A kind of high-voltage equipment state prediction method

Abstract

The invention discloses a high voltage equipment state predicting method. The method comprises the following steps: (1) actually detecting historical data of gas components of at least one kind of transformer oil dissolved gas and / or SF6 gas in high voltage equipment, namely, detecting data of gas components in various periods on the condition that the equipment is an electrification mode; (2) utilizing the historical data in the step (1) and adopting a least squares curve fitting method to legally obtain the component data of corresponding gas in the high voltage equipment at certain time in future, namely, obtaining a characteristic curve changing along with time, and eventually obtaining long-time discharging or overheating stability data of the equipment, wherein the numerical value of the characteristic curve possibly continuously tends to be stable; (3) through a vector supporting machine, utilizing the component data, of the corresponding gas in the high voltage equipment at certain time in future, obtained in the step (2) to predict the failure state type of the high voltage equipment. The high voltage equipment state predicting method has the advantages of being low in dependence on experiences, strong in anti-interference ability and generalization ability, high in accuracy and capable of solving a multi-classification problem.

Description

technical field [0001] The invention mainly relates to the detection field of high-voltage equipment, in particular to a state prediction method suitable for high-voltage equipment. Background technique [0002] High-voltage equipment is subjected to electrical, thermal, and mechanical loads during operation, as well as the influence of the natural environment (temperature, humidity, etc.). Long-term work will cause aging, fatigue, and wear, resulting in a gradual decline in performance and reliability. Under the long-term action of high voltage and high temperature, the composition and structure of the insulation material of the equipment will change, the insulation performance will decrease or even be destroyed, and the final equipment failure rate will increase, thus endangering the operation safety of the power system. Therefore, it is very necessary to detect and grasp the state of high-voltage equipment. If the state of high-voltage equipment can be predicted in advanc...

AI Technical Summary

Problems solved by technology

[0005] (1) Local minimization problem: From a mathematical point of view, the traditional neural network method is a local search optimization method, which solves a complex nonlinear problem. Gradually adjusted, this will cause the algorithm to fall into a local extremum, and the weights will converge to the local minimum point, which will cause the network training to fail

[0006] (2) The convergence speed of the artificial neural network algorithm is slow: since the neural network algorithm is essentially a gradient descent method, the objective function to be optimized by it is very complex, therefore, there will inevitably be a "zigzag phenomenon", resulting in low efficiency of the algorithm

[0007] (3) The traditional neural network method adopts the principle of empirical risk minimization (ERM), which has a strong dependence on experience;

[0009] (1) The SVM algorithm is difficult to implement for large-scale training samples

Since SVM uses quadratic programming to solve support vectors, and solving quadratic programming will involve the calculation of m-order matrix (m is the number of samples), when the number of m is large, the storage and calculation of the matrix will consume a lot of machines memory and computing time

[0010] (2) There are difficulties in solving multi-classification problems with SVM

[0011] In addition to the shortcomings of the above two methods, because the historical data of the gas composition in the equipment is often affected by some interference factors or the historical data itself is unstable, this single method often cannot accurately predict the high-voltage equipment. state

Images