Abstract:The Backpropagation (BP) neural network model is optimized in this research to predict NOx concentrations under dynamic conditions. Various activation functions (elliotsig, logsig, poslin, radbas, satlin, satlins, and tansig) were explored, and hidden neuron counts were adjusted to determine the optimal configuration for achieving maximal prediction accuracy. Out of the 8791 dataset entries, 7033 were used for training, and 1758 were used for testing. The model was evaluated using four primary metrics: Mean Absolute Error (MAE), Root Mean Square Error (RMSE), the coefficient of determination (R 2 ), and Model Efficiency (ME). It was observed that activation functions had a significant impact on the computational time of the BP model. Satlins, poslin, and satlin functions operated swiftly, with a running time of approximately 11 ms for an ANN model with 100 neurons. Due to the computational complexity of logsig, tansig, and radbas, their computation time was more than three times longer compared to the previous functions. Overall, this research illustrates the effectiveness of the BP neural network in predicting NOx concentrations when optimized with the appropriate architecture. It highlights the elliotsig function for its combination of accuracy and efficiency. During a 10 ms operating time of the microcontroller, the combination of elliotsig function with 68 hidden neurons resulted in the highest R 2 and ME values, achieving 0.9934 and 0.9861, respectively. Furthermore, this configuration demonstrated exceptional accuracy, with the lowest RMSE and MAE values recorded at 45.4654 and 23.2959, respectively.

An Artificial Neural Network with Improved Activation Function and Its Application

Back-propagation Artificial Neural Networks for Water Supply Pipeline Model

Improved BP algorithm for neural network and its application in welding process control

BP Neural Network Applied in Modeling of Water Distribution Network

Evolving Neural Networks Using the Hybrid of Ant Colony Optimization and Bp Algorithms

The Prediction for the Activity of Herbicide Based on Back-Propagation Artificial Neural Network

Application of the Improved BP Neural Network in Ship and Ocean Engineering

Application of Improved BP Neural Network Based on Matlab

Application Research of BP Neural Network Optimization Based on Firefly Algorithm

Optimizing activation functions and hidden neurons in Backpropagation neural networks for real-time NOx concentration prediction

Brief Introduction of Back Propagation (BP) Neural Network Algorithm and Its Improvement

Performance Analysis of Various Activation Functions in Artificial Neural Networks

Study on Improved FIR Filters Using BP Neural Networks

Activation functions selection for BP neural network model of ground surface roughness

Neural Networks with Activation Networks

Importance of Optimizing Neural Activation Function Types

Safety assessment of system based on artifical neural network

Research and application on improved BP neural network algorithm

A Method on Searching Better Activation Functions

Programme of an Improved BP Neural Network and Its Application

Optimizing the Laser-Cutting Parameters by Using Improved Artificial Neural Network