Adaptive neuro fuzzy inference system to enhance the classification performance in smart irrigation system. (6th December 2021)

• Record Type: Journal Article
• Authors: Vivekanandhan, VijayaRangan; Sakthivel, Subramaniam; Manikandan, Muthaiyan
• Is Part Of: Computational intelligence. Volume 38:Number 2(2022)
• Page Start: 308
• Page End: 322
• Publication Date: 2021-12-06
• Subjects: adaptive neuro‐fuzzy inference system (ANFIS) -- environment data collection -- fuzzy logic controller -- linear regression model -- recursive adaptive filter -- wireless sensor nodes; Artificial intelligence -- Periodicals; Computational linguistics -- Periodicals; 006.3
• DOI: 10.1111/coin.12492
• Languages: English
• ISSNs: 0824-7935

Abstract

Abstract: Traditional agricultural systems require huge amounts of money for on‐site irrigation power. Irrigation is the process of giving water to plants for growth and development. The cost of global warming creates the potential effects of water adaptation measures to ensure food production and water availability for consumption. It requires continuous monitoring of soil moisture content at the root zone and initiating irrigation as per the pre‐programmed schedule depending upon the plant's nature, growth, soil type, and environment. Thus, the cost to farmers using conventional drip irrigation has become so high that they have to go manually and monitor the land frequently. So when the previous method was used for irrigation, it did not give correct classification results. Agricultural data mining technology provides the best crops for travel water to increase the number of crops produced. Agricultural data can be collected more efficiently using data collection and capable analytics sensors. This proposed work introduces an adaptive neuro‐fuzzy inference system (ANFIS) technique for analyzing agricultural plant growth based on soil, water level, temperature, and moisture conditions. The proposed technique has three phases: environment data collection using sensor node, preprocessing, feature selection, and classification. The initial phase captures the plant's environment condition via a sensor device, then transmits data to servers with the help of an edge node or gateway. This recursive adaptive filter method normalizes the data and removes irrelevant noise data from the sensor node. The feature selection method to extract the environment feature value with the help of a linear regression model. The ANFIS method's plant environment feature value has trained the neurons with the help of the stochastic gradient descent method. The classification method introduces a fuzzy rule to compute and validate the input parameters (e.g., soil moisture, temperature, and humidity) to predict any environment value changes. The comparison results prove that efficient monitoring is obtained through the proposed smart irrigation system.