The invention discloses an
ultraviolet partial discharge detection and prediction method and
system based on a
gallium nitride sensor, and relates to the technical field of
semiconductor firmware production and detection. The
system comprises a detection
system body, and the detection system body comprises an intelligent sensing acquisition module, a
signal conditioning and
noise reduction module, a multi-dimensional
feature extraction module, a mode recognition and classification module, a
data management trend prediction module and a decision output man-
machine interaction module. According to the invention, high-quality acquisition and conversion are carried out on
ultraviolet sub-signals through the intelligent sensing acquisition module, the
signal conditioning and
noise reduction module adopts adaptive filtering and
background noise deduction technologies to effectively suppress environmental
radiation and
electromagnetic interference, and the multi-dimensional
feature extraction module comprehensively extracts
signal features from a
time domain, a
frequency domain and a
time domain. The mode identification and classification module utilizes a
machine learning
algorithm to distinguish real
partial discharge and
noise pulses, so that the detection accuracy is improved, and the defects of high
false alarm rate and low signal-to-noise ratio caused by noise interference of a traditional system are overcome.