微生物降解动力学参数估计新算法的比较

为了有效地估计较大范围变化的微生物降解动力学参数,发展了基于瞬时精英保护策略的遗传算法(IEPGA)和简单改进遗传算法(IGA),并和多次在参数区间内获取随机初值联用Matlab的lsqnonlin搜寻的算法进行了比较.这些算法利用模拟数据和文献数据,估计了积分形式的Monod模型参数.结果表明,虽然三者均能较好解决这一问题,但考虑到实际运行规模、时间以及最终结果精度,对于较大范围变化的微生物降解参数估计问题,多次(>20次)随机初值联用lsqnonlin的方法相对两种遗传算法更为可行.

Comparisons among New Algorithms of Parameter Estimation for Microbial Biodegradation Kinetics

To effectively estimate parameters varying within wide range for microbial biodegradation kinetics,a genetic algorithm with Instantaneous Elitist Protection Strategy(IEPGA) and an improved genetic algorithm(IGA) were developed and compared with the method of repetitious lsqnonlin procedure(Matlab software) initialized by stochastically-generated values.These algorithms were used to estimate parameters of the integrated-form Monod model based on synthetic data and published data,respectively.The results show that the three proposed algorithms are all capable of dealing with this problem.However,in view of operation scale,runtime and final results precision,the method of repetitious(>20 times) lsqnonlin function initialized by stochastic values is more feasible than GAs to estimate microbial biodegradation kinetics parameters.