In the context of structural health monitoring it is of great importance to be able to use the hundreds of sensors nowadays installed on a structure to surveil the deterioration of its single constituting parts that will hence determine its behavior as a whole.
The ability of detecting early warning signs allows to promptly intervene, avoiding catastrophic failures, increasing safety and extending the lifespan of the structure itself.
Implementing an effective structural health monitoring system is very demanding on several fronts.
Sensors malfunctions and huge amounts of data makes it very hard for the subject matter experts to perform manual checks, and the intrinsic complexity of the structures makes the analysis very challenging.
Moreover, to fully take advantage of the IoT revolution and to enable real time insights, these monitoring systems need to work in continuous operation mode, and require live, compute-intensive data processing.
Technical challenges include:
- Collecting data coming from different acquisition systems;
- Analyzing data with different sampling rates;
- Manual data correction that requires extensive experience, focus and time;
- Sensor malfunctions that generate missing and misleading data;
- The complexity of developing complex models describing relations between hundreds of signals, such as temperature and stress or wind speed and acceleration;
- The development of real-time methods to automatically detect structural dynamic changes requires several different technologies and expertise, including IoT, signal processing, artificial neural networks and web applications.
Our platform allows to easily create real-time systems capable of automatically detecting early signs of structural weakness and to issue warnings accordingly.
Sharpsense combines the information coming from hundreds of sensors, the knowledge of subject-matter experts and cutting edge Artificial Intelligence models to provide next generation monitoring and early warning systems tailored for the specific client use cases.
Technical solutions include:
- A software agent to easily connect several different sensor types, acquisition systems or databases to our platform;
- A centralised cloud platform to manage all data, analyses and reports;
- Automatic data validation methods to check incoming data quality (e.g. periodicity and cross-correlation analysis);
- Easy to use graphical interface to train complex models (no coding required);
- Real-time analyses and anomaly detection methods based on complex AI models.