Tardid is pioneering the application of some Innovative Technologies to customers problems


How Artificial Intelligence helps Brainbox deliver real-time insight on industrial assets health

We start with the machines and structures every time we invent. We have been working with applied AI over four years, but we are only in the beginning stages of understanding the true potential of AI. We are researching further on how to improve health of industrial assets and help our customers solve big problems such as productivity losses due to unplanned downtime, unfortunate situations etc.


We work on real industrial problems that have a tangible impact on our customer businesses and world, and the benefit are growing by the day.


We have developed a strong chain of distributed Physics-Aware Neural network for effective state estimation


Due to the rapid introduction of volatile and renewable energy sources and controllable loads, asset conditions are challenged by unusual fluctuations in operational conditions. Therefore, accurate real-time monitoring of assets becomes increasingly pivotal in order to ensure reliable operations and understanding the continuous stresses causing failures. 

Brainbox has inbuilt classifying models which continuously observes physical phenomena and predicts outcomes of health and new phenomenon. Brainbox processes four types of models to simplify (1) Objects (2) Interactions between objects (3) Systems of objects together with their interactions, and/or (4) processes


Its all about monitoring environmental data in real-time under various cyclic loadings  like, ingress or loss, temperature changes, wind, waves, current, seismic, flows etc.


Brainbox considers these influencers to develop neural models and then applies the knowledge to determine the cyclic responses. Furthermore, these models are further categorized into discrete and continuum. These metrics are generated to assess the influencing environmental impact on the assets or its activities. The primary impacts are measured (renewable or non-renewable) to assess situations for reasoning purpose. The ultimate sustainability goal of this technology is to optimize operations and minimize downtimes. Since the complete elimination of these factors is hardly possible, but with advanced tracking and simulation Brainbox can bring an impeccable improvement in future design of these assets.


Limiting behaviours and positivity


The most important part of Artificial Intelligence is to create a mere-self of the subject. At Tardid we have been working extensively on Quasi-Local Mass to understand the asset behaviours and subsequent changes limiting to their behaviours and failures. The primary goal of Brainbox is to provide the "Time-To-Failure" matrix, and to do the same digitized data and models are not enough as there are certain invisible factors which has its own relevance stronger than what we perceive. 

Brainbox closely observes

  • Asymptotically Flat Manifolds

  • Quasi Local Mass in AH Manifolds, which includes curvature estimates

  • Positivity factors of the Masses

event horizon.jpg


Event Horizon for Effective Signal Processing

We have developed a quantum machine learn based filtering system which processes every frequency response as flat as possible in the passband. The event horizon is machine trained model which observes sequential data (streaming data) in real-time for anomaly detection and streaming responses. The AI Algorithms are embodied into various signal spaces (AE, various signatures) with autoencoders (GANs) .

This allows event horizon to listen to what it has been couched to, thus resulting rapid speed and accuracy.



AI based Geometric State Model Generator

We have built an algorithmic method of developing automated state equations for generating geometric models for individual assets.

These state models are live representation of assets under monitoring. The model includes the mechanical behaviour, influencing patterns, kinematics of the physical asset. This helps the AI to understand the current conditions and pertaining threats.



AI based Health Predictions for Pipes Under Insulation

By its very nature, CUI is very difficult to detect since corrosion occurs beneath the insulation, hence making the continous monitoring process very complicated.

To address the above challenge, we have developed Brainbox Adaptive Neural based Fuzzy Inference System (BANFIS). It gives result with the minimum errors without removing or disturbing the insulation layer, when compared to existing techniques. 

Subscribe to the Tardid Lookout

You'll get interesting stories about our employees, customers and the latest hot topics in Artificial Intelligence

Thanks for submitting!