采用遗传算法对AVR片内RC校频处理

Frequency Calibration for On-chip RC Oscillator of AVR Using Genetic Algorithms

介绍了一种使用遗传算法对AVR片内RC振荡器进行实时校频的方法。以AVR片内高精度时钟单元RTC为时基控制单元,以标准时间内MCU定时器的计数值为控制目标,采用遗传控制策略搜索对应MCU理想时钟频率的OSCCAL寄存器最佳设置值,快速准确地对RC振荡频率进行校准,为CPU提供具有高抗干扰特性的精确时钟。OSCCAL寄存器参数码的初始种群通过C语言库中的rand()函数随机产生,通过复制、交叉、变异等遗传操作获得各代的适应值。其中,种群复制按向最优解迫近的原则操作,交叉匹配按随机配对交换码符的方法操作,变异概率取值0.01以保证遗传算法的稳定性。给出了实时校频遗传算法的基本操作程序以及10次迭代的搜索结果,表明该算法的优化效果十分明显。

This paper described a method to use Genetic Algorithms to calibrate the on-chip RC oscillator of AVR real-time.Take the high precision on-chip RTC(real time clock) as the time base control unit,using Genetic Algorithms to search the best OSCCAL register value which will make the on-chip RC oscillator to generate a high precision clock for the MCU.This method can be used fast and real-time,so that the CPU will get a stable clock source.The initial population of OSCCAL register participation digital was randomly generated by the function of rand() from C language library.They get all generations of fitness through genetic manipulation such as reproduction,crossover,mutation and so on.Among them,the population reproduce according to the principle that close to the optimal solution,cross-matching pair exchanged code randomly at the method of operation,mutation probability values of 0.01 to ensure the stability of genetic algorithm.Given the basic operating procedures of real time genetic algorithm and 10 iterations of search results,show that the algorithm optimization effect is very obvious.