Bayesian network-based power distribution network outage node prediction method

Abstract

The invention relates to a Bayesian network-based power distribution network outage node prediction method. The method comprises the steps of constructing a tree-shaped Bayesian network corresponding to a power distribution network; giving a key node set and an evidence node set in the tree Bayesian network, and calculating the joint probability P1 of the evidence node set; selecting a shutdown combination from the key node set, merging the shutdown combination with the evidence node set, and calculating the overall joint probability P2, thereby obtaining the posterior probability P2 / P1 of the shutdown combination; circularly calculating the posterior probability of each outage combination in the key node set; and selecting the outage combination with the maximum posterior probability as the most possible outage state, namely the power distribution network outage node prediction result. Compared with the prior art, the possibility of various outage combinations is fully considered, and the result reliability is high; the multivariate joint probability problem of the original index complexity is converted into a linear complexity problem through a recursion thought by combining the characteristics of the tree-shaped Bayesian network; the curse of dimensionality is overcome, and the calculation time is shortened.

Description

technical field [0001] The invention relates to the technical field of distribution network outage node prediction, in particular to a distribution network outage node prediction method based on a Bayesian network. Background technique [0002] Extreme disasters have repeatedly triggered power outages in distribution networks, resulting in huge economic losses. How to make the distribution network better cope with extreme disasters, improve the resilience of the distribution network, and ensure the continuous power supply of the distribution network when disasters strike have become the common goals of electric power practitioners. [0003] The invention with the publication number CN111932122A discloses a distribution network disaster prevention material allocation method, device, terminal and storage medium. This application considers the uncertainty of the disaster situation and uses the result of the Bayesian network disaster situation inference, that is, the node shutdo...

AI Technical Summary

Problems solved by technology

[0003] The invention with the publication number CN111932122A discloses a distribution network disaster prevention material allocation method, device, terminal and storage medium. This application considers the uncertainty of the disaster situation and uses the result of the Bayesian network disaster situation inference, that is, the node shutdown of each node Then, based on the distribution network importance information such as the weight coefficient of the node and the load, the outage probability of the node is fused, and the elastic risk index of the overall distribution network is used as the objective function of judgment to obtain when the elastic risk index satisfies When optimizing the requirements, the node material allocation amount of each node realizes the optimization of the global material allocation of the distribution network, and solves the problems of unreasonable allocation of disaster prevention materials and high waste of resources in the existing distribution network disaster prevention material allocation method question

[0004] This method proposes to determine the node outage probability of each node through the Bayesian network, but does not propose the specific calculation process of the Bayesian network. The existing other node outage prediction methods are all calculations of the failure probability of a single node, which do not fully consider The possibility of various outage combinations, and the calculation of the Bayesian network also has a large amount of data that is difficult to load, and the disaster of dimensionality often occurs, which makes the time cost of problem solving unacceptable.

Images