Genetic algorithm solves thermistor problem
Rick Nelson, Editor in Chief -- Test & Measurement World, 5/1/2008
Given the circuit topology in Figure a, find the resistor and NTC thermistor values that generate the voltage vs. temperature curve shown in Figure b, using only standard resistor values. That’s the problem Craig Lyon, a senior member of the technical staff and strategic applications engineer at Maxim Integrated Products, presents in an article posted at EDN.com.
a)  |
b)  |
| A genetic algorithm can determine the circuit values (a) that match a specific voltage/temperature characteristic (b). | |
Some engineers, Lyon writes, may be able to solve the problem using a scratch pad and numerous equations, while others would probably painstakingly plug values into a spreadsheet and try to adjust them to obtain the desired characteristic.
A better, quicker alternative, he writes, is to employ an optimization algorithm that mimics a genetic process to find an optimal solution to a multivariable problem. He notes that other optimization algorithms can be faster than the genetic algorithm, but he explains that they are often more complicated to program and will frequently converge on a local minimum—difficulties the genetic algorithm avoids.
Lyon presents a simple routine he developed using a free Visual Basic compiler. To learn more about Lyon’s approach, visit the complete article at www.edn.com/CA6542464.html. The article includes a link to the source code and an executable file.
