The invention discloses a power distribution network
master station fault positioning method and
system based on multi-dimensional
information fusion, and relates to the technical field of intelligent power grids, and the method comprises the steps: obtaining FTU electrical quantity data,
circuit breaker monitoring terminal switching value data and
fault indicator state
signal data; constructing a power distribution
network structure matrix D in a normal operation state, and generating a fault information matrix G based on the state
signal data; performing normalization
processing on the electrical quantity and switching value data, and converting the electrical quantity and switching value data into a time axis image; inputting the
image sequence into a
convolutional neural network for
deep learning, extracting fault features and outputting an inter-node fault initial probability matrix Q; and fusing Q, D, G and three types of
original data, performing multi-dimensional
information fusion by adopting a D-S evidence theory, generating a final
fault probability matrix S, and outputting a fault section and type with the highest probability. According to the method, CNN
feature extraction is guided by a topological matrix D / G, and the misjudgment rate of fault positioning is reduced through D-S fusion of electrical features, topological association and
equipment state weights.