neural network technologies
Today abilities of human brain considerably exceed those of modern PC-s in terms of processing, analysis and storage of information. Researches and experiments have proven that human brain is able to accumulate from 3 up to 1000 terabyte of data during its life.
These very “super-abilities” are attributed to unique cells – neurons, which are interconnected by synapses and thus form a neural network.
The issue of neural networks is widely discussed today since such developments as machine learning, intelligent algorithms and artificial intellect enable us to solve a wide range of tasks.
The Software for Structural Health Monitoring Systems (hereinafter referred to as SHM) allows to select multi-channel programmable detectors and is designed based on principles of neural networks. This allows to improve overall performance in the course of vibration impacts, damping decrement and bearing elements tilt control.
This type of structure enables high-precision recording of the events relating to the impacts upon the controlled object in heavy interference environment. It is also possible to adjust the parameters of the multi-functional detector, so that to avoid false alarm notifications.
SHM Software enables registered signals data transfer to third-party systems both in normal operation mode and in the case if the limits of acceptable values are exceeded. Operator’s workstation displays notification message in the case if an event is registered. Each message is to be confirmed – this allows to make sure that operator has checked the notification message.
“Danger” signal can be transferred to monitoring service of the facility or to other corresponding local emergency authorities.
SHM Software enables displaying of the registered event at operator’s display as well as its positioning on the structural scheme or a particular location.