太阳能是新兴的可再生能源之一，可将其转化为电能以供无线传感器网络(Wireless Sensor Networks，WSN)使用，对太阳能进行预测可以有效地利用能量，从而达到节省能源、维持网络持续稳定运行的目的。提出了一种新的组合预测模型来预测太阳能辐照强度，其中改进的粒子群优化(Particle Swarm Optimization，PSO)算法被引入寻找长短期记忆(Long Short Term Memory，LSTM)神经网络模型的最优参数。选取自回归差分移动平均(Auto-Regressive In-tegrated Moving Average，ARIMA)模型来预测太阳辐照数据中的线性分量;采用PSO算法来优化LSTM神经网络模型的超参数，有助于提高模型预测的精度和鲁棒性;采用优化的LSTM神经网络模型来预测数据中的非线性分量;最后将两个模型的预测结果进行叠加。实验结果表明，新的组合模型比ARIMA、LSTM等模型，具有更高的预测精度。
Solar Intensity Prediction Based on ARIMA-PSO-LSTM Model
Solar energy is one emerging renewable energy source,which can be converted into electricity for the use of wireless Sen-sor Networks(WSN),and prediction of solar energy can effectively use energy to save energy consumption and maintain continuous and stable operation of network.In this paper,a new combined energy prediction model is proposed to predict solar radiation intensity,in which an improved algorithm Particle Swarm Optimization(PSO)is introduced to find optimal parameters of a Long Short Term Memory(LSTM)model.Auto-Regressive Integrated Moving Average(ARIMA)is initially employed to distill and forecast linear elements of solar radiation data.Secondly,PSO is used to optimize hyperparameters of the LSTM model,which helps to improve the accuracy and robust-ness of the model prediction.Then,the optimized LSTM model is used to predict nonlinear components in the data.Finally,forecast out-comes of both models are combined.Experiments show that the new combined model has higher prediction accuracy than ARIMA,LSTM and other models.
国家科技期刊平台
NSTL
万方数据
