A Novel Framework for Integration of Abstracted Inspection Data and Structural Health Monitoring for Damage Prognosis of Miter Gates

Abstract

Operational condition assessments, using a discrete rating system, are frequently used by field engineers to assess inland navigation assets and components. Challenges such as the occasional inability to perform inspections (such as the case with locks watered in an operational state) and protocol requirements requiring ratings even when they aren't inspected lead to highly abstracted inspection data, which are also very prone to human error and misinterpretations due to inspections protocol. On the other hand, some navigational locks are equipped with structural health monitoring (SHM) systems to continuously perform assessments from data obtained in situ . This paper aims to develop a novel hybrid damage prognosis framework for miter gate component of navigational locks, by mitigating effects of human errors on the condition assessment and integrating the highly abstracted inspection data with the SHM. It overcomes two main challenges, namely (1) there is no physical or empirical model available to model the loss-of-contact degradation in the gate, and (2) the mismatches between the inspection data and the SHM system due to data abstraction. A practical case of monitoring loss-of-contact quoin block demonstrates the efficacy of the proposed framework.

Diagnosis, Prognosis, and Maintenance Decision Making for Civil Infrastructure: Bayesian Data Analytics and Machine Learning

“…gap length). Then, the posterior probability density function of the gap length at time step conditioned on strain measurements is estimated using Bayesian inference method recusively as follows (9) which is defined as follows (10) in which is the likelihood function, obtained from the measurement equation, of observing for given at time step , and the term represents the probability distribution of for a given obtained using the state equation, which describes the damage evolution over time.…”

“…2.3. To tackle this issue, an approach has been proposed in [10] to map the reported transition matrix to a more useful transition matrix that has eliminated some of the effects of the human errors. In order to map the reported rating transition matrix to the underlying "true" transition matrix , the underlying true transition rating is defined at time t as and that at t+1 as .…”

“…4.1 after mitigating human error bias, a continuous degradation model can be obtained through Bayesian or optimization-based calibration. More details of the mapping from inspection data to a degradation model are available in [10]. Here, a brief summary of the…”

“…It is also applicable to other degradation models and can be integrated into a Bayesian framework. More detailed discussions of this approach can be found in [10]. Failure diagnostics and prognostics of civil infrastructure usually require the usage of multiple models including degradation model [39][40][41] and physics-based model [12][13][14], as discussed earlier.…”

“…Prognostics and Health Management (PHM) is the notion of augmenting current structural state diagnostic information gleaned by inspections or SHM to make predictions of the future state and reliability of the system based on degradation models or historic degradation/failure data [5]. When such prior data is available (not very common in the civil infrastructure domain), a data-driven degradation model is possible, but more commonly physics-based approaches [6,7] or empirical approaches to build the model [8][9][10] are required.…”