What is Structural Health Monitoring?

Structural health monitoring (SHM) is a type of infrastructure monitoring that employs a sensor network to monitor the physical integrity of a structure. The network may consist of a wide variety of sensor node types, strategically distributed, about the structure to maximize data collection. Network data is collected by the end user and statistically analyzed, using specialized damage detection algorithms and feature extraction, to determine if the structure is performing as intended, or if it has become damaged over time. By observing the data trends provided by an SHM system, the end-user can detect minute structural changes that would otherwise not be discernible by visual inspection alone. SHM data analysis has proved crucial in determining the overall safety of structural elements in bridges, tunnels and dams, as well as providing for early warning systems that can alert system operators prior to a catastrophic failure. It is also being applied to aeronautical and automotive monitoring, especially on structural components such as airframes, fuselages, landing gears, rotor blades and bulkheads for both fixed wing and rotor wing aircraft.

Types of structures that can benefit from SHM:

• Bridges, Viaducts, Tunnels, Dams, Levees
• Buildings, Piers
• Turbines
• Fixed and Rotor Wing Aircraft
• Landing Gear
• Tarmacs
• Large construction equipment, like cranes
• Transmission Towers
• Power Plants
• Pipes, Pipelines, Shafts
• Ships, Oil Tankers

Components of an SHM System:

Structure: Bridge, building, turbine, aircraft, etc, to be monitored

Sensor Network: Sensor type, distribution and network architecture for an SHM are application specific. A SHM network may require a variety of sensor types (strain, temperature, pressure, vibration, acceleration, etc) and may include a combination of both wired and wireless sensor nodes, and energy harvesting power generation

Data Acquisition, Storage and Transfer Systems: Embedded hardware modules and software systems that collect, store, process and transmit sensor data to the required end point.

Data Management System: Personal computer or cloud computing with an interface that allows the end user to manipulate data and apply various statistical analysis techniques. Data management capabilities may include:

­ Viewing sensor node data
­ Sending commands to the SHM network.
­ Statistically analyzing data
­ Assessing condition of structure
­ Assessing condition of a structural model
­ Predicting remaining service life of structure

Monitoring of Structural Responses:

SHM networks allow for the measurement and monitoring of structural responses to force. For example, a bridge is subject to the constant force produced by its own weight, as well as variable forces from vehicle traffic, construction, wind and possible micro-failures occurring within its structural members. Some common responses include bow, settling, sway, tilt, torsion and vibration. Collecting structural response data is crucial for determining how a structure is coping with the forces acting upon it. Structural failure can rapidly occur when loading exceeds the original design parameters.

There are two major types of force (or loading) situations that are commonly measured as part of an SHM system:

1. Static Loads: Force is constant over time. Static forces include the “dead weight” of the structure itself, as well the weight of any objects that are upon the structure in a constant manner.
2. Dynamic Loads: Force varies over time. Dynamic forces include conditions like traffic on a bridge, in-flight loads on an aircraft, and hurricane force winds impacting a skyscraper
3. Other Structural Response Factors: Other factors that can affect structural health and should be taken into consideration when designing a SHM network, include the sub-structural support (soil, rock, concrete platforms) that the structure is built upon, as well as creep, shrinkage and corrosion of the structural members.
4. Objects in constant motion: Due to the forces generated by constant motion or rotation (as found in an engine, turbine or rotor) an added layer of complexity is added to the SHM design. Sensor node placement, as well as measurement parameters, sampling rates and network control must be carefully considered within the network design

SHM Statistical Models:

Since sensor nodes themselves cannot determine if actual structural damage has occurred, the data collected by the node array must be converted into meaningful information via statistical analysis. By properly analyzing the SHM data, ”damage-sensitive features” can be extracted to determine the current state of a structure’s health.

Types of SHM Statistical Model Algorithms include:

Supervised Learning (Baseline): Algorithms that are applied when SHM data is available from both the undamaged, and damaged, parts of a structure. Group classification and regression analysis are examples of algorithms applied in this situation.

Unsupervised Learning (Non-Baseline): Algorithms that are applied when undamaged structure data is not available, for comparison to damaged structure data. Outlier, or novelty detection, is the primary class of algorithms applied in this situation.

AI and Fuzzy Logic Techniques: Applied to engineering issues within SHM that would be nearly impossible to solve by conventional analysis techniques. Conventional Artificial Intelligence (CAI) and Computational Intelligence (CI) have been used to create “intelligent” SHM data analysis systems for feature extraction.