基于大数据统计的土壤养分含量预测模型设计
2020-08-03王廷超王国伟安宇
王廷超 王国伟 安宇



摘 要: 针对传统预测模型预测土壤养分含量精度低的问题,设计基于大数据统计的土壤养分含量预测模型。使用光谱采集仪采集土壤养分含量数据,对数据进行平滑、标准化和正交化处理,消除影响预测精的数据。对处理后的光谱图进行分析,总结土壤养分大数据规律,明确在反射曲线突变处的不同波段对应不同含量的土壤养分。重复多次后,完成对模型精度的校验,实现基于大数据统计的土壤养分含量预测模型的设计。通过与传统模型的对比实验,验证设计的模型能够提高大约3倍的预测精度,更适用于对土壤养分含量进行预测。
关键词: 预测模型; 土壤养分含量; 大数据统计; 模型设计; 光谱分析; 模型测试
中图分类号: TN911.34?34; TP273 文献标识码: A 文章编号: 1004?373X(2020)08?0012?03
Design of soil nutrient content prediction model based on big data statistics
WANG Tingchao, WANG Guowei, AN Yu
(College of Information and Technology Science, Jilin Agricultural University, Changchun 130118, China)
Abstract: A soil nutrient content prediction model based on big data statistics is designed to improve the low accuracy of traditional prediction model in predicting soil nutrient content. The data of soil nutrient content is collected with the spectral acquisition instrument, and the data is smoothed, standardized and normalized to eliminate the influence of prediction precision. The processed spectrogram is analyzed to summarize the rule of big data of soil nutrients, and it is clear that different bands at the abrupt change of reflection curve correspond to different contents of soil nutrients. The checking of model precision is completed after repeated for many times, and the design of soil nutrient content prediction model based on big data statistics is realized. The experimental results verify that, in comparison with the traditional model, the designed model can increase the prediction accuracy by about 3 times, and is more suitable for the prediction of soil nutrient content.
Keywords: prediction model; soil nutrient content; big data statistics; model design; spectrum analysis; model measurement
0 引 言
土壤養分含量影响当地植被、作物的生长态势,进而影响土壤所在地的水土条件。充分了解土壤养分含量,对于农业发展意义重大。土壤养分含量决定土地生产力,通过对其的预测,可以更科学地决策土地使用方式,选择适宜技术手段对土壤养分含量进行调整[1]。随着现代科学技术快速发展,遥感、地统计学等方法手段,亦可实现对土壤养分含量的预测。传统土壤养分含量预测模型,由于使用实验室分析数据,不仅在数据采集阶段耗费时间长、成本高,而且因为预测模型预测依据数据不具有实时性,导致预测精度低、预测周期长。
大数据统计是海量数据分析数据间规律的基础。……
