采用改进HPO-LSTM-Attention算法的太阳辐射散射预测

采用改进HPO-LSTM-Attention算法的太阳散射辐射预测

摘要:

针对现有对于太阳散射辐射的预测方法准确度低等问题，构建了HPO-LSTM-Attention组合模型，为了进一步提高模型预测的准确度为HPO设计了一种新的自适应动态权重，从而可以平衡算法的全局探索性和局部开发性。经过注意力机制和HPO两种算法的优化使得LSTM的预测性能有了极大的提升。实验结果表明新提出的HPO-LSTM-Attention模型优于LSTM、BiLSTM和HPO-LSTM模型，在MAE、MAPE、R2和MSE评价指标下表现更出色，与未改进模型相比其均方差减少了近40%。证明了HPO-LSTM-Attention模型在预测太阳散射辐射方面的有效性。

Abstract:

In order to solve the problem of low accuracy of existing prediction methods for solar scattered radiation, an HPO-LSTM-Attention combination model was constructed. In order to further improve the accuracy of model prediction, a new adaptive dynamic weight was designed for HPO, which can balance Global explorability and local development of the algorithm. After the optimization of the attention mechanism and HPO algorithms, the prediction performance of LSTM has been greatly improved. Experimental results show that the newly proposed HPO-LSTM-Attention model is better than the LSTM, BiLSTM and HPO-LSTM models, and performs better under the MAE, MAPE, R2 and MSE evaluation indicators. Compared with the unimproved model, its mean square error is reduced by nearly 40%. The effectiveness of the HPO-LSTM-Attention model in predicting solar scattered radiation is demonstrated.