Cost-Benefit Optimization of Sensor Networks for SHM Applications

Giovanni Capellari, Eleni Chatzi, Stefano Mariani2017 · DOI

TL;DR

This paper introduces a novel framework for optimizing the design of sensor networks for Structural Health Monitoring (SHM). It proposes two methods: one maximizing the information gained from sensors, and another maximizing the information gained per unit cost, offering engineers a powerful tool for cost-effective structural monitoring.

What Problem Does This Paper Address?

Designing effective Structural Health Monitoring (SHM) systems involves a critical challenge: determining the optimal configuration, number, and accuracy of sensors. Simply adding more sensors or using the most accurate ones isn't always the best approach due to increasing costs and data processing demands. The core problem is to balance the desire for comprehensive structural integrity information with practical constraints like budget and technological limitations. This paper tackles the question of how to design an SHM sensor network that maximizes its informational value while considering its associated costs.

Methods

The authors leverage information theory, specifically the concept of Shannon information gain, to quantify the "utility" of a sensor network. This utility represents how much new information about the structure's mechanical properties is obtained from sensor measurements. They then formulate two optimization problems:

  1. Maximizing Information Gain: This approach aims to find the sensor network configuration (spatial placement, number, and accuracy) that yields the highest expected Shannon information gain, subject to constraints like technological feasibility (minimum sensor accuracy), identifiability (minimum number of sensors needed to uniquely determine structural parameters), and budget.
  2. Maximizing Utility-Cost Index (UCI): This second, more cost-conscious approach focuses on maximizing the ratio of information gain to the total cost of the SHM system. This effectively seeks the most "bang for your buck" in terms of structural monitoring.

Both formulations utilize surrogate models and stochastic optimization algorithms to find the optimal sensor network design.

Datasets

The method is applied to a large-scale structural problem: the Pirelli tower, a 130m tall building in Milan. The associated finite element model of the tower is used, featuring 4106 nodes and 6 degrees of freedom per node. This serves as a realistic benchmark for evaluating the proposed optimization framework. No new datasets were introduced; rather, an existing structural model was utilized.

Key Insights and Ideas

  • Information Theory as a Design Tool: Using Shannon information gain provides a principled way to quantify the value of sensor measurements for estimating structural parameters.
  • Diminishing Marginal Utility: The paper highlights that the benefit of adding each additional sensor decreases as the network grows. This is analogous to economic principles of diminishing marginal utility, suggesting that there's a point where adding more sensors yields less and less additional information value.
  • Trade-off between Information and Cost: The two proposed formulations clearly illustrate the fundamental trade-off between maximizing the sheer amount of information and maximizing the efficiency of information acquisition relative to cost.
  • Budget as a Key Driver: The contour plots clearly show how budget constraints significantly influence the optimal sensor network design, dictating the achievable number and accuracy of sensors.
  • Pareto Front for Decision Making: The introduction of a Pareto front approach offers a visual and practical tool for designers to explore various optimal trade-offs between information gain and cost savings.

Key Results

The paper demonstrates the application of its optimization framework to the Pirelli tower model.

  • Maximizing Information Gain: This approach leads to designs that prioritize higher sensor accuracy and a greater number of sensors, especially when budgets are generous.
  • Maximizing Utility-Cost Index (UCI): This approach tends to favor selecting the most accurate sensors available and then determining the optimal number based on the cost model and budget. It shows that for a given budget, there's an optimal number of sensors beyond which adding more sensors becomes economically inefficient due to diminishing returns in information gain.
  • The figures illustrate how different budget levels (e.g., €2000, €2500, €3000) lead to distinct optimal sensor network configurations in terms of sensor accuracy ($\sigma$) and number ($n_y$).

Limitations and Future Work

  • Cost Model Simplification: The paper acknowledges that the cost model used is a simplification, particularly the assumption of a constant cost per unit sensor ($c(\sigma)$). Future work could explore more complex cost models that account for installation, maintenance, and data transmission costs, which might vary with sensor type and location.
  • Environmental Effects: The current framework assumes environmental effects are neglected. Incorporating these complexities would lead to more robust SHM system designs.
  • Computational Cost: While the paper aims to reduce computational cost by optimizing sensor networks, the optimization process itself can be computationally intensive. Further research into more efficient optimization algorithms or surrogate model techniques could be beneficial.
  • Generalizability: While applied to a large-scale structural problem, further validation on a wider range of structural types and SHM applications would strengthen the generalizability of the proposed method.

Summary updated

Abstract

Structural health monitoring (SHM) is aimed to obtain information about the structural integrity of a system, e.g., via the estimation of its mechanical properties through observations collected with a network of sensors. In the present work, we provide a method to optimally design sensor networks in terms of spatial configuration, number and accuracy of sensors. The utility of the sensor network is quantified through the expected Shannon information gain of the measurements with respect to the parameters to be estimated. At assigned number of sensors to be deployed over the structure, the optimal sensor placement problem is ruled by the objective function computed and maximized by combining surrogate models and stochastic optimization algorithms. For a general case, two formulations are introduced and compared: (i) the maximization of the information obtained through the measurements, given the appropriate constraints (i.e., identifiability, technological and budgetary ones); (ii) the maximization of the utility efficiency, defined as the ratio between the information provided by the sensor network and its cost. The method is applied to a large-scale structural problem, and the outcomes of the two different approaches are discussed.
BibTeX
@misc{capellari2017cost,
  author = {Giovanni Capellari and Eleni Chatzi and Stefano Mariani},
  title = {{Cost-Benefit Optimization of Sensor Networks for SHM Applications}},
  year = {2017},
  doi = {10.3390/ecsa-4-04891},
}

Loading related content...

Figure