Detection and Classification of Wear Fault in Axial Piston Pumps: Using ANNs and Pressure Signals

Detection and Classification of Wear Fault in Axial Piston Pumps: Using ANNs and Pressure Signals

Jessica Gissella Maradey Lázaro (Universidad Autónoma de Bucaramanga, Colombia) and Carlos Borrás Pinilla (Universidad Industrial de Santander, Colombia)
Copyright: © 2020 |Pages: 31
DOI: 10.4018/978-1-7998-1839-7.ch012

Abstract

Variable displacement axial piston hydraulic pumps (VDAP) are the heart of any hydraulic system and are commonly used in the industrial sector for its high load capacity, efficiency, and good performance in the handling of high pressures and speeds. Due to this configuration, the most common faults are related to the wear and tear of internal components, which decrease the operational performance of the hydraulic system and increase maintenance costs. So, through data acquisition such as signals of pressure and the digital processing of them, it is possible to detect, classify, and identify faults or symptoms in hydraulic machinery. These activities form the basis of a condition-based maintenance (CBM) program. This chapter shows the developed methodology to detect and classify a wear fault of valve plate taking into account six conditions and the facilities providing by wavelet analysis and ANNs.
Chapter Preview
Top

Introduction

Reliability and safety of industrial processes, with an efficient asset management, have taken a very important role to achieve greater competitiveness and productivity of global companies which working hard to ensure reliability, quality services, safety and maintainability of hydraulic equipment. The manufacturer's usual advice regarding with maintenance is to implement a management strategy of maintenance that includes preventive and predictive activities (Aly, 2015), whose results providing high reduction in maintenance and operational costs, increasing the productivity of the processes, as well as the stability of the system in general. On the other hand, methodologies for on-line and off-line fault diagnosis have been developed for this purpose, allowing to taking actions that avoid either long unscheduled stops or permanent damage of machine’s components and as well as preserve their useful life. Hydraulic axial piston pumps are commonly used for high fluid power applications, which are severely affected by valve plate wear failure.

Wavelet analysis has been extensively studied for feature extraction (Guillén, Paredes & Quintero, 2004; Peng & Chu, 2004), which has shown an advantage in the processing of non-linear and non-stationary signals. This allows the management in the time and frequency domain and the decomposition of the series in terms of the approximation coefficients (cA) and detail (cD), separating the information into low and high frequencies respectively. The generalized Wavelet Packet Transform (WPT) algorithm is also a very effective method to apply, which after the decomposition of the signal allows extracting and normalizing the characteristic vectors according to the energy of each component (Tao, Lu, Lu, & Wan, 2013; Tao, Wang, Ma & Fan, 2012).

Furthermore, artificial neural networks (ANNs), has proven their capacity of training - learning, self-organization, accuracy, convergence, performance and fault tolerance (Lázaro, Pinilla & Prada, 2016), as well as their robustness in the solution of complex and nonlinear dynamic systems. ANNs have been especially used in tasks of patterns recognition problems, diagnosis and classification, model predictive faults, to handle analytical and heuristic information, incomplete data among others.

Therefore, combining ANN technique with wavelet data analysis achieve a methodology that enables to detect and classify wear failure in an automatic and effective way. The objectives of this chapter is to show every step followed in the proposed methodology, analyze the results obtained as of pressure signals taken of five wear conditions and verify the performance of test stage.

Complete Chapter List

Search this Book:
Reset