Optimized Neural Network Parameters Using Stochastic Fractal Technique to Compensate Kalman Filter for Power...

Area of Science:

• Electrical Engineering
• Control Systems
• Computational Intelligence

Background:

• Power system state estimation is crucial for real-time monitoring and control.
• Traditional Kalman filtering (KF) can suffer from divergence and inaccuracies due to model uncertainties and noise.
• Existing methods for anomaly detection in power systems require further enhancement in accuracy and robustness.

Purpose of the Study:

• To develop an advanced tracking-state estimation technique for power systems.
• To improve the accuracy and suppress filter divergence in Kalman filtering.
• To effectively detect and identify various anomalies in power system operations.

Main Methods:

• Kalman filtering (KF) enhanced with a multilayer perceptron (MLP) optimized via stochastic fractals search (SFS).
• Replacing KF gain (mismodeling error) and measurement noise with the optimized MLP-SFS.
• Applying the proposed KF-MLP-based SFS to detect anomalies like linear load fluctuations, bad data, and component outages.

Main Results:

• The KF-MLP-based SFS demonstrated improved accuracy and suppressed filter divergence compared to standalone KF and radial basis function compensation.
• The proposed method effectively detected and identified anomalies in the IEEE 57-bus system under various fault conditions.
• Accurate state estimation was achieved even with linear load fluctuations and sudden component losses.

Conclusions:

• The KF-MLP-based SFS offers a robust and accurate solution for power system tracking-state estimation.
• This technique enhances the reliability of power system monitoring by effectively handling noise and uncertainties.
• The proposed method shows significant potential for real-world applications in smart grids and anomaly detection.