Fault Diagnosis Method for Industrial Robots based on Dimension Reduction and Random Forest

With the development of modern industry, sophisticated and complex industrial robots have been used more and more widely. However, it is still difficult to carry out accurate and efficient fault detection for robots. This paper applies a method by combining dimension reduction and Random Forest. In this method, vibration signals are acquired by using accelerometers, current sensors as well as angle encoder, and processed to extract multi-features according to time domain and frequency domain. After some pre-process steps, dimension reduction methods are applied with an optimized threshold. Three different classifiers have been evaluated: the Support Vector Machine (SVM), the Random Forest (RF), the eXtreme Gradient Boosting (XGBoost) and RF proves itself the best classifier.