Prediction on Energy Consumption of Semi-Closed Greenhouses Based on Self-Accelerating PSO-GA
Chen Jiaoliao,Chen Jiaoxuan,Yang Jiangxin,Xu Fang,Shen Zhen
DOI: https://doi.org/10.11975/j.issn.1002-6819.2015.24.028
2015-01-01
Abstract:In order to manage the energy and improve the heating/cooling efficiency, a hybrid method with mechanism modeling and system identification is proposed to predict the energy consumption of semi-closed greenhouses. Based on the balance of energy and mass balance, a physical model for energy consumption in the semi-closed greenhouse is developed, which takes into account the energy flux due to solar radiation, long wave thermal radiation, crop transpiration, heat conduction and convection. According to the measurement difficulty and time-variation, besides the environmental parameters, 11 uncertain parameters of the model are selected such as leaf area index of crop, aerodynamic resistances of the leaves and heat transfer coefficient of greenhouse cover. Self-accelerating particle swarm optimization and genetic algorithm (SPSO-GA) is presented to adjust the uncertain parameters of the energy consumption model. To speed up the convergence, the acceleration factor is considered to change the inertia weight in the optimization process. The population is evolved by particle swarm optimization (PSO) and the local optimal particle population is retained. Then, the local worst particle population from PSO is evolved by implementing crossover and mutation genetic algorithm (GA) operators. Combining the local best particle population by PSO and the particles by GA,thebest particles are remained. The data measurement is conducted in a semi-closed greenhouse in the Chongming agricultural demonstration base (31°57′N, 121°7′E), located in Shanghai City from November 1, 2014 to May 31, 2015. SPSO-GA is utilized to calibrate the uncertain parameters of energy consumption model by using the measured data in the experimental greenhouse from March 1 to March 5, 2015. The identification results illustrate that SPSO-GA can provide the advantage of rapid convergence, global optimization and strong robustness. The root mean square error (RMSE) by the GA and the PSO at final convergence is respectively 21.3% and 12.9% higher than that by the SPSO-GA. Hence, these results indicate that the SPSO-GA is effective to adjust the uncertain parameters of greenhouse energy model with high accuracy. According to the validation by using the continual three-day data, the RMSE by the GA and the PSO is respectively 32.2% and 10.7% higher than that by the SPSO-GA. The predicted heat power consumption by the SPSO-GA in the semi-closed greenhouse is well fitted with the measured data, and the total energy consumption performs a better accuracy of 98.6%. For checking the extensive applicability of energy consumption model in the semi-closed greenhouses, the energy forecasting model is then utilized to further predict the daily energy consumption in January, 2015. The deference of predicted and measured daily energy consumption is 1.58%-27.05%; however, all the relative error between predicted and measured daily energy consumption can be less than 14% when the trend of actual outside temperature and solar radiation is similar to that set for simulation. It proves that the energy forecasting model can be reliable to estimate the energy consumption of semi-closed greenhouse. The developed model for energy consumption can act as a tool to design the cooling/heating load and manage the greenhouse energy for financial and energy savings.