Search

The high energy consumption of northern greenhouses, which allow the cultivation of plants year-round in cold regions, can considerably impact production costs. With the goal of increasing the production of fruits and vegetables in the greenhouse globally, it is evident to reduce these costs and improve the energy efficiency of these greenhouses. This work aims to optimize energy management in a greenhouse. To achieve this objective, we generate a linear discrete-time thermal model for capturing the dynamics of the greenhouse. Specifically, parameters influencing energy consumption and crop production, such as indoor temperature, humidity, weather conditions and energy consumption by the heater, are considered. Ridge regression, a popular parameter estimation method, is employed to estimate the coefficients of the greenhouse model. Notably, the ridge regression is performed on a dataset of the greenhouse with 15.6 square meters of surface area and Trois-Rivières weather data in a typical meteorological year. The dataset is generated by analytical equations governing the greenhouse function consisting of winter conditions since it is the period of concern for the utility service. The performance of the calibrated model is evaluated using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) statistical indicators to estimate the error and accuracy of the model with the dataset. As a result, we achieved the model with an accuracy of 94%. Thus, incorporating the significant elements affecting energy consumption permits the utilization of the proposed model for developing a convex optimization technique to reduce energy consumption and maximize the production inside the greenhouse. A convex objective function is devised to minimize the energy cost and maintain the setpoint preferences of temperature and humidity within the desired limit. As a case study, the setpoint preferences are set according to the day and night requirements for planting tomatoes in, Résumé de la communication présentée lors du congrès international tenu conjointement par Canadian Society for Mechanical Engineering (CSME) et Computational Fluid Dynamics Society of Canada (CFD Canada), à l’Université de Sherbrooke (Québec), du 28 au 31 mai 2023.

In recent years, with the progress of computer technology, artificial intelligence has been rapidly developed and begun to be applied in industry, economy and other aspects. Besides, with the pursuit of green hydrogen, hydrogen-based hybrid renewable energy systems have become the focus of the development of the hydrogen industry. This paper compares different artificial intelligence applications in hydrogen-based hybrid renewable energy systems and carries out a case study in a typical area. Firstly, this paper summarizes important works in literature, which use artificial intelligence methods to predict the supply chain of the renewable energy system, including the prediction of renewable energy system resources, output power, load demand and terminal electricity price. Secondly, main articles about artificial intelligence optimization algorithms used in renewable energy systems are also summarized, including swarm and non-swarm biological heuristics, physical or chemical heuristics and hybrid optimization algorithms. Finally, a case study is carried out in Tikanlik, Xinjiang, China. Tikanlik's weather and load data train the artificial neural network to predict system output power. It shows that 99.32% of the relative error of the test set is less than 3%, which proves that this model can achieve good prediction results.