The application discloses a UHF
signal dynamic pre-
correction method and
system for ground
digital television, relates to the field of
signal pre-correction, and comprises the following steps: collecting a detection
signal sequence of a UHF
frequency band in real time, and obtaining an actual output signal processed by a power
amplifier; separating the actual output signal into a detection signal component and a television
data signal component; performing
distortion feature extraction, and calculating and obtaining a
nonlinear distortion parameter matrix; inputting the
nonlinear distortion parameter matrix into a model as input features for learning and training based on a
time domain-
frequency domain double channel of a double-loop
convolutional neural network, and obtaining a pre-
distortion function; inputting the television
data signal into the pre-
distortion function, and outputting a pre-correction signal; and dynamically adjusting the pre-correction strength through a closed-
loop control mechanism. The application has the advantages that: through
deep learning technology, signal components are accurately separated, distortion features are extracted, and the pre-correction strength is dynamically adjusted, so that the
transmission quality and stability of the ground
digital television signal are significantly improved.