A Scalable Algorithm for Identifying Multiple-Sensor Faults Using Disentangled RNNs

Abstract

The problem of detecting and identifying sensor faults is critical for efficient, safe, regulatory-compliant and sustainable operations of modern systems. Their increasing complexity brings new challenges for the Sensor Fault Detection and Isolation (SFD-SFI) tasks. One of the key enablers for any SFD-SFI methods employed in modern complex sensor systems, is the so-called analytical redundancy, which is nothing but building an analytical model of the sensors observations (either derived from first principles or identified from historical data in a data-driven fashion). In a nutshell, SFD amounts to generate and to monitor residuals by comparing the sensor measurements with the model predictions with the idea that the faulty sensors will result in large residuals (i.e. the defective sensors generate measurement that are inconsistent with their expected behavior represented by the model). In this paper we introduce a disentangled Recurrent Neural Network (RNN) with the objective to cope with the smearing-out effect, i.e. the propagation of a sensor fault to the non-faulty sensors resulting in large misleading residuals. Moreover, the introduction of a probabilistic model for the residual generation allows us to develop a novel procedure for the identification of the faulty sensors. The computational complexity of the proposed algorithm is linear in the number of sensors as opposed to the combinatorial nature of the SFI problem. Finally, we empirically verify the performances of the proposed SFD-SFI architecture using a real data set collected at a petrochemical plant.

A Machine-Learning Architecture for Sensor Fault Detection, Isolation, and Accommodation in Digital Twins

“…A different rationale is pursued in [31], where a single Deep belief network (a Bayesian type of NNs) has been trained (in a supervised fashion) to detect a faulty condition whereas sensor identification is naively carried out based on the maximum deviation from data mean-value. Along the same lines, a general approach is presented to detect and identify sensor faults using either a single Recurrent NN (RNN) or an MLP [39] for predicting next-step measurements and comparing with actual ones. A disentanglement regularization term on the NN loss function is introduced to help the algorithm coping with propagation of faults to non-faulty sensors in the identification stage.…”

“…Secondly, part of the literature evaluates corresponding proposals on private (e.g. [39], [40]) or simulated (e.g. [28], [36], [37]) measurement data, thus precluding reproducibility and convincing evaluation, respectively.…”

“…Thirdly, a number of the discussed works evaluate their proposals only on a single fault type (e.g. bias [21], [39] or drift [22]). Equally important, some architectures are only limited to fault detection [29], [30].…”

Exploring a Modular Architecture for Sensor Validation in Digital Twins

“…Data-driven SFDIA approaches have gained attention due to their ability to handle complex systems without the need for exact knowledge of the underlying model. Popular approaches build upon principal component analysis [12], support vector machine [13] and neural network (NN) based methods [14]- [17]. A modular SFDIA (M-SFDIA) scheme has been recently proposed in [18], [19] based on multi-layer perceptron (MLP) blocks connected in three layers.…”

A Random Forest and Current Fault Texture Feature–Based Method for Current Sensor Fault Diagnosis in Three-Phase PWM VSR

“…Gao et al (2020) proposed a CNN-based FDD method for micro-electromechanical system (MEMS) inertial sensors, in which the time-domain features of temperature-related sensor faults were adopted to train the data-driven FDD classifier. Haldimann et al (2020) proposed a disentangled RNN and residual analysis-based SFDD method and developed a novel procedure to identify the fault sensors. Chen et al (2021b) proposed an NN-based fault estimation method, which can obtain the accurate estimation of sensor faults.…”