Transformer apparent power prediction method

Abstract

The invention discloses a transformer apparent power prediction method, and belongs to the technical field of prediction. The method comprises the following steps: S1, acquiring historical load data of a transformer; s2, starting from a low-order improved ridge polynomial neural network, training and updating the weight; s3, if the observed error change is lower than a predefined threshold value, adding a higher-order feedforward neural network unit; s4, adjusting the threshold value of the error gradient and the learning efficiency through learning factors; and S5, continuously learning and updating the improved ridge polynomial neural network, and finally calculating and outputting a predicted value. According to the method, the historical data can be accurately and quickly trained, the change rule of the data is obtained, the prediction precision of the data can be effectively improved, and accurate prediction of the data is realized.

Description

technical field [0001] A transformer apparent power prediction method belongs to the field of prediction technology. Background technique [0002] At present, the construction of smart grid is a direction of great concern in the construction of power systems, which can effectively ensure the safe and reliable operation of the network. In addition, while the power grid operates safely, operating economy is indispensable. Power system apparent power prediction focuses on the real-time operating state of apparent power, is committed to the safe operation and data analysis of apparent power, and provides a technical basis for mining and understanding the operating situation and working conditions of apparent power. [0003] Under the current new power system with electricity as the main body, the power prediction of TVs is interrelated with multiple factors. Under the influence of uncertain power output, the prediction error of the algorithm is relatively large, which cannot m...

AI Technical Summary

Problems solved by technology

[0003] Under the current new power system with electricity as the main body, the power prediction of TVs is interrelated with multiple factors. Under the influence of uncertain power output, the prediction error of the algorithm is relatively large, which cannot meet the high-precision scenarios

If the apparent power prediction error is large, it will affect the control of the power dispatching system, resulting in a mismatch between power generation, power transmission, power distribution, power consumption, and reserve capacity, resulting in waste of resources, and in severe cases, it will lead to power consumption of users. Unsatisfied demand, low operating efficiency of power generation, transmission and distribution equipment, seriously affecting the operating economics of the grid

Therefore, in view of the fact that the current algorithm often cannot guarantee the accuracy and accuracy of the data when the apparent power time series has significant randomness, complexity and nonlinear characteristics, it is urgent to study and develop a cloud-based day-ahead In the power prediction method and system, the apparent power characteristics can be accurately grasped to ensure the accuracy of the predicted data

Images