In the production environment of the Internet of Things (IoT), sensors of various qualities generate a large amount of multivariate time series (MTS) data. The long-term prediction of time series data generated by various IoT devices provides longer foresight and help execute necessary resource scheduling or fault alarms in advance, thus improving the efficiency of system operation and ensuring system security. In recent years, deep learning models like Transformers have achieved advanced performance in multivariate long-term time series forecasting (MLTSF) tasks. However, many previous research attempts either overlook the inter-series dependencies or ignore to model strict temporal order of MTS data. In this paper, we introduce GT-former, a graph-based temporal-order-aware transformer model. We propose an adaptive graph learning method specifically designed for MTS data to capture both uni-directional and bi-directional relations. In addition, we generate positional encoding in a sequential way to emphasize the strict temporal order of time series. By adopting these two components, our model can have a better understanding of the inter-series and intra-series dependencies of MTS data. We conduct extensive experiments on eight real-world datasets and the results show that our model achieves better predictions compared with state-of-the-art methods.

中文翻译：

---

GTformer：用于长期序列预测的基于图的时间顺序感知变压器


在物联网（IoT）的生产环境中，各种质量的传感器会产生大量的多元时间序列（MTS）数据。对各种物联网设备产生的时间序列数据进行长期预测，提供更长久的预见性，有助于提前执行必要的资源调度或故障报警，从而提高系统运行效率并保证系统安全。近年来，像 Transformers 这样的深度学习模型在多元长期时间序列预测（MLTSF）任务中取得了先进的性能。然而，许多先前的研究尝试要么忽视了系列间的依赖性，要么忽略了对 MTS 数据的严格时间顺序进行建模。在本文中，我们介绍了 GT-former，一种基于图的时间顺序感知变压器模型。我们提出了一种专门为 MTS 数据设计的自适应图学习方法，以捕获单向和双向关系。此外，我们以顺序方式生成位置编码，以强调时间序列的严格时间顺序。通过采用这两个组件，我们的模型可以更好地理解 MTS 数据的系列间和系列内依赖性。我们对八个真实世界的数据集进行了广泛的实验，结果表明，与最先进的方法相比，我们的模型实现了更好的预测。