Scribd
Upload
25 views1 page

MPC Info Sheet

Model predictive control (MPC) is an advanced process control technique that uses dynamic models to predict plant responses and compute optimal control actions. However, solving the underlying optimization problem of MPC is computationally intensive and infeasible for systems with fast dynamics or limited computing resources. This document proposes using deep neural networks to approximate MPC control laws since DNNs can efficiently approximate functions in real-time. The researchers aim to design MPC control laws using deep learning techniques and parallel computing to enable real-time implementation of MPC on fast systems while maintaining control performance. This would allow MPC to be used for applications like building energy management that traditionally could not implement MPC.

Uploaded by

hinsermu
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
25 views1 page

MPC Info Sheet

Model predictive control (MPC) is an advanced process control technique that uses dynamic models to predict plant responses and compute optimal control actions. However, solving the underlying optimization problem of MPC is computationally intensive and infeasible for systems with fast dynamics or limited computing resources. This document proposes using deep neural networks to approximate MPC control laws since DNNs can efficiently approximate functions in real-time. The researchers aim to design MPC control laws using deep learning techniques and parallel computing to enable real-time implementation of MPC on fast systems while maintaining control performance. This would allow MPC to be used for applications like building energy management that traditionally could not implement MPC.

Uploaded by

hinsermu
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 1

Deep Learning-based

Model Predictive
Control
Professor Bhushan Gopalun

Model Predictive Control (MPC) is an advanced process control technique that is implemented in many industrial
processes. MPC utilizes a dynamic model for predicting plant responses and computes the optimal control action.
More recently, MPC is finding novel applications in building energy management and optimization.

Despite its popularity in industry, solving the underlying optimization problem of MPC is computationally
intensive, which can be infeasible for systems with fast dynamics or local devices with limited computing resources.

As a popular machine learning technique, deep neural networks (DNN) exhibit excellent function approximation
capability, and can be used to approximate the MPC control law. Since the online implementation of a well-trained
DNN only entails simple function evaluation, the DNN controller can be executed efficiently in real-time.

Using techniques from parallel computing and distributed optimization, we solve large-scale nonlinear
optimization problems to design MPC control laws via deep learning techniques. These novel DNN-based
controllers will allow for real-time implementation of MPC on systems with fast dynamics while maintaining
satisfactory control performance.

Our research will enable engineers to efficiently implement advanced process control and optimization strategies
such as MPC on non-traditional applications such as building energy management systems, battery management
systems or energy production devices.

You might also like