Online Feedback Optimization of Compressor Stations with Model Adaptation using Gaussian Process Regression

Efficient adaptive deep gradient RBF network for multi-output nonlinear and nonstationary industrial processes

“…In real-time process operation optimization and control applications, without an up-to-data and accurate surrogate model, there is a high risk that an industrial plant is operated in a suboptimal manner due to plant-model mismatch and unknown uncertainties [49,50,51]. The chief advantage of our proposed adaptive deep MGRBF network is that it provides high prediction accuracy and fast adaptation capability as well as low online computational complexity, which makes it particularly desirable for real-time optimization/control of nonlinear and time-varying processes.…”