INTEGRATED NEURO-FUZZY EXPERT SYSTEM FOR THE CONTROL OF HYDRO-POWER RESERVOIR SYSTEM

This paper introduces an Integrated Neuro-Fuzzy Expert System for controlling hydro-power reservoir systems, combining neural networks and fuzzy logic to improve decision-making in water level and flow rate management.Python libraries were utilized for data preprocessing, training, and prediction.A triangular membership function established 15 rule-base for input and output variables, with the dataset split into 70% for training and 30% for testing, ensuring reproducibility.A convolutional neural network was built using TensorFlow's Keras API, featuring 64 neurons and ReLU activation in the input and hidden layers, and a double neuron output layer.The model was trained for 150 epochs with an Adam optimizer, a batch size of 50, and a 30% validation split.The training set achieved a Mean Absolute Error (MAE) of 35.3623 and Mean Squared Error (MSE) of 2935.7104, while the test set had MAE and MSE values of 35.0812 and 1458.8833,respectively, with an RMSE of 38.2.Lower test loss and MSE compared to training MSE suggest the model generalizes reasonably well, which is crucial for reliable water reservoir control decisions.To prevent overflow, specific water levels and rates of change were set, predicting reservoir ranges of 0.07%, 0.10%, and 0.17% for the respective conditions.