CVT measurement error correction by double regression-based particle swarm optimization compensation algorithm

Abstract

The capacitive voltage transformer (CVT) is a widely-used voltage measuring instrument directly related to the safety and security of the power system operation. As the CVT’s range of application scenarios is getting wider, its environmental adaptability requirements become more stringent. Therefore, the CVT measurement error generated from the atmospheric environment is becoming a matter. This paper proposes a CVT measurement error correction method by employing a double regression-based particle swarm optimization compensation algorithm (DR-PSO). In DR-PSO, the double regression algorithm is used to calculate the rough result, and the theoretical formula is used to calculate the multi-type disturbance. Finally, the particle swarm optimization is used to determine the optimal influence factor matrix and obtain accurate results. Based on the transformer’s multidimensional operating data, the DR-PSO can be trained by the empirical data, and then the algorithm shall be used to correct CVT errors based on the real-time data. The test by actual data of State Grid proved that the root mean squared error of the proposed algorithm’s phase error was only 0.13, and the mean absolute error of the amplitude error is only 0.0009. Compared with other algorithms such as the equal weight method, information entropy method, and multiple regression method, the proposed algorithm can calculate the measurement error with the highest accuracy.

Keywords

• Capacitive voltage transformer
• Measurement error correction
• Nonlinear regression
• Particle swarm optimization