Boosting operational optimization of multi-energy systems by artificial neural nets

The operation of multi-energy systems has to be optimized repeatedly, e.g., react changing energy prices. Thus, operational optimization problems need solved in a reliably short time. Reliably computations are challenging for optimizing due complex time-coupling constraints. These constraints reflect effects such as component start costs or storage. However, increase the computational effort. Here, we present decomposition method solve using artificial neural nets efficiently. decomposes into single-time-step optimizations. optimizations incorporate predictions from networks trained on long-term In two case studies, provides high-quality solutions all less than 2 min. is significantly faster up factor 375 directly solving problem while practically retaining quality solution.