Photovoltaic power missing data filling based on multiple matching and long- and short-term memory network

To address the problem of data missing caused by equipment faults, abnormal transmission, and improper storage in data acquisition systems of photovoltaic (PV) power plants, we hereby propose a method for PV power missing data filling based on multiple matching and a long short-term memory network. First, test samples were divided into three different weather sequences according to the weather type. According to the missing data types, the test samples were divided into two major categories, namely partial missing and complete missing. Each resulting set continued to be divided into four subsets according to the missing proportion. Then, a 450-day data source with 15-minutes intervals was established. The maximal information coefficient was used to reduce the data source and the redundant data space. Finally, coupling meteorological factors were eliminated according to the mutual information theory. A relationship between meteorological factors and short-wave radiation was developed to improve multiple matching accuracy. The correctness of the proposed method and its validity were verified by filling missing power data from a PV power station in Jilin, China.