Research On The Optimization Method Of Soil Moisture Sensor Layout In Tea Plantations Under Complex Weather

Soil quality is an important factor in determining whether tea trees can survive and thrive,and the moisture provided by soil for tea trees is an essential material condition for it.Therefore,the timely and accurate acquisition of soil moisture is particularly important for the growth of tea trees and the decision of irrigation in tea plantations.Tea plantations have a large slope,and the rate of water infiltration or transpiration in the same soil layer varies between the top,bottom and gap of the slope,and is related to precipitation and temperature.Therefore,the layout of soil moisture sensors in tea plantations must take into account weather factors.In this study,we use the K-medoids algorithm,the improved K-medoids algorithm and the Greedy-Dijkstra algorithm to optimize the layout of soil moisture sensors in tea plantations.The main research of this thesis is as follows:(1)Optimization of sensor layout based on K-medoids algorithm.Thirty-two sensors were arranged in a tea garden of about 20 acres in the National Hi-Tech Agricultural Park of Anhui Agricultural University,and the relative soil moisture content under various weather conditions was collected under the premise of ensuring full coverage of the sensor network,and clustering was performed by K-medoids to obtain the cluster centers and the clusters to which each node belonged,and the sensors were arranged in the cluster centers to eliminate other redundant nodes.The experimental results show that the optimization reduces the number of sensors from 32 to 4,and the node dissimilarity is 26.3,which reduces the data redundancy of soil moisture sensors.K-medoids layout optimization reduces the number of sensors and data redundancy,but the clustering results are unstable and the nodes are poorly represented under different weather conditions,for this reason,an improved K-medoids algorithm is proposed to optimize the sensor layout.(2)Optimize sensor layout based on improved K-medoids algorithm.The selection of initial clustering centers is optimized by variance to improve the stability of the results,and the similarity function suitable for high-dimensional data with temporal correlation is defined based on the profile alignment algorithm to improve the representativeness of the clustering results.The experimental results show that the node dissimilarity of the improved K-medoids optimization results is 32.5,the data redundancy is lower,and the deviation of the mean relative water content of the clustering centers from the overall mean value of the experimental area is about 1%,which improves the representativeness of the nodes under complex weather.Improved K-medoids layout optimization further improves node representation,reduces data redundancy,and shortens path length,but the sensor sensing range is limited and some nodes cannot communicate directly.(3)Optimization of sensor layout based on Greedy-Dijkstra algorithm for tea garden wireless sensor network.Using Greedy-Dijkstra algorithm,the clustering results of improved K-medoids are used as the path starting point,the set of mandatory intermediate nodes and the end point to search for an optimal sensor deployment path with a small number of sensors,low data redundancy and short transmission path.The experimental results show that the number of sensors in the optimal deployment path searched by Greedy-Dijkstra algorithm is 7,the node dissimilarity is 23.7,each sensor node can communicate directly,the nodes have the best representation in complex weather,and the transmission path length is 144.84 m which is better than the result of K-medoids.