Complex Event Processing on uncertain data streams in product manufacturing process

With the development of automatic production, manufacturing factories record tremendous amounts of data with sensor devices deployed in a factory. Because of the inherent inaccuracy of sensor readings, these data is of high level of uncertainty. How to use Complex Event Processing (CEP) to get useful information for quality monitoring of products from a lot of uncertain raw data continually generated from the production lines is becoming a challenging research. Therefore, in this paper, we propose a model of uncertain complex event processing system for real-time monitoring in product manufacturing process. And then we define the probabilistic event model and propose a probabilistic event detection algorithm based on rNFA and its optimization plan by event filtering. At the same time, we introduce Conditional Probability Matrix (CPM) and describe the calculation of probability of complex events with the multiplication theorem of probability. The experimental results show that our proposed method is efficient to detect complex events over probabilistic event streams with better event throughput capabilities and lower time consumption.