IoT Based Water Quality Measurement Using Hybrid Sensors and Data Mining

The purposes of this research were 1) to develop the prototype for water quality measurement by using hybrid sensors which is the combination between e-tongue and e-nose together to work of IoT; and 2) to present algorithms for classification by using data mining techniques. In e-tongue for measurement of water chemical properties consist of pH, Electrical Conductivity (EC), Turbidity, Total Dissolved Solids (TDS), Salinity, Dissolved Oxygen (DO) and Temperature. And e-nose for measurement the response of different types of gases consist of MQ2, MQ3, MQ4, MQ5, MQ6, MQ7, MQ8, MQ9 and MQ135. Samples were collected at inlet and outlet areas from water quality control plants of Bangkok in 7 zones including SiPhraya, Rattana Kosin, Chong nonsi, Chatuchak, Din daeng, Nong kham, and Thung khru. Classification was done by algorithm such as NaiveBayes, NaiveBayesMultinominal, Logistic, SimpleLogigtic and IBK. Training and Testing were done by 10-fold cross validation and compared by accuracy. The results showed that the neural network model 15-10-1 with GainRatio include Salinity, Terbidity, TDS, MQ8, MQ4, DO, MQ135, pH, MQ5, MQ9, MQ7, MQ2, MQ6, EC and MQ3 have accuracy at 95.12%. Therefore, it can be concluded that hybrid sensors can be applied to measurement and monitoring water quality around communities in real environments.