End-to-End Deep Neural Network Based Nonlinear Model Predictive Control: Experimental Implementation on Diesel Engine Emission Control
نویسندگان
چکیده
In this paper, a deep neural network (DNN)-based nonlinear model predictive controller (NMPC) is demonstrated using real-time experimental implementation. First, the emissions and performance of 4.5-liter 4-cylinder Cummins diesel engine are modeled DNN with seven hidden layers 24,148 learnable parameters created by stacking six Fully Connected one long-short term memory (LSTM) layer. This then implemented as plant in an NMPC. For implementation LSTM-NMPC, open-source package acados quadratic programming solver HPIPM (High-Performance Interior-Point Method) employed. helps LSTM-NMPC run real time average turnaround 62.3 milliseconds. prototyping, dSPACE MicroAutoBox II rapid prototyping system used. A Field-Programmable Gate Array employed to calculate in-cylinder pressure-based combustion metrics online time. The developed was tested for both step smooth load reference changes, which showed accurate tracking while enforcing all input output constraints. To assess robustness data outside training region, speed varied from 1200 rpm 1800 rpm. results illustrate disturbance rejection out-of-training region. At 5 bar indicated mean effective pressure rpm, comparison between production proposed 7.9% fuel consumption reduction, also decreasing nitrogen oxides (NOx) Particle Matter (PM) up 18.9% 40.8%.
منابع مشابه
Safe end-to-end imitation learning for model predictive control
We propose the use of Bayesian networks, which provide both a mean value and an uncertainty estimate as output, to enhance the safety of learned control policies under circumstances in which a test-time input differs significantly from the training set. Our algorithm combines reinforcement learning and end-to-end imitation learning to simultaneously learn a control policy as well as a threshold...
متن کاملExplicit Approximate Nonlinear Predictive Control Based on Neural Network Models
Nonlinear Model Predictive Control (NMPC) algorithms are based on various nonlinear models. Among others, an on-line optimization approach for NMPC based on neural network models can be found in the literature. Nevertheless, NMPC with on-line optimization is time consuming. On the other hand, an explicit solution to the NMPC problem would allow efficient on-line computations as well as verifiab...
متن کاملNeural Network Based Model Predictive Control
Greg Martin Pavilion Technologies Austin, TX 78758 [email protected] Mark Gerules Pavilion Technologies Austin, TX 78758 [email protected] Model Predictive Control (MPC), a control algorithm which uses an optimizer to solve for the optimal control moves over a future time horizon based upon a model of the process, has become a standard control technique in the process industries over the past two ...
متن کاملSupervised Hashing with End-to-End Binary Deep Neural Network
Image hashing is a popular technique applied to large scale content-based visual retrieval due to its compact and efficient binary codes. Our work proposes a new end-to-end deep network architecture for supervised hashing which directly learns binary codes from input images and maintains good properties over binary codes such as similarity preservation, independence, and balancing. Furthermore,...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Energies
سال: 2022
ISSN: ['1996-1073']
DOI: https://doi.org/10.3390/en15249335