基于温度感知RFID标签的冷链厢体中温度监测

针对温度感知RFID(radio frequency identification)标签应用于冷链物流温度监测中缺乏有效数据验证的问题,该研究通过将42个温度感知RFID标签部署于冷链模拟平台中,划分了7个横截面、3个纵截面和两个层,设置了机械降温-冷链维持-自然回温3个不同阶段,同时在42个监测位点中选择7个位点同步部署了便携式温度记录仪,获取了不同条件下的温度监测数据,并与便携式温度记录仪数据和CFD(computational fluid dynamics)模拟数据进行了比较。7个温度感知RFID标签与便携式温度记录仪同步监测位点的数据表明,两种监测方法温差分布于±0.5℃范围内的数据点最多,占43.6%,温差分布于-1.0~-0.5℃区间的数据占了24.6%,考虑到2种设备自身的温度采集精度,温差在±0.8℃范围内可接的,其比例占71.3%,因此利用温度感知RFID标签进行冷链温度监测是可行的。对42个位点在3个不同阶段的温度监测数据表明,机械降温阶段各位点用时在1 h以内、冷链维持阶段大部分位点表现出温度在在0~4℃之间振荡的特征、自然回温阶段用时约49 h。深入分析机械降温阶段及冷链维持阶段不同截面的温度监测数据,结果表明3种截面均表现为降温初始阶段温度差值不稳定、稳定后具有明显的分布特征且离出风口较近降温较快的特点。以横截面2和横截面6平均温度为例,将温度感知RFID标签数据采集数据与CFD模拟数据进行比较,去除测量精度的干扰,截面2的均方根误差为0.73℃、平均相对误差为13.58%、截面6的均方根误差为0.56℃、平均相对误差为10.94%,具有较好的空间一致性。研究结果可为实现冷链物流中低成本、连续的温度监测奠定基础。

Temperature monitoring in cold chain chamber based on temperature sensing RFID labels

Temperature is the core of cold chain and its monitoring plays a fundamental role on controlling temperature and reducing energy consumption comfortably. With the development of information and communication technology, the monitoring of cold chain temperature has been developed from single point to multipoint, from wired to wireless, and from delayed to real-time. Compared with the temperature monitoring based on WSN (wireless sensor network), the temperature monitoring based on RFID (radio frequency identification) has the characteristics of low cost, low energy consumption and high flexibility. For the problem of lack of effective data validation with RFID temperature tags applied in the processing of cold chain logistics temperature monitoring, a cold chain simulation chamber was used for this study. The chamber's wall thickness is 15 cm and the refrigeration unit is fixed at the lower part of a circulating fan with a diameter of about 0.30 m. Twelve boxes of apples were placed in the middle of the chamber to provide a stable source of respiratory heat. Temperature monitoring experiment was designed through deploying 42 temperature sensing RFID tags in the chamber. The chamber space was divided into 3 types of virtual sections, i.e. 7 cross surfaces, 3 longitudinal sections and 2 layers. The experiment was lasting for 69 h from 14:00 on December 7 to 11:00 on December 10 in 2016. It was split into 3 stages according to the temperature's diversification, which were mechanical cooling, cold chain keeping and temperature recovering. Simultaneously, 7 portable temperature recorders were deployed in the sites selected from 42 temperature sensing RFID tags points to record the temperature, and the difference between the 2 temperature monitoring methods was compared. In addition, temperature spatial simulation data with CFD (Computational Fluid Dynamics) were compared with the temperature sensing RFID tag data to analyze spatial difference. The recording interval was 4 min. Characteristics of temporal and spatial variation were analyzed through the experiment and using Microsoft Excel 2010 software, CFD simulation software and Fluent 15.0 solver. At each point 1035 data were collected, and 43470 RFID tags data for 42 points were collected in the experiment. Seven synchronous data from portable temperature recorders and temperature sensing RFID tags showed that temperature difference between the 2 temperature monitoring methods in the range from -1.0 to -0.5 ℃ and from -0.5 to +0.5 ℃ accounted for 24.6% and 43.6%respectively. The result indicates that it is feasible to use RFID tags for temperature monitoring in cold chain processing. Further, temperature monitoring data from 42 RFID tags in the 3 different stages showed that, time consumption in the mechanical cooling stage was less than 1 h, temperature fluctuation in the cold chain keeping stage between 0 to 4 ℃ was obvious, and time consumption in the temperature recovering stage was about 49 h. Through deeper analysis on the temperature monitoring data of different virtual sections from the mechanical cooling stage and the cold chain keeping stage, the result showed that temperature difference in the initial cooling stage was unstable and spatial distribution in the temperature calm stage was significant. Specially, the feature that the site nearer to the outlet had faster cooling was evident. Taking the average temperature of T2 (No.2 cross surface) and T6 (No.6 cross surface) as an example, the data acquired by temperature sensing RFID tag were compared with CFD simulation data. Except for the interference of measurement precision, RMSE (root-mean-square-error) and ARD (average-relative-deviation) of T2 were 0.73 ℃ and 13.58%. RMSE and ARD of T6 were 0.56℃ and 10.94%. Spatial consistency was verified. The research result can provide the reference for achieving low cost and continuous temperature monitoring in cold chain logistics. The follow-up studies can be enhanced by improving package design label or the use of high precision sensor, with better application in the whole cold chain of tracking.