Overhead power transmission line running state assessment method based on bidirectional Bayesian network

Abstract

The invention discloses an overhead power transmission line running state assessment method based on a bidirectional Bayesian network. The method can be used for conducting a real-time assessment on the running state of an overhead power transmission line. According to the method, a Bayesian network structure for the assessment of the running state of the power transmission line is constructed with various factors which influence the running state of the power transmission line serving as a condition attribute set and the running state of the line serving as a decision attribute, a conditional probability table is obtained according to sample training, and by utilizing the bidirectional reasoning technology dedicated to the Bayesian network, the running state of the line can be judged by means of causal reasoning, and the hidden danger of the state can also be recognized by means of diagnostic reasoning; when an assessment error exists, a self-feedback system can be used for conducting early warning and correction, an assessment database, the network structure and parameters can be modified dynamically in real time so as to be adapted to an update, and therefore healthy running of the power transmission line is truly guaranteed.

Description

technical field [0001] The invention relates to a power system operation safety technology, in particular to a self-feedback bidirectional Bayesian network-based overhead transmission line operation state evaluation method. Background technique [0002] With the rapid growth of my country's electric power construction, the scale of the power network is increasing day by day, which puts forward higher requirements for the safe and stable operation, monitoring and protection of overhead transmission lines. Since the overhead transmission line is put into operation, it must ensure its power transmission capacity as well as the level of safety and reliability of operation. The safe operation of the entire power system has brought different degrees of influence and threat. Therefore, how to improve the operation efficiency of overhead transmission lines, reduce operation and maintenance costs, and ensure the long-term normal and stable operation of the lines has become a key iss...

AI Technical Summary

Problems solved by technology

[0005] 1. During the operation of the overhead transmission line, there are various operating states, which have great randomness and uncertainty. However, most traditional state assessment methods only care about the fault state of the line operation, which is easy to ignore the fault state before the occurrence of the fault state. Some unnoticed situations and abnormal phenomena of some lines,

[0006] 2. Most traditional state evaluation methods are difficult to describe polymorphic events, and state evaluation is only divided into two cases of normal and fault analysis, such as fault tree analysis method and rough set method: fault tree analysis method requires two states of event state The certainty of the relationship between nature and fault logic; rough sets can not deal with the preference multi-attribute decision-making classification problem, and there is no corresponding processing method for the fuzziness of the original data itself

In addition, the logic gates in the fault tree all describe deterministic logical relationships, but for the operation of transmission lines, there are many possibilities to cause them to fail, and there may not be a definite relationship between upper and lower events. It is more appropriate to use the method of probability to describe the situation, but the logic gate does not have the ability to describe the probability

[0007] 3. In terms of evaluation efficiency, evaluation accuracy and evaluation scale, various traditional evaluation methods are quite different, such as artificial neural network method and Monte Carlo method: the evaluation or prediction of artificial neural network requires high original statistical data, and training Too few samples will affect the accuracy of evaluation results, and too large training sample set will cause complex network structure, long training time, and reduce evaluation efficiency; Monte Carlo method is opposite to artificial neural network method, it is effective for large systems, but Its calculation accuracy is inversely proportional to the square of the calculation time, the convergence speed is slow and the error is probabilistic;

[0008] 4. For the evaluation process of transmission line operating status, traditional methods can only qualitatively give the weak link of each load point, but generally cannot quantitatively give a certain component or a few components in the reliability of the entire system. status, and the calculation model is complex, and the calculation workload increases exponentially with the system scale. When the state of some components in the system is known, it is difficult for existing methods to calculate the conditional probability of these components affecting the entire system or part of the system. These conditional probabilities are very helpful for improving and enhancing the reliability of power systems

Images