Maximum likelihood estimation of synchronous machine parametersfrom flux decay data

A time-domain system identification procedure to estimate the parameters of a 5 kVA salient pole synchronous machine from standstill test measurements is proposed. The test consists of a DC flux decay signal applied to the d-axis and q-axis of the machine. From the recorded responses to this signal, the admittance transfer function models and the standstill frequency response equivalent circuit models are identified. The maximum likelihood algorithm is used to estimate the model parameter values, and the Akaike Criterion is used to select the best-fit model. The performance of the standstill models in the dynamic environment is studied through simulation of an on-line small-disturbance test. The results are compared with measured data     

Sliding-Mode Flux Observer With Online Rotor Parameter Estimation for Induction Motors

Field orientation techniques without flux measurements depend on the parameters of the motor, particularly on the rotor resistance or rotor time constant (for rotor field orientation). Since these parameters change continuously as a function of temperature, it is important that the value of rotor resistance is continuously estimated online. A fourth-order sliding-mode flux observer is developed in this paper. Two sliding surfaces representing combinations of estimated flux and current errors are used to enforce the flux and current estimates to their real values. Switching functions are used to drive the sliding surfaces to zero. The equivalent values of the switching functions (low-frequency components) are proven to be the rotor resistance and the inverse of the rotor time constant. This property is used to simultaneously estimate the rotor resistance and the inverse of the time constant without prior knowledge of either the rotor resistance or the magnetizing inductance. Simulations and experimental results prove the validity of the proposed approach     

Neural network based modeling of a large steam turbine-generatorrotor body parameters from on-line disturbance data

A novel technique to estimate and model rotor-body parameters of a large steam turbine-generator from real time disturbance data is presented. For each set of disturbance data collected at different operating conditions, the rotor body parameters of the generator are estimated using an output error method (OEM). Artificial neural network (ANN) based estimators are later used to model the nonlinearities in the estimated parameters based on the generator operating conditions. The developed ANN models are then validated with measurements not used in the training procedure. The performance of estimated parameters is also validated with extensive simulations and compared against the manufacturer values     

Color image segmentation with genetic algorithm in a raisin sorting system based on machine vision in variable conditions

This study was undertaken to develop machine vision-based raisin detection technology for various lighting conditions. Supervised color image segmentation using a permutation-coded genetic algorithm (GA) identifying regions in hue–saturation–intensity (HSI) color space (GAHSI) for desired and undesired raisin detection in various conditions was successfully implemented. Images from two extreme intensity lighting and dense conditions: under weak lighting and high-density product and under suitable lighting and low-density product, were mosaicked to explore the possibility of using GAHSI to locate desired raisin and undesired raisin regions in color space when these two extremes were presented simultaneously. The GAHSI results provided evidence for the existence and separability of such regions. In the experiment, GAHSI performance was measured by comparing the GAHSI-segmented image with a corresponding hand-segmented reference image. When compared with cluster analysis-based segmentation results, the GAHSI method showed no significant difference.     

The effects of noise on frequency-domain parameter estimation ofsynchronous machine models

The authors present results of a study conducted to evaluate the effects of measurement noise on the estimation of machine parameters from standstill frequency response test data. Results obtained indicate that, because the test data are inherently noise-corrupted, multiple solution sets can be obtained. Moreover, some of the estimated machine parameters could be unrealistic. It is shown that the results are very sensitive to the value of armature resistance used in the data analysis. Even a 0.5% error in the value of armature resistance could result in unrealistic estimation of the machine parameters     

On-line evaluation of a round rotor synchronous machine parameterset estimated from standstill time-domain data

This paper presents a method for identifying the best linear circuit model parameters of a three-phase, round rotor synchronous machine. Standstill time-domain test data and the maximum likelihood estimator are used to identify the values for the equivalent circuit models. The estimated models are validated against standstill data and an online test. A steady-state error adjustment procedure is introduced and the results are analyzed. The final dand q-axis model selections are based on the minimization of the cost function, the concept of parsimony and how well the models predict the online dynamics of the machine. Issues related to the necessity of the L_fld differential leakage inductance and the necessity of the Z_fe eddy current branch are also discussed     

Control of distributed generation systems - Part II: Load sharing control

This work is concerned with the control strategy for the parallel operation of distributed generation systems (DGS) in a standalone ac power supply. The proposed control method uses only low-bandwidth data communication signals between each generation system in addition to the locally measurable feedback signals. This is achieved by combining two control methods: droop control method and average power control method. The average power method with slow update rate is used in order to overcome the sensitivity about voltage and current measurement errors. In addition, a harmonic droop scheme for sharing harmonic content of the load currents is proposed based on the voltages and currents control algorithm. Experimental and simulation studies using two parallel three-phase pulsewidth modulation (PWM) inverters are presented to show the effectiveness of the proposed control.     

Implementation of indirect vector control on an integrated digitalsignal processor-based system

This paper presents an implementation of indirect vector control of an induction machine on an integrated DSP (digital signal processor) system manufactured by dSPACE GmbH. The system integrates into a single board the computational power of a TMS320C31 DSP with extra peripherals needed in vector control application, and therefore requires minimal hardware development. The induction machine parameters required for the vector control operation are obtained through an offline parameters identification using the maximum likelihood estimation technique with DC voltage source excitation. It is shown through extensive experimental study that the offline identified parameters yield in a reliable field orientation of the induction machine     

Neural network based modeling of round rotor synchronous generatorrotor body parameters from operating data

It is generally accepted that in order to account for the effect of eddy currents in the solid rotor-iron of a round-rotor synchronous machine, two or more fictitious rotor-circuits are to be used in each axis of the d- and q-axis equivalent circuit representations of the machine model. This paper presents a novel technique to estimate the parameters of these rotor-circuits (hereinafter referred to as rotor body parameters) from measurements collected online at several operating conditions. The effects of generator saturation, rotor position and loading are included in the estimation process. Tests conducted on a round-rotor synchronous generator reveal that certain rotor-body parameters are nonlinear functions of generator operating condition. A novel artificial neural network (ANN) based technique is used to map variables representative of generator operating condition to each parameter being modeled. The developed ANN models are validated with measurements not used in the modeling process