Data-Enhanced Infrastructure Engineering Committee

Preamble

The field of data analytics offers opportunities for innovation, growth and job creation. Understanding the challenges is essential to delineate policies that encourage appropriate use of data analytics, artificial intelligence and behavior modeling by the public and private sectors, thus protecting the interests of all involved. This ISHMII committee will focus on topics that increase such understanding.

Vision

Instrument, sense, collect, store, manage and interpret data to evaluate civil infrastructure, with and without the use of behavior modelling, through implementation of recent advances in artificial intelligence. Impacts include safer and more resilient infrastructure, increased sustainability through replacement avoidance and less invasive interventions, support for construction and asset management decision making and ultimately, better design codes for monitored civil infrastructure.

Mission

To advise the engineering community by providing perspective and evaluating data structures, strategies, methodologies, frameworks and algorithms that lie on the interface between Big-Data analytics, behavior modelling and artificial intelligence. The committee will concentrate on those aspects that provide greater understanding while satisfying engineering needs and through recognizing civil- engineering contexts.

Deliverables

  1. Define a benchmark study and share data to compare different data-driven and model-based technologies
  2. Provide guidelines on how to extract quality data from data acquired by monitoring networks

Data and Model Engine

Define the capabilities of the personnel, process and technology that lead to appropriate instrumentation, sensing, collection and storage, management and interpretation of data to increase understanding and to improve predictions of real behavior of civil infrastructure. Technology evaluations will be made in the context of existing non-destructive evaluation technologies. Also, the interaction between these technologies and the potential performance of visual inspection will be included for a range of situations.

Data and Model Usage

Define skills and capabilities of organizations to generate and evaluate ideas for using data, gather standard datasets, ensure data privacy and security, evaluate and compare algorithms, develop behavior models, surrogate models, verify model relevance.

Data and Modeling Ecosystem

Define skills and conditions that lead to creation of partnerships that utilize data analytics for behavior modeling and artificial intelligence strategies.

Interactions

Identify promising representations and methodologies that combine data interpretation with behavior modelling and new developments in artificial intelligence. For example, the Committee will define the challenges and risks of using data-only methods and identify situations where engineers are well supported in their decision making. Also, the costs of creating accurate behavior models, as well as the extra uncertainty introduced by cheaper surrogates, need to be included in the choice of strategy.

Predictions

Define knowledge that lead to creation of appropriate predictive strategies and validate with full-scale case studies. For example, data-only strategies are best used for interpolation only; physical principle behavior models are needed for extrapolation. Also, the costs of erroneous predictions (type I and type 2 errors) and the subsequent needs for explanation, eventually in a legal context, will be evaluated.