基于STM32与机器视觉检测的粮仓环境监管系统设计

Design of a Grain Storage Environment Monitoring System Based on STM32 and Machine Vision Inspection

摘要: 粮食安全是农业生产与仓储管理的核心诉求，当前粮仓环境监管中，数据检测不准确、不及时及人工抽样检测易遗漏等问题频发，不仅影响监管效率，还可能引发粮食霉变、虫害泛滥等风险，直接威胁粮食储存安全与经济效益。为破解这一难题，本文设计了一种基于STM32F103系列微控制器的主控单元监管系统，通过DHT11温湿度传感器、MQ-2烟雾传感器及SGP30二氧化碳模块，精准采集粮仓内温湿度、烟雾浓度、二氧化碳含量等关键环境数据，再借助ESP8266-01S无线Wi-Fi模块，基于稳定高效的MQTT协议将本地传感器数据上传至云端平台，并同步更新至手机APP端实现实时可视化显示。此外，在K230视觉识别模块上部署基于改进YOLOv8n算法自训练的虫害检测模型，该模型优化了检测精度与实时性，可精准识别粮仓内老鼠、麻雀、蟑螂三类常见有害生物，检测结果同步至手机端供管理人员查看。当粮仓内任意环境参数超出预设安全阈值时，系统将自动触发声光报警，为粮仓环境的协同监管提供全流程、智能化的有力保障。

Abstract: Food security constitutes the core imperative of agricultural production and storage management. Current grain storage environment monitoring frequently encounters issues such as inaccurate and untimely data detection, alongside the risk of omission in manual sampling inspections. These shortcomings not only compromise regulatory efficiency but may also precipitate risks including grain moulding and pest infestations, directly jeopardising storage safety and economic viability. To address this challenge, this paper proposes a monitoring system centred on an STM32F103 series microcontroller. Utilising the DHT11 temperature and humidity sensor, MQ-2 smoke sensor, and SGP30 carbon dioxide module to precisely collect critical environmental data including temperature, humidity, smoke concentration, and carbon dioxide levels within the silo. Utilising an ESP8266-01S wireless Wi-Fi module, the system transmits local sensor data to a cloud platform via the stable and efficient MQTT protocol, synchronising updates to a mobile application for real-time visualisation. Furthermore, the K230 visual recognition module deploys a self-trained pest detection model based on an enhanced YOLOv8n algorithm. This model optimises detection accuracy and real-time performance, precisely identifying three common pests within the silo: rodents, sparrows, and cockroaches. Detection results are synchronised to the mobile interface for management personnel to review. Should any environmental parameter within the silo exceed preset safety thresholds, the system automatically triggers audible and visual alarms. This provides comprehensive, intelligent safeguards for collaborative monitoring of the silo environment throughout its entire operational lifecycle.