| 本文已被：浏览 911次 下载 781次 |
码上扫一扫！ |
| 基于微波反射法的冬笋探测器设计 |
|
王俊楠1,吕艳1,2,倪忠进1,2,黄政晖1,倪益华1,3*
|
|
|
| (1.浙江农林大学 工程学院, 杭州 311300;2.浙江省竹资源与高效利用协同创新中心, 杭州 311300;3.国家林业局林业感知技术与智能装备重点实验室, 杭州 311300) |
|
| 摘要: |
| 针对目前我国冬笋采收只能依靠传统人工经验法，没有合适的探测设备的问题，根据高频电磁波作用下冬笋与土壤的介电特性差异导致反射系数显著不同的原理，设计基于微波反射法的冬笋探测器。该探测器通过2块贴片天线向土壤发射并接收电磁波信号；构建基于卷积神经网络的回波序列检测模型，用于所设计冬笋探测器接收回波信号的分类，利用回波信号的电磁波衰减强度判断土壤内是否有冬笋存在。试验结果表明，本研究提出的回波序列检测模型通过冬笋探测器回波信号判断土壤内有无冬笋的准确率为79.08%，漏警率为21.47%，虚警率为20.37%，与支持向量机以及频谱阈值法相比，回波序列检测模型在冬笋探测的准确率、虚警率、漏警率方面均为最优。该探测器能对地下冬笋位置进行预测，可提高冬笋采收效率。 |
| 关键词: 冬笋探测器 介电常数 微波反射法 卷积神经网络(CNN) |
| DOI：10.11841/j.issn.1007-4333.2021.09.19 |
| 投稿时间：2020-11-24 |
| 基金项目:“十三五”国家重点研发计划(2018YFD0701103)；浙江省重大科技专项(2016C02G2100540),浙江省自然科学基金项目(LQ16E050013)；浙江省竹资源与高效利用协同创新中心项目(2017ZZY2-04) |
|
| Design of winter bamboo shoot detector based on microwave reflection method |
|
WANG Junnan1,LV Yan1,2,NI Zhongjin1,2,HUANG Zhenghui1,NI Yihua1,3*
|
| (1.College of Engineering, Zhejiang A & F University, Hangzhou 311300, China;2. Zhejiang Province Bamboo Resources and Efficient Utilization Collaborative Innovation Center, Hangzhou 311300, China;3.Key Laboratory of Forestry Sensing Technology and Intelligent Equipment of State Forestry Administration, Hangzhou 311300, China) |
| Abstract: |
| At present, the harvest of winter bamboo shoots in China can only rely on traditional manual work, and there is no suitable detection equipment. According to the principle that the difference in dielectric properties between winter bamboo shoots and soil under the action of high-frequency electromagnetic waves results in significantly different reflection coefficients, a winter bamboo shoot detector is designed based on the microwave reflection method. The detector transmits and receives electromagnetic wave signals to the soil through two patch antennas. It constructs an echo sequence detection model based on a conventional neural network used to classify the echo signals received by the designed winter bamboo shoot detector and uses the echo signal's electromagnetic waves. The attenuation intensity judges whether there are winter bamboo shoots in the soil. The results show that the echo sequence detection model proposed in this study uses the winter bamboo shoot detector's echo signal to judge the presence of winter bamboo shoots in the soil with an accuracy rate of 79. 08%, missed alarm rate is 21. 47%. The false alarm rate is at 20. 37%. Compared with the SVM and the spectral threshold method, the performance of the echo sequence detection model is the best in accuracy, false alarm rate, and missed alarm rate of winter bamboo shoots detection. The detector can predict underground winter bamboo shoots' location and improve winter bamboo shoots' harvesting efficiency. |
| Key words: winter bamboo shoots detector dielectric constant microwave reflection method convolutional neural network(CNN) |
