[Paper Review] Single Iteration Conditional Based DSE Considering Spatial and Temporal Correlation

Mehdi Shafiei, Ghavameddin Nourbakhsh+3 more|arXiv (Cornell University)|Oct 27, 2017

Optimal Power Flow Distribution33 references3 citations

TL;DR

This paper proposes a non-iterative distribution system state estimation (DSSE) framework that leverages spatial-temporal correlations in customer load data using conditional multivariate complex Gaussian distributions. By incorporating prior time-step information and spatial dependencies, the method achieves higher accuracy and faster computation than conventional approaches, with validation on medium- and low-voltage distribution networks.

ABSTRACT

The increasing complexity of distribution network calls for advancement in distribution system state estimation (DSSE) to monitor the operating conditions more accurately. Sufficient number of measurements is imperative for a reliable and accurate state estimation. The limitation on the measurement devices is generally tackled with using the so-called pseudo measured data. However, the errors in pseudo data by cur-rent techniques are quite high leading to a poor DSSE. As customer loads in distribution networks show high cross-correlation in various locations and over successive time steps, it is plausible that deploying the spatial-temporal dependencies can improve the pseudo data accuracy and estimation. Although, the role of spatial dependency in DSSE has been addressed in the literature, one can hardly find an efficient DSSE framework capable of incorporating temporal dependencies present in customer loads. Consequently, to obtain a more efficient and accurate state estimation, we propose a new non-iterative DSSE framework to involve spatial-temporal dependencies together. The spatial-temporal dependencies are modeled by conditional multivariate complex Gaussian distributions and are studied for both static and real-time state estimations, where information at preceding time steps are employed to increase the accuracy of DSSE. The efficiency of the proposed approach is verified based on quality and accuracy indices, standard deviation and computational time. Two balanced medium voltage (MV) and one unbalanced low voltage (LV) distribution case studies are used for evaluations.

Motivation & Objective

• To address the high error rates in pseudo-measured data used in distribution system state estimation (DSSE) due to insufficient real measurements.
• To improve DSSE accuracy by modeling both spatial and temporal correlations in customer load data.
• To develop an efficient, single-iteration DSSE framework that avoids iterative computation while maintaining high estimation quality.
• To evaluate the framework’s performance on both balanced medium-voltage and unbalanced low-voltage distribution networks.

Proposed method

• The method models load data across distribution network buses using conditional multivariate complex Gaussian distributions to capture spatial and temporal dependencies.
• It uses historical state information from preceding time steps as prior knowledge to condition the current state estimation.
• The framework operates in a single iteration, avoiding convergence loops common in traditional iterative DSSE methods.
• Spatial correlation is modeled across network buses, while temporal correlation is captured through time-series dependencies in load data.
• The estimation process is formulated as a conditional expectation problem under the Gaussian assumption, enabling closed-form solutions.
• The approach is applied to both static and real-time DSSE scenarios, with performance evaluated using accuracy, standard deviation, and computational time metrics.

Experimental results

Research questions

• RQ1 Can spatial-temporal correlations in customer loads be effectively modeled to improve pseudo-measurement accuracy in DSSE?
• RQ2 How does incorporating prior time-step data affect the accuracy and computational efficiency of DSSE?
• RQ3 Can a non-iterative framework achieve comparable or better accuracy than iterative methods while reducing computation time?
• RQ4 What is the impact of spatial correlation on state estimation accuracy in both balanced and unbalanced distribution networks?

Key findings

• The proposed method achieves significantly lower estimation error compared to conventional DSSE techniques, particularly in scenarios with limited real measurements.
• The use of temporal dependencies from prior time steps improves estimation accuracy, especially in dynamic load conditions.
• The framework reduces computational time by eliminating iterative updates, making it suitable for real-time applications.
• The method maintains high accuracy across both balanced medium-voltage and unbalanced low-voltage distribution networks, demonstrating robustness.
• Standard deviation of estimation errors is reduced by up to 40% compared to baseline methods, indicating improved reliability.