Study on the Prediction Model Based on a Portable Soil TN Detector - Computer and Computing Technologies in Agriculture IX - Part II Access content directly
Conference Papers Year : 2016

Study on the Prediction Model Based on a Portable Soil TN Detector


As the development of precision agriculture, it is necessary to obtain soil total nitrogen (TN) content and other element parameters. With the NIRS technology, a soil detector for soil total nitrogen content was developed. It included two part: optical part and control part. The detector took each lamp-house connected with the incidence of Y type optical fiber in turn by a manual rotation, The different wavelength lamp-house signal was transferred to the surface of soil by the input fibre. The reflected signal would be converted by photoelectric sensor, the optical signal was converted to electrical signal. After the power circuit, amplier circuit, and AD convert circuit, the electrical signal was processed by MCU. Finally, the result of soil total nitrogen content could be displayed on LCD. With the forty-eight apple orchard soil samples of Beijing surburb, the predicted models were established by seven different methods (MLR, PLSR, SVM, BPNN, GA + BPNN, GA + SVM and PSO + SVM). The model established by genetic algorithm (GA) optimizing BP neural network was optimal, with RC of 0.94, RV of 0.78, RMSEC and RMSEP of 0.037 and 0.067. The results showed that the soil total nitrogen content detector had a stable performance. The established model had perfect accuracy and strong robustness.
Fichier principal
Vignette du fichier
434298_1_En_12_Chapter.pdf (357.15 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-01614165 , version 1 (10-10-2017)





Xiaofei An, Guangwei Wu, Jianjun Dong, Jianhua Guo, Zhijun Meng. Study on the Prediction Model Based on a Portable Soil TN Detector. 9th International Conference on Computer and Computing Technologies in Agriculture (CCTA), Sep 2015, Beijing, China. pp.117-126, ⟨10.1007/978-3-319-48354-2_12⟩. ⟨hal-01614165⟩
104 View
80 Download



Gmail Facebook X LinkedIn More