Hong Mei Liu

Performance Assessment of Hydraulic Servo System Based on Radial Basis Function Neural Network and Mahalanobis Distance

The performance assessment of hydraulic servo systems has attracted an increasing amount of attention in recent years. However, only a few studies have focused on practical approaches in this field. A performance assessment method based on radial basis function (RBF) neural network and Mahalanobis distance (MD) is proposed in this study; the method is quantized by the performance confidence value (CV). An observer model based on RBF neural network is designed to calculate the residual error between the actual and estimated outputs. The root mean square (RMS), peak value, and average absolute value are then extracted as the features of residual error, which serve as the coordinates of the feature points. Lastly, the MD between the most recent feature point and the constructed Mahalanobis space is calculated. The condition of the system is assessed by normalizing MD into a CV. The proposed method is proven to be effective by a simulation model in which leakage faults are injected. Experimental results show that the proposed method can assess the performance of hydraulic servo systems effectively.

Health Assessment and Fault Diagnosis for the Heat Exchanger of the Aircraft Environmental Control System Based on STF-LR

The aircraft environmental control system (ECS) is a critical aircraft system that provides the appropriate environmental conditions to ensure the safe transport of air passengers and equipment. The functionality and reliability of ECS have elicited an increasing amount of attention in recent years. The heat exchanger is a particularly significant component of ECS because its failure reduces the system’s efficiency and can lead to catastrophic consequences. Health assessment and fault diagnosis for the heat exchanger are necessary to perform maintenance and prevent risks in a timely manner. This paper presents fault-related parameter estimation methods based on strong tracking filter (STF) and logistic regression (LR) algorithm for heat exchanger health assessment and root cause classification, respectively. Heat exchanger fault simulation is conducted to generate performance degradation data, through which the proposed methods are validated. Results demonstrate that the proposed methods are capable of providing stable, effective, and accurate heat exchanger health assessment and root cause classification.

Real-Time Life Prediction for Rolling Bearings Based on Nonparametric Bayesian Updating Method

Real-time life prediction for rolling bearings contributes to maintenance decision-making and optimization based on the health state. Real-time life prediction based on Bayesian methods usually require that the priori distribution of the product be obtained; however, this task is extremely difficult to implement for new products or small sample sizes. To solve this problem, a nonparametric Bayesian updating method is proposed in this study. Kernel density estimation is employed to estimate the priori and posterior distribution of parameters by integrating real-time performance degradation information. Thus, bearing real-time life prediction based on nonparametric Bayesian updating is realized. In addition, this study investigates the calculation and normalization process of the working condition conversion factor. The effectiveness of the proposed method is verified by bearing run-to-failure experiments.

Real-Time Reliability Evaluation and Life Prediction for Bearings Based on Normalized Individual State Deviation

Most of the existing methods for bearing real-time reliability evaluation employ real-time transformation of traditional reliability indices, performance degradation trajectory analysis, and performance degradation distribution, which are usually limited in terms of accuracy and applicability. A method for real-time reliability evaluation and life prediction for bearings based on normalized individual state deviation is proposed in this study. First, a self-organizing map neural network is utilized to obtain the individual state deviation of a running rolling bearing. Second, individual state deviation is normalized into a state deviation degree, which is used to formulate a modified real-time reliability model for the realization of real-time reliability evaluation and residual life prediction. The proposed method combines population information with real-time monitoring information of individual bearings, and thus avoids the negligence of the real-time transformation of the monitored individual. The errors caused by the randomness of the individual bearing operational process are also reduced. Finally, the feasibility and efficiency of the proposed method is validated by performing run-to-failure experiments on bearings.

Performance Degradation Prediction for a Hydraulic Servo System Based on Elman Network and Support Vector Regression

Hydraulic servo system is highly nonlinear. Building an accurate model of the system and predicting its remaining life are difficult. Thus, this study focuses on the prediction of the Hydraulic servo System based on Support vector regression (SVR). Elman neural network is utilized to build an observer to estimate the normal state output. The residual that contains a large amount of fault information is obtained, by calculating the difference between the estimated and actual values. Then we defined degradation index (DI) value which reflect the health of the system to normalize the residual. Lastly, a prediction model based on SVR established. The algorithm is verified by experiment.

Feature Extraction Technology for Rolling Bearings Based on Local Tangent Space Alignment

Health assessment and fault diagnosis for rolling bearings mostly adopt traditional methods, such as time-frequency, spectral, and wavelet packet analyses, to extract the feature vector. These methods are suitable for processing data with a linear structure. However, for the non-linear and non-stationary signal, the result of these methods is not ideal. Thus, this study proposes a suitable method to extract the feature vector in nonlinear signals. Local tangent space alignment of a manifold algorithm is employed to extract the feature vector from the rolling bearings. Results verify the advantage of the manifold algorithm for non-linear and non-stationary signals.

Fault Detection for an Aileron Actuator under Variable Conditions Based on Bi-Step Neural Network

Fault detection for aileron actuators mainly involves the enhancement of reliability and fault tolerant capability. Considering the complexity of the working conditions of aileron actuators, a fault detection method for an aileron actuator under variable conditions is proposed in this study. A bi-step neural network is utilized for fault detection. The first neural network, which is employed as the observer, is established to monitor the aileron actuator and generate the residual error. The other neural network generates the corresponding adaptive threshold synchronously. Faults are detected by comparing the residual error and the threshold. In considering of the variable conditions, aerodynamic loads are introduced to the bi-step neural network. The training order spectrums are designed. Finally, the effectiveness of the proposed scheme is demonstrated by a simulation model with different faults.

Health Assessment of Hydraulic Servo System Based on Fractal Analysis and Gaussian Mixture Model

A hydraulic servo system is a typical feedback control system. Health assessment of a hydraulic servo system is usually difficult to realize when traditional methods based on sensor signals are utilized. An approach for health assessment of hydraulic servo systems based on multi-fractal analysis and Gaussian mixture model (GMM) is proposed in this study. A GRNN neural network is employed to establish a fault observer for the hydraulic servo system. The observer is utilized to simulate the system output under normal state. The residue is then generated by subtracting the estimated output from the actual output. The residue’s feature is extracted by fractal analysis. After the feature extraction, the overlap between the current feature vectors and the normal feature vectors is obtained by applying GMM. The confidence value (CV) can be obtained in advance; this value is employed to characterize the health degree of the current state and consequently implement the health assessment of the hydraulic servo system. Lastly, two common types of fault, namely, burst and gradual, are applied to validate the effectiveness of the proposed method.

Fault Detection and Isolation for Hydraulic Servo System Based on Adaptive Threshold and SOM Neural Network

A fault detection and diagnosis method for the hydraulic servo system based on adaptive threshold and self-organizing map (SOM) neural network is proposed in this study. The nonlinear, time-varying, fluid-solid coupling properties of the hydraulic servo system are considered. Fault detection is realized based on a two-stage radial basis function (RBF) neural network model. The first-stage RBF neural network is adopted as a fault observer for the hydraulic servo system; the residual error signal is generated by comparing the estimated observer output with the actual measurements. To overcome the drawback of false alarms when the traditional fixed fault threshold is used, an adaptive threshold producer is established by the second-stage RBF neural network. Fault occurrence is detected by comparing the residual error signal with the adaptive threshold. When a system fault is detected, the SOM neural network is employed to implement fault classification and isolation by analyzing the features of the residual error signal. Three types of common faults are simulated to verify the performance and effectiveness of the proposed method. Experimental results demonstrate that the proposed method based on adaptive threshold and SOM neural network is effective in detecting and isolating the failure of the hydraulic servo system.

Bearing Performance Degradation Signal Simulation Based on Health Confidence Value Injection and FPGA

A digital simulation method for the performance degradation signal of rolling bearings is developed based on the analysis of experimental data. A self-organizing map neural network is utilized to build the performance degradation assessment model of the rolling bearings based on characteristic parameter extraction. Wavelet packet decomposition is then implemented to extract the wavelet coefficients in the corresponding performance degradation sensitive band. Different health confidence values are injected into the extracted wavelet packet coefficients, and signals are reconstructed according to the simulation needs to obtain rolling bearing vibration data under different degradation degrees. Understanding the exact mathematical model of the measured object is unnecessary in this method; the method is simple and reliable and helps solve the problem of performance degradation data simulation. Finally, an FPGA-based performance degradation signal simulator is designed by combining the analogy procedure, employed to support the verification process of fault diagnosis and prediction capability.