| 초록 |
Artificial neural networks are revolutionizing the chemical industry with rapid implementations in process design, optimization, and control. However, researchers spend considerable amount of time through trial-and-error methods to build and train these neural networks resulting in lower accuracies. To eliminate this challenge, the current study proposes a python-based software platform for automatically building and training neural networks with higher accuracies by optimizing the network’s hyperparameters such as number of neurons, activation functions, learning rates, cost functions, and optimizers using a genetic algorithm embedded structure. The developed neural Network can further be employed for searching suitable operating conditions that optimize the case understudy. The software can also be used to plot both 2D and 3D graphs. The software platform was benchmarked against 5 published cases with only a marginal error of less than 3% and higher accuracies in most of the cases. |
