How to Smartly Manage Power in Small Grids with Renewable Energy

This patent describes a method for quickly and efficiently planning how different energy sources (like solar, wind, and batteries) work together in a small, local power grid, especially when renewable energy output changes unexpectedly.

What it covers

The patent outlines a four-step method (Claim 1) for real-time scheduling of "multi-energy complementary micro-grids." First, it sets up a "moving-horizon Markov decision process model" to represent the micro-grid's scheduling problem, considering unpredictable "new-energy outputs" and defining various "constraint conditions" like battery limits and power exchange with the main grid. Second, it establishes a "target function" to minimize the micro-grid's operating cost. Third, it divides the scheduling into smaller intervals and finds an initial, basic plan using a "greedy algorithm." Finally, it refines this basic plan using a "Rollout algorithm" to achieve a high-speed, efficient real-time schedule. For example, in a small community micro-grid with solar panels, wind turbines, and a battery, this method could instantly decide whether to use solar power, charge the battery with wind, or buy/sell electricity from the main grid, all while keeping costs low and adapting to sudden changes in weather.

What it doesn't cover

• —Does not cover scheduling methods that do not use a "moving-horizon Markov decision process model" for the micro-grid.
• —Does not cover scheduling methods that do not establish a "target function" specifically aiming for minimum operating cost.
• —Does not cover scheduling methods that do not use a "greedy algorithm" to find an initial basic feasible solution.
• —Does not cover scheduling methods that do not use a "Rollout algorithm" to find the final solution based on the basic feasible solution.
• —Does not cover micro-grid scheduling that ignores specific constraints like battery operating limits or power exchange limits with the main grid.

The clever bit

The clever bit is combining a "greedy algorithm" for an initial feasible solution with a "Rollout algorithm" to quickly refine it within a "moving-horizon Markov decision process model." This approach allows for high-speed, efficient real-time scheduling of complex micro-grids, even when new energy sources fluctuate unpredictably, overcoming the slowness of traditional methods.

Why it matters

The patent addresses a crucial challenge in integrating renewable energy: its unpredictable nature. By offering a faster and more efficient scheduling method, it helps micro-grids manage fluctuating wind and solar power more effectively. This is important for reducing reliance on fossil fuels and improving grid stability in local energy systems. Efficient real-time scheduling can lower operational costs for micro-grids, making them more economically viable for communities and businesses.

Real-world examples

• 1.University campus micro-grids
• 2.Military base micro-grids
• 3.Remote island power systems
• 4.Industrial park micro-grids
• 5.Community energy projects with renewables and storage