No abstract

No abstract

Neural networks (NNs) are an effective tool to model nonlinear systems. However, their forecasting performance significantly drops in the presence of process uncertainties and disturbances. NN-based prediction intervals (PIs) offer an alternative solution to appropriately quantify uncertainties and disturbances associated with point forecasts. In the present work, a NN ensemble procedure is proposed to construct quality PIs. A recently developed lower upper bound estimation (LUBE) method is applied to develop NN-based PIs. Then, constructed PIs from the NN ensemble members are combined using a weighted averaging mechanism. Simulated annealing and genetic algorithm are used to optimally adjust the weights for the aggregation mechanism. The proposed method is examined for three different case studies. Simulation results reveal that the proposed method improves the average PI quality of individual NNs by 22%, 18% and 78% for first, second and third case studies, respectively. Simulation study also demonstrates that 3-4% improvement in the quality of PIs can be achieved using the proposed method compared to the simple averaging aggregation method.

No abstract

Modelling polymerization processes involves considerable uncertainties due to the intricate polymerization reaction mechanism involved. The complex reaction kinetics results in highly nonlinear process dynamics. Available conventional models are limited in applicability and cannot describe accurately the actual physico-chemical characteristics of the reactor dynamics. The usual practice for operating polymerization reactors is to optimize the reactor temperature profile because the end use properties of the product polymer depend highly on temperature. However, to obtain accurate models in order to optimize the temperature profile, the kinetic parameters (i.e. frequency factors and activation energies) for a specific reactor must be determined accurately. Kinetic parameters vary considerably in batch reactors because of its high sensitivity to other reactor design and operational variables such as agitator geometry and speed, gel effects, heating systems, etc. In this work, the kinetic parameters were estimated for a styrene-free radical polymerization conducted in an experimental batch reactor system using a nonlinear least squares optimization algorithm. The estimated kinetic parameters were correlated with respect to reactor operating variables including initial reactor temperature (T o ), initial initiator concentration (I o ) and heat duty (Q) using artificial neural network (ANN) techniques. The ANN kinetic model was then utilized in combination with the conventional mechanistic model. The experimental validation of the model revealed that the new model has high prediction capabilities compared withother reported models. 