This invention belongs to the field of aeronautical
acoustics and flight
simulation technology, and relates to a method for cabin multi-source sound source evolution and
sound field reconstruction under flight state constraints. It solves the problems of traditional blind
source separation relying on statistical characteristics,
sound field reconstruction relying on measured data, and low sound fidelity in flight simulators. This invention constructs a multi-source mixed acoustic
signal-flight parameter synchronous dataset as input, constructs a flight state-sound source feature prior
knowledge base as prior constraints, and constructs an objective function including
reconstruction error, source independence regularization, and flight
state constraint terms. The time-domain signals of each independent sound source are separated using the alternating direction
multiplier method, and the sound source feature parameters are extracted as output. Flight state parameters are used as input, and an evolutionary mapping function is constructed based on the mean function. The target flight state is input into the evolutionary mapping function to obtain predicted sound source feature values, driving the equivalent source method to achieve
sound field reconstruction, and outputting a three-dimensional sound field distribution. This invention achieves predictive, high-fidelity real-time reconstruction of the sound field within the entire
flight envelope.