A pattern recognition method using big data for device behaviour with linear features

In recent years, big data technology has rapidly developed, yet research on device behavior recognition based on big data monitoring remains relatively scarce. To address this gap, a big data-based pattern recognition method for device behavior with linear characteristics is proposed. This research aims to identify the behavior and activities of devices through big data-based pattern recognition algorithms to ensure their normal operation. Initially, the required sequence data, window data, and corresponding formal structural expressions are constructed from the raw data. Following preprocessing, the big data is cleaned, and rules and calculation methods for handling abnormal data, such as mutations and losses, are formulated. The device behavior identification method classifies data into three basic states based on the changing characteristics of the data. The logical relationship between device behavior and data state changes is studied, revealing that the change in data state flow is crucial for accurately judging or determining device behavior. A graphical simulation of big data on device behavior is provided, and the simulation experiment visually displays the data changes associated with device behavior activities. The results effectively verify the proposed method's effectiveness.