Independent Component Analysis Appling To Electric Power Harmonic Current Estimation

With the development of telecommunication and measurement technology, data storage and power calculation, power systems will be monitored more extensively. HSE (Harmonic State Estimation) technology can estimate the entire electric network based on the limited measured datum, thus monitor and manage the harmonic. However, due to the constraint of harmonic measuring equipment and harmonic datum, it is highly difficult to obtain the harmonic network parameter and topology. Consequently, HSE cannot be used to estimate the load curve. To solve this problem, ICA (Independent Component Analysis) is applied to the harmonic current estimation in this dissertation.First, this dissertation summarizes the theory of the harmonic estimation and knowledge of harmonic model, which elaborates the methods of HSE. Then, the model, theory and application conditions of ICA are described in detail; the principle and algorithm implementation of FastICA (Fast Independent Component Analysis) , which performs better in convergence rate and performance, is especially investigated in this dissertation.Second, the feasibility of applying ICA to estimate the power system harmonic is analyzed from the aspects of harmonic current model and the application conditions of ICA.Finally, the experiment is emulated in the electric circuit. In the experiment, harmonic voltage is regarded as the known measured signal, the system parameters and topology of power system is handled as the unknown hybrid matrix, and the harmonic current is treated as the unknown source signal. FastICA is applied to the harmonic current estimation. By decreasing the number of experiment sample, ICA is explored to estimate the harmonic state in the error tolerance range, which reduces the cost of power system investment significantly. The result shows that even in the real power system where noise exists inevitably, ICA can still perform well in the harmonic estimation.