Application of functional link artificial neural networks constructed with least squares support vector machine in fault diagnosis of rolling bearings

作者：Sun, Lin^*; Yang, Shi Yuan

来源：Proceedings of the CSEE, 2010, 30(8): 82-87.

• Diagnosis systems
• Functional link artificial neural networks
• Learning models
• Least squares support vector machines
• Rolling bearings
• Strong robustness
• System diagnosis
• Time consumption

A diagnosis system for rolling bearing fault is presented based on functional link artificial neural networks (FLANN) inverse system constructed with least squares-support vector machine (LS-SVM). The principle and algorithms are introduced and the system diagnosis model for rolling bearing fault is presented. First, the model avoids the randomness for choosing external function because it replaces the external function of common FLANN function with the kernel function of LS-SVM. Second, the FLANN model weight coefficients can be obtained with LS-SVM learning model, avoiding the disadvantages embodied by the BP method such as long time-consumption, local minimum, and dependence on experience while setting initial values. Last, a multi-layer LS-SVM-FLANN structure is built to diagnose the long-time consumption rolling bearing faults. The result of experiment shows that the diagnosing method for rolling bearing fault with LS-SVM-FLANN has the characteristics of high precision, strong robustness, and is easy to realize.

• 出版日期2010
• 单位合肥工业大学