An integrated algorithm-based detection and prediction calculation method for abnormal power fluctuations

Abstract

This application relates to a power abnormal fluctuation detection and prediction calculation method based on an integrated algorithm of catastrophe point analysis and hidden Markov model. Using the collected electricity consumption data, Bayesian inference is used to realize the mutation point analysis of abnormal fluctuations; the mutation point data is predicted, and the hidden Markov model is used to process the time-related information, and the time series information is processed. , using the hidden Markov forward algorithm and Baum-Welch algorithm to obtain the likelihood parameters, get the prediction data of the mutation point, and obtain the maximum likelihood value corresponding to the sliding window through the training feature. Based on the improvement of the basic algorithm of mutation point analysis and hidden Markov model, it can detect the power consumption in real time, and can predict and calculate the reasonable range of the power consumption interval in the future. If it exceeds the range, it will be judged as abnormal power consumption.

Description

technical field [0001] The present application relates to the Intelligent Internet of Things (AIoT) technology, in particular to an intelligent Internet of Things power abnormal fluctuation detection and prediction calculation method based on mutation point analysis and hidden Markov model integration algorithm. Background technique [0002] It is a common method in industrial and engineering projects to judge whether instruments and equipment are running normally through abnormal power fluctuations. Therefore, improving the accuracy of abnormal power fluctuation detection is of great significance for ensuring the normal operation of instruments and equipment and the sub-item measurement of building energy consumption. [0003] With the rise of artificial intelligence (AI) algorithms in recent years, many scholars have begun to explore the application of time series algorithms in power load forecasting, but there are few studies on the application of AI algorithms in the det...

AI Technical Summary

Problems solved by technology

[0005] 1. There is no clear definition index for power abnormalities;

[0006] 2. There is a lack of labeled data for power abnormalities;

Images