At most one email a week. Unfollow anytime with one click.
Transcript
The full narration of this video. The current sentence is highlighted while the video plays; click any sentence to jump there.
Show transcript
Introduction
Here is the final result, MPC is run every hour throughout the year, 8,736 optimizations, only about 20 seconds. Even with inaccurate PV and load forecasts, the cost is almost the same as with perfect forecasts. Let's get started. Hello, welcome to episode five of the MPC-based microgrid energy management series. The 24-hour plan is not executed entirely. MPC only executes the decision for the first hour, then one hour later, it replans with the latest data and forecasts. This is Receding Horizon. Today we are simulating it for an entire year, with perfect forecasts or imperfect forecasts.
Step 1: Perfect Forecast & Continuous Flow
Create a simulation file underscore mpc.py. The forecast function generates the net load for the next hours. Without a random generator, the forecast is perfect. With a random generator, the PV forecast is given a growing error of 5% in the first hour up to 30% after 12 hours, and the load is given an error of 2 to 6%. This pattern of growing errors over time is similar to real weather forecasts.
Step 2: Receding Horizon
The MPC simulation function creates two batteries, one for the controller, one for the real battery. Every hour, a forecast is created, MPC plans for 24 hours, and then only the decision for the first hour is executed on the real battery. Import and export are calculated from the actual net load, not from the forecast, so the forecast error truly has an impact. The real battery's SoC becomes the initial SoC of the next hour. The main part runs both versions and saves the results to an NPZ file for the next episode.
Step 3: One-Year Simulation
Run Each version solves 8,736 optimizations in about 21 seconds thanks to the DPP parameter. With perfect forecasts, the annual cost is 258.9 million rupiah. With noisy forecasts, 260.2 million, only about 1.3 million more expensive. The battery shifts 154 MW hours, 2.5 times more than rule-based. Why is MPC resilient to forecast errors?
Step 4: Why MPC is Robust Against Forecast Errors
Since the plan is updated every hour, forecasts for several hours ahead are never executed, and the first hour's decision uses the most accurate forecast. This is the main advantage of Residing Horizon over making a daily schedule.
Done!
Done We have been running the Residing Horizon MPC throughout the year, comparing perfect and noisy forecasts, and observing that MPC remains reliable even when forecasts are off. In the last episode, we compared all strategies, costs, PLN usage, and battery lifespan. Currently working on a thesis, dissertation, or research in the field of microgrid, batteries, and renewable energy? Cerdas Inovasi Digital helps with energy optimization implementation, from solar panel, wind turbine, and battery modeling, rule-based strategies, up to model predictive control with CVXPY, as well as offering Python programming courses for electrical engineering. Want to master the entire series, from weather data to cost comparison? All 6 episodes, complete with PDF step-by-step guides and source code, are available to premium tutorial subscribers. Subscribe starting at Rp69.000 per month, save 28% with the 12-month package, or pay once Rp999.000 for lifetime access. Want just one class? Buy a single class with lifetime access. Subscribe now at the premium page cerdasinovasidigital.com Contact us via WhatsApp or visit cerdasinovasidigital.com Thank you for watching, see you in the next episode.
Ask the class material
Type a question or keyword; we find the part of this episode that explains it and jump the video there.
My notes
Notes are saved with the current video time and are only visible to you.
Questions & answers
Stuck on a step? Ask here; the instructor and other learners can answer.
No questions yet.
Your instructor
Cerdas Inovasi Digital Team
Software developers, BIM & 3D engineers, and AI practitioners
52 video classes1,277 episodes
We build software, BIM models, games and AI automations for clients every day, and turn that hands-on work into step-by-step classes: every step is shown on screen and comes with a PDF guide you can keep.
