Distribution Network State Estimation Using Hidden Markov Load Flow
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current state estimation methods for electrical energy distribution networks face challenges in accuracy and efficiency due to incomplete measurement data, reliance on pseudo-measured values, and convergence issues, especially in distribution networks with high uncertainty from distributed energy sources like solar and wind.
Innovation Solution
A method utilizing a hidden Markov model and Monte Carlo sampling with load flow calculations to estimate network states, leveraging Gaussian distributions for load and generator data, and employing the Viterbi algorithm for state determination, which can handle incomplete data and reduce the need for extensive training and model management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional state estimation methods are used with pseudo-measured values, then state estimation can be performed, but accuracy deteriorates due to lack of real measurement data at 80% of network nodes
Solution Approach 1:
The patent introduces a data imputation module that uses Gaussian process regression as an intermediary method to estimate missing measurement values at unmonitored nodes. This intermediary approach generates pseudo-measurements with quantified uncertainty, enabling state estimation to proceed while maintaining awareness of data quality limitations.
Solution Approach 2:
The patent transforms the measurement data by converting raw measurements into standardized residuals and uncertainty metrics. The measurement data is reparameterized to include confidence intervals, allowing the state estimation algorithm to weigh measurements appropriately based on their reliability rather than treating all measurements equally.
2Reliability
If load scaling methods are used to initialize load information, then convergence can be achieved, but device complexity increases due to numerous case distinctions required
Solution Approach 1:
The patent develops a universal initialization algorithm based on Gaussian process regression that can handle all types of measurement configurations and network topologies through a single unified approach. This eliminates the need for multiple specialized case-handling routines while maintaining convergence reliability across diverse operational scenarios.
Solution Approach 2:
The algorithm performs self-initialization by automatically detecting the available measurement configuration and adapting the Gaussian process regression parameters accordingly. The system self-adjusts to different network states without requiring external configuration or manual intervention to select appropriate case-specific methods.
3Measurement precision
If more measurement devices are deployed to improve observability, then state estimation accuracy improves, but cost increases due to additional infrastructure requirements
Solution Approach 1:
The patent creates virtual copies of measurement data through Gaussian process regression, generating synthetic measurements at unmonitored nodes based on correlations with nearby measured nodes. This copying approach provides the benefits of additional measurements without the physical infrastructure costs of deploying actual sensors at every node.
Solution Approach 2:
The system performs preliminary data processing and imputation to pre-fill missing measurement values before state estimation is executed. By preparing synthetic measurements in advance based on historical patterns and spatial correlations, the system maximizes the utility of existing measurements and reduces the urgent need for additional physical sensors.
4Reliability
If conventional state estimation algorithms are used, then computation can be performed, but computational time increases due to convergence issues with incomplete data
Solution Approach 1:
The patent performs preliminary Gaussian process regression to pre-estimate missing measurements and their uncertainties before the state estimation iteration begins. This preliminary action provides the state estimation algorithm with initialized values that are already statistically optimized, significantly reducing the number of iterations required to converge compared to conventional initialization methods.
Solution Approach 2:
The patent replaces the traditional iterative numerical optimization mechanics with a statistically-based Gaussian process approach for handling missing data. This substitution eliminates the convergence struggles of conventional iterative methods by using closed-form statistical solutions that naturally handle incomplete data without requiring repeated numerical optimization cycles.
Data Source
Figure 1~2
Figure 3~4
AI summary
The present invention relates to a method for state estimation of an electrical power distribution network, characterized in that load flow calculations are performed for a plurality of load states and for a plurality of switching states of the switching devices using a load flow calculation device based on a network model that takes into account nodes, switching devices and measuring points, and the respective load flow result is stored in a load flow data set, wherein the load flow data set provides a probability distribution for each node in the network model for at least one value of a first electrical quantity, and that a state estimation, which includes the respective voltage values at the nodes, is determined for the electrical power distribution network using a state estimation device.wherein, taking into account the load flow data set and current switching states of switching devices and current measured values acquired at the measurement points, a hidden Markov model is used to determine a most probable value for a second electrical quantity for each node. Furthermore, the invention relates to a corresponding state estimation arrangement and a corresponding computer program product.