The Study And Application Of Multisensor Data Fusion Method For Granary Monitoring System

This paper introduces a new Granary Monitoring System based on ARM embedded system, which realizes real-time monitoring of temperature, humidity, grain quality and grain storage safety information, at the same time realizes centralized management of data for grain and remote queries. An algorithm of analysis grain information based on D-S Evidence Theory and Multi-Sensor Information Fusion technology is designed, has a security to grain storage.The system proposes a three-tier architecture based on distributed Granary Monitoring System, including PC, ARM controller and data collection system. It builds communication network using Ethernet and CAN Bus to realize network management of data. Data management system using C/S and B/S type of hybrid architecture model separated function of client, application side and database-side, reducing the mutual coupling and increasing system reusability, scalability, and security. Multi-sensor information with many characteristics is accessed through a variety of transformations to collected data of real-time temperature, humidity and image. Then this information which have some relevance are fused using D-S Evidence Theory to find fusion methods of temperature, humidity and image information and fusion algorithms in certain constraints. An algorithm is proposed based on feature statistics and adaptive weight parameter of D-S evidence theory, which improves the adaptability and accuracy of multi-sensor fusion.The Granary Monitoring System introduced in this paper used a can bus network based on ARM. It realized real-time collection of temperature, humidity, and image data in granary. The data is analyzed and dealt with by Multi-Sensor Information Fusion and D-S evidence theory, to achieve intelligent control of granary environment and stored grain pest warning. It ensures national grain storage safety.