Enhancing Decentralized Energy Storage Investments with Artificial Intelligence-Driven Decision Models
Academic Background
As the global energy structure transitions towards renewable energy, the importance of decentralized energy storage is becoming increasingly prominent. Unlike traditional centralized energy storage systems, decentralized energy storage localizes the energy production and storage processes, reducing the risk of large-scale system failures and enhancing the continuity and flexibility of energy supply. However, the complexity and limited resources of decentralized energy storage projects make it difficult for companies to determine strategic priorities, which may lead to investment failures or inefficiencies.
To address this issue, the authors propose an artificial intelligence (AI)-driven decision model aimed at providing effective strategic guidance for decentralized energy storage investments. The study not only focuses on optimizing investment decisions but also improves the consistency and efficiency of decision-making by introducing information gain and mass expert selection techniques.
Source of the Paper
This paper was co-authored by Gang Kou, Hasan Dinçer, Edanur Ergün, Serkan Eti, Serhat Yüksel, and Ümit Hacıoğlu, and published in the 2025 issue of the journal Artificial Intelligence Review, with the DOI 10.1007/s10462-025-11204-y. The research team is affiliated with multiple institutions, including renowned universities and research centers in China, Turkey, and Europe.
Research Process
1. Mass Expert Selection
The study first screened three of the most representative experts from a pool of eight using information gain techniques. This process analyzed input data such as educational background, work experience, salary, and age to calculate their impact on decision criteria, thereby selecting the most relevant experts.
2. Balancing Expert Evaluations
Using the Q-learning algorithm, the opinions of the best expert were balanced with the evaluations of other experts. This process adjusted the weights of the experts through a reward and punishment mechanism, ultimately achieving consistent evaluation results.
3. Molecular Fuzzy Cognitive Maps (MF Cognitive Maps)
The study employed Molecular Fuzzy Cognitive Maps (MF-CM) to analyze and prioritize key factors in decentralized energy storage investments. This method combines fuzzy logic with molecular geometry to better handle uncertainties in complex systems.
4. Multi-Objective Particle Swarm Optimization (MF-MOPSO)
Finally, the study used Molecular Fuzzy Multi-Objective Particle Swarm Optimization (MF-MOPSO) to rank alternative decentralized energy storage investment strategies. This method optimizes multiple objective functions by simulating the movement of particles in a search space, ultimately determining the optimal investment solution.
Key Results
1. Expert Selection and Evaluation
Through information gain techniques, the study successfully screened three of the most representative experts from a pool of eight and balanced their evaluations using the Q-learning algorithm. This process significantly improved the consistency and efficiency of decision-making.
2. Prioritization of Key Factors
The analysis using Molecular Fuzzy Cognitive Maps revealed that customer expectations (weight: 0.2577) and financial issues (weight: 0.2513) are the most critical factors affecting the performance of decentralized energy storage investments. This finding provides important guidance for companies in formulating investment strategies.
3. Ranking of Investment Strategies
Using the MF-MOPSO method, the study ranked five decentralized energy storage investment strategies. The results showed that hydrogen-based energy storage (average score: 0.1878) and distributed battery swapping stations (average score: 0.1877) are the most promising investment strategies.
Conclusions and Value
This study successfully addressed the complexity and uncertainty in decentralized energy storage investments by introducing an AI-driven decision model. The main contributions of the study include: 1. Improved Decision Efficiency: The use of information gain and the Q-learning algorithm significantly enhanced the consistency and efficiency of decision-making. 2. Optimized Investment Strategies: The Molecular Fuzzy Cognitive Maps and MF-MOPSO methods provided companies with scientific and rational investment strategies. 3. Promotion of Sustainable Development: The application of hydrogen-based energy storage and other strategies helps achieve sustainable energy goals and reduces reliance on fossil fuels.
Research Highlights
- Innovative Methods: The study pioneered the integration of molecular fuzzy logic with multi-objective optimization techniques, proposing the MF-CM and MF-MOPSO methods, which significantly improved the accuracy and reliability of decision-making.
- Practical Application Value: The research results are directly applicable to decentralized energy storage investments, providing actionable strategic guidance for companies.
- Interdisciplinary Approach: The study integrates multiple disciplines, including artificial intelligence, fuzzy logic, and energy management, demonstrating the immense potential of interdisciplinary research.
Other Valuable Information
The study also conducted sensitivity analysis by altering criterion weights, verifying the stability and reliability of the results. This analysis further enhanced the credibility of the research conclusions.
This paper not only provides a scientific basis for decentralized energy storage investments but also offers new insights and methods for decision optimization in other complex systems. Its innovation and practicality make it a significant research achievement in the fields of artificial intelligence and energy management.