In this paper, we revisit the problem of robust diagnosability of Discrete-Event Systems (DES), and present a comparative analysis between the following notions of robust diagnosability existing in the literature: (i) diagnosability of DES subject to permanent sensor failures, assuming that sensors may fail only before the first occurrence of the events they are supposed to record; (ii) diagnosability of DES subject to permanent sensor failures, assuming that sensors may fail at any time; (iii) diagnosability of DES against intermittent loss of observations; (iv) diagnosability of partially observed DES; (v) generalized robust diagnosability. We show that all of the robust diagnosability definitions are particular cases of the generalized robust diagnosability by presenting transformation mechanisms for each one of the analyzed robust diagnosability notions so as to convert it into an equivalent generalized robust diagnosability problem. We also compare the use of projections and masks in the context of language diagnosability and show that there is no loss of generality in using projections in place of masks by presenting a map that transforms the language diagnosability problem with observation mask into an equivalent one with projection.

在本文中，我们重新审视了离散事件系统 (DES) 的鲁棒可诊断性问题，并对文献中存在的以下鲁棒可诊断性概念进行了比较分析：(i) DES 受永久性传感器故障影响的可诊断性，假设传感器可能只在它们应该记录的事件第一次发生之前发生故障；(ii) DES 受永久性传感器故障影响的可诊断性，假设传感器随时可能发生故障；(iii) DES 对间歇性观察丢失的可诊断性；(iv)部分观察到的 DES 的可诊断性；(五)广义鲁棒可诊断性。我们通过为每个分析的鲁棒可诊断性概念呈现转换机制以将其转换为等效的广义鲁棒可诊断性问题，表明所有鲁棒可诊断性定义都是广义鲁棒可诊断性的特例。我们还比较了在语言可诊断性上下文中投影和掩码的使用，并通过展示将具有观察掩码的语言可诊断性问题转换为具有投影的等效问题的地图，表明使用投影代替掩码并没有损失一般性.