A sewage plant biochemical tank control quantity prediction method based on space-time prediction

By constructing a deep learning model with CNN, LSTM and Attention mechanisms, the shortcomings of traditional sewage treatment systems in terms of data quality and model adaptability are solved, and collaborative prediction of multiple control units is realized, thereby improving the prediction accuracy and response capability of the sewage treatment system.

CN121386637BActive Publication Date: 2026-07-03AI WO TE ZHI NENG SHUI WU (AN HUI) YOU XIAN GONG SI
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
AI WO TE ZHI NENG SHUI WU (AN HUI) YOU XIAN GONG SI
Filing Date
2025-09-03
Publication Date
2026-07-03

AI Technical Summary

Technical Problem

Traditional wastewater treatment systems have shortcomings in data quality assurance, model adaptability, control accuracy and response capability. They are unable to achieve spatiotemporal joint modeling of multi-source high-frequency time series data and full-process collaborative control, resulting in low prediction accuracy and difficulty in adapting to complex operating conditions.

Method used

We employ a deep learning model based on CNN, LSTM and Attention mechanisms. We extract spatial features through a one-dimensional convolutional neural network, capture temporal dependencies through a long short-term memory network, and introduce an attention mechanism for weighted feature fusion to achieve collaborative prediction by multiple control units.

Benefits of technology

It achieves minute-level high-frequency prediction, reduces energy and chemical consumption by 10%-20%, improves the intelligence level and control accuracy of the sewage treatment system, and adapts to sudden changes in complex working conditions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121386637B_ABST
    Figure CN121386637B_ABST
Patent Text Reader

Abstract

This invention discloses a method for predicting control quantities in the biological treatment tanks of wastewater treatment plants based on spatiotemporal prediction, comprising the following steps: S1, collecting multi-source time-series data from key control units of the wastewater treatment plant; S2, cleaning the collected raw time-series data; S3, using a one-dimensional convolutional neural network (CNN) to perform sliding convolution on the preprocessed time-series data to extract local spatial features and generate a spatial feature vector; S4, inputting the spatial features into a long short-term memory network (LSTM) to extract temporal dependency information, capturing long-short-term temporal dependencies, and generating a temporal feature vector; S5, constructing an attention mechanism to establish a correlation between the temporal features output by the LSTM and the spatial features output by the CNN, and calculating weight coefficients; S6, mapping the fused and weighted spatiotemporal feature vector to the required current control quantity prediction result and outputting it. By constructing a fusion model, a control paradigm shift from "passive response" to "active prediction" in the operation of wastewater treatment systems is realized.
Need to check novelty before this filing date? Find Prior Art