摘要
在算法优化层面,研究大气颗粒污染物浓度物联网监测方法。使用差值升维算法治理PM_(10)和PM_(2.5)物联网探头采集的单列数据,并沿时间戳指针变量,对每个数据记录分别进行神经网络分析,同时利用最小二乘法的R^(2)值数据作为循环控制变量,使输出值的精度达到R^(2)>0.990的精度水平。经过该算法处理后,三种探头的非线性结果R2值均超过0.990,该方法可有效提升低精度探头的测量精度,提高低精度探头的实际测量结果的误差控制能力,且具有更好的污染物监测数据处理精度。
The monitoring method of atmospheric particle pollutant concentration by Internet of Things is studied in the aspect of algorithm optimization.The single-column data collected by PM_(10) and PM_(2.5) IoT probes were treated with the difference rising dimension algorithm,and the neural network analysis was performed on each data record by pointing variables along the time stamp.Meanwhile,the R^(2) value data of least square method was used as the cyclic control variable,so that the accuracy of the output value reached the accuracy level of R^(2)>0.990.After the algorithm processing,the nonlinear R2 values of the three probes are all more than 0.990.This method can effectively improve the measurement accuracy of the low-precision probes,improve the error control ability of the actual measurement results of the low-precision probes,and have better processing accuracy of pollutant monitoring data.
作者
李文敏
Li Wenmin(Gansu Longnan Ecological Environment Monitoring Center, Longnan 746000, China)
出处
《环境科学与管理》
CAS
2021年第6期130-134,共5页
Environmental Science and Management
