In order to realize the condition monitoring and early fault diagnosis of the dry-type transformer in the power system, combined with acoustic emission sensors, temperature sensors, humidity sensors and other kinds of sensors to carry out the condition monitoring and fault diagnosis of the dry-type transformer, a set of dry-type transformer fault diagnosis system based on LabVIEW is developed. For acoustic emission, temperature, humidity and other data, the variational mode decomposition method based on Golden jackal optimization algorithm is used to extract the data trend, and the potential early fault is identified according to the change law of the trend; Aiming at the high sampling rate of acoustic emission data, a feature extraction method based on multi-scale Teo is used to extract the weak features of early faults. Simulation and application have shown that this system has superior recognition and diagnostic capabilities for weak early fault signals with a signal-to-noise ratio of around -10dB. The advantage of this system is that, combined with the sensitivity and trend of acoustic emission to material degradation and the analysis method of weak characteristics, early warning can be carried out in the early stage of failure, without completely relying on the index threshold.