A method and system for detecting anomalies in water conservancy projects

Abstract

The invention provides a method and system for detecting anomalies in water conservancy projects. The present invention combines the Hilbert-Huang transform with the deep forest model to detect abnormal points. Firstly, the Hilbert-Huang transform conducts a preliminary analysis on the time series data to determine the abnormal time series data, and then analyzes the abnormal time series based on the deep forest model. Data monitoring reduces the scale of data processed by the deep forest model, improves the accuracy of abnormal data detection, and ensures the safety of water conservancy projects.

Description

technical field [0001] The invention relates to the technical field of water conservancy project monitoring, in particular to a method and system for detecting abnormality of water conservancy projects. Background technique [0002] With the rapid development of society and economy in the new era, the requirements for the safe and efficient operation of water conservancy projects have also increased. Traditional water conservancy projects have been unable to meet the requirements of specialization, informatization, and intelligence required for economic and social development in the new era. The water conservancy industry in China has covered functions such as data collection, query browsing, and statistical analysis, but in terms of anomaly detection of time series data, its accuracy and comprehensiveness need to be improved. The abnormal detection of time series data has important applications in the comprehensive management of large rivers, the reinforcement of dangerous ...

AI Technical Summary

Problems solved by technology

[0004] The outlier detection based on the neural network mainly utilizes the characteristics of the artificial neural network to be effective in dealing with small-scale problems, but for large-scale problems, the construction of the artificial neural network will be very complicated

Moreover, due to the nature of the "black box" of the neural network, the theoretical explanation in the hidden layer is not rigorous enough, and the output of the hidden layer and the final output cannot be perfectly controlled, so it cannot be well extended to large-scale problems.

[0005] Outlier detection based on support vector machines only has the same processing power, and the calculation efficiency is also greatly improved, but support vector machines are relatively complicated in theoretical modeling, such as the selection of kernel functions, etc., so in practice It is often difficult to achieve ideal results in applications, and it is impossible to extract more comprehensive abnormal points

Images