The invention relates to the technical field of
welding quality online monitoring, and discloses an
exothermic welding quality monitoring method based on
sensing data. Arranging a temperature sensor at a
welding station to be aligned with a heat release center, and collecting a calibration temperature and an original
voltage sample to generate a conversion slope and an intercept; determining sampling starting and ending according to starting and working step ending signals, forming a
voltage sequence according to a sampling interval, converting temperature, generating temperature rise by taking the first temperature as a reference, and normalizing the temperature rise; self-adaptive triangular
smoothing is carried out according to the number of sampling points, a computable interval is limited, and differential change rate and local second-order bending rate energy samples are constructed; samples are sorted to obtain self-baseline statistics, an abnormal index, a quality value and an abnormal time period are generated in combination with maximum energy, and a quality grade is output and filed. Cross-device comparability is achieved,
noise is resisted, invalid sampling is reduced, and compared with a
peak value method, process abnormity is easier to recognize and locate and
traceability is achieved.