J. Xiang

Reliability analysis for wind turbines with incomplete failure data collected from after the date of initial installation

Reliability has an impact on wind energy project costs and benefits. Both life test data and field failure data can be used for reliability analysis. In wind energy industry, wind farm operators have greater interest in recording wind turbine operating data. However, field failure data may be tainted or incomplete, and therefore it needs a more general mathematical model and algorithms to solve the model. The aim of this paper is to provide a solution to this problem. A three-parameter Weibull failure rate function is discussed for wind turbines and the parameters are estimated by maximum likelihood and least squares. Two populations of German and Danish wind turbines are analyzed. The traditional Weibull failure rate function is also employed for comparison. Analysis shows that the three-parameter Weibull function can obtain more accuracy on reliability growth of wind turbines. This work will be helpful in the understanding of the reliability growth of wind energy systems as wind energy technologies evolving. The proposed three-parameter Weibull function is also applicable to the life test of the components that have been used for a period of time, not only in wind energy but also in other industries.

Direct inversion of the apparent complex-resistivity spectrum

Cole‐Cole model parameters are widely used to interpret electrical geophysical methods and are obtained by inverting the induced polarization (IP) spectrum. This paper presents a direct inversion method for parameter estimation based on multifold least-squares estimation. Two algorithms are described that provide optimal parameter estimation in the least-squares sense. Simulations demonstrate that both algorithms can provide direct apparent spectral parameter inversion for complex resistivity data. Moreover, the second algorithm is robust under reasonably high noise.

Smart Monitoring of Wind Turbines Using Neural Networks

In this paper, we present a neural networks model-based method to monitor the condition of a generator bearing in a wind turbine. The temperature of the generator bearing is modelled as a function of the generator speed in the wind turbine. The difference between the temperature measurement and the model is calculated and compared to logbook records to verify the method. Other methods are also given for comparison.

Wind turbine availability analysis based on statistical data

Availability is an important performance index for wind turbines. To predict wind turbine availability, failure rate and repair rate have to be known. There are some sources of wind turbine failure data that can be used to estimate parameters of the failure rate function and the repair rate. With repair rate assumed to be constant, this paper first presents maximum likelihood estimates for failure rate, then gives a method to predict instantaneous and long term availability for wind turbines. By keeping a record of failure time for each subassembly of a wind turbine, maintenance could be more effectively targeted with more detail in terms of what is likely to fail. Failure data from a wind farm is analyzed as an example, which shows results predicted by the technique presented are close to the wind farm statistics. This work will be helpful in planning timely and cost-effective maintenance of wind turbines.

An algorithm for common quadratic Lyapunov function

Using Lyapunov mapping, in this paper a necessary and sufficient condition for a set of linear systems to share a common quadratic Lyapunov function is presented. Based on this condition an algorithm has been developed to search such a Lyapunov function. Theoretical proof for the executability of the algorithm is completed. The algorithm has been coded as a software, which shows the algorithm is very powerful. Several numerical examples, that are produced by the software, are presented to support the algorithm.

Control requirements of direct drive wave energy converters

In a direct drive wave energy converter the electrical generator is directly coupled to the reciprocating motion of the waves, resulting in a variable frequency, variable voltage output with zero power dips within each wave cycle. Variable reluctance permanent magnet machines are capable of reacting the large forces required in a direct drive wave energy converter, but they suffer from low power factor. A power control strategy is presented for optimizing the electrical power from VRPM machines and for satisfying the grid requirements of constant voltage, frequency and power. Frequency tuning, control of wave energy devices using the electrical system is demonstrated by simulation of a test rig designed to emulate a wave energy converter.

Theoretical analysis and experimental investigation of induced polarization effect of rocks/minerals

This paper considers the polarization phenomena and measurement techniques of rock materials from the view of system identification and parameter estimation. Induced polarization method is used to measure the rock materials, which is a scheme for using electric fields in the Earth to estimate physical structure. These fields are generated by the deliberate injection of electrical currents into the Earth. The effective conductivity of rock materials for alternating current is not in general constant, but is variable and complex. The complex conductivity in the frequency domain gives rise to the overvoltage or induced polarization effect in the transient time domain. However it is difficult to discriminate between a valid induced polarization (IP) response and electromagnetic (EM) coupling effects, which are caused by IP measurement and the complexity of the rock materials. Therefore we use a system identification method to find an adequate IP system model. Summarizing the analysis of the IP system model, it is concluded that the proposed IP system model fits the field data much better than traditional models. EM coupling effects can be removed using system identification method and the induced polarization effect in the transient time domain can be presented in Simulink via the IP system model.

Signal processing using Matlab and control toolbox for linear electrical machine modelling

This paper presents the method of filter design using Matlab, Simulink, and control toolbox. Further, a derivation of an armature circuit model is used to investigate the performance of a linear permanent magnet electrical machine. The armature circuit model is estimated through system identification techniques and digital filter technologies from the data of a 2D finite element model that is generated to determine the flux linkage map for a prototype made at Durham University. To explain the modelling process, in this paper, we demonstrate a Simulink model that includes a look up table for the 2D finite element model and digital filters. The models have been experimentally verified.

Batch signal processing using Matlab and Excel

This paper presents a general method of calculation and numerical treatment of batch experimental data using Matlab and Excel. The data sets used are results from experimentation with a linear generator connected to an active inverter. The results suffer from high frequency noise that is removed using harmonic analysis. Although the method is used for the calculation of electrical power from current and voltage results, it can equally be applied to other large data sets collected.