Anomaly detection technique based on signal processing for structural durability testing of an excavator component is presented in this study. The proposed method utilizes the time-series skewness of output strain signals and input force as the features. Anomaly index for each strain gauge sensor is calculated in the feature space based on the distance from the data cluster at healthy condition. Results showed that the time and position when and where crack presumably occurred were successfully specified with the proposed anomaly detection technique.