YangHao,YangBiaolong,LiuHaiqing,et al.Data-Model Driven Forecasting and Analysis of Influent Data in Wastewater Treatment Plant[J].China Water & Wastewater,2026,42(13):55-62.
Click Copy

Data-Model Driven Forecasting and Analysis of Influent Data in Wastewater Treatment Plant

China Water & Wastewater[ISSN:1000-4062/CN:12-1073/TU] volume: 第42卷 Number: 第13期 Page: 55-62 Column: Date of publication: 2026-07-01
Author(s):
Yang　Hao1; 2; 3;  Yang　Biaolong1;  Liu　Haiqing4;  Wang　Baoshan1; 2; 3;  Zhou　Tianhong1; 2; 3;  Wu　Nan1; 2; 3;  Zhang　Tingting1
（1. School of Environmental and Municipal Engineering， Lanzhou Jiaotong University， Lanzhou 730070， China； 2. Ministry of Education Engineering Research Center of Water Resource Comprehensive Utilization in Cold and Arid Regions， Lanzhou Jiaotong University， Lanzhou 730070， China； 3. Key Laboratory of Yellow River Water Environment in Gansu Province， Lanzhou 730070， China； 4. Gansu Water Investment Zhuanglang Environmental Protection Limited Liability Company， Pingliang 744000， China）
Keywords:
urban wastewater treatment plant;  influent data forecasting;  Prophet model;  energy saving and emission reduction
Abstract:
The raw influent data from March 2020 to August 2023 of a county-level wastewater treatment plant (WWTP) in Northwest China were preprocessed and subjected to exploratory analysis. Using this dataset as the training set, the influent data of the plant were forecasted with the open-source Prophet model based on R language. The predicted results were then compared and validated against the actual influent data. The results showed that the raw influent quantity and quality data exhibited significant autocorrelation, indicating the presence of temporal dependencies, and the overall distributions influent quantity and quality data approximate normality. The Prophet model demonstrated a certain level of validity and accuracy in interpreting and forecasting influent data, with particularly excellent performance in predicting influent quantity achieving a mean absolute percentage error (MAPE) of 4.55%. The prediction results revealed that the influent quantity of the WWTP had significant multi-seasonal variation characteristics and a notable holiday effect, with an overall increasing trend. Influent COD and TP showed a general decreasing trend, whereas NH3-N and TN displayed an overall increasing trend. COD and TN exhibited significant seasonal variations, peaking in winter, while NH3-N and TP showed smaller fluctuations and less obvious periodic patterns. The forecast indicates that from September 2027 to September 2028, the average influent quantity of the plant will reach 18 541 m3/d, with influent COD, TN, NH3-N, and TP concentrations of 1 334.4 mg/L, 184.6 mg/L, 146.6 mg/L, and 15.7 mg/L, respectively. These data can provide an important basis for further investment planning of the plant.
Last Update: 2026-07-01