Event discovery method and device, storage medium and terminal

Abstract

An event discovery method and device, a storage medium and a terminal are provided, the method comprising: obtaining a historical distribution sequence, the historical distribution sequence comprising a plurality of historical distributions, each historical distribution representing a difference degree of data distribution of a plurality of historical monitoring data in a monitoring data sequence in an adjacent preset time window; inputting at least one part of historical distribution in the historical distribution sequence into a preset LSTM model to obtain prediction distribution, wherein the prediction distribution is predicted according to the change trend of at least one part of historical distribution along with time, and the prediction distribution is a prediction value of the difference degree of the to-be-processed monitoring data and the historical monitoring data in an adjacent preset time window; calculating actual distribution; and when the deviation between the actual distribution and the predicted distribution is greater than a preset threshold, determining that an event is found and giving prompt. Through the scheme provided by the invention, the time discovery threshold of the sensor can be automatically adjusted, so that the event can be better discovered, and the precision ratio and recall ratio of event discovery are improved.

Description

technical field [0001] The present invention relates to the technical field of big data processing, in particular to an event discovery method and device, a storage medium, and a terminal. Background technique [0002] At present, the concept of smart city and artificial intelligence is very hot, and the development of smart city is also very fast, which provides a lot of convenience for people's life and work. The development of a smart city is inseparable from sensor networks. Through sensor networks deployed throughout the city, data from various parts of the city can be obtained in real time to discover various events. [0003] Usually, most incidents are discovered based on the values ​​returned by sensors and artificially set thresholds. For example, the smart devices (such as smart bracelets) worn by the elderly in nursing homes can obtain the heart rate and breathing data of the elderly. The current common practice is to manually set a fixed threshold based on medic...

AI Technical Summary

Problems solved by technology

[0005] Secondly, for the discovery of some events, such as the smoke sensor and temperature sensor in the corridor, the existing technology relies entirely on manual threshold setting mode, which cannot take into account the precision and recall of event discovery, and cannot meet both requirements. Discovering real events without false positive user expectations

[0006] Finally, in the smart city scenario, there are many sensors installed in the city. If different thresholds are artificially set for each sensor, the labor cost will be extremely high, time-consuming and labor-intensive.

Images