Search

This work solves the localization estimation of dynamic multi-agent systems (MASs) with sensor multiplicative failures, which is more general yet challenging to address than static sensor networks with ideal conditions. Barycentric coordinate is introduced to characterize the relative positions between agents. A new linear data model is constructed to represent the relationship between barycentric coordinates and relative distance. Based on the linear model, an adaptive parameter estimation algorithm is designed, and then it is applied to solve the relative distance compensation problem of the MASs with sensor multiplicative failures. Using the estimated parameter, a data-driven adaptive distributed localization estimation scheme based on iterative learning is proposed, in which only the measured relative distance data are available instead of the system model information. A key to obtaining accurate localization is overcoming the difficulties from inaccurate relative distance variables due to sensor failure via the data-driven adaptive relative distance compensation method. The numerical examples and experimental results verify the effectiveness of the proposed methods.

Despite the successful application of convolutional neural networks (CNNs) in object detection tasks, their efficiency in detecting faults from freight train images remains inadequate for implementation in real-world engineering scenarios. Existing modeling shortcomings of spatial invariance and pooling layers in conventional CNNs often ignore the neglect of crucial global information, resulting in error localization for fault objection tasks of freight trains. To solve these problems, we design a spatial-wise dynamic distillation framework based on multilayer perceptron (MLP) for visual fault detection of freight trains. We initially present the axial shift strategy, which allows the MLP-like architecture to overcome the challenge of spatial invariance and effectively incorporate both local and global cues. We propose a dynamic distillation method without a pretraining teacher, including a dynamic teacher mechanism that can effectively eliminate the semantic discrepancy with the student model. Such an approach mines more abundant details from lower level feature appearances and higher level label semantics as the extra supervision signal, which utilizes efficient instance embedding to model the global spatial and semantic information. In addition, the proposed dynamic teacher can jointly train with students to further enhance the distillation efficiency. Extensive experiments executed on six typical fault datasets reveal that our approach outperforms the current state-of-the-art detectors and achieves the highest accuracy with real-time detection at a lower computational cost.