FIDO Differential Analysis
FIDO’s patented algorithm detects and analyses minute nuances and variations in sound, vibration, speed and turbulence, depending on the data available. It filters out extraneous data and is unaffected by background noise or pipe material.
FIDO is an implementation of scientist Neil Edward’s white paper Applying Differential Analysis and Machine Learning Techniques to Temporal Data Series in Order to Determine Fluid Leakage from Pipelines.
The FIDO algorithm
FIDO’s patented hybrid deep-learning multi-tiered algorithm analyses the vibrational frequencies and wave functions of any audio and kinetic file in eight distinct ways.
First, it decides if the file it’s analysing is a leak, then it calculates its likely size relative to a baseline.
Fully hardware agnostic, FIDO can work with data from any 3rd party sensor and can incorporate other available utility company data points such as flow, pressure and leak resolution information.
FIDO is compliant with industry data sharing protocols because it processes and learns remotely.
Constantly evolving to meet new challenges
As FIDO acquires more and more real fix data, the machine-learning algorithm is improving and re-applying old data to look for other data patterns.
Leak-sizing is an example of this. As FIDO gathers and verifies relevant vibrational signatures with flow data, the accuracy and precision of the leak size results will improve to the point that absolute leak volume/hour prediction will be possible for each leak sound.
We’re also developing the application of specific waveform analysis to narrow leak location to a single defined point between sensor locations.
FIDO is designed to act as both a stand-alone product or a decision support and verification tool to supplement top level network analysis tools and identify with the best accuracy possible the physical locations of leaks and their size.
The diagram shows how all the software and hardware components in the FIDO solution interact with each other and customer platforms.
FIDO’s cloud-based algorithm is at the heart of it all. Audio and vibration files from any client asset such as sensors, third party apps or FIDO’s own smartphone app are automatically uploaded to the FIDO platform for analysis. Client data is accessed via secure data exchange and not held by FIDO Tech.
FIDO instantly compares new files against more than 1.7 million verified leak and non-leak samples in its growing library to give its decision on leak likelihood and size, which, once verified, further adds to its deep learning knowledge and improves accuracy.
Quarterly software updates go through extensive offline concept testing and development before being made available to clients during beta testing as a release candidate (RC) on the platform. At RC stage, FIDO uses real data to rapidly learn and improve the accuracy of new functions before they go live.