Adaptive quantum inspired deep reinforcement learning for multi objective low carbon CCHP optimization | JoVE Visualize

Area of Science:

Energy Systems Engineering
Artificial Intelligence
Optimization Theory

Background:

Combined Cooling, Heating, and Power (CCHP) systems offer energy efficiency but face complex scheduling challenges.
Low-carbon energy management requires sophisticated optimization for greenhouse gas emission reduction and energy saving.
Integrating advanced computational methods like deep reinforcement learning and digital twins is crucial for optimizing CCHP operations.

Purpose of the Study:

To propose a novel deep reinforcement learning (DRL)-based multi-objective optimization framework for low-carbon CCHP systems.
To enhance scheduling efficiency by incorporating adaptive Quantum-Inspired Evolutionary Algorithms (QIEA) and digital twin technology.
To address carbon emission flow, carbon trading, and demand response within the CCHP scheduling framework.

Main Methods:

Developed an Adaptive Quantum-DRL Multi-Objective Optimization (AQ-DRLMO) model integrating QIEA and digital twins.
Utilized Physics-Informed Neural Networks (PINNs) within a hierarchical digital twin for predictive optimization.
Introduced enhanced control strategies (ETEF, EEF, TEF) with attention-based transformer networks and a Carbon-Conscious Optimal Power Flow (C-OPF) model.

Main Results:

The AQ-DRLMO framework achieved significant reductions: 40.08% in greenhouse gas emissions, 34.04% in primary energy saving, and 24.44% in cost.
Quantum-inspired optimization demonstrated 67.3% faster convergence (45 iterations vs. 137) compared to standard genetic algorithms.
The framework maintained solution diversity on the Pareto front across different control strategies and seasonal conditions.

Conclusions:

The proposed AQ-DRLMO framework is a promising simulation-based solution for day-ahead scheduling in low-carbon CCHP systems.
It offers substantial improvements in energy efficiency, cost reduction, and environmental impact.
Further field validation is recommended for real-time control applications in smart grids and distributed energy systems.