Multi-layer hierarchical optimisation of greenhouse climate setpoints for energy conservation and improvement of crop yield

Energy conservation is an increasingly important issue in greenhouse production. Since higher energy input usually produces higher crop yields, crop yield and energy consumption must be carefully traded off. Generally, at any instant, the heating energy demand and crop yield depend on an interior temperature setpoint. Good, dynamically determined greenhouse climate setpoints not only improve crop yield, but also reduce the total energy consumption of the greenhouse. The greatest uncertainty in determining setpoints is the long-term weather. To deal with this uncertainty, this paper presents a multi-layer hierarchical optimisation framework for greenhouse climate setpoints. This framework has two optimisation layers with different timescales, and the whole production cycle is divided into several crop development stages, with each stage including several days. In the first layer, off-line multi-objective optimisation is carried out using historical weather data to maximise the crop yield and minimise the total energy consumption of the greenhouse heating. The aim is to optimise the mean temperature for each crop development stage. Based on the constraint of the optimal mean temperatures of the development stages, this work used a surrogate-assisted constrained single-objective method for online optimisation of the daily temperature using current short-term weather forecasts. Since the proposed method does not directly optimise the greenhouse climate setpoint, but instead optimises the mean temperature of the stages and the daily mean temperature, this work develops an effective indoor daily mean temperature serialisation method to generate the temperature setpoint series. The simulation results show that, compared to a widely used commercial system, the energy efficiency can be improved by 24.9%, 16.8% and 14.6% under real weather conditions of three example years.