College of Mechanical and Electrical Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China.
School of Engineering, The University of Edinburgh, Edinburgh EH9 3JL, United Kingdom.
Rev Sci Instrum. 2020 Dec 1;91(12):124104. doi: 10.1063/5.0029491.
A hybrid Genetic Algorithm (GA) and Levenberg-Marquardt (GA-LM) method is proposed for cell suspension measurement with electrical impedance spectroscopy. This algorithm combines the GA with global search ability and Levenberg-Marquardt (LM) algorithm with local search ability, which has the advantages of high accuracy and high robustness. First, GA-LM is compared with GA and LM algorithm separately by ideal simulation. Second, Gaussian noise is added to the ideal simulation data. The anti-noise ability of the GA-LM is discussed. Finally, experiments are conducted to verify the practicability of the proposed GA-LM method. In the experiment, GA-LM is used to fit the impedance spectrum of yeast suspensions with different volume fractions and active states. The results show that the GA-LM algorithm can converge to the real value that is set in the simulation under ideal numerical simulation conditions. In the simulation within 2% noise level, the mean relative error of the parameter solution is less than 4%, and the root mean square error of the fitting is less than 0.4. This method also performs well in fitting of the experimental data. In addition, the electric double layer resistance and cell membrane capacitance are selected as the main indicators for the identification of yeast suspension concentration and activity, respectively.
一种混合遗传算法（GA）和列文伯格-马夸尔特（GA-LM）方法被提出用于细胞悬浮液的阻抗谱测量。该算法将 GA 的全局搜索能力与 LM 算法的局部搜索能力相结合，具有高精度和高鲁棒性的优点。首先，通过理想模拟将 GA-LM 与 GA 和 LM 算法分别进行比较。其次，向理想模拟数据中添加高斯噪声。讨论了 GA-LM 的抗噪声能力。最后，通过实验验证了所提出的 GA-LM 方法的实用性。在实验中，使用 GA-LM 拟合具有不同体积分数和活性状态的酵母悬浮液的阻抗谱。结果表明，在理想数值模拟条件下，GA-LM 算法可以收敛到模拟中设置的真实值。在噪声水平为 2%的模拟中，参数解的平均相对误差小于 4%，拟合的均方根误差小于 0.4。该方法在实验数据的拟合中也表现良好。此外，双电层电阻和细胞膜电容分别被选为酵母悬浮液浓度和活性的主要识别指标。
