Research on Algorithm of big data sensor anomaly diagnosis

With the decrease of sensor cost and the increase of performance, the application of sensor is becoming more and more popular. Sensors in different locations and devices produce different data streams. We expect to analyze the data or events we are interested in in in real time from these data streams. With the increase of the number of sensors, a large number of sensors are organized into the form of networks. How to carry out data mining in sensor network data stream will bring a new challenge. As one of the most important data flow technologies, sliding window technology has been widely studied and applied. By accurately estimating the data distribution in the sliding window, we can carry out anomaly detection and other important applications.