Identification of polynomial models of static load characteristics based on passive experiment results

Abstract

Introduction. Dependence of a load power from a voltage in the form of static load characteristics permits to improve the accuracy of regime calculations significantly. This fact is extremely important, when it is necessary to define the area of permissible operating regimes for power systems. The most technologically advanced approach to collect data for the static load characteristics identification is a passive experiment, but the statistical processing of data does not allow for the polynomial models identification. The problems in the statistical processing connect with the small range of voltage variations in a passive experiment and an influence of grid responsiveness to the voltage on load busses.Goal - development of a technique that allows to determine polynomial models of static characteristics in terms of voltage from linear models.Materials and methods. In the paper, the technique based on the initial identification of the linear model, defined by EM-algorithm, and continued by the Lagrange multiplier method optimization with iterations by the Newton method is suggested.Results and discussion. As a result, for the large industrial consumer the polynomial models of the static load characteristics have been obtained. The polynomial model coefficients are installed into the software for regime calculations, and, then, the regime modeling with serial changes of active and reactive power nominal values, recording calculated values of powers on load busses and the comparison with standard models of static load characteristics are completed.Conclusion. The dispersion of initial data in reference to calculated data shows, that polynomial models of static load characteristics, identified by suggested technique, simulate the regime with the greater accuracy than standard models of static load characteristics.

Keywords

• static load characteristics
• network responsiveness
• voltage-regulating effect of a load
• clustering
• em-algorithm
• optimization
• lagrange multiplier method
• newton method