The invention discloses an inland river ship lock fault risk grade assessment method based on
machine learning, and the method comprises the following steps: collecting
water level, gate opening, current, vibration and strain data, and generating a
monitoring data set; constructing an improved
spectrogram wavelet network comprising a physical
simulation graph fusion
branch, a heterogeneous structure fusion
branch, a scale crossing fusion
branch and a
hypergraph structure fusion branch; inputting a monitoring graph structure and a physical
simulation graph structure, and generating physical prior
graph embedding; inputting a physical part diagram structure and a sensor diagram structure, and generating heterogeneous
structure diagram embedding; extracting high-frequency and low-frequency image embedding, and generating scale
fusion image embedding; constructing a coarse-grained function
hypergraph and a fine-grained monitoring
hypergraph, and generating coarse-fine fusion
graph embedding; and fusing four types of
graph embedding, generating a risk
feature vector, inputting the risk
feature vector into a
risk assessment network, and outputting a
risk level result. According to the method, multi-
structure chart information and
frequency domain features are fused, and the modeling capability of fault
risk assessment is improved.