The invention discloses an
infrared film absorptivity prediction method and
system based on
machine learning, and relates to the technical field of
optical film design and
performance prediction.The
infrared film absorptivity prediction method comprises the following steps that
feature data of a to-be-measured
film material and incident light are obtained,
electromagnetic field distribution and local absorption
power density are output, and according to power integration and boundary conditions, an
infrared film absorptivity prediction result is obtained; obtaining initial absorptivity distribution; performing
modal analysis on the initial absorptivity distribution, constructing a low-dimensional absorptivity
response model, and outputting absorptivity response data after dimension reduction; constructing a physical constraint neural network by taking
energy conservation and Fresnel boundary conditions as constraints; on the basis of a physical constraint neural network, absorptivity prediction is carried out on the infrared thin film under different conditions, thin
film structure data are optimized through a multi-objective optimization
algorithm, and an optimal structure parameter combination of the infrared thin film is obtained; according to the invention, through multi-
physics field modeling and the physical constraint neural network, the problems of low infrared film absorptivity prediction precision and poor optimization efficiency are solved.