Prediction of Dissolved Oxygen Concentration in Sewage Treatment Process Based on Data Recognition Algorithm.

Area of Science:

• Environmental Engineering
• Computational Intelligence
• Process Control

Background:

• Accurate real-time prediction of dissolved oxygen (DO) is crucial for effective sewage treatment.
• Traditional methods face challenges in handling complex data characteristics and achieving high accuracy.

Purpose of the Study:

• To develop a novel data recognition algorithm for real-time and accurate prediction of DO concentration in sewage treatment.
• To enhance prediction model performance by improving data selection and optimization strategies.

Main Methods:

• A data identification algorithm incorporating a new sample similarity measure for representative data extraction.
• Integration of the chaos algorithm with the fireworks algorithm (FWA) to create a chaotic fireworks algorithm (CFWA).
• Implementation of a two-stage optimization process with simultaneous group and individual search mechanisms.

Main Results:

• The CFWA algorithm demonstrated superior performance compared to the basic FWA algorithm.
• CFWA effectively utilized chaotic search, avoiding random initial weight selection and leveraging strengths of both FWA and chaos optimization.
• CFWA model achieved lower training and generalization errors in soft sensor simulations, validating its effectiveness.

Conclusions:

• The proposed data recognition algorithm, CFWA, effectively predicts dissolved oxygen concentration in wastewater treatment.
• This approach offers a new measurement method for difficult-to-measure process variables in complex chemical processes.
• The study validates the efficacy of integrating chaotic mechanisms into optimization algorithms for enhanced predictive modeling.