The problem of automatic reliability monitoring and reliability-centered maintenance is increasingly important today. In this paper, we compare the accuracy of four machine learning approaches for fault detection in a hydraulic system. The first three approaches are based on SVM classifiers with linear, polynomial and RBF kernels and the last one is a gradient boosting on oblivious decision trees. We evaluate algorithms on the synthetic dataset generated by our simulation model of the helicopter hydraulic system and show that high accuracy fault detection can be achieved.
This paper examines the effectiveness of neural network algorithms for hydraulic system fault detection and a novel neural network architecture is suggested. The proposed gated convolutional autoencoder was trained on a simulated training set augmented with just 0.2% data from the real test bench, dramatically reducing the time needed to spend with the actual hardware to build a high-quality fault detection model. Our fault detection model was validated on a test bench and showed accuracy of more than 99% of correctly recognized hydraulic system states with a 10-s sampling window. This model can be also leveraged to examine the decision boundaries of the classifier in the two-dimensional embedding space.
This article discusses a sampling algorithm for machine learning in order to capture the trend of the cumulative deterioration of the characteristics of a hydraulic pump (cumulative degradation), which affects the efficiency of its operation and manifests itself in the form of a drop in volumetric efficiency. To generate data, a simulation model of a typical station for the supply of working fluid in technological complexes, developed in the SimulationX program, is used. The transient processes of pressure change in the system are described, from the analysis of which a tendency of a decrease in the average component of the pressure signal is traced, which is used as a diagnostic feature - an indicator of the state of the system. An example is also considered that describes the possibility of assessing the residual life of the system based on data characterizing the past state of the system, and can be adapted when forming a more complex base, taking into account the use of artificial neural networks.
Gas pressure regulators are widely used in gas transportation and distribution systems. They are designed for deep pressure reduction and maintainance with high accuracy over a wide flow range. Operation at a high pressure drop is accompanied by a high level of noise, for reduction of which, silencers are used. However, installation of a noise suppressor into the regulator design has a significant impact on its static and dynamic characteristics. This can lead to a decrease of accuracy, loss of stability and occurrence of self-oscillations of the valve. These, in turn, lead to increasing noise and vibration, wear of contact surfaces and premature failure of the regulator. The paper presents results of a study of dynamic characteristics of a modernized serial regulator with a built-in noise suppressor. A mathematical model was compiled and its study was carried out in the SimulationX software package. The joint influence on the system stability of the parameters of the muffler and the block of throttles, designed to adjust the static characteristic of the regulator, is considered. It is shown that the proper choice of throttle resistances can ensure the stability of the control system in a wide range of gas flow rates. The results can be used when designing regulators with built-in noise suppressors.
Results of calculations and experimental studies of the acoustic characteristics of the gas pressure regulator with a silencer are reported here. A mathematical model and a method for obtaining a stable solution of the original system of equations was developed. The possibility of reducing the noise of the pressure regulator, when one orifice plate is installed, is presented here. We obtained a good convergence of theoretical and experimental data. The optimization of the flow area silencer consisting of several orifice plates was developed.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.