摘要：

电梯作为特种设备之一, 其运行安全风险预测至关重要。当前对于电梯相关的研究多基于电梯部件数据, 并且预测方法在变换应用场景的情况下会出现预测精度低、泛化能力差等问题。为此, 提出一种基于领域自适应与注意力机制的电梯安全风险预测方法。该方法基于对抗领域自适应网络, 并且使用注意力机制优化网络的特征提取能力。方法包括特征提取器、标签分类器和领域分类器3个部分, 输入数据为同时包含源域与目标域数据的电梯安全风险因素, 经由注意力机制优化的特征提取器, 自适应提取并保留源域和目标域之间的公共关键特征, 然后将关键特征同时输入至标签分类器和领域分类器, 通过领域自适应实现由源域至目标域的迁移学习, 通过标签分类器输出电梯运行状态。实验结果表明, 所提出的方法在迁移至目标域应用场景的情况下, 预测精度可以达到86.9%, 相较于优化前提高了2.6百分点, 与LSTM-AE、CNN-LSTM、TrAdaBoost.R2、深度子领域自适应网络(DSAN)相比分别高出9.5、8.3、3.7和1.2百分点, 能够有效地对电梯安全风险进行预测。

关键词: 电梯, 安全风险预测, 注意力机制, 对抗领域自适应网络, 迁移学习

Abstract:

As special equipment, the operational safety risk prediction of elevators is crucial. Currently, most research on elevators is based on their component data, and the prediction method has problems such as low prediction accuracy and poor generalization ability in the case of changing application scenarios. Therefore, a method for elevator safety risk prediction based on domain adaptation and attention mechanisms is proposed. This method is based on an adversarial domain adaptive network and uses an attention mechanism to optimize the feature extraction ability of the network. The method includes three parts: feature extractor, label classifier, and domain classifier. The input data are the elevator safety risk factor, containing both source domain and target domain data. The feature extractor optimized by the attention mechanism adaptively extracts and retains the common key features between the source and target domains. The key features are simultaneously input to the label classifier and the domain classifier. Transfer learning from the source domain to the target domain is realized through domain adaptation, and the elevator operation status is output through the label classifier. The experimental results show that the prediction accuracy of the proposed method can reach 86.9% when it is transferred to the target domain application scenario, which is 2.6 percentage points higher than that before optimization, and it is 9.5, 8.3, 3.7, and 1.2 percentage points higher than that of LSTM-AE, CNN-LSTM, TrAdaBoost.R2, and Deep Subdomain Adaption Network(DSAN), respectively. Therefore, it can effectively predict elevator safety risks.

Key words: elevator, safety risk prediction, attention mechanism, adversarial domain adaptive network, transfer learning