基于傅里叶变换的注意力机制负载预测模型

Load prediction is crucial for effective management and optimization of resources in cloud computing. Accurate prediction helps ensure the stability and performance of the system under high load conditions. Although load prediction has made some progress, there are still many research directions worth exploring, such as improving the accuracy of prediction models and studying new prediction methods and technologies. Therefore, this paper proposes a load prediction model combining Fourier transform, LSTM, and Transformer, which converts input data into the frequency domain for processing, further captures relevant information of the input data, reduces redundant information, and enables the model to process input data more efficiently. Finally, compared with other neural network prediction models on the same dataset, the proposed model achieves better performance and accuracy.