针对半封闭式温室环境参数众多且难以测量的问题,提出了一种机理建模与系统辨识建模相结合的温室能耗建模方法。采用自加速遗传粒子群算法(self-accelerating hybrid algorithm of particle swarm optimization and genetic algorithm,SPSO-GA)对温室物理模型中难以确定的参数进行辨识,建立半封闭式温室能耗预测模型。根据上海半封闭式玻璃试验温室的气象数据和测量的能耗值,分别采用遗传算法(genetic algorithm,GA)、粒子群算法(PSO,particle swarm optimization)和SPSO-GA进行参数辨识与能耗预测比较分析。采用SPSO-GA获得的温室能耗预测结果与实测数据的相对误差为1.4%,分别比GA和PSO减少了2.9%和13.7%。根据日太阳光照辐射总量、室外日均温度2个参数及相应的变化曲线,预测的温室能耗值精确度大于86%。试验与模拟结果验证了基于SPSO-GA的温室能耗预测模型有效,可为半封闭式温室能量负载设计、管理和控制提供理论依据。
英文摘要:
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, the best 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.