Soft Sensor Modeling of Self-Organizing Interval Type-2 Fuzzy Neural Network Based on Adaptive Quantum-Behaved Particle Swarm Optimization Algorithm

In this study, a self-organizing interval type-2 fuzzy neural network (SOIT2FNN) based on the Adaptive Quantum-behaved Particle Swarm Optimization (AQPSO) algorithm is proposed as a solution to the soft sensor modeling problem of effluent total phosphorus in the wastewater treatment process. An AQPS...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:International journal of fuzzy systems Ročník 26; číslo 5; s. 1716 - 1729
Hlavní autori: Wang, Panchao, Zhao, Taoyan, Cao, Jiangtao, Li, Ping
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: Heidelberg Springer Nature B.V 01.07.2024
Predmet:
Accuracy
Adaptive algorithms
Analytical chemistry
Artificial neural networks
Eutrophication
Fuzzy logic
Genetic algorithms
Modelling
Network topologies
Neural networks
Optimization algorithms
Parameter robustness
Particle swarm optimization
Phosphorus
Process parameters
Sensors
Wastewater treatment
Water treatment
ISSN:1562-2479, 2199-3211
On-line prístup:Získať plný text
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Popis
Shrnutí:In this study, a self-organizing interval type-2 fuzzy neural network (SOIT2FNN) based on the Adaptive Quantum-behaved Particle Swarm Optimization (AQPSO) algorithm is proposed as a solution to the soft sensor modeling problem of effluent total phosphorus in the wastewater treatment process. An AQPSO algorithm is proposed to train and optimize the SOIT2FNN soft sensor model's structure and parameters in order to improve forecasting accuracy. Firstly the number of rules and parameters of SOIT2FNN are mapped to the spatial position of each particle of AQPSO, which is convenient for the parameter and structure optimization of the network model. The weighted average optimal position and the adaptive shrinkage coefficient are introduced in order to enhance the global search ability of PSO and ensure the robustness and convergence of the algorithm during the optimization process. By dynamically adjusting the network structure, the network size of each particle approximates the network size of an ideal particle, thereby improving the search efficiency and optimization results of the model. Finally, the proposed AQPSO-SOIT2FNN was applied to the Mackey–Glass chaotic time series prediction and soft sensor modeling of total phosphorus in wastewater treatment effluent. The results of the experiments show that the proposed AQPSO-SOIT2FNN is capable of producing a relatively compact network topology with high identification and prediction accuracy. address uncertainty satisfactorily, which is computationally more implementable.
Bibliografia:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1562-2479
2199-3211
DOI:10.1007/s40815-024-01701-7