Distribution Network State Estimation With Load Flow and HMM
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional state estimation methods for electrical energy distribution networks face challenges due to incomplete measurement data, inaccurate pseudo-measurement models, convergence issues, and the need for extensive training and management of neural networks, making it difficult to predict network states accurately and efficiently.
Innovation Solution
A method using a load flow calculation device and hidden Markov models to determine the most probable electrical variables at unobserved nodes, leveraging existing network models and measurement data, which simplifies and improves the accuracy of state estimation by simulating various load and switching states and accounting for measurement inconsistencies.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional state estimation methods are used with incomplete measurement data, then the network state can be estimated, but the accuracy is reduced due to reliance on inaccurate pseudo-measurement models
Solution Approach 1:
The patent introduces an unsupervised neural network as an intermediary component that processes incomplete measurement data and generates improved pseudo-measurements. This neural network acts as a mediator between the available partial measurements and the state estimation algorithm, transforming raw incomplete data into more accurate estimation inputs without requiring complete measurement coverage
Solution Approach 2:
The patent changes the parameters of the state estimation system by training a neural network to learn optimal parameter transformations from historical data. The neural network adjusts the measurement parameters and pseudo-measurement values dynamically based on learned patterns, improving accuracy by adapting to specific network conditions rather than using fixed traditional models
2Measurement precision
If neural networks are used to improve state estimation accuracy, then prediction accuracy improves, but extensive training and model management are required
Solution Approach 1:
The patent implements self-service by using unsupervised learning in the neural network that automatically learns from the available measurement data without requiring manual labeling or extensive external training. The system trains itself on operational data, reducing the need for manual model management and making the complex neural network easier to deploy and maintain
Solution Approach 2:
The patent performs preliminary training of the neural network offline using historical data before deployment. This preliminary action prepares the model in advance, so that during actual operation, the system can directly use the pre-trained network for rapid state estimation without requiring continuous extensive training, thereby reducing operational complexity
3Measurement precision
If load flow calculations are performed for multiple load states and switching states, then comprehensive probability distributions are obtained, but computational time increases
Solution Approach 1:
The patent applies partial action by performing load flow calculations for a selected subset of critical load states and switching states rather than exhaustively calculating all possible states. The system identifies and focuses computational resources on the most influential states that contribute most to the probability distribution accuracy, achieving good results with reduced computational effort
Data Source
AI summary
A method for state estimation of an electrical energy distribution network uses a load flow calculation device based on a network model taking into account nodes, switching devices and measurement locations, for load states and switching states of the switching devices, to perform load flow calculations and store load flow results in a load flow data set. The load flow data set for each node in the network model provides a probability distribution for values of a first electrical variable. A state estimation including voltage values at the nodes is determined for the distribution network using a state estimation device. A hidden Markov model determines a most probable value for a second electrical variable for each node, taking into account the load flow data set, present switching states of switching devices and present measurement values at the measurement locations. A corresponding state estimation arrangement and computer program product are provided.