Dani Juricic

Fuel Cell Condition Monitoring System Based on Interconnected DC–DC Converter and Voltage Monitor

This paper presents a comprehensive solution for condition monitoring (CM) of proton exchange membrane (PEM) fuel cell systems. It comprises a modular dc-dc converter, a 90-channel fuel cell voltage monitor, and embedded diagnostics algorithms. Besides providing its basic functionality, the dc-dc converter is designed to perform diagnostic probing of fuel cells by injecting current excitation waveforms. The designed voltage monitor enables the precise voltage measurement of all individual cells in the stack. The interconnection between the dc-dc converter and the voltage monitor provides a platform for an embedded CM system. It employs a novel algorithm for fast electrochemical impedance estimation and automatic tuning of the thresholds used in the fault detection algorithm. The final output is a unit-free condition indicator that describes the overall condition of the fuel cell stack. As such, the designed CM system allows the seamless integration and optimal exploitation of fuel cell power systems. The complete solution has been evaluated on an 8.5-kW PEM fuel cell power system.

An end-quality assessment system for electronically commutated motors based on evidential reasoning

Abstract Automatic end-quality assessment is a mean that helps reaching zero-fault products at the end of the manufacturing process. In this paper we present a system for assessing the quality of electronically commutated motors. The system consists of two major parts: feature extraction and overall quality assessment. The feature extraction part consists of signal processing algorithms tailored for mechanical fault detection. The quality assessment part, aimed for fault isolation and final quality decision, employs evidential reasoning for multi-attribute decision analysis. A prototype version of the system is validated on a test batch of 130 electronically commutated motors, demonstrating high diagnostic resolution and accuracy.

Prediction of the remaining useful life: An integrated framework for model estimation and failure prognostics

Machine failure prognostic is concerned with the generation of long term predictions and the estimation of the probability density function of the remaining useful life. Nowadays, a commonly used approach for this task is to make the prediction using a dynamical state-space model of the fault evolution. However, the main limitation of this approach is that it requires the values of the model parameters to be known. This work aims to alleviate the need for extensive prior efforts related to finding the exact model. For this we propose a framework for data-driven prediction of RUL with on-line model estimation. This is achieved by combining the state estimation algorithm with Maximum-Likelihood parameter estimation in the form of the Expectation-Maximization algorithm. We show that the proposed algorithm can be used with different classes of both black-box and grey-box models. First, a detailed solution for linear black-box models with the Kalman filter is presented followed by the extension to nonlinear models using either the Unscented Kalman filter or the particle filter. The performance of the algorithms is demonstrated using the experimental data from a single stage gearbox.

Condition prognosis of mechanical drives based on nonlinear dynamical models

Forecasting the condition of the equipment is becoming an important ingredient of the advanced maintenance and asset management systems. In this paper a probabilistic approach to the prognosis of damage progression in gearboxes is presented. It is based on a stochastic nonlinear grey-box model of the underlying wear phenomena. Model parameters are estimated from the available vibration records by using an iterative Maximum Likelihood procedure. The procedure relies on the Expectation-Maximization algorithm and the Unscented Kalman filter for estimation of hidden system states. The algorithm has been used to predict the normal operating horizon of a single-stage gearbox system. Several test runs of the system have been preformed to validate the algorithm.

Development of an Intelligent Rotating Machinery Diagnostics Programme

Rotating machines are one of the most wide spread items of equimpnet in the industrial plants; hence the reliable operation is of great practical importance. Analyses show that when a run-to-failure philosophy is adopted in rotating machinery maintenance, their downtime is usually three to four times longer comparing to a periodic or proactive maintenance approach. A successful proactive maintenance program requires an integration of several diagnostic procedures into an intelligent data processing system. Such a system allows detection of a broad range of faults in an early stage. The main aim of this paper is to present current results of our development of an intelligent rotating machinery diagnostics program for detecting a broad range of faults from signals which can be measured non-destructively and on-line. The main motivation is to develop computationally efficient algorithm that can be implemented on a standard (low-cost) platform. In that respect we have developed a test rotating machine equipped with accelerometers, temperature sensors and sensors for lubricating oil characterization. In this paper we focus on gear-box faults and a feature extraction procedure based on non-parametric statistical concepts as suggested and demonstrated on experimental data.Copyright

System identification of nonlinear dynamical models: Application to wastewater treatment plant

The paper presents a novel approach to identification of stochastic nonlinear dynamic systems using efficient approximation methods. The motivation behind this work is to develop a computationally efficient and robust algorithm for estimation of wastewater treatment plant model parameters. The mathematical model of the plant is required for the application of advanced predictive control algorithms and condition monitoring. The presented algorithm employs the Expectation-Maximization algorithm to compute the Maximum likelihood estimates of the unknown model parameters. The algorithm uses the Unscented Transformation (UT) to approximate the posterior distribution of the random variable that undergoes a nonlinear transformations. The advantage of this approach lies in efficient approximation methods that greatly reduce the computational load of the algorithm and is therefore suitable for on-line implementation.

A System for Automated Final Quality Assessment in the Manufacturing of Vacuum Cleaner Motors

In this paper we present a system for fault detection and isolation of electrical motors for vacuum cleaners, which serves for on-line quality assessment at the end of the production line. The main focus of the paper is on detection of incipient mechanical faults by means of sound analysis. A detailed description of the procedures for detection and isolation of faults with mechanical origin is given. One of the contributions concerns innovative implementation of the feature extraction algorithm in a way that spectral procedures used in system analysis are substituted by much simpler and computationally more effective algorithms. Herewith the system is able to accurately isolate three different types of mechanical faults. Additional contribution regards diagnostic system integration which generates highest product end quality and traceability. In order to fully utilize a huge amount of diagnostic data, possibilities for the upgrade of the present diagnostic system with advanced production line supervision support are indicated.

Using Filter Diagonalization for Fault Detection in Low-Speed Rotational Machinery

: The paper presents the application of Filter Diagonalization (FD) method for fault detection in low-speed rotational machinery. FD serves to extract frequency components from short time signals. The length of the signal can be significantly shorter than the length required by methods based on computing the similarity between the signal and the sine wave, e.g. Fourier Transform (FT) and Wavelet Transform (WT). Brief mathematical background of the FD method is outlined along with a short comparison of its performance with respect to the FT. Finally, an example of utilization of FD for fault detection is described.