Search

This paper presents a novel speed estimation scheme for induction motors (IMs) based on back electromotive-force model reference adaptive system (back-EMF MRAS). The scheme is employed for the purpose of sensorless fault-tolerant torque-controlled drives used in a limp-home mode operation in electric vehicle (EV) applications. The proposed scheme was experimentally tested on a laboratory dynamometer using a 19-kW IM and a 29-kW controller, which are both currently used in the automotive industry for EV applications. The scheme was also implemented on an electric golf buggy which was equipped with a 5-kW IM. A performance comparison was carried out between the proposed and conventional back-EMF MRAS schemes for starting from standstill, sensitivity to parameter variations and constant speed operation with load variations. Utilizing the golf buggy, the behaviors of the new scheme was separately investigated for vehicle starting from standstill, wide speed range including field weakening region, and hill-starting operations. The proposed scheme is computationally easy to implement, robust against sensitivity to parameters variations, inverter nonlinearity and errors due to digitization in the field weakening region. This scheme is not only consistent for vehicle starting from standstill, it also provides a reliable vehicle-drive in the field weakening region and during vehicle hill-starting. The dynamometer and vehicle test-drive results show the suitability of the proposed scheme for the purpose of EV fault-tolerant limp-home mode operation. [ABSTRACT FROM PUBLISHER]

An induction generator always absorbs lagging reactive power from the power network. In most previous power system optimization studies involving induction generator based distributed generation, this generator reactive power demand is calculated using an approximate empirical formula. In addition, shunt compensation capacitors have not been considered as part of the network optimization study. In this article, the use of the per phase equivalent circuit of the induction generator is proposed to compute its reactive power requirement, thus providing a more accurate estimation. A discrete particle swarm optimization algorithm is then employed to address the problem of simultaneous integration of the induction generators and shunt compensation capacitors. The proposed algorithm has the advantage of being able to cope with a mixed search space optimization problem of integer, discrete, and continuous variables. The study is carried out on a standard 69-bus benchmark distribution network. Results show that the use of the approximate empirical formula leads to an underestimation of the machine reactive power demand. The inclusion of shunt compensation capacitors in the optimization process results in lower network reactive power flows with improved network power losses, improved voltage profile, and increased levels of distributed generation integration. [ABSTRACT FROM PUBLISHER]

Abstract—An optimization algorithm based on a novel discrete particle swarm optimization technique is proposed in this article for optimal sizing and location of distributed generation in a power distribution network. The proposed algorithm considers distributed generation size and location as discrete variables substantially reducing the search space and, consequently, computational requirements of the optimization problem. The proposed algorithm treats the generator sizes as real discrete variables with uneven step sizes that reflect the sizes of commercially available generators, meaning that it can handle a mixed search space of integer (generator location), discrete (generator sizes), and continuous (reactive power output) variables while substantially reducing the search space and, consequently, computational burden of the optimization problem. The validity of the proposed discrete particle swarm optimization algorithm is tested on a standard 69-bus benchmark distribution network with four different test cases. Two optimization scenarios are considered for each test case: a single objective optimization study where network real power loss is minimized and a multi-objective study in which network voltages are also considered. The proposed algorithm is shown to be effective in finding the optimal or near-optimal solution to the problem at a fraction of the computational cost associated with other algorithms. [ABSTRACT FROM PUBLISHER]

Abstract: This paper presents a modified V/f control strategy using neural network with an improved flux pattern using third harmonic injection for multiphase induction machines. The control objective is to generate a nearly rectangular air–gap flux, resulting in an improved machine power density for the required speed range. If just a proportional relation is used between the third harmonic and fundamental plane voltage magnitudes with zero phase shift, variable misalignment between fundamental and third air–gap flux components occurs with varying mechanical loading as a result of stator voltage drop. Due to this misalignment, saturation may take place which affects the total flux and increases machine iron losses. Neural network is used to obtain the required injected voltage phasors magnitudes and angles to ensure that the air–gap flux is near rectangular with a maximum value of 1pu for all loading conditions. Simulations are carried out on an eleven-phase induction machine to validate the proposed controller using MATLAB/Simulink. [Copyright &y& Elsevier]