Evaluation and performance prediction of a hybrid solar-based cycle based on trough collector and PCM storage using artificial intelligence

It is essential to develop a trustworthy and meticulous output power forecasting method to certify solar multigeneration systems stability, credibility, and power dispatchability. Therefore, this study focuses on improving the conventional forecasting tools using an evolutionary algorithm PSO. At first, a dataset is provided by simulating the proposed hybrid system in TRNSYS software. Then, intelligent forecasting approaches like adaptive neuro-fuzzy inference system (ANFIS) and multilayer perceptron (MLP) neural networks, are modeled using MATLAB software. The MLP and ANFIS networks are optimized via the PSO algorithm during the training process with specific inputs and targets. The evaluated input parameters consist of solar radiation, dry ambient temperature, and wet bulb. The total efficiency of the proposed system is determined as the target variable of all intelligent networks. Sensitivity analysis estimated the optimal dataset division as 60 % for ANN and 70 % for ANFIS. PSO optimization reduced prediction errors by 99.9 %. The ANN-PSO model had the best accuracy (MSE: 0.026 train, 0.025 test), while ANN achieved the highest correlation (R = 0.893 train, 0.873 test). The results demonstrate that the PSO algorithm works as intended for optimizing the forecasting tools and the comparison results indicate that the ANN-PSO method outperforms the other developed methods.