A New Algorithm Applied to the Evaluation of Self Excited Induction Generator Performance Aissa KHELDOUN, Hamid BENTARZI, and Larbi REFOUFI Signals & Systems Laboratory Institute of Electronics and Electrical Engineering Boumerdes university Avenue of independence, 35000, ALGERIA Aissa1973@gmail.com Abstract: - The paper presents the application of DIRECT algorithm to analyse the performances of the Self-excited induction generator. It is used to minimize the induction generator admittance yielding the solution which consists of the magnetizing reactance and the frequency. These parameters are the keys to find out the self excitation process requirements in terms of the prime mover speed, the capacitance and the load impedance and finally the output performances such as the voltage, output power, etc. A comparison with other powerful optimization algorithms is investigated to obtain DIRECT algorithm performances. Key-Words: - SEIG, optimization, DIRECT algorithm. 1 Introduction Owing to its many features such as: inherent ruggedness, lower unit cost, maintenance free, self protection against overload and short-circuit faults and its ability to generate power at various speeds the induction machine dominates in wind energy conversion applications particularly in remote area (standalone) [1], [2]. Besides, the induction machine may generate a voltage across it stator windings if it’s driven by prime mover with sufficient speed and a capacitor bank with required VAr is connected to the stator. This process, known as self excitation, is considered as an optimization problem where the cost function to be minimized is the equivalent circuit’s admittance (or impedance) with two constraints, the first one involves that the induction generator must operate in the saturation region which means the magnetizing reactance is always less than the unsaturated value and the second constraint involves that the obtained frequency must be less than the prime mover’s speed, in other words negative slip. The cost function is minimized using an optimization algorithm in a way to find out the required values of the external parameters namely speed, load impedance and capacitance leading to values of frequency and magnetizing reactance which satisfy the previous constraints. Newton-Raphson algorithm has been mostly used to solve this optimisation problem. First the impedance (or the admittance) must be resolved to real and imaginary parts then solved for one of the following configurations: frequency and magnetizing reactance, frequency and speed, frequency and load impedance or frequency and capacitive reactance. The solution yields to the required value of the parameter at hand for the self excitation process. However, despite the good optimization accuracy, separate the admittance’s complex function to real and imaginary parts and the Jaccobian matrix calculation are tedious tasks. Besides, the solution is very sensitive to the initial solution, that’s the optimization diverges in case of incorrect initial guess. MATLAB built-in minimization routines [3] and MathCAD functions [4] have been used to minimize the impedance. The advantage is that no algebraic derivations are required for optimization process, but both of them need the knowledge of the initial guess. Global optimization as Genetic Algorithm (GA) and Simulated Annealing (SA) has been used to solve the above optimization problem, the advantage of GA is the no need of both cost function derivative and the initial guess. However, the optimization accuracy is not satisfied. For this reason, mostly GA is used to bring the values of the unknown variables close to the region of the optimality then a classical constrained optimizer will take over to fine tune these unknowns [5]. Singaravelu S. et all. proposed a fuzzy logic- based optimization approach [6] to determine the capacitive VAr requirements for voltage regulation of three phase induction generator, the algorithm avoids the tedious and erroneous manual work of segregating the real and imaginary components of the impedance. However, the algorithm requires the calculation of the determinant to check the matrix singularity. In this paper a new global search optimization algorithm, known as DIRECT algorithm, is employed to minimize the admittance of the self excited induction generator. The convergence of the algorithm is guaranteed when the Proceedings of the 12th WSEAS International Conference on AUTOMATIC CONTROL, MODELLING & SIMULATION ISSN: 1790-5117 82 ISBN: 978-954-92600-1-4