The invention discloses a phase
recovery method, device and
system for non-proximity field measurement based on a physical enhanced neural network, and belongs to the technical field of
antenna measurement, and the method comprises the following steps: S1, preprocessing: carrying out the collection of non-phase amplitude data of a # 1 surface and a # 2 surface through a high-precision
radio frequency sensor array, and carrying out the size reforming of the non-phase amplitude data as
source data; s2, model construction: constructing a neural
network model; s3, model training: inputting data into a trainer, setting training rounds, extracting data of a source reconstruction part in the model when training is stopped, i.e., an equivalent source of the antenna, and further obtaining an
initial phase of a # 1 surface; s4, fine adjustment of the model: combining the measured amplitude and the
initial phase of the # 1 surface, and performing fine adjustment on the combined amplitude and the amplitude of the # 2 surface by using an alternate projection strategy; and S5, phase
recovery: setting an adjustment round, and when fine adjustment is stopped, obtaining a recovered phase of the # 1 surface. According to the near field-free measurement phase
recovery method, device and
system based on the physical enhanced neural network, the required data volume is small, the calculation speed is high, and the recovery precision is good.