Fault Detection and Classification in Cantilever Beams Through Vibration Signal Analysis and Higher-Order Statistics
- PDF / 1,117,119 Bytes
- 7 Pages / 595.276 x 790.866 pts Page_size
- 47 Downloads / 192 Views
Fault Detection and Classification in Cantilever Beams Through Vibration Signal Analysis and Higher-Order Statistics Tássio S. Barbosa2 · Danton D. Ferreira1 · Daniel A. Pereira1 · Ricardo R. Magalhães1 · Bruno H. G. Barbosa1
Received: 21 November 2015 / Revised: 20 April 2016 / Accepted: 15 June 2016 © Brazilian Society for Automatics–SBA 2016
Abstract A method for detecting and classifying faults in an aluminum cantilever beam is proposed in this paper. The method uses features based on second-, third- and fourthorder statistics, which are extracted from the vibration signals generated by the cantilever beam. Fisher’s discriminant ratio (FDR) is used for feature selection, and an artificial neural network is used for fault detection and classification. Three different degrees of faults (low, medium and high) were applied to the cantilever beam, and the proposed pattern recognition system was able to classify the faults, reaching performances ranging from 88 to 100 %. Moreover, the use of higher-order statistics-based features combined with FDR led to a compact feature space and provided satisfactory results. Keywords Cantilever beam · Vibration analysis · Higher-order statistics
1 Introduction There is an increased need for the early detection of structural faults or damage, i.e., the field of structural health monitoring (SHM). The majority of research performed in this area has been based on applying signal processing techniques, par-
B
Danton D. Ferreira [email protected]
1
Engineering Department, Federal University of Lavras, Lavras, MG, Brazil
2
CEFET-MG, Campus IX Nepomuceno, Nepomuceno, MG, Brazil
ticularly those from the pattern recognition field, to analyze vibration signals (Farrar and Worden 2012). This topic is also related to predictive maintenance, also known as conditionbased maintenance (Liao and Wang 2013; Compare and Zio 2014). In this context, the analysis of vibration signals has been efficiently used for the early detection of structural damage in industrial machines (Farrar and Worden 2012; Zarei et al. 2014; Abed et al. 2015; Schmitt et al. 2015). Most industrial equipment vibrates during operation, and the vibration signals can provide valuable information regarding the different operating stages and can be used to indicate the occurrence of damage. According to Farrar and Worden (2012), a small damage in a structure changes its structural properties, such as stiffness and mass, and its dynamic properties, such as natural frequencies, vibration modes and damping rate. Therefore, the presence of a fault or damage modifies the vibratory response of a system. Many components of machines, vehicles, and structures are subjected to cyclic loadings or are under the influence of many other sources of vibration. In general, these vibration sources do not induce stresses that exceed the material’s yield point. However, if they persist for a long period of time, structural damage can occur, most of which are losses in the structural stiffness (Cawley and Adams 1979). In this context, many
Data Loading...
