Optimized deep belief network based on an improved Blood-sucking Leech Optimizer algorithm for wastewater quality...

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Machine Learning

Background:

• The Blood-Sucking Leech Optimizer (BSLO) algorithm faces challenges with limited exploration and premature convergence.
• Effective optimization is crucial for complex problem-solving in various scientific domains.

Purpose of the Study:

• To introduce an Improved Blood-Sucking Leech Optimizer (IBSLO) algorithm to overcome the limitations of the original BSLO.
• To enhance the algorithm's exploitative and exploratory capabilities for more accurate and efficient optimization.
• To validate the IBSLO algorithm's performance on benchmark functions and a practical water quality prediction task.

Main Methods:

• Developed a directional leeches switching mechanism using an inverted S-shaped nonlinear perceived distance.
• Incorporated a dynamic perception signal to guide the search and optimization process.
• Integrated a memory sharing mechanism to improve search efficiency and ensure global optimal solutions.

Main Results:

• The IBSLO algorithm demonstrated superior performance on 23 benchmark functions and the CEC-2017 test set.
• Convergence analysis confirmed the algorithm's enhanced capabilities.
• The IBSLO-Deep Belief Network model achieved superior accuracy in predicting crucial water quality parameters compared to other optimization strategies.

Conclusions:

• The proposed IBSLO algorithm effectively addresses the limitations of the BSLO, offering improved exploration and convergence properties.
• IBSLO shows significant potential for application in complex optimization problems, including environmental monitoring and predictive modeling.
• The enhanced optimization approach provides a robust solution for accurate prediction of water quality parameters in wastewater treatment.