Contemporary real time system monitoring appears to be the same in each discipline and field of application. Whether it is an aircraft, a chemical plant, a hospitalized patient or an assembly line, the logic is always the same:
Deploy a set of sensors
Establish admissible upper and lower bounds for each sensor (channel)
Define a nominal value for each channel
Monitor deviation from nominal
Trigger alarm if value approaches one of the bounds
The surprising thing is that only on rare occasions are the relationships or inter-dependencies between data channels analyzed. And yet, if one has N data channels, the number of possible channel inter-dependencies is (N x N – N)/2. For N=100 this is nearly 5000. These inter dependencies establish structure, a map, or a graph, which, at the end of the day, illustrates how things work. Knowing which channels vary together, and how they do it (in a…
Artificial Intuition 
Contemporary real time system monitoring appears to be the same in each discipline and field of application. Whether it is an aircraft, a chemical plant, a hospitalized patient or an assembly line, the logic is always the same:
The surprising thing is that only on rare occasions are the relationships or inter-dependencies between data channels analyzed. And yet, if one has N data channels, the number of possible channel inter-dependencies is (N x N – N)/2. For N=100 this is nearly 5000. These inter dependencies establish structure, a map, or a graph, which, at the end of the day, illustrates how things work. Knowing which channels vary together, and how they do it (in a…
View original post 419 more words