Abstract:The process-based water system models have been transitioning from single-functional to integrated multi-objective and multi-functional since the worldwide digital upgrade of urban water system management. The proliferation of model complexity results in more significant uncertainty and computational requirements. However, conventional model calibration methods are insufficient in dealing with extensive computational time and limited monitoring samples. Here we introduce a novel machine learning system designed to expedite parameter optimization with limited data and boost efficiency in parameter search. MLPS, termed the machine learning parallel system for fast parameter search of integrated process-based models, aims to enhance both the performance and efficiency of the integrated model by ensuring its comprehensiveness, accuracy, and stability. MLPS was constructed upon the concept of model surrogation + algorithm optimization using Ant Colony Optimization (ACO) coupled with Long Short-Term Memory (LSTM). The optimization results of the Integrated sewer network and urban river model demonstrate that the average relative percentage difference of the predicted river pollutant concentrations increases from 1.1 to 6.0, and the average absolute percent bias decreases from 124.3% to 8.8%. The model outputs closely align with the monitoring data, and parameter calibration time is reduced by 89.94%. MLPS enables the efficient optimization of integrated process-based models, facilitating the application of highly precise complex models in environmental management. The design of MLPS also presents valuable insights for optimizing complex models in other fields.

A novel supervised multi-model modeling method based on k-means clustering

Multi-Hierarchical Model Predictive Control Based on K-Means Clustering Algorithms

Multi-Model Predictive Control Based on a New Clustering Modeling Method

A New Switching Scheme for Multi-model Predictive Control Using Clustering Modeling

Research on multi-mode MPC based on clustering multi-modeling

Nonlinear adaptive control based on RBF networks and multi-model method

Nonlinear semi-parametric modeling method based on GA-ANN

Nonlinear Semi-Parametric Modeling Mothed Based on GA-ANN

Adaptive Soft Sensor Modeling Method Based on Multi-Model Dynamic Fusion and Its Industrial Application.

Soft Sensor Modeling of Sewage Disposal Process Based on Multi-model Fuzzy Kernel Clustering Method

Research on Model Switching Method of Multi-hierarchical Model Predictive Control System

A Multi-Model Modeling Approach to MIMO Nonlinear Systems

Nonlinear Model Predictive Control Based on Support Vector Machine with Multi-kernel

Novel Multimode Process Soft Sensing Methods Based on the Dynamic Mixture Variational Autoencoder Regression Model

A Multi-Model Modeling Approach To Nonlinear Systems Based On Lazy Learning

Multiple Neural Network Model Based on Data Partition Using Feature Clustering

Multi-Relational Data Semi-Supervised K-Means Clustering Algorithm

Multi-View Clustering and Semi-Supervised Classification with Adaptive Neighbours

Machine Learning Parallel System for Integrated Process-Model Calibration and Accuracy Enhancement in Sewer-River System

A novel monitoring method based on multi-model information extraction and fusion

Application of novel hybrid deep leaning model for cleaner production in a paper industrial wastewater treatment system