Think of any complex system - your factory, a machine, a business, even a society like the orange below. Yes even the orange is a complex system of atoms, cells, segmented structures and layers.

Each of these complex systems contains information that can be used to describe its properties. when these properties are given degrees of freedom in space and time, they become dimensions.

When an external event acts on the system the system will respond according to its properties. this may be a power fluctuation, a huge unexpected order when your near 100% capacity, or even a competitors new product displacing yours. By understanding the response of the system under a few known conditions the systems properties can be determined, just like we would measure the properties of a material. Or like sending a sonar ping and measuring the response, or an ultrasound measurement, the internal structures can be observed and simulated.

The problem with traditional machine learning techniques to collect data and try to predict is that they need to understand how every atom in the orange above will respond by collecting lots of data to train the system. But not every event or circumstance will be captured or will there be any understanding of correlated versus causal events.

We don’t need to model every atom in the orange to understand how it will fail under different scenarios. Having measured the systems properties, we are able to see where a problem is or stress test under different scenarios under controlled conditions. Finding weak points in the system or even the structure of the business under different market scenarios allows actions to be taken before they cause an expensive reaction.

The project aims to develop a holistic tool to monitor factory and system health with the potential for stress testing capability to forward predict problems under different production scenarios. Decision makers will be able at a glance, see a single visual of the overall operational health of the facility and allow a drill down to localize any problems under current or future simulated ‘stress test’ events. The method offers potential to include the wider business strategy, organizational structure and organizational health providing several optimization strategies to build a roadmap to growth.

How might it look?

Well every system will have its own structure based on its complexity, but this naturally forms from the properties to describe it found during the development phase. An example might look like the below images where smaller complex systems (spheres) are nested within larger systems in a fractal manner. So when a visual distortion and color change are observed, one can drill down to a more localized visual of the problem to root cause problems faster to see which inputs and outputs are the effected.

Interested to find out more or potentially get involved? We are preparing a talk for the upcoming SVC TechCon 2023 Conference where you can speak directly with Frank Papa.

A video description by World Formula Apps 16-year-old intern is provided here: Decision Optimization