This study focused on the measurement and diagnostic analysis of partial discharge in power capacitors operating under prolonged periods of high voltage. An off-line test was conducted on two power capacitors, which were subjected to defect preprocessing before the experiment. The partial discharge test of the terminal to the outer casing was carried out using a 24kV high voltage transformer. The voltage of the power discharge, after which the voltage signals of the test voltage and the partial discharge were extracted. The characteristics of the discharge signal were identified using an oscilloscope. The chaos synchronization detection method was then employed to establish a chaotic error scatter plot for the discharge voltage he gravity (chaos eye) of the error scatter plot was used as the fault diagnosis feature. The advantage of this approach was the drastic reduction of the amount of data for feature extraction, and the effective detection of the slight discharge signal variation in power capacitors, thereby anticipating their operation and enabling the implementation of preemptive contingency measures to avoid major accidents.