Spatial-Temporal Event Detection Method with Multivariate Water Quality Data

With the rapid development of the society, water contamination events cause great loss if the accidents happen in the water supply system. A large number of sensor nodes of water quality are deployed in the water supply network to detect and warn the contamination events to prevent pollution from speading. If all of sensor nodes detect and transmit the water quality data when the contamination occurs, it results in the heavy communication overhead. To reduce the communication overhead, the Connected Dominated Set construction algorithm-Rule K, is adopted to select a part fo sensor nodes. Moreover, in order to improve the detection accuracy, a Spatial-Temporal Abnormal Event Detection Algorithm with Multivariate water quality data (M-STAEDA) was proposed. In M-STAEDA, first, Back Propagation neural network models are adopted to analyze the multiple water quality parameters and calculate the possible outliers. Then, M-STAEDA algorithm determines the potential contamination events through Bayesian sequential analysis to estimate the probability of a contamination event. Third, it can make decision based on the multiple event probabilities fusion. Finally, a spatial correlation model is applied to determine the spatial-temporal contamination event in the water supply networks. The experimental results indicate that the proposed M-STAEDA algorithm can obtain more accuracy with BP neural network model and improve the rate of detection and the false alarm rate, compared with the temporal event detection of Single Variate Temporal Abnormal Event Detection Algorithm (M-STAEDA).