Abstract:Population-based structural health monitoring (PBSHM), seeks to address some of the limitations associated with data scarcity that arise in traditional SHM. A tenet of the population-based approach to SHM is that information can be shared between sufficiently-similar structures in order to improve predictive models. Transfer learning techniques, such as domain adaptation, have been shown to be a highly-useful technology for sharing information between structures when developing statistical classifiers for PBSHM. Nonetheless, transfer-learning techniques are not without their pitfalls. In some circumstances, for example if the data distributions associated with the structures within a population are dissimilar, applying transfer-learning methods can be detrimental to classification performance -- this phenomenon is known as negative transfer. Given the potentially-severe consequences of negative transfer, it is prudent for engineers to ask the question `when, what, and how should one transfer between structures?'. The current paper aims to demonstrate a transfer-strategy decision process for a classification task for a population of simulated structures in the context of a representative SHM maintenance problem, supported by domain adaptation. The transfer decision framework is based upon the concept of expected value of information transfer. In order to compute the expected value of information transfer, predictions must be made regarding the classification (and decision performance) in the target domain following information transfer. In order to forecast the outcome of transfers, a probabilistic regression is used here to predict classification performance from a proxy for structural similarity based on the modal assurance criterion.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to quantify and optimize the value of information transfer in Population - based Structural Health Monitoring (PBSHM) to minimize the risk of negative transfer. Specifically, the research aims to develop a decision - making framework to guide when, under what circumstances, and how to transfer information between structures by introducing the concept of "Expected Value of Information Transfer" (EVIT), thereby improving the performance of classification tasks and ensuring the safety and economy of operation and maintenance decisions. ### Problem Background 1. **Data Scarcity Problem**: One of the main challenges faced by traditional SHM is the scarcity of data for training and validating models. PBSHM overcomes this problem by sharing information between sufficiently similar structures. 2. **Risk of Negative Transfer**: Although transfer learning techniques such as domain adaptation can effectively improve prediction performance in some cases, when the data distribution differences between structures are large, it may lead to negative transfer, that is, a decline in prediction performance. This not only increases unnecessary inspection and maintenance costs but may also miss critical maintenance interventions, threatening the safety and integrity of the structure. ### Research Objectives - **Quantify the Value of Information Transfer**: Quantify the impact of information transfer on classification tasks by introducing EVIT. - **Optimize Information Transfer Strategies**: Based on EVIT, develop a decision - making framework to select the optimal source domain and transfer algorithm to minimize the risk of negative transfer. ### Method Overview 1. **Construct Numerical Cases**: Use a population of 20 10 - degree - of - freedom systems to simulate the similarity and damage states between different structures. 2. **Define Classification Tasks**: Predict the damage state of each structure and classify based on natural frequency characteristics. 3. **Evaluate Prediction Quality**: Evaluate the classification performance through indicators such as True Prediction Rate (TR), False Positive Rate (FPR), and False Negative Rate (FNR). 4. **Calculate Structural Similarity**: Use the Modal Assurance Criterion (MAC) to evaluate the structural similarity between the source domain and the target domain. 5. **Establish Mapping Relationships**: Predict the classification performance from structural similarity through a probabilistic regression model (such as Multi - Layer Perceptron MLP). 6. **Optimize Transmission Strategies**: Select the optimal transmission strategy according to EVIT to ensure the maximization of classification performance in the target domain. ### Summary of Mathematical Formulas - **Definition of EVIT**: \[ \text{EVIT}(T)=EU(Q|T)-EU(Q|T_0)=P(Q|T)\cdot U(Q)-P(Q|T_0)\cdot U(Q) \] where \(EU\) represents expected utility, \(T\) represents a given transmission strategy, \(T_0\) represents the null strategy of no transmission, \(Q\) represents the prediction quality standard, and \(U(Q)\) represents the utility function related to prediction quality. - **Calculation of MAC**: \[ \text{MAC}(i,j)=\frac{|\phi_s^{\top,i}\phi_{t,j}|^2}{\phi_s^{\top,i}\phi_{s,i}\cdot\phi_t^{\top,j}\phi_{t,j}} \] where \(\phi_s^i\) and \(\phi_t^j\) represent the \(i\) - th and \(j\) - th modal vectors of the source domain and the target domain respectively. - **Proxy for Structural Similarity**: \[ \varsigma(\text{MAC})=\frac{\text{tr}(\text{MAC})}{N_m} \] where \(\text{tr}(\cdot)\) represents the trace of a matrix, and \(N_m\)

Quantifying the value of information transfer in population-based SHM

A decision framework for selecting information-transfer strategies in population-based SHM

Quantifying the value of positive transfer: An experimental case study

On the topology and geometry of population-based SHM

Towards risk-informed PBSHM: Populations as hierarchical systems

An Improved Multi-Task Approach for SHM Missing Data Reconstruction Using Attentive Neural Process and Meta-Learning

Domain-adapted Gaussian mixture models for population-based structural health monitoring

Improving decision-making via risk-based active learning: Probabilistic discriminative classifiers

Towards an active-learning approach to resource allocation for population-based damage prognosis

On risk-based active learning for structural health monitoring

Value of information from vibration-based structural health monitoring extracted via Bayesian model updating

Monitoring-Supported Value Generation for Managing Structures and Infrastructure Systems

A framework for quantifying the value of vibration-based structural health monitoring

Unsupervised transfer learning for structural health monitoring of urban pedestrian bridges

Foundations of population-based SHM, Part IV: The geometry of spaces of structures and their feature spaces

A value-of-information-based optimal sensor placement design framework for cost-effective structural health monitoring (with application to miter gate monitoring)

Probabilistic Inference for Structural Health Monitoring: New Modes of Learning from Data

On statistic alignment for domain adaptation in structural health monitoring

A probabilistic risk-based decision framework for structural health monitoring

Structural State Translation: Condition Transfer between Civil Structures Using Domain-Generalization for Structural Health Monitoring

Modelling variability in vibration-based PBSHM via a generalised population form