Cultural coyote optimization algorithm applied to a heavy duty gas turbine operation

•A grey-box simulation model of a heavy duty GT is designed and explored.•The normative knowledge of Cultural Algorithms improves the COA’s performance.•The CCOA is a powerful optimizer for benchmarks and the heavy duty GT application.•The CCOA outperformed other state-of-the-art metaheuristics for the GT application. In the past decades, the quantity of researches regarding industrial gas turbines (GT) has increased exponentially in terms of number of publications and diversity of applications. The GTs offer high power output along with a high combined cycle efficiency and high fuel flexibility. As consequence, the energy efficiency, the pressure oscillations, the pollutant emissions and the fault diagnosis have become some of the recent concerns related to this type of equipment. In order to solve these GTs related problems and many other real-world engineering and industry 4.0 issues, a set of new technological approaches have been tested, such as the combination of Artificial Neural Networks (ANN) and metaheuristics for global optimization. In this paper, the recently proposed metaheuristic denoted Coyote Optimization Algorithm (COA) is applied to the operation optimization of a heavy duty gas turbine placed in Brazil and used in power generation. The global goal is to find the best valves setup to reduce the fuel consumption while coping with environmental and physical constraints from its operation. In order to treat it as an optimization problem, an integrated simulation model is implemented from original data-driven models and others previously proposed in literature. Moreover, a new version of the COA that links some concepts from Cultural Algorithms (CA) is proposed, which is validated under a set of benchmarks functions from the Institute of Electrical and Electronics Engineers (IEEE) Congress on Evolutionary Computation (CEC) 2017 and tested to the GT problem. The results show that the proposed Cultural Coyote Optimization Algorithm (CCOA) outperforms its counterpart for benchmark functions. Further, non-parametric statistical significance tests prove that the CCOA’s performance is competitive when compared to other state-of-the-art metaheuristics after a set of experiments for five case studies. In addition, the convergence analysis shows that the cultural mechanism employed in the CCOA has improved the COA balance between exploration and exploitation. As a result, the CCOA can improve the current GT operation significantly, reducing the fuel