Identifying Uncertainties In Air Temperature Data Of An Indoor Farming System

A small growth chamber was used to grow lettuce for 5 weeks during four air temperature trials, and an automated control system was used to control the environmental conditions of the air and root zone for the plants. A sensor array made up of low-cost arduino connected sensors would collect aerial (air temperature, relative humidity, CO₂ concentration) and root-zone data (water temperature, pH, EC, DO) and control the hydroponic system and carbon dioxide enrichment, while a reference sensor was used to collect aerial environmental conditions (air temperature, relative humidity, CO₂ concentration) to compare with the low-cost data sets. We hypothesized that our alternative decomposition method would successfully identify uncertainty occurrences in the data collected throughout this experiment since this data had many gaps in data when the data collection system would stop functioning or for other uncertainties. The small size of the growth chamber would also make any agricultural operations (any actions where humans would enter/exit and be inside the small chamber for any time), sensor failures, or other such uncertainties have a significant impact on system operations and reliability, making this decomposition method necessary for data quality control. Only the air temperature data from the low-cost and reference sensors was used to test the alternative decomposition method since the standard decomposition methods failed to successfully de-seasonalize the data.

本研究采用小型生长箱开展生菜栽培试验，设置4组空气温度处理，栽培周期为5周；同时通过自动化控制系统调控植株的空气与根际环境参数。由低成本且经Arduino（Arduino）连接的传感器组成的传感器阵列，可采集空气环境参数（空气温度、相对湿度、二氧化碳浓度）与根际环境参数（水温、pH值、电导率（EC）、溶解氧（DO）），并调控水培系统与二氧化碳增施流程；同时配置参考传感器采集空气环境参数（空气温度、相对湿度、二氧化碳浓度），用于与低成本传感器阵列的数据集进行比对。本研究假设，所提出的替代分解方法可成功识别本试验采集数据中的不确定性异常事件；究其原因，当数据采集系统停机或存在其他不确定性因素时，该数据集存在大量数据缺失。由于该生长箱体积较小，任何农事操作（即人员进入、离开并在箱内停留的各类操作）、传感器故障或其他类似不确定性事件，均会对系统运行与可靠性造成显著影响，因此该分解方法对于数据质量控制而言不可或缺。鉴于标准分解方法无法有效去除该数据的季节性趋势，本研究仅采用低成本传感器与参考传感器采集的空气温度数据，对替代分解方法进行测试。