Electric Submersible Pump (ESP) is currently widely employed to help enhance production for nonlinear-flowing well with high production and high water cut well. However, ESP broken shaft is common in the oil industry. The broken shaft leads to production disruptions, resulting in significant economic losses. The objective of this paper is to evaluate Principal Component Analysis (PCA) as an unsupervised machine learning technique to detect the cause of the breakage of the ESP shaft. This method was successfully applied in the Penglai block of Bohai Oilfield in China to detect the ESP shaft fracture in real-time. A two-dimensional plot of scores of Principal Component 1 and Principal Component 2 can be used to identify different clusters of the stable region, unstable region and failure region. By this means, potential ESP shaft fracture will be found when the cluster starts deviating away from the stable region. Moreover, a PCA diagnostic model is built to predict the time at which the ESP shaft fracture occurs and to determine the main decision variable most responsible for ESP broken shaft. This paper demonstrates that the application of the PCA method performs well in monitoring the ESP operation system and predicts the impending breakage of ESP shaft with high accuracy.