Replies to post #32688 on Sigma Labs Inc (SASI)

[jjsmith77]

Thank you, EDMGUY.

The INSPECT closed loop link you used

Do you mean the diagram with controller 14, 16, 17? I got that from Sigma Lab's patent PDF. I didn't draw it but put labels and arrows to follow their explanation. The numbers are from their patent invention descriptions, btw.

Yes, experts systems are not perfect. Cola and Vivek also mentions this. The weakness in IPQA is its training. However, what makes SGLB's closed-loop process unique is higher order control systems. Notice control 16 & database 17 having a feedback loop to controller 14, allowing controller 14 to "learn" from higher functions. Controller 14 can then "better" control the printer. Each instance of anomaly can then be recognized and make the "fingerprinting" process more efficient.
http://investorshub.advfn.com/boards/read_msg.aspx?message_id=112635685

The weakness in IPQA is having a "true fingerprint" in database 17. This is where all parts are compared and matched. So for each layer of the build there must be a "true fingerprint" and this is the most difficult and critical goal to accomplish.

I'm on the same page with you on this. There is a weak link.

I am learning from you're explanations and this is what I enjoy. Thanks, EDMGUY.

[jjsmith77]

As an aside, EDMGUY.

Notice from SGLB's diagram that there are 2 loops:
1. 14 -> 15 -> 10-> 11&12 -> 13 -> 14
2. 14 -> 16 -> 17 -> 16 -> 14
Notice each loop starts and begins with the same number. It is, however, 17 at the highest order. Without 17, the optimizer (13) cannot run iterations to search for the correct printer parameters to fit "pre-determined" tolerances, the "fingerprint".

[0033] Based on one or more such thermal measurements, the weld pool volume can be directly estimated and controlled as follows: ...

[0041] The optimizer module can have a fitness function… defined by comparing the measured thermal response to the predicted thermal response…and iterates on parameter sets until this fitness is within a pre-determined tolerance…

[0054] When the weld pool volume varies from its desired setpoint, this can be detected by the thermal sensors, the weld pool frequency sensors, or combinations thereof…

[0055] Once a shift is detected, the real-time controller 14 can have an internal logic table that analyzes the nature… Based on pre-determined rules.., the controller 14 can decide which machine variable will be the best choice

[0056] The controller 14 can then make changes to the process variables welding machine tool 15...

[0057] The process continues indefinitely as long as the welding process 10 is in progress.

[0058] A supervisory controller 16 can also ensure that the chosen response vector produced the desired effect--otherwise it will seek an alternative vector, update the response vector database 17, or both. Updating the response vector database can allow the controller 14 to "learn" the process as it encounters more instances and examples of anomalous behaviour