| 小麦品质分级数字化无损检测方法研究 |
| |
| 引用本文： | 刘光宗,曹成茂,张金炎,周润东,李文宝.小麦品质分级数字化无损检测方法研究[J].安徽农业大学学报,2020,47(4):655. |
| |
| 作者姓名： | 刘光宗 曹成茂 张金炎 周润东 李文宝 |
| |
| 作者单位： | 安徽农业大学工学院，合肥 230036; 安徽电气工程职业技术学院，合肥 230051 |
| |
| 基金项目： | 安徽省重大科技专项（18030701204）和安徽省重大科技专项(17030701046)共同资助。 |
| |
| 摘 要： | 为解决小麦品质分级数字化检测的问题，通过对小麦的含水率、容重、杂质、不完善粒4个方面的检测建立小麦品质分级检测方法。分别设计了电容式传感器对小麦进行含水率检测，建立含水率检测数学模型，验证含水率检测最大误差范围为±0.4%。利用称重传感器测定小麦定容下的重量，标定容重参数，建立容重检测数学模型，验证误差范围±4 g·L-1。基于机器视觉对小麦不同形态杂质进行分类识别，构建SVM算法分类模型，鉴别不完善粒、各类杂质，其评价参数准确率、精准率、召回率、F1-Score数值分别达到96.5 %、96.0 %、96.4 %、96.2 %。通过对比验证表明，小麦含水率、容重、杂质、不完善籽粒四个方面的检测最大误差分别为±0.4%，±4 g·L-1，±0.15 %，±0.06 %，误差范围在合理区间，可以准确快速对小麦品质实现分级判定，为小麦品质分级判定提供了新的技术和方法。
|
| 关 键 词： | 小麦品质 数字化 机器视觉 含水率 容重 |
Research on digital non-destructive testing method for wheat quality grading |
| |
| LIU Guangzong,CAO Chengmao,ZHANG Jinyan,ZHOU Rundong,LI Wenbao.Research on digital non-destructive testing method for wheat quality grading[J].Journal of Anhui Agricultural University,2020,47(4):655. |
| |
| Authors: | LIU Guangzong CAO Chengmao ZHANG Jinyan ZHOU Rundong LI Wenbao |
| |
| Institution: | School of Engineering, Anhui Agricultural University, Hefei 230036; Anhui Electrical Engineering Professional Technique College, Hefei 230051 |
| |
| Abstract: | In order to solve the problem of digitized detection of wheat quality classification, a method for grading detection of wheat quality was established by detecting the moisture content, bulk density, impurities, and imperfect grains of wheat. Capacitive sensors were designed for moisture content detection in wheat, and a mathematical model for moisture content detection was established to verify that the maximum error range for moisture content detection is ±0.4%. The weight under constant volume of wheat was measured with a load cell, the volumetric parameters were calibrated, and a mathematical model for volumetric weight detection was established to verify the error range of ±4 g·L-1. Classification and recognition of different morphological impurities of wheat based on machine vision, construction of SVM algorithm classification model, identification of imperfect kernels and various impurities, the evaluation parameters accuracy rate, accuracy rate, recall rate, F1-Score value reached 96.5%, 96.0 % ,96.4 %, 96.2 %. The comparison and verification show that the maximum errors in the four aspects of wheat moisture content, bulk density, impurities, and imperfect grains are ±0.4%, ± 4 g·L-1, ±0.15 %, and ±0.06 %. The error ranges are within a reasonable range and can be accurate. The rapid classification of wheat quality is achieved, which provides new techniques and methods for the classification of wheat quality. |
| |
| Keywords: | wheat quality digitization machine vision water content bulk density |
|
| 点击此处可从《安徽农业大学学报》浏览原始摘要信息 |
| 点击此处可从《安徽农业大学学报》下载免费的PDF全文 |
|
