Related Experiment Video
Updated: Sep 3, 2025
10:45
Time-dependent Increase in the Network Response to the Stimulation of Neuronal Cell Cultures on Micro-electrode Arrays
Published on: May 29, 2017
10.0K
Improved Artificial Neural Network Training Based on Response Surface Methodology for Membrane Flux Prediction
Syahira Ibrahim1, Norhaliza Abdul Wahab1
1School of Electrical Engineering, Faculty of Engineering, Universiti Teknologi Malaysia, Johor Bahru 81310, Malaysia.
Membranes
|July 27, 2022
Summary
This study optimizes artificial neural network (ANN) training for membrane flux prediction using response surface methodology (RSM). This approach significantly enhances training performance and reduces time for predicting palm oil mill effluent permeate flux.
Area of Science:
- Chemical Engineering
- Computational Modeling
Background:
- Membrane technology is crucial for treating palm oil mill effluent.
- Accurate prediction of membrane permeate flux is essential for process optimization.
- Conventional artificial neural network (ANN) training is often time-consuming due to trial-and-error methods.
Purpose of the Study:
- To improve ANN training efficiency for membrane flux prediction.
- To optimize ANN parameters using response surface methodology (RSM) and design of experiments (DoE).
- To predict permeate flux of palm oil mill effluent using airflow and transmembrane pressure as inputs.
Main Methods:
- Implemented a feed-forward neural network (FFNN) structure.
- Utilized Levenberg-Marquardt (lm) and gradient descent with momentum (gdm) training functions.
- Incorporated central composite design (CCD) within the ANN methodology for parameter optimization.
Main Results:
- Achieved over 50% improvement in training performance compared to conventional methods.
- Reduced the number of repetitions required for accurate model prediction.
- Demonstrated high model accuracy with smaller generalization errors for the FFNN-RSM.
Conclusions:
- RSM-based DoE significantly enhances ANN training performance and efficiency.
- The proposed FFNN-RSM model provides accurate permeate flux predictions.
- This optimized approach offers a faster and more reliable method for membrane process modeling.
