Thank you for sharing. However, I have a question, please. I am currently implementing an MPC to control the temperature inside a room. To model the system, I used a neural network that takes as input a window of data (disturbance_w, control_w, output_w) to predict the output over a prediction horizon. Then, I use these predictions to calculate an objective function in order to obtain the first command to apply to my system to get the first output. For this, I use scipy, but the control proposed by this library remains constant regardless of the output values (the output does not follow the reference). Do you have any advice to improve this?
So MPC saved money for that brief period after the price of electricity went up and after that it would be the same as any other control system? It would be huge savings if considered across all of households but for singular examples is it really worth the money?
Bear in mind this is an educational video. Thermal storage is becoming more attractive, at least in large commercial HVAC systems, so this kind of dynamic optimization over a long time horizon is going to be more and more relevant, even if this is one-house example does not show it.
Very nice demonstration thanks. What are those 'future sample extractor' blocks in the Simulink model? I can't find any documentation for them or see them in my Simulink app (R2021b). Can anyone point me to the documentation on this?
Hi @BillTubbs, Thanks for your comment. The example used in this video is now available in our documentation, which can be accessed here: bit.ly/HouseHeatingSystem On this page, you can also open up the model in MATLAB Online to investigate any blocks in the Simulink model. The one you’re asking about is not a built-in Simulink block but it’s masked subsystem. By clicking the arrow in the left bottom corner, you can look under the mask to see what this block implements. You’ll notice the subsystem is using a MATLAB Function block to generate the Outdoor Temperature forecast data.
@@MATLAB Thanks but I don't have R2023a. I am still using R2021b and I can't open this simulation file. My main question is what is the signal between Future Sample Extractor and md (seq). Is it a scalar representing the predicted disturbance at a future time, or a sequence representing the full predicted sequence of disturbances from time t to t + Hp, where Hp is the prediction horizon?
Hi @@BillTubbs , that’s correct. The sequence generated by that subsystem is the sequence of disturbances from t to t+prediction horizon. Although you might be running a previous version of MATLAB, you can explore the following example in your browser using MATLAB Online (click the blue button on the example page), which always runs latest version of MATLAB: www.mathworks.com/help/mpc/ug/use-multistage-mpc-with-neural-state-space-prediction-model-for-house-heating.html
Thank you for sharing. However, I have a question, please. I am currently implementing an MPC to control the temperature inside a room. To model the system, I used a neural network that takes as input a window of data (disturbance_w, control_w, output_w) to predict the output over a prediction horizon. Then, I use these predictions to calculate an objective function in order to obtain the first command to apply to my system to get the first output. For this, I use scipy, but the control proposed by this library remains constant regardless of the output values (the output does not follow the reference). Do you have any advice to improve this?
So MPC saved money for that brief period after the price of electricity went up and after that it would be the same as any other control system?
It would be huge savings if considered across all of households but for singular examples is it really worth the money?
MPC is also applied to commercial/community housing network as you said
Bear in mind this is an educational video. Thermal storage is becoming more attractive, at least in large commercial HVAC systems, so this kind of dynamic optimization over a long time horizon is going to be more and more relevant, even if this is one-house example does not show it.
Very nice demonstration thanks. What are those 'future sample extractor' blocks in the Simulink model? I can't find any documentation for them or see them in my Simulink app (R2021b). Can anyone point me to the documentation on this?
Hi @BillTubbs, Thanks for your comment. The example used in this video is now available in our documentation, which can be accessed here: bit.ly/HouseHeatingSystem
On this page, you can also open up the model in MATLAB Online to investigate any blocks in the Simulink model. The one you’re asking about is not a built-in Simulink block but it’s masked subsystem. By clicking the arrow in the left bottom corner, you can look under the mask to see what this block implements. You’ll notice the subsystem is using a MATLAB Function block to generate the Outdoor Temperature forecast data.
@@MATLAB Thanks but I don't have R2023a. I am still using R2021b and I can't open this simulation file. My main question is what is the signal between Future Sample Extractor and md (seq). Is it a scalar representing the predicted disturbance at a future time, or a sequence representing the full predicted sequence of disturbances from time t to t + Hp, where Hp is the prediction horizon?
Hi @@BillTubbs , that’s correct. The sequence generated by that subsystem is the sequence of disturbances from t to t+prediction horizon. Although you might be running a previous version of MATLAB, you can explore the following example in your browser using MATLAB Online (click the blue button on the example page), which always runs latest version of MATLAB: www.mathworks.com/help/mpc/ug/use-multistage-mpc-with-neural-state-space-prediction-model-for-house-heating.html