Piano key weirs (PKWs) are acquired and developed for free surface control structures which improve their performance by increasing the storage capacity and flood evacuation. In this study, the potential combinations of two popular artificial intelligence data-driven models (AI-DDMs) of multi-layer perceptron neural network (MLPNN) and adaptive neuro-fuzzy inference system (ANFIS) with four meta-heuristic optimization algorithms (particle swarm optimization, genetic algorithm, firefly algorithm & moth-flame optimization) are assessed for predicting the PKW’s flow rate. Comparing the outcomes of the ten standard and hybrid AI-DDMs with three empirical relations based on several statistics and diagnostic analysis (such as the Taylor diagram) for estimating the flow rate shows that the AI-DDMs can predict the passing flow more accurately. In addition, the particle swarm optimization and firefly algorithm meta-heuristic algorithms improve the performance of ANFIS and MLPNN, respectively. The Mann-Whitney test for investigating the differences between two independent applied models indicates a significant difference between the AI-DDMs and two of the empirical relations at the 95% confidence level.
Read full abstract