农田水势软测量中的辅助变量选择与数据预处理  

Secondary Variables Selection and Data Pretreatment in Soft-sensing for the Farmland Water Potential

作　　者：卢胜利[1] 王菁华[1] 储健[1] 邹杨[1] 潘艳梅[1] 孙文君[2] 

机构地区：[1]天津工程师范学院自动化系,天津300222 [2]天津津南国家农业科技园区,天津300350

出　　处：《现代科学仪器》2009年第4期69-72,共4页Modern Scientific Instruments

基　　金：国家自然科学基金项目(农田水势软测量技术及装置研究60772167)

摘　　要：连续自动监测土壤-植物-大气连续体中的水势并据此了解农作物水分胁迫状态是真正精准灌溉(PI)的重要前提。鉴于作物活体水势自动监测较为困难,提出基于作物活体微环境信息的软测量技术。基于SPAC水分传输机理和作物微环境数据特征分析,选择若干种易测微环境信息软测量的辅助变量;数据预处理则采用了聚类分析等新方法,在侦测异常数据和滤除随机噪声方面取得了良好的效果。Continuous automatic measurement of wate.r potential of the soil-plant-atmosphere continuum （SPAC） and understanding of crop water-stress is the most important premise of genuine precision irrigation （PI）. In the view of the difficulty in auto-measurement of living crop water potential, the soft-sensing technique based on the microenvironment information around living crop is proposed. Based on mechanism of water transfer in SPAC and the sampled data analysis for microenvironment information around living crop, some easily measurable microenvironment information are selected as the secondary variables of the soft-sensing model. Some new methods such as cluster analysis are applied in dada pretreatment and proved to have good effect for abnormal data recognition and random noise filtering.

关 键 词：SPAC 农田水势 微环境 软测量 辅助变量 数据预处理 

分 类 号：TP274.4[自动化与计算机技术—检测技术与自动化装置]