SEARCH · Search NASA
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Fault isolation and sensor placement are vital for monitoring and diagnosis. A sensor conveys information about a system's state that guides troubleshooting if problems arise. We are using machine learning methods to uncover behavioral patterns over snapshots of system simulations that will aid fault isolation and sensor placement, with an eye towards minimality, fault coverage, and noise tolerance.
An approach is presented to evaluating sensor placements to maximize monitorability of the target system while minimizing the number of sensors. The approach uses a model of the monitored system to score potential sensor placements on the basis of four monitorability criteria. The scores can then be analyzed to produce a recommended sensor set. An example from our NASA application domain is used to illustrate our model-based approach to sensor placement.
Guidance on Sensor Placement was identified as the top research priority for hydrogen sensors at the 2018 HySafe Research Priority Workshop on hydrogen safety in the category Mitigation, Sensors, Hazard Prevention, and Risk Reduction. This paper discusses the initial steps (Phase 1) to develop such guidance for mechanically ventilated enclosures. This work was initiated as an international collaborative effort to respond to emerging market needs related to the design and deployment equipment for hydrogen infrastructure that is often installed in individual equipment cabinets or ventilated enclosures. The ultimate objective of this effort is to develop guidance for an optimal sensor placement such that, when integrated into a facility design and operation, will allow earlier detection at lower levels of incipient leaks, leading to significant hazard reduction. Reliable and consistent early warning of hydrogen leaks will allow for the risk mitigation by reducing or even eliminating the probability of escalation of small leaks into large and uncontrolled events. To address this issue, a study of a real-world mechanically ventilated enclosure containing GH2 equipment was conducted, where CFD modeling of the hydrogen dispersion (performed by AVT and UQTR, and independently by the JRC) was validated by the NREL Sensor laboratory using a Hydrogen Wide Area Monitor (HyWAM) consisting of a 10-point gas and temperature measurement analyzer. In the release test, helium was used as a hydrogen surrogate. Expansion of indoor releases to other larger facilities (including parking structures, vehicle maintenance facilities and potentially tunnels) and incorporation into QRA tools, such as HyRAM is planned for Phase 2. It is anticipated that results of this work will be used to inform national and international standards such as NFPA 2 Hydrogen Technologies Code, Canadian Hydrogen Installation Code (CHIC) and relevant ISO/TC 197 and CEN documents.
There are multiple different algorithms to perform modal test sensor placement optimization: effective independence, residual kinetic energy, iterative Guyan reduction, genetic algorithms, or a brute-force methodology. However, any of these methods may be computationally expensive, especially for structural models with a large number of degrees of freedom. Given the high-cost and the need to optimize the solution, modal sensor placement is a great application for machine learning (ML) algorithms. In this paper, we will apply ML algorithms to determine the optimal sensor locations for simple and complex structures. We will also discuss the benefits and drawbacks of using machine learning over other sensor placement algorithms.
This paper presents an approach to evaluating sensor placements on the basis of how well they are able to discriminate between a given fault and normal operating modes and/or other fault modes. In this approach, a model of the system in both normal operations and fault modes is used to evaluate possible sensor placements upon the basis of three criteria. Discriminability measures how much of a divergence in expected sensor readings the two system modes can be expected to produce. Accuracy measures confidence in the particular model predictions. Timeliness measures how long after the fault occurrence the expected divergence will take place. These three metrics then can be used to form a recommendation for a sensor placement. This paper describes how these measures can be computed and illustrated these methods with a brief example.
Guidance on Sensor Placement remains one of the top priorities for the safe deployment of hydrogen and fuel cell equipment in the commercial marketplace. Building on the success of Phase 1 work reported at ICHS2019 and published in IJHE, this paper discusses the consecutive steps to further develop and validate such guidance for mechanically ventilated enclosures. The key step included a more in-depth analysis of sensitivity to variation of physical parameters in a small enclosure, and finally, expansion of the developed approach to confined spaces in an outdoor environment.
The problem of selecting optimal locations of actuators and sensors for the control of flexible structures is addressed. Novel geometrical and physical interpretations are given for both controllability and observability grammians which enhances the intuitive aspects of the actuator and sensor placement problem. The explicit form of controllability and observability grammians for flexible structures is used to define modal grammian coefficients, which provides the basis for an actuator and sensor placement technique. The relationship of the proposed metric to three different existing criteria are given. The method is demonstrated on a model of an experimental large flexible structure.
Nuclear power plants (NPPs) require continuous monitoring of various systems, structures, and components to ensure safe and efficient operations. The critical safety testing of new fuel compositions and the analysis of the effects of power transients on core temperatures can be achieved through modeling and simulations. They capture the dynamics of the physical phenomenon associated with failure modes and facilitate the creation of digital twins (DTs). Accurate reconstruction of fields of interest (e.g., temperature, pressure, velocity) from sensor measurements is crucial to establish a two-way communication between physical experiments and models. Sensor placement is highly constrained in most nuclear subsystems due to challenging operating conditions and inherent spatial limitations. This study develops optimized data-driven sensor placements for full-field reconstruction within reactor and steam generator subsystems of NPPs. Optimized constrained sensors reconstruct field of interest within a tri-structural isotropic (TRISO) fuel irradiation experiment, a lumped parameter model of a nuclear fuel test rod and a steam generator. The optimization procedure leverages reduced-order models of flow physics to provide a highly accurate full-field reconstruction of responses of interest, noise-induced uncertainty quantification and physically feasible sensor locations. Accurate sensor-based reconstructions establish a foundation for the digital twinning of subsystems, culminating in a comprehensive DT aggregate of an NPP.
This paper examines the sensitivity of experimentally measured modal frequencies and mode shapes to structural reassembly and sensor placement errors on a suspended three-bay truss. The statistical variations of identified mode shapes and frequencies are measured by repeated experiments. Both parameters are shown to vary measurably more with reassembly than sensor placement errors. Also, a directional stiffness in the truss joints was found to cause a parameter dependence on member orientation during reconstruction.
The objective of this work is to quantify the reconstruction error in sparse inverse problems with measures and stochastic noise, motivated by optimal sensor placement. To be useful in this context, the error quantities must be explicit in the sensor configuration and robust with respect to the source, yet relatively easy to compute in practice, compared to a direct evaluation of the error by a large number of samples. In particular, we consider the identification of a measure consisting of an unknown linear combination of point sources from a finite number of measurements contaminated by Gaussian noise. The statistical framework for recovery relies on two main ingredients: first, a convex but non-smooth variational Tikhonov point estimator over the space of Radon measures and, second, a suitable mean-squared error based on its Hellinger–Kantorovich distance to the ground truth. To quantify the error, we employ a non-degenerate source condition as well as careful linearization arguments to derive a computable upper bound. This leads to asymptotically sharp error estimates in expectation that are explicit in the sensor configuration. Thus they can be used to estimate the expected reconstruction error for a given sensor configuration and guide the placement of sensors in sparse inverse problems.
Systems health management, and in particular fault diagnosis, is important for ensuring safe, correct, and efficient operation of complex engineering systems. The performance of an online health monitoring system depends critically on the available sensors of the system. However, the set of selected sensors is subject to many constraints, such as cost and weight, and hence, these sensors must be selected judiciously. This paper presents an offline design-time sensor placement approach for complex systems. Our diagnosis method is built upon the analysis of model-based residuals, which are computed using structural model decomposition. Sensor placement in this framework manifests as a residual selection problem, and we aim to find the set of residuals that achieves single-fault diagnosability of the system, uses the minimum number of sensors, and corresponds to the best model decomposition for the best distribution of the diagnosis system. We present a set of algorithms for solving this problem and compare their performance in terms of computational complexity and optimality of solutions. We demonstrate the approach using a benchmark multi-tank system.
Stochastic fluctuations in power injections from distributed energy resources (DERs) combined with load variability can cause constraint violations (e.g., exceeded voltage limits) in electric distribution systems. To monitor grid operations, sensors are placed to measure important quantities such as the voltage magnitudes. Here, in this paper, we consider a sensor placement problem which seeks to identify locations for installing sensors that can capture all possible violations of voltage magnitude limits. We formulate a bilevel optimization problem that minimizes the number of sensors and avoids false sensor alarms in the upper level while ensuring detection of any voltage violations in the lower level. This problem is challenging due to the nonlinearity of the power flow equations and the presence of binary variables. Accordingly, we employ recently developed conservative linear approximations of the power flow equations that overestimate or underestimate the voltage magnitudes. By replacing the nonlinear power flow equations with conservative linear approximations, we can ensure that the resulting sensor locations and thresholds are sufficient to identify any constraint violations. Additionally, we apply various problem reformulations to significantly improve computational tractability while simultaneously ensuring an appropriate placement of sensors. Lastly, we improve the quality of the results via an approximate gradient descent method that adjusts the sensor thresholds. We demonstrate the effectiveness of our proposed method for several test cases, including a system with multiple switching configurations.
Previous research into object localization has shown that sensor placement and alignment plays an important role in achieving higher accuracy levels of the estimated location of a tracked Urban Air Mobility Vehicle. In general, a near-orthogonal intersection between the ground node observation vectors results in the highest accuracy due to a smaller overlapping uncertainty region between both. This applies to triangulation by means of ground node camera angle observations as well as trilateration by means of ground node distance measurements. However, this simple concept is not easily fulfilled with a network of a limited number of static ground nodes and a moving object to be localized. This case study performs sensitivity analyses and explores practical ways on how to achieve higher estimate accuracy levels in this context.
Previous research into object localization has shown that sensor placement and alignment plays an important role in achieving higher accuracy levels of the estimated location of a tracked Urban Air Mobility Vehicle. In general, a near-orthogonal intersection between the ground node observation vectors results in the highest accuracy due to a smaller overlapping uncertainty region between both. This applies to triangulation by means of ground node camera angle observations as well as trilateration by means of ground node distance measurements. However, this simple concept is not easily fulfilled with a network of a limited number of static ground nodes and a moving object to be localized. This case study performs sensitivity analyses and explores practical ways on how to achieve higher estimate accuracy levels in this context.
Optimal design of experiments for Bayesian inverse problems has recently gained wide popularity and attracted much attention, especially in the computational science and Bayesian inversion communities. An optimal design maximizes a predefined utility function that is formulated in terms of the elements of an inverse problem, an example being optimal sensor placement for parameter identification. The state-of-the-art algorithmic approaches following this simple formulation generally overlook misspecification of the elements of the inverse problem, such as the prior or the measurement uncertainties. This work presents an efficient algorithmic approach for designing optimal experimental design schemes for Bayesian linear inverse problems such that the optimal design is robust to misspecification of elements of the inverse problem. Specifically, we consider a worst-case scenario approach for the uncertain or misspecified parameters, formulate robust objectives, and propose an algorithmic approach for optimizing such objectives. Furthermore, both relaxation and stochastic solution approaches are discussed with detailed analysis and insight into the interpretation of the problem and the proposed algorithmic approach. Extensive numerical experiments to validate and analyze the proposed approach are carried out for sensor placement in a parameter identification problem.
Advances in sensor technology have increased our ability to monitor a wide range of environments. However, even as the cost of sensors decline, only a limited number of sensors can be installed at any given site. The physical placement of sensors, along with the sensor technology and operating conditions, can have a large impact on our ability to adequately monitor environmental change. This paper introduces a new open-source Python package, called Chama, that determines optimal sensor placement and technology to improve a sensor network’s detection capabilities. Additionally, the methods are demonstrated using site-specific methane emission scenarios that capture uncertainty in wind conditions and emission characteristics. Mixed-integer linear programming formulations are used to determine sensor locations and detection thresholds that maximize detection of the emission scenarios. The optimized sensor networks consistently increase the ability to detect leaks, as compared to sensors placed near each potential emission source or along the perimeter of the site.
For various reasons as discussed for instance in, the selection of actuator and sensor positions is still ad hoc. This is especially true for flexible structures where many candidate configurations can exist. This study is an attempt to make the selection process more methodical. One approach to actuator and sensor placement is to optimize a closed loop performance metric directly by selecting the actuators, sensors, and controller gains simultaneously. This direct approach makes sense if the desired closed loop performance is well defined. Since the individual actuator and sensor contributions to the closed loop performance metric is complex, the solution strategy usually employs non linear programming with many design and numerical iterations. A second approach is to select actuators and/or sensors based on open loop properties so that closed loop performance is indirectly optimized. Since the individual sensor and actuator contributions to the open loop metric is simple, nonlinear optimization is usually not needed. This approach will suggest efficient actuator and sensor configurations for any type of control law. The method suggested in this study falls into the latter class of approaches.
The challenge of reconstructing spatial fields that change over time from limited sensor data has been a focal point for many research studies. Various machine learning methods have been used in attempts to address this complex issue, including convolutional neural networks. All of the proposed methods share a common requirement that the user needs to manually determine the sensor positions. This requirement remains a limiting factor in the ongoing quest for efficient learning and accurate field reconstruction. This study aims to present a method that enables a model to optimize sensor positions via backpropagation, thereby facilitating the model’s exploration of the spatial domain and enhancing sensor positioning effectively. Indexing naturally incorporates discrete decisions. This operation is nondifferentiable which is a requirement for the application of gradient-based optimization methods. We showcased its effectiveness by training an attention-based neural network, which achieved top-tier performance on two separate datasets. To our knowledge, this represents the first fully end-to-end differentiable workflow for enhancing sensor placement within a neural network model.