A methodology for detection of wear in hydraulic axial piston pumps

Abstract

An effective asset management has a direct impact on maintenance costs, reliability, and equipment availability, especially in hydraulic machinery. Variable displacement axial piston pump is a major component used in the industry due to its load capacity ratio, pressure management, and high performance. Some of the main faults are wear and abrasion of the valve plates, increasing pressure losses as well as temperature, decreasing volumetric efficiency, and abnormal vibration. The off-line methodology implemented includes preprocessing of the vibration signals taken from the test bench available for this study, the feature extraction using wavelets, a stage of detection and classification through the use of artificial neural networks. Several networks were assessment, such as Adaline, nonlinear, and multilayer perceptron networks. Classification percentages greater than 90% are obtained taking into consideration 5 wear conditions related to the loss of volumetric efficiency.