The invention provides an over-temperature detection
system, and the
system comprises an air inlet detection mechanism which is disposed at an air inlet of a centralized
photovoltaic inverter and is used for detecting real-time air inlet data; the temperature detection unit is used for detecting first temperature data at the IGBT unit and second temperature data at the radiator; the
data processing unit is electrically connected with the air inlet detection mechanism and the temperature detection unit and used for receiving the real-time air inlet data, the first temperature data and the second temperature data and obtaining the real-time operation power of the
inverter, the real-time rotating speed of the exhaust fan and the comparison air inlet data; the real-time air inlet data, the first temperature data, the second temperature data, the real-time operation power, the real-time rotating speed and the comparison air inlet data are subjected to multi-
modal data fusion
processing of a BP neural network; and determining whether the
inverter is over-temperature or not according to the
processing result, and determining a heat dissipation fault reason causing the over-temperature. The invention also provides a centralized
photovoltaic inverter and an over-temperature detection method. According to the invention, the heat dissipation fault reason of the
inverter can be accurately determined.