it is a measure to take measures to eliminate the fault rate.In this passage, determining the settlement of PV array fault diagnosis strategy, put forward a kind of research. Through reading and sorting the classical fault diagnosis technology of relevant information, to determine the solution of PV array fault diagnosis strategy, proposed a research method, without using other hardware devices, the use of the PV array current, the output power of the PV array monitoring data, temperature sensor is in ambient temperature, solar irradiance, acquisition environment tester, micro inverter fault indicator data the key point is that the cleaning and processing of massive data, using the processed data to establish the model of decision tree, C5.0 algorithm, to solve the problem of fault diagnosis of photovoltaic array. At the same time and the CART algorithm, the results of the former is superior to the latter.