| In recent years,with the rapid development of agricultural Internet of Things technology,semantic Internet of Things has become a hot topic in many agricultural fields.Semantic Internet of Things analyzes and processes agricultural information data through intelligent decision-making and management technology,improves crop cultivation levels,optimizes agricultural production and resource allocation processes,effectively improves the quality and efficiency of agricultural information production,and reduces resource waste and environmental pollution.With the rapid development of intelligent agriculture,the complexity of farmland moisture collection networks is becoming increasingly high,with a wide variety of equipment and large number of equipment,increasing the difficulty of farmland information management.Therefore,correct maintenance methods and the construction of semantic networks suitable for agricultural equipment are also crucial elements in the work of the agricultural Internet of Things.The existing farmland moisture monitoring system is difficult to popularize due to various drawbacks.In addition,there is a problem of integrating heterogeneous Internet of Things data information in farmland,making it difficult for agricultural maintenance personnel to fully utilize moisture information,thereby making it difficult to improve production efficiency.Therefore,the work of constructing a farmland moisture sensing network using semantic Internet of Things technology has been carried out.The main research contents are as follows:(1)Propose an FSSN ontology model that integrates semantic sensor networks and farmland moisture ontology,design 9 key entity types in the field of agricultural moisture collection equipment,and construct a top-level ontology in the field of farmland moisture to expand the application model of FSSN;Then,an improved remote supervised semantic extraction model is proposed,incorporating the Trans-E graph embedding mechanism into the remote supervised model,and a PCNN-Trans-E model is designed for relationship extraction in the field of agricultural soil moisture equipment;Then,complete the training through the DBpedia dataset,and verify 12 relationship types in the self-made FMSR dataset;Finally,the experimental results show that the proposed PCNN-Trans-E extraction method outperforms other traditional models,with an AUC value of 0.5875.(2)A lightweight design method is proposed to simplify the FSSN model.Firstly,using Tf-Idf algorithm to analyze the semantic weight of farmland ontology,FSSNSDRM algorithm and semantic annotation method are proposed to construct a lightweight ontology model of farmland moisture content;Then,analyze the crop growth environment,set suitability inference rules,and use Jena API to infer annotation ontology models;Finally,experiments were conducted on the lightweight middleware NVIDIA TX2 platform based on the soil moisture information collected by farmland nodes to query the correctness of the inference information in the database and compare the response time of the devices;The experimental results show that the lightweight FSSN annotation ontology reduces the average response time by 41.4%,and the time in TX2 decreases by 12.81%.It can quickly and accurately judge the suitability of the crop growth environment,providing a new idea for agricultural information collection.(3)Finally,based on the above research content,a semantic platform for farmland moisture content is designed.Combining the perceptual layer,middleware,and software application layer,the semantic platform has been built,and the obtained entities and relationships have been stored.A visual management platform for farmland moisture collection equipment with visual display,knowledge query,and knowledge management functions has been developed.It is concluded that this system can achieve the desired results. |
PDF Full Text Request