Sensor Breakdown Research Articles

Integrating process dynamics in data-driven models of chemical processing systems

Data-driven models require high-fidelity data of sufficient quantity and granularity. This is challenging in a complex chemical processing system due to frequent sensor breakdown, process shutdown, malfunctioning of equipment, random fluctuations, miscalibration, inconsistent sampling frequencies, and data entry errors. Thus many models scoring well on training data flounder on the real-time data of industrial systems. This work presents a process dynamics-guided neural network (PDNN) model to improve model generalization that can maintain higher performance in sparse and low-quality data. This has been enacted by adding an additional layer in the neural network architecture to incorporate process dynamics such as material and energy balance equations, universal laws, standard correlations, and field knowledge. We evaluated the proposed model against a standard neural network on a regression and a classification tasks representing a steady state and transient behavior of processing systems. The proposed model yielded improved outcomes on reduced sample-sized data and in extrapolated regimes implying a higher generalization capability of the PDNN model. The proposed process dynamics-guided neural network can be employed as a robust model for handling generalization issues of data-driven methods in processing systems.

IoMT-Enabled Fusion-Based Model to Predict Posture for Smart Healthcare Systems

Smart healthcare applications depend on data from wearable sensors (WSs) mounted on a patient’s body for frequent monitoring information. Healthcare systems depend on multi-level data for detecting illnesses and consequently delivering correct diagnostic measures. The collection of WS data and integration of that data for diagnostic purposes is a difficult task. This paper proposes an Errorless Data Fusion (EDF) approach to increase posture recognition accuracy. The research is based on a case study in a health organization. With the rise in smart healthcare systems, WS data fusion necessitates careful attention to provide sensitive analysis of the recognized illness. As a result, it is dependent on WS inputs and performs group analysis at a similar rate to improve diagnostic efficiency. Sensor breakdowns, the constant time factor, aggregation, and analysis results all cause errors, resulting in rejected or incorrect suggestions. This paper resolves this problem by using EDF, which is related to patient situational discovery through healthcare surveillance systems. Features of WS data are examined extensively using active and iterative learning to identify errors in specific postures. This technology improves position detection accuracy, analysis duration, and error rate, regardless of user movements. Wearable devices play a critical role in the management and treatment of patients. They can ensure that patients are provided with a unique treatment for their medical needs. This paper discusses the EDF technique for optimizing posture identification accuracy through multi-feature analysis. At first, the patients’ walking patterns are tracked at various time intervals. The characteristics are then evaluated in relation to the stored data using a random forest classifier.

A portable multi-sensor module for monitoring external ventricular drains.

External ventricular drains (EVDs) are used clinically to relieve excess fluid pressure in the brain. However, EVD outflow rate is highly variable and typical clinical flow tracking methods are manual and low resolution. To address this problem, we present an integrated multi-sensor module (IMSM) containing flow, temperature, and electrode/substrate integrity sensors to monitor the flow dynamics of cerebrospinal fluid (CSF) drainage through an EVD. The impedimetric sensors were microfabricated out of biocompatible polymer thin films, enabling seamless integration with the fluid drainage path due to their low profile. A custom measurement circuit enabled automated and portable sensor operation and data collection in the clinic. System performance was verified using real human CSF in a benchtop EVD model. Impedimetric flow sensors tracked flow rate through ambient temperature variation and biomimetic pulsatile flow, reducing error compared with previous work by a factor of 6.6. Detection of sensor breakdown using novel substrate and electrode integrity sensors was verified through soak testing and immersion in bovine serum albumin (BSA). Finally, the IMSM and measurement circuit were tested for 53days with an RMS error of 61.4 μL/min.

Vibration Monitoring for Position Sensor Fault Diagnosis in Brushless DC Motor Drives

A non-invasive technique for condition monitoring of brushless DC motor drives is proposed in this study for Hall-effect position sensor fault diagnosis. Position sensor faults affect rotor position feedback, resulting in faulty transitions, which in turn cause current fluctuations and mechanical oscillations, derating system performance and threatening life expectancy. The main concept of the proposed technique is to detect the faults using vibration signals, acquired by low-cost piezoelectric sensors. With this aim, the frequency spectrum of the piezoelectric sensor output signal is analyzed both under the healthy and faulty operating conditions to highlight the fault signature. Therefore, the second harmonic component of the vibration signal spectrum is evaluated as a reliable signature for the detection of misalignment faults, while the fourth harmonic component is investigated for the position sensor breakdown fault, considering both single and double sensor faults. As the fault signature is localized at these harmonic components, the Goertzel algorithm is promoted as an efficient tool for the harmonic analysis in a narrow frequency band. Simulation results of the system operation, under healthy and faulty conditions, are presented along with the experimental results, verifying the proposed technique performance in detecting the position sensor faults in a non-invasive manner.

PCA-based missing information imputation for real-time crash likelihood prediction under imbalanced data

ABSTRACTAs an important research topic, real-time crash likelihood prediction has been studied for many years. However, few research focuses on the missing data imputation in real-time crash likelihood prediction, although missing values are commonly observed due to breakdown of sensors or external interference. Besides, classifying imbalanced data is also a critical issue in real-time crash likelihood prediction, since the number of crash-prone cases is much smaller than that of non-crash cases. In this paper, three principal component analysis (PCA) based approaches are established for imputing missing values, while two kinds of solutions are developed to tackle the issue of imbalanced data. The results show that the proposed methods can help the classifiers achieve better predictive performance under situations with missing data. The two solutions, i.e. cost-sensitive learning, and synthetic minority oversampling technique (SMOTE), can help improve the sensitivity by adjusting the classifiers to pay more attention to the minority class.

Arcades: A deep model for adaptive decision making in voice controlled smart-home

In a voice controlled smart-home, a controller must respond not only to user’s requests but also according to the interaction context. This paper describes Arcades, a system which uses deep reinforcement learning to extract context from a graphical representation of home automation system and to update continuously its behavior to the user’s one. This system is robust to changes in the environment (sensor breakdown or addition) through its graphical representation (scale well) and the reinforcement mechanism (adapt well).The experiments on realistic data demonstrate that this method promises to reach long life context-aware control of smart-home.

Does PHM Make an Engineered System Resilient under Sensor Faults?

The use of Prognostics and Health Management (PHM) technology enables engineered systems resilient under adverse events. Adverse events include unexpected system failures, anomaly operation, manufacturing defects, etc. As engineered systems become more complex, resilience thus becomes an emerging engineering feature, which offers an attractive ability to resist and recover from adverse events. However, most resilience studies were conducted under the assumption of no false alarms. This study thus concerns how engineering resilience can be formulated while taking into account false alarm due to sensor faults. Sensor faults can be interpreted in various manners: sensor breakdown, calibration change, sensor noise, etc. This study proposes a new formulation of engineering resilience measure considering sensor faults. Sensor faults can be implemented by changing a sensor gain. The effectiveness of the proposed resilience measure is demonstrated by implementing the PHM into the electro-hydrostatic actuator (EHA) and assessing the resilience under sensor faults. This study is also conducted to see how the proposed resilience measure behaves in the course of the sensor degradation.

Statistical non-parametric mapping in sensor space.

Establishing the significance of observed effects is a preliminary requirement for any meaningful interpretation of clinical and experimental Electroencephalography or Magnetoencephalography (MEG) data. We propose a method to evaluate significance on the level of sensors whilst retaining full temporal or spectral resolution. Input data are multiple realizations of sensor data. In this context, multiple realizations may be the individual epochs obtained in an evoked-response experiment, or group study data, possibly averaged within subject and event type, or spontaneous events such as spikes of different types. In this contribution, we apply Statistical non-Parametric Mapping (SnPM) to MEG sensor data. SnPM is a non-parametric permutation or randomization test that is assumption-free regarding distributional properties of the underlying data. The method, referred to as Maps SnPM, is demonstrated using MEG data from an auditory mismatch negativity paradigm with one frequent and two rare stimuli and validated by comparison with Topographic Analysis of Variance (TANOVA). The result is a time- or frequency-resolved breakdown of sensors that show consistent activity within and/or differ significantly between event or spike types. TANOVA and Maps SnPM were applied to the individual epochs obtained in an evoked-response experiment. The TANOVA analysis established data plausibility and identified latencies-of-interest for further analysis. Maps SnPM, in addition to the above, identified sensors of significantly different activity between stimulus types.

Sensitivity and 3 dB Bandwidth in Single and Series-Connected Tunneling Magnetoresistive Sensors.

As single tunneling magnetoresistive (TMR) sensor performance in modern high-speed applications is limited by breakdown voltage and saturation of the sensitivity, for higher voltage applications (i.e., compatible to 1.8 V, 3.3 V or 5 V standards) practically only a series connection can be applied. Thus, in this study we focused on sensitivity, 3 dB bandwidth and sensitivity-bandwidth product (SBP) dependence on the DC bias voltage in single and series-connected TMR sensors. We show that, below breakdown voltage, the strong bias influence on sensitivity and the 3 dB frequency of a single sensor results in higher SBP than in a series connection. However, the sensitivity saturation limits the single sensor SBP which, under 1 V, reaches the same level of 2000 MHz∙V/T as in a series connection. Above the single sensor breakdown voltage, linear sensitivity dependence on the bias and the constant 3 dB bandwidth of the series connection enable increasing its SBP up to nearly 10,000 MHz∙V/T under 5 V. Thus, although by tuning bias voltage it is possible to control the sensitivity-bandwidth product, the choice between the single TMR sensor and the series connection is crucial for the optimal performance in the high frequency range.

Model based compensation for overcoming sensor breakdown in a piezo-based wireless sensor network

Abstract Modern control systems with increased sensor and actuator wiring have the problem of high system cost and increased sensor noise. A wireless sensor network (WSN) based control system is an alternative solution. However, a WSN based control system also encounters possible data loss and irregular sampling period due to the possibility of sensor breakdown. In this paper, a model based compensation was used in the piezo-based multi-node wireless control system to overcome sensor breakdown. Experiment results validated the proposed fault tolerant algorithm.

Wireless sensing and vibration control with increased redundancy and robustness design.

Control systems with long distance sensor and actuator wiring have the problem of high system cost and increased sensor noise. Wireless sensor network (WSN)-based control systems are an alternative solution involving lower setup and maintenance costs and reduced sensor noise. However, WSN-based control systems also encounter problems such as possible data loss, irregular sampling periods (due to the uncertainty of the wireless channel), and the possibility of sensor breakdown (due to the increased complexity of the overall control system). In this paper, a wireless microcontroller-based control system is designed and implemented to wirelessly perform vibration control. The wireless microcontroller-based system is quite different from regular control systems due to its limited speed and computational power. Hardware, software, and control algorithm design are described in detail to demonstrate this prototype. Model and system state compensation is used in the wireless control system to solve the problems of data loss and sensor breakdown. A positive position feedback controller is used as the control law for the task of active vibration suppression. Both wired and wireless controllers are implemented. The results show that the WSN-based control system can be successfully used to suppress the vibration and produces resilient results in the presence of sensor failure.

A Research of Multimode Microbending Optical Fiber Pressure Sensors Based on Different Gasket Structure

Microbending optical fiber pressure sensor is designed with the sensing unit based on multimode fiber in this paper. It consists of the pressure gasket, the microbending regulator and the metal clamp with fine adjustment structure. The measuring range of the sensor could be adjusted by choosing different gasket. When pressure acts on the fixture, a little displacement may occur, and luminous flux of the fiber will be lost after going through microbending part. According to the principle of light field detection, the pressure measurement can be achieved by detecting the loss of light intensity caused by different load. The basic structure of the sensor and its sensing mechanism are well introduced in this paper. In addition, for improving the effectiveness of the Microbending optical fiber pressure sensors, we set up mechanical zero adjustment and the maximal displacement limitation in order to avoid sensor breakdown caused by excess displacement.

Robust Fault Location Using Least-Absolute-Value Estimator

This paper presents a robust alternative to the previously developed method to reliably locate power-system faults using simultaneously recorded data from multiple locations. Automatic removal of corrupted measurements resulting from various factors (e.g., sensor breakdowns and cyberattacks) will be accomplished via the use of a least-absolute-value (LAV) estimator for fault location. Furthermore, inherent limitations of the approach imposed by sensor configurations as well as the effect of quantization errors incurred by low-precision sensors on the accuracy of estimated fault locations will be described. The performance of the overall algorithm will be tested and verified by fault scenarios on the IEEE 118-bus transmission grid.

A Lower Bound of Direction-of-Arrival Estimation for an Acoustic Vector Sensor Subject to Sensor Breakdown

The work presented here derives an approximate lower bound (ALB) for the error variance of direction-finding using a single acoustic vector sensor subject to random breakdown in its sensors. This ALB equals a weighted sum of Cramer-Rao bounds (CRBs), conditioned on disjoint events of sensor breakdown, that together define the overall phenomenon of random sensor breakdown. Its tightness is verified by Monte Carlo simulation of a maximum likelihood estimator.

Development of multi-guard ring-equipped p+–n Si microstrip sensors for the SiD detector at the ILC

Si sensors in future high-energy physics experiments will encounter an extremely harsh radiation environment. In lepton colliders, the dominant background from electron–positron pairs can be managed with modern Si technology, but there is also a potentially dangerous neutron background. Specifically, the estimated neutron background is around 1–1.6 × 1010 1-MeV equivalent neutrons cm−2 year−1 for the Si microstrip sensors to be used in the innermost vertex detectors of the International Linear Collider (ILC). This causes bulk damage in the Si sensors due to the non-ionization energy loss and is of much greater importance in determining the viability of their long-term operation. For reliable operation, such sensors will need to operate at high reverse biases, far beyond the full depletion voltages, which can lead to the sensor breakdown. Multi-guard ring structures have evolved as a powerful technique to improve the breakdown performance of these sensors. In this work, a comprehensive approach towards the understanding of multi-guard ring structures is followed by simulating several structures with different layouts. The potential and field distributions help in optimizing design parameters.

On Sensor Switching Visual Servoing

This article investigates dynamic camera switching during visual servoing tasks using multi-camera systems considering image-based and position-based Jacobian transpose control schemes. Stability is discussed using a common Lyapunov function assuming ideal target and camera models and multiple Lyapunov functions under parameter perturbations. It is shown that asymptotic stability cannot be achieved for arbitrary switching under parameter perturbations. Therefore, an energy supervised switching scheme is introduced guaranteeing asymptotic stability. The contributions are stable visual servoing strategies which facilitate instantaneous adjustments of control performance and dynamical device switches in case of task requirements or sensor breakdown.

Automated system for magnetic monitoring of active volcanoes

In order to provide a basis for short-term decision-making in the forecasting and monitoring of volcanic activity, we developed an entirely automated system of data acquisition and reduction for magnetic data. The system (Mag-Net) is designed to provide monitoring and analysis of magnetic data on Etna volcano at large distances from the central observatory. The Mag-Net system uses data from an array of continuously recording remote stations spread over the volcanic area and linked by mobile phone to the control center at the local observatory. At this location a computer receives the data and performs data sorting and reduction as well as limited evaluation to detect abnormal behavior or breakdown of remote sensors. Communication software, called MagTalk, is also designed to provide data to distant users. With a view to using continuous magnetic observations in advanced analysis techniques for volcano monitoring, the Mag-Net system also delivers two graphical user interface based applications to provide an interpretation capability. The former, called MADAP, speeds up all the data reduction processes in order to evaluate the reliability of magnetic signals. The latter, called VMM, is a procedure for modeling magnetic fields associated with tectonic and volcanic activity to facilitate the identification and interpretation of the sources of a wide spectrum of magnetic signals.

Identification Based Sensor Fault Detection and Localisation. Application to an Inverted Pendulum

In this paper a method for instrument fault detection based on identification techniques is presented. The aim of this FDI method is to detect slow acting faults rather than sudden sensor breakdowns. Periodic identification is performed in order to obtain state-space model parameters. The principle is that sensor gain and offset faults act differently on the parameters of state space models. The method can be applied on continuous and discrete time models, but beside closed-loop considerations, the use of discrete time models provides robustness to system component variations. This method is applied on an inverted pendulum.

Robustness of interconnected systems to structural disturbances in structural controllability and observability

The robustness to structural changes of interconnected systems described by a structured state model is considered. This corresponds to coefficient cancellations in matrices A, B or C, i.e. to breaking of interconnections, breakdown of actuators or sensors and transducers. According to the representation of the system, the rupture conditions in which the controllability and observability properties are maintained, may be deduced.