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.
The article deals with diagnostics of hydraulic systems using phase portraits. A brief review of the existing methods for diagnosing hydraulic units identifying their advantages and disadvantages is given. An approach based on the analysis of dynamic characteristics of a hydraulic system and phase portraits of hydro-mechanical units in their operational and faulty conditions is proposed. As an example, we consider a dynamic model of a simplified hydraulic system consisting of standard components. By adjusting the model parameters characteristic faults typically occurring in operation, such as internal leaks in the pump, contamination of the hydraulic fluid with mechanical impurities, sticking of the valve, etc. were artificially introduced in hydro-mechanical units. A family of phase portraits of a hydraulic system for the operational condition and various faulty ones was constructed. A quantitative estimate of their changes, based on calculating the difference in the areas of the figures restricted by their graphs, is proposed. As a result, it was established that failures and malfunctions introduce changes in the phase portraits of hydro-mechanical units, which makes it possible to apply the proposed approach as a basis for diagnosing the technical condition of hydraulic systems.
А method based on comparing oscilloscope patterns of operational parameters with reference curves is one of the most promising methods of diagnosing hydraulic systems among the existing ones. Its implementation does not allow accurate localization of the faulty unit in the system and quantitative estimation of the magnitude of the fault. To eliminate these shortcomings, it is advisable to use simulation models of hydraulic units, taking into account typical faults of a hydraulic system. Their use makes it possible to evaluate the effect of a particular malfunction on the change of dynamic parameters at the stage of mathematical modeling. As a result of the analysis of statistical information and literary sources, characteristic faults of hydraulic systems are identified. Their causes and the impact on the operation of hydraulic units are examined. Simulation models of units taking into account typical faults are described in the Matlab / Simscape software package. They are implemented using a typical hydraulic system as an example. Dynamic characteristics of a hydraulic system in a healthy condition and those of a system with one of the characteristic faults are compared.
It is necessary to ensure appropriate information content of the measuring instruments used for intelligent diagnosing systems of energy and technological complexes based on the measurement of dynamic parameters. Sensors and measuring equipment should possess sufficient accuracy, reliability, speed and consistency of performance. Types of sensors for measuring dynamic parameters are selected depending on the systems structure. They can be, for example, sensors for the electrohydromechanical systems of these complexes, pressure sensors, as well as sensors of flow and temperature of the working media, displacement of moving elements and vibration of the base members. The type of sensor intended for use in the diagnostic system is largely determined by the dynamics of the processes taking place in it. It is necessary that the sensors satisfy their performance requirements. If the sensors have lower speed than is necessary for the process dynamics in the electrohydromechanical system, it can lead to dynamic measurement error and an error in the diagnostics of technical condition. In technical literature, the requirement for the sensor speed is indicated by the fact that it should be an order of magnitude higher than the dynamics of the processes occurring in the system. This approach is not acceptable for choosing the type of sensors for diagnostic systems, taking into account the process dynamics. Firstly, sensors for measuring with this required parameter may not be available. Secondly, even if there is a sensor with a parameter close in speed to the dynamics of the system processes, it is necessary to know in advance what dynamic error it can lead to and how it will affect the accuracy of the diagnostic system. An analytically generalized dependence of the dynamic measurement error of electrohydromechanical system parameters on the relative sensor speed is obtained in this paper. This dependence allows you to choose a sensor with a dynamic error that does not exceed a given value. The calculation of the dynamic measurement error is shown using the MI-8 helicopter hydraulic system as an example.
In order to increase efficiency of diagnostics of electro-hydro-mechanical systems (EHMS) it is advisable to have simulation models of typical faults. Such approach makes it possible to estimate in advance, even at the stage of mathematical modeling, the impact of different faults on functioning of hydraulic systems. This work is aimed at creating a database containing complexes of diagnostic features, which allow distinguishing types of faults, their causes and stages of development. In the paper, typical faults of EHMS are presented on the basis of statistical information from literature sources and experimental research. They include internal and external leakages, spool and sleeve sticking, degradation of power fluid. The causes of faults and their impact on hydraulic systems functioning are considered. Simulation models of typical faults are implemented and studied in the SimulationX software package. The static and dynamic characteristics of the systems are investigated in order to identify diagnostic signs of various faults. The impact of typical faults on various system parameters is discussed. During the research, the tasks of selecting the rational location of sensors of different types (pressure, flow, displacement, or force sensors), their quantity for recognition of a typical fault are solved. The results of theoretical and experimental studies of serviceable and faulty systems for cases of control and disturbance actions are presented. Comparative analysis of transient processes of serviceable and faulty EHMS is presented with assessment of difference between theoretical and experimental data. The results of the work allow to more rationally designing the diagnostic complex for more accurate identification of the type of fault, stage of its development and prediction of residual service life of EHMS.
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.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.