can anyone give idea how to use optimization (like pso/gwo etc) in a circuit. as i am new to this field so i need ideas

 

(like pso/gwo etc)

Please define what you are talking about.

 

i mean that i have a converter topology & i want to optimize my circuit with any of the optimization techniques for better performance. by using any technique parameter values can be defined so that performance can be better.

 

Let’s step back a bit. We do not even know what the circuit you want to optimize is supposed to do. The words “converter topology” suggest DC to DC converters to me, but there may very well be other types of circuits that these words apply to.

Tell us first what type of circuit you are talking about. Then what pso/gwo means, because I have no idea.

 

An internet search comes up with:
particle swarm optimization (PSO)
grey wolf optimizer (GWO)
The internet did not define what either is.

 

These guys might point you at some useful approaches:

https://www.frontiersin.org/journals/energy-research/articles/10.3389/fenrg.2024.1476638/full

 

i mean that i have a converter topology & i want to optimize my circuit with any of the optimization techniques for better performance. by using any technique parameter values can be defined so that performance can be better.

What does "better performance" mean?

More accurate and precise output? With temperature? With load? Over time?
More efficient?
Lower cost? For the BOM? For the manufacturing costs? For volume production? Higher yield?
Smaller footprint?
More reliable?

You have little hope of optimizing anything unless you know what it is you are trying to optimize.

 

Generally you can do statistically evaluate 3sigma on all prototypes and in production and compute margins to upper and lower/or control limit specs for a dozen design specs.

That stuff does features for correlation like random forest (primary), support vector machines, k-nearest neighbors, and logistic regression.

LTspice can do Monte Carlo methods ,

- also you can consider Taguchi multivariable Analysis by change all variables +/- X% like L, C, DCR , ESR R , loop gain phase margin, efficiency and measure step up/down overshoot settling time, hot cold thermals, Noise P-P. Cload stability range, etc and conducted emissions. This is called Design of Experiments DoE and is used on process controls like fluxing process variables and SMD solder controls too.

 

You start by defining all the specs of a power supply.

 

Back in the early days of black and white television, I was given to understand that the designers took the prototype circuit and removed parts, one at a time, until it stopped working. When they put back the last part they had removed, they had "optimized" the cost of the circuit.

 

Back in the early days of black and white television, I was given to understand that the designers took the prototype circuit and removed parts, one at a time, until it stopped working. When they put back the last part they had removed, they had "optimized" the cost of the circuit.

One of my professors worked for RCA back in the day (mid-1950s, I think). He said that one of the highest-paid engineers there had a single job -- to reduce the cost of the flyback transformer by fives cents each year. As long as he managed to do that, he had a guaranteed job for the following year, plus a significant bonus for each cent above fives cents (enough to make it worthwhile not to sandbag). IIRC, he was making something like $20k/yr and RCA was making something like 1.5 million TV sets a year. So a five cent savings per unit would add something in the ballpark of $75k to the bottom line. I would imagine that by the time they first made him that offer that it was already pretty close to being as optimized as everyone else could make it, so it probably took all of his effort to meet the goal. I don't recall my prof saying how long he was able to keep meeting that target.