An Adaptive Optimization Spiking Neural P System for Binary Problems.

Area of Science:

• Membrane computing
• Artificial intelligence
• Optimization algorithms

Background:

• The Optimization Spiking Neural P System (OSNPS) is a pioneering model for combinatorial optimization, particularly the 0/1 knapsack problem.
• OSNPS utilizes parallel Spiking Neural P Systems (SNPS) for solution generation and a Guider algorithm for probability adjustment.
• However, OSNPS's effectiveness diminishes in high-dimensional problems, limiting its practical application.

Purpose of the Study:

• To enhance the performance of OSNPS for complex combinatorial problems.
• To introduce a novel Dynamic Guider algorithm with adaptive learning and diversity-based adaptation.
• To propose the Adaptive Optimization Spiking Neural P System (AOSNPS) as a more robust optimization model.

Main Methods:

• Development of a novel Dynamic Guider algorithm incorporating adaptive learning and diversity-based adaptation.
• Implementation of the Adaptive Optimization Spiking Neural P System (AOSNPS) model.
• Testing AOSNPS on 0/1 knapsack problems of varying dimensionality and power system fault estimation case studies.

Main Results:

• AOSNPS demonstrates superior effectiveness in solving 0/1 knapsack problems, outperforming existing algorithms, especially in high-dimensional instances.
• The model shows significant efficacy in estimating fault sections within power systems (IEEE 39 and 118 bus systems).
• AOSNPS successfully handles various fault scenarios, including multiple and uncertain faults.

Conclusions:

• AOSNPS represents a significant advancement in membrane computing for optimization, overcoming the limitations of previous models.
• The proposed adaptive strategies in the Dynamic Guider algorithm are crucial for improved performance.
• AOSNPS is a versatile and effective tool for both combinatorial optimization and critical power system analysis.