Research On Small Water Qualitv Data Acquisition System And Modified KNN Early Warning Algorithm

With the rapid development of economy and society in China, pollution of water resources tends to be worse and hence water quality early warning technique is particularly important. Water quality early warning can acquire water quality data in real time and transmit them to water quality monitoring center, and predict the future water quality information according to the acquired data. In conclusion, water quality early warning technique is of great importance for water quality center to obtain the information and trend of water quality in time.The prerequisite of water quality early warning is water quality data acquisition. Manual acquisition and automatic monitoring station acquisition are the main methods to acquire water quality data. However, the manual acquisition is time-consuming, and the application of automatic monitoring station is limited due to its high cost. Therefore, it is of great significance to develop a low-cost and small water quality data acquisition system. Except for acquiring and transmitting water quality data, water quality early warning system needs to process the acquired data. However, with the rapid development of technology and national heavy investment in water quality monitoring, the acquired water quality data have been increasing rapidly due to the increase of water quality monitoring stations and the method of water quality data acquisition which has been improved from manual acquisition to automatic monitoring station acquisition. Consequently, there is an urgent need for a new prediction algorithm, which can deal with the large amount of data.Based on the literature review, this thesis focuses on the research of small water quality data acquisition system and water quality prediction algorithm that can be applied to dealing with large amount of data. The main contents and innovations are summarized as follows:1. Small water quality data acquisition system is developed and then is verified by carrying out experiments preliminarily. The system is composed of a mainboard and a function expansion board. The mainboard consists of single-chip module and power module and can implement functions such as power control, time control etc. The function expansion board is an open platform and can access several basic water quality parameter acquisition modules, which can also be further expanded according to actual requirement. The system can be established with low cost and can work as an unattended system for a long time by using solar energy.2. A water quality prediction method based on piecewise linear representation k nearest neighbor (PLR-kNN) algorithm is proposed to reduce the data processing time to some extent. With the combination of piecewise linear representation algorithm and k nearest neighbor (kNN) algorithm, this prediction method considers the influence of historical information and current trend of water quality to the prediction results. Compared with traditional kNN algorithm, the new prediction method can reduce computational burden and execution time effectively when dealing with large amount of water quality data.3. The water turbidity prediction experiments are carried out to verify the effectiveness of the proposed PLR-kNN algorithm. At the same time, second exponential smoothing algorithm and traditional kNN algorithm are applied respectively to the water turbidity prediction. Finally this thesis carries out a comparative study between kNN algorithm, second exponential smoothing and PLR-kNN algorithm.The related techniques of water quality early warning system are studied in this thesis. The designed small water quality data acquisition system and the proposed water quality prediction method based on PLR-kNN algorithm are carried out to verify their effectiveness. Experimental results show that the techniques proposed in this thesis are feasible.