Marine predator algorithm incorporating hybrid search operators and competitive learning and applications
-
摘要:
针对经典海洋捕食者算法(MPA)在迭代后期种群多样性丧失、求解精度不高且难以跳出局部最优解等问题,提出了一种融合混合搜索算子和竞争学习的海洋捕食者算法PSMPA。引入随机动态重心反向学习机制,增强算法后期的种群多样性,扩展搜索空间,提升算法逃离局部最优解并加速收敛的能力;融合动态随机搜索和模式搜索作为混合搜索算子,增强算法的局部搜索能力;将竞争学习行为模式引入捕食者当中,改善种群平均适应度值,有效促进算法的快速收敛,显著提高解的质量。选用12个CEC2017测试函数进行仿真实验,结果表明:PSMPA在寻优性能、收敛速度和稳定性方面均取得了较大程度的改善。在太阳能光伏模型参数优化设计问题上的应用进一步验证了PSMPA在实际工程优化问题中的应用价值和有效性。
-
关键词:
- 海洋捕食者算法 /
- 混合搜索算子 /
- 随机动态重心反向学习 /
- 竞争学习 /
- 工程优化问题
Abstract:The predator search marine predators algorithm PSMPA is a new algorithm that incorporates hybrid search operators and competitive learning to address the problems of population diversity loss, low solution precision, and difficulty escaping local optima in the later iterations of the classic marine predator algorithm (MPA). The introduction of a stochastic dynamic centroid opposition-based learning mechanism enhances population diversity in the later stages of the algorithm, expands the search space, and improves the algorithm’s ability to escape local optima and accelerate convergence. By combining random search and pattern search as hybrid search operators, the algorithm’s local search capability is enhanced. The average population fitness is increased when predators engage in competitive learning behavior, which effectively promotes rapid convergence and greatly enhances solution quality. Simulations using 12 CEC2017 benchmark functions demonstrate that PSMPA achieves substantial improvements in optimization performance, convergence speed, and stability. Furthermore, its application in optimizing parameters for solar photovoltaic models further validates PSMPA’s practical value and effectiveness in solving real-world engineering optimization problems.
-
表 1 CEC2017测试函数
Table 1. CEC2017 test functions
类型 标号 函数名称 最优值 单峰函数 F1 Shifted and Rotated Bent Cigar Function 100 简单多峰函数 F2 Shifted and Rotated Zakharov Function 300 F3 Shifted and Rotated Rosenbrock’s Function 400 F4 Shifted and Rotated Schwefel’s Function 1000 混合函数 F5 Hybrid Function 2(N=3) 1200 F6 Hybrid Function 3(N=3) 1300 F7 Hybrid Function 5(N=4) 1500 F8 Hybrid Function 6(N=4) 1600 组合函数 F9 Hybrid Function 6(N=6) 2000 F10 Composition Function 1 (N=3) 2100 F11 Composition Function 2 (N=3) 2200 F12 Composition Function 8 (N=6) 2800 算法 参数设置 MPA[1] FAD效应概率为0.2,步长控制因子为0.5 FA[17] 光吸收系数为1,最大吸引力为2,随机扰动因子为0.2,扰动衰减因子为0.98 GABC[16] 位置更新步长因子为0.5,均值学习权重为0.5 CLPSO[19] 种群规模为100,认知因子为1,社会因子为2,惯性权重为$ [0.4{,}0.9] $,惯性权重下限为$ [0.4{,}0.9] $ SaDE[18] 学习周期为20 PSMPA FAD效应概率为0.2,步长控制因子为0.5 NMPA[21] FAD效应概率为0.2,步长控制因子为0.5 QQLMPA[20] FAD效应概率为0.2,步长控制因子为0.8,局部搜索权重为0.8 RDPSO[22] 种群规模为100, 认知因子为2,社会因子为2,惯性权重下限为$ [0.4{,}0.9] $ PSO_sono[23] 种群规模为100,惯性权重下限为$ [0.4{,}0.9] $,收敛控制因子为0.1×问题维度/种群规模,社会学习因子为0.5 SHADE[24] 交叉概率均值为0.5,缩放因子均值为0.5 LMHHO[25] 长期记忆限制为10 表 3 经典算法测试函数结果统计
Table 3. Statistical results of test functions for classical algorithms
函数 算法 平均值 标准差 F1 PSMPA 1.00×102 2.45×10−1 FA 7.19×103 6.78×103 GABC 1.53×109 6.41×108 CLPSO 3.60×103 2.22×103 SaDE 3.18×103 2.99×103 F2 PSMPA 3.00×102 1.18×10−6 FA 3.26×102 1.37×102 GABC 1.69×105 4.13×104 CLPSO 6.96×104 1.40×104 SaDE 2.35×104 7.86×103 F5 PSMPA 1.43×104 3.50×104 FA 1.10×105 1.02×105 GABC 8.61×107 4.99×107 CLPSO 9.25×105 5.36×105 SaDE 5.22×105 7.35×105 F6 PSMPA 1.99×103 3.09×102 FA 2.03×104 1.99×104 GABC 7.32×105 5.56×105 CLPSO 1.47×104 1.07×104 SaDE 1.62×104 1.40×104 F7 PSMPA 1.59×103 4.12×101 FA 8.41×103 8.16×103 GABC 1.39×105 2.51×105 CLPSO 1.47×104 1.11×104 SaDE 2.33×103 1.19×103 F8 PSMPA 2.29×103 2.49×102 FA 2.34×103 3.32×102 GABC 3.82×103 2.23×102 CLPSO 2.63×103 2.55×102 SaDE 2.74×103 1.93×102 F9 PSMPA 2.18×103 7.13×101 FA 2.29×103 1.68×102 GABC 3.04×103 2.16×102 CLPSO 2.46×103 1.41×102 SaDE 2.41×103 1.10×102 F10 PSMPA 2.35×103 1.34×101 FA 2.37×103 2.07×101 GABC 2.57×103 2.42×101 CLPSO 2.43×103 3.58×101 SaDE 2.44×103 1.51×101 F11 PSMPA 2.30×103 1.37×101 FA 2.66×103 1.11×103 GABC 9.56×103 2.11×10 CLPSO 2.53×103 1.10×103 SaDE 4.35×103 2.95×103 注:黑体数据表示最佳结果。 表 4 近期变体算法测试函数结果统计
Table 4. Statistics results of test function for recent variants algorithms
函数 算法 平均值 标准差 排名 函数 算法 平均值 标准差 排名 F1 PSMPA 1.001×102 3.919×10−1 1 F7 PSMPA 1.617×103 4.251×101 1 MPA 3.306×105 3.283×105 3 MPA 1.701×103 4.967×101 2 NMPA 9.765×107 7.434×107 7 NMPA 1.696×103 4.377×101 3 QQLMPA 5.083×107 5.473×107 6 QQLMPA 1.728×103 5.939×101 4 PSO_sono 4.380×106 6.529×106 4 PSO_sono 1.135×104 8.577×103 6 RDPSO 1.120×107 1.543×107 5 RDPSO 1.024×104 1.214×104 7 LMHHO 5.421×1010 6.351×109 8 LMHHO 4.611×108 3.300×108 8 SHADE 4.120×102 3.709×102 2 SHADE 1.979×103 6.128×102 5 F2 PSMPA 3.000×102 1.059×10−6 1 F8 PSMPA 2.262×103 3.001×102 1 MPA 4.010×103 2.292×103 2 MPA 2.301×103 2.254×102 2 NMPA 5.633×103 1.995×103 4 NMPA 2.444×103 2.405×102 3 QQLMPA 1.117×104 6.931×103 5 QQLMPA 2.806×103 2.315×102 6 PSO_sono 2.139×104 6.702×103 6 PSO_sono 2.527×103 3.037×102 5 RDPSO 2.948×105 1.599×105 8 RDPSO 3.197×103 4.939×102 7 LMHHO 8.865×104 2.452×103 7 LMHHO 6.399×103 1.481×103 8 SHADE 5.114×103 3.303×103 3 SHADE 2.519×103 2.222×102 4 F3 PSMPA 4.360×102 4.590×101 1 F9 PSMPA 2.164×103 8.774×101 1 MPA 4.997×102 1.940×101 3 MPA 2.225×103 1.080×102 2 NMPA 5.246×102 3.145×101 6 NMPA 2.285×103 1.247×102 3 QQLMPA 5.146×102 2.398×101 5 QQLMPA 2.394×103 8.926×101 6 PSO_sono 5.805×102 5.618×101 7 PSO_sono 2.389×103 1.250×102 5 RDPSO 5.066×102 4.683×101 4 RDPSO 2.610×103 2.954×102 7 LMHHO 1.447×104 2.742×103 8 LMHHO 3.214×103 1.530×102 8 SHADE 4.846×102 1.882×101 2 SHADE 2.339×103 1.380×102 4 F4 PSMPA 4.301×103 8.085×102 2 F10 PSMPA 2.350×103 1.248×101 1 MPA 4.300×103 6.234×102 1 MPA 2.354×103 6.429×101 2 NMPA 4.799×103 6.658×102 4 NMPA 2.389×103 7.870×101 5 QQLMPA 6.358×103 8.571×102 6 QQLMPA 2.423×103 4.602×101 6 PSO_sono 5.294×103 7.823×102 5 PSO_sono 2.362×103 1.535×101 3 RDPSO 9.179×103 3.448×102 8 RDPSO 2.371×103 2.892×101 7 LMHHO 8.436×103 4.809×102 7 LMHHO 2.797×103 6.066×101 8 SHADE 4.472×103 3.837×102 3 SHADE 2.369×103 1.162×101 4 F5 PSMPA 8.999×103 1.290×104 1 F11 PSMPA 2.301×103 1.185×100 1 MPA 4.662×105 2.080×106 3 MPA 2.309×103 4.311×100 2 NMPA 2.061×106 1.878×106 5 NMPA 2.371×103 4.826×101 4 QQLMPA 1.142×106 1.613×106 4 QQLMPA 2.337×103 2.332×101 3 PSO_sono 6.307×106 1.437×107 7 PSO_sono 2.412×103 4.995×102 5 RDPSO 5.139×106 4.596×106 6 RDPSO 1.009×104 1.529×103 8 LMHHO 1.534×1010 3.777×109 8 LMHHO 1.004×104 6.296×102 7 SHADE 4.523×104 4.530×104 2 SHADE 3.638×103 1.798×103 6 F6 PSMPA 1.980×103 3.467×102 1 F12 PSMPA 3.111×103 3.266×101 1 MPA 3.731×103 6.768×102 3 MPA 3.236×103 2.136×101 3 NMPA 3.257×103 7.251×102 2 NMPA 3.266×103 2.716×101 5 QQLMPA 4.737×103 1.210×103 5 QQLMPA 3.257×103 2.528×101 4 PSO_sono 2.102×104 1.896×104 6 PSO_sono 3.315×103 5.774×101 7 RDPSO 2.334×104 3.136×104 7 RDPSO 3.299×103 3.860×100 6 LMHHO 9.444×109 4.915×109 8 LMHHO 7.194×103 6.037×102 8 SHADE 4.307×103 6.049×103 4 SHADE 3.219×103 4.205×101 2 注:黑体数据表示最佳结果。 表 5 Wilcoxon秩和检验结果
Table 5. Wilcoxon rank sum test results
函数 MPA NMPA QQLMPA PSO_sono RDPSO LMHHO SHADE $ p $ $ R $ $ p $ $ R $ $ p $ $ R $ $ p $ $ R $ $ p $ $ R $ $ p $ $ R $ $ p $ $ R $ F1 3.02×10−11 + 3.02×10−11 + 3.02×10−11 + 3.02×10−11 + 3.02×10−11 + 3.02×10−11 + 3.69×10−11 + F2 3.02×10−11 + 3.02×10−11 + 3.02×10−11 + 3.02×10−11 + 3.02×10−11 + 3.02×10−11 + 3.02×10−11 + F3 2.00×10−6 + 9.26×10−9 + 5.09×10−8 + 1.96×10−10 + 2.15×10−6 + 3.02×10−11 + 1.25×10−4 + F4 8.65×10−1 − 1.56×10−2 + 6.72×10−10 + 2.28×10−5 + 3.02×10−11 + 3.02×10−11 + 4.73×10−1 − F5 3.69×10−11 + 3.02×10−11 + 3.34×10−11 + 3.69×10−11 + 3.02×10−11 + 3.02×10−11 + 3.35×10−8 + F6 6.07×10−11 + 9.76×10−10 + 4.08×10−11 + 3.02×10−11 + 1.09×10−10 + 3.02×10−11 + 1.29×10−6 + F7 4.11×10−7 + 2.15×10−6 + 1.60×10−7 + 3.02×10−11 + 4.50×10−11 + 3.02×10−11 + 3.16×10−10 + F8 5.79×10−1 − 2.61×10−2 + 1.56×10−8 + 4.64×10−3 + 9.26×10−9 + 3.02×10−11 + 1.17×10−3 + F9 3.78×10−2 + 9.21×10−5 + 3.47×10−10 + 5.97×10−9 + 1.78×10−10 + 3.02×10−11 + 7.22×10−6 + F10 9.21×10−5 + 9.51×10−6 + 1.43×10−8 + 2.75×10−3 + 2.25×10−4 + 3.02×10−11 + 2.00×10−6 + F11 2.37×10−10 + 3.02×10−11 + 3.02×10−11 + 8.10×10−10 + 3.02×10−11 + 3.02×10−11 + 5.61×10−5 + F12 4.50×10−11 + 3.02×10−11 + 3.34×10−11 + 3.02×10−11 + 3.02×10−11 + 3.02×10−11 + 6.72×10−10 + 表 6 可扩展性分析结果统计
Table 6. Statistics on results of scalability analysis
函数 算法 50维 100维 函数 算法 50维 100维 平均值 标准差 平均值 标准差 平均值 标准差 平均值 标准差 F1 PSMPA 3.528×103 2.586×103 9.475×103 5.908×103 F7 PSMPA 7.431×103 6.688×103 3.781×103 2.112×103 MPA 4.413×107 2.724×107 3.621×109 1.275×109 MPA 3.761×103 1.423×103 5.665×104 3.002×104 NMPA 1.688×109 1.090×109 2.198×1010 5.534×109 NMPA 3.632×103 1.651×103 5.331×106 3.410×106 QQLMPA 1.077×109 8.709×108 2.412×1010 1.670×1010 QQLMPA 3.869×103 8.206×102 1.088×106 8.724×105 PSO_sono 2.765×108 3.648×108 4.326×109 3.658×109 PSO_sono 1.437×104 8.904×103 3.411×104 2.164×104 RDPSO 4.574×109 2.362×109 1.170×1011 2.987×1010 RDPSO 8.495×104 1.530×105 1.069×107 1.212×107 LMHHO 1.045×1011 7.368×109 2.557×1011 7.525×109 LMHHO 7.694×109 2.075×109 2.650×1010 3.869×109 SHADE 5.992×103 8.188×103 7.043×106 6.060×106 SHADE 5.485×103 4.463×103 5.522×103 3.890×103 F2 PSMPA 4.241×102 1.502×102 7.288×104 1.155×104 F8 PSMPA 2.687×103 3.133×102 5.093×103 6.557×102 MPA 4.229×104 9.827×103 2.149×105 2.338×104 MPA 3.067×103 3.994×102 5.774×103 8.139×102 NMPA 4.673×104 1.010×104 2.255×105 1.743×104 NMPA 3.246×103 3.650×102 6.925×103 6.222×102 QQLMPA 6.874×104 1.438×104 2.491×105 2.292×104 QQLMPA 3.840×103 4.836×102 9.162×103 1.116×103 PSO_sono 1.217×105 3.301×104 3.211×105 4.582×104 PSO_sono 3.325×103 3.807×102 6.231×103 7.215×102 RDPSO 1.238×106 6.525×105 7.936×106 5.217×106 RDPSO 5.450×103 7.098×102 1.226×104 7.306×102 LMHHO 2.663×105 3.055×104 3.620×105 8.152×103 LMHHO 9.951×103 2.149×103 2.554×104 2.986×103 SHADE 6.177×104 4.587×104 3.169×105 6.728×104 SHADE 3.515×103 2.778×102 7.377×103 4.900×102 F3 PSMPA 4.408×102 5.482×101 5.577×102 1.117×102 F9 PSMPA 2.609×103 3.124×102 4.591×103 4.973×102 MPA 6.316×102 5.690×101 1.438×103 1.917×102 MPA 2.752×103 1.964×102 4.700×103 3.455×102 NMPA 7.913×102 8.247×101 3.065×103 6.870×102 NMPA 2.833×103 2.725×102 5.104×103 3.625×102 QQLMPA 7.409×102 1.047×102 2.598×103 1.318×103 QQLMPA 3.306×103 1.971×102 6.292×103 5.526×102 PSO_sono 8.578×102 1.770×102 2.214×103 8.400×102 PSO_sono 3.117×103 3.186×102 5.186×103 5.874×102 RDPSO 1.051×103 2.280×102 1.462×104 4.245×103 RDPSO 4.284×103 3.812×102 8.342×103 4.666×102 LMHHO 3.500×104 4.117×103 1.030×105 1.041×104 LMHHO 4.476×103 2.627×102 8.021×103 3.604×102 SHADE 5.523×102 5.424×101 7.432×102 5.248×101 SHADE 3.299×103 2.035×102 6.318×103 3.726×102 F4 PSMPA 6.753×103 9.441×102 1.408×104 1.183×103 F10 PSMPA 2.420×103 2.410×101 2.662×103 5.785×101 MPA 6.926×103 5.894×102 1.618×104 1.381×103 MPA 2.493×103 3.603×101 2.898×103 5.965×101 NMPA 8.805×103 9.060×102 2.183×104 1.621×103 NMPA 2.590×103 3.761×101 3.217×103 9.728×101 QQLMPA 1.155×104 1.505×103 2.766×104 2.135×103 QQLMPA 2.631×103 6.070×101 3.257×103 1.096×102 PSO_sono 9.186×103 1.209×103 2.356×104 3.102×103 PSO_sono 2.450×103 3.485×101 3.034×103 1.020×102 RDPSO 1.593×104 5.371×102 3.413×104 7.713×102 RDPSO 2.495×103 4.477×101 3.286×103 1.689×102 LMHHO 1.516×104 8.697×102 3.233×104 1.057×103 LMHHO 3.326×103 1.161×102 5.053×103 1.854×102 SHADE 8.898×103 4.419×102 2.372×104 9.486×102 SHADE 2.504×103 1.847×101 3.120×103 6.968×101 F5 PSMPA 2.029×105 4.414×105 5.435×106 1.679×107 F11 PSMPA 4.338×103 2.976×103 1.028×104 8.180×103 MPA 2.338×107 1.704×107 5.443×108 2.520×108 MPA 6.153×103 3.178×100 1.944×104 1.412×103 NMPA 1.157×108 6.637×107 2.624×109 1.192×109 NMPA 7.298×103 4.065×103 2.386×104 5.707×103 QQLMPA 7.038×107 5.181×107 2.307×109 1.730×109 QQLMPA 1.074×104 4.017×103 2.900×104 4.624×103 PSO_sono 5.911×107 6.068×107 7.337×108 4.766×108 PSO_sono 1.077×104 1.583×103 2.573×104 2.831×103 RDPSO 2.658×108 2.703×108 1.221×1010 4.671×109 RDPSO 1.764×104 3.756×102 3.623×104 6.093×102 LMHHO 8.692×1010 1.581×1010 1.991×1011 1.682×1010 LMHHO 1.742×104 6.857×102 3.490×104 9.788×102 SHADE 1.191×106 1.043×106 1.574×107 1.151×107 SHADE 1.034×104 1.624×103 2.612×104 1.247×103 F6 PSMPA 4.764×103 3.238×103 9.421×103 5.318×103 F12 PSMPA 3.296×103 3.243×101 3.383×103 5.182×101 MPA 5.149×104 3.775×104 3.225×105 4.482×105 MPA 3.423×103 4.764×101 4.685×103 5.006×102 NMPA 8.226×104 7.141×104 8.662×107 1.232×108 NMPA 3.645×103 1.883×102 6.396×103 8.676×102 QQLMPA 8.509×104 5.927×104 3.817×107 1.990×107 QQLMPA 3.531×103 1.308×102 6.539×103 1.858×103 PSO_sono 4.731×104 2.871×104 1.314×106 4.375×106 PSO_sono 3.834×103 3.138×102 7.005×103 1.543×103 RDPSO 1.430×106 1.291×106 1.403×108 1.120×108 RDPSO 3.300×103 1.853×100 3.300×103 1.297×10−4 LMHHO 4.842×1010 1.538×1010 4.720×1010 5.695×109 LMHHO 1.336×104 9.151×102 2.960×104 1.694×103 SHADE 1.299×104 1.015×104 1.238×104 8.830×103 SHADE 3.303×103 1.698×101 3.519×103 4.364×101 注:黑体数据表示最佳结果。 表 7 待识别参数统计
Table 7. Statistics of parameters to be identified
模型 $ {I}_{\mathrm{ph}}/\mathrm{A} $ $ {I}_{\text{sd}},{I}_{\mathrm{sd}1},{I}_{\mathrm{sd}2}/\text{μA} $ $ {R}_{\mathrm{S}}/\Omega $ $ {R}_{\text{sh}}/\Omega $ $ n,{n}_{1},{n}_{2} $ 下限 上限 下限 上限 下限 上限 下限 上限 下限 上限 单二极管/双二极管 0 1 0 1 0 0.5 0 100 1 2 光伏组件 0 2 0 50 0 2 0 2000 1 50 表 8 不同算法在单二极管模型求解结果
Table 8. Solution results of different algorithms in single diode model
算法 $ {I}_{\text{ph}}/\mathrm{A} $ $ {I}_{\text{sd}}/\text{μA} $ $ {R}_{\mathrm{S}}/\Omega $ $ {R}_{\text{sh}}/\Omega $ $ n $ RMSE PSMPA 0.7609 9.2596 ×10−70.0317 92.2450 1.59551 9.8602 ×10−4MPA 0.7627 9.2230 ×10−70.0315 56.0962 1.5951 9.8622 ×10−4NMPA 0.7665 7.9000 ×10−70.0285 9.9306 1.5769 1.4576 ×10−3QQLMPA 0.7632 2.8926 ×10−70.0357 23.8530 1.4759 1.4926 ×10−3PSO_snono 0.7608 3.2302 ×10−70.0364 53.7185 1.48119 9.8602 ×10−4RDPSO 0.7611 3.3144 ×10−70.0363 53.4172 1.48374 1.0094 ×10−3LMHHO 0.7627 9.2230 ×10−70.0315 56.0962 1.5951 2.8042 ×10−3SHADE 0.7603 5.0222 ×10−70.0347 81.2078 1.52685 9.8602 ×10−4表 9 不同算法在双二极管模型求解结果
Table 9. Solution results of different algorithms in double diode model
算法 $ {I}_{\text{ph}}/\mathrm{A} $ $ {I}_{\mathrm{sd}1}/\text{μA} $ $ {R}_{\mathrm{S}}/\Omega $ $ {R}_{\text{sh}}/\Omega $ $ {n}_{1} $ $ {I}_{\mathrm{sd}2}/\text{μA} $ $ {n}_{2} $ RMSE PSMPA 0.7575 3.3312 ×10−70.0341 41.9237 1.4975 5.4355 ×10−71.7851 9.8261 ×10−4MPA 0.7604 6.6848 ×10−70.0331 89.2116 1.5587 7.1148 ×10−82.0000 1.0432 ×10−3NMPA 0.7734 1.0000 ×10−60.0268 11.0175 1.6161 5.6194 ×10−71.9079 1.3933 ×10−3QQLMPA 0.7575 3.3312 ×10−70.0341 41.9237 1.4975 5.4355 ×10−71.7851 1.4467 ×10−3PSO_snono 0.7604 8.7006 ×10−70.0365 63.9692 1.9858 2.2910 ×10−71.4536 1.0232 ×10−3RDPSO 0.7607 4.9749 ×10−70.0347 78.1739 1.5258 1.0000 ×10−92.0000 1.3667 ×10−3LMHHO 0.7618 6.7570 ×10−70.0319 32.7746 1.5735 8.3032 ×10−71.9140 7.5806 ×10−3SHADE 0.7627 6.0857 ×10−70.0320 48.4662 1.5609 6.3290 ×10−71.8478 9.8248 ×10−4表 10 不同算法在光伏组件模型求解结果
Table 10. Solution results of different algorithms in PV module models
算法 $ {I}_{\text{ph}}/\mathrm{A} $ $ {I}_{\text{sd}}/\text{μA} $ $ {R}_{\mathrm{S}}/\Omega $ $ {R}_{\text{sh}}/\Omega $ $ n $ RMSE PSMPA 1.0392 4.8979 ×10−61.1563 766.7714 50.0000 2.4251 ×10−3MPA 1.0414 4.9747 ×10−61.1661 1676.9654 50.0000 2.4546 ×10−3NMPA 1.0222 4.8724 ×10−61.1594 2000.0000 50.0000 2.6906 ×10−3QQLMPA 1.0265 3.5349 ×10−61.2351 1430.5212 48.6941 2.6027 ×10−3PSO_snono 1.0305 3.4823 ×10−61.2013 981.9822 48.6429 2.4251 ×10−3RDPSO 1.0311 4.3026 ×10−61.1747 993.0811 49.4755 2.5684 ×10−3LMHHO 1.0287 4.8548 ×10−61.1666 1557.9994 49.9487 1.1013 ×10−2SHADE 1.0361 4.8511 ×10−61.1303 622.4671 50.0000 2.4251 ×10−3表 11 不同算法在光伏组件模型结果统计
Table 11. Results statistics of different algorithms in PV module model
模型 算法 RMSE 最小值 平均值 最大值 标准差 单二极管模型 PSMPA 9.8602 ×10−41.1959 ×10−31.9791 ×10−32.9787 ×10−4MPA 9.8622 ×10−43.0772 ×10−38.6551 ×10−31.9661 ×10−3NMPA 1.4576 ×10−34.5834 ×10−31.3668 ×10−22.9732 ×10−3QQLMPA 1.4926 ×10−31.4412 ×10−24.1147 ×10−21.3134 ×10−2PSO_snono 9.8602 ×10−41.7923 ×10−32.4480 ×10−36.4516 ×10−4RDPSO 1.0094 ×10−31.2591 ×10−21.1689 ×10−12.2583 ×10−2LMHHO 2.8042 ×10−33.3661 ×10−28.9889 ×10−22.3615 ×10−2SHADE 9.8602 ×10−49.8602 ×10−49.8602 ×10−43.4431 ×10−17双二极管模型 PSMPA 9.8261 ×10−41.2621 ×10−34.7446 ×10−37.4253 ×10−4MPA 1.0432 ×10−32.5841 ×10−36.4684 ×10−31.3126 ×10−3NMPA 1.3933 ×10−33.4807 ×10−38.9248 ×10−31.6362 ×10−3QQLMPA 1.4467 ×10−31.1673 ×10−25.4201 ×10−21.5256 ×10−2PSO_snono 1.0232 ×10−33.1527 ×10−33.9488 ×10−26.8902 ×10−3RDPSO 1.3667 ×10−36.5507 ×10−33.2181 ×10−26.8343 ×10−3LMHHO 7.5806 ×10−33.8441 ×10−21.9368 ×10−13.4146 ×10−2SHADE 9.8248 ×10−49.8650 ×10−41.0385 ×10−31.1249 ×10−5光伏组件模型 PSMPA 2.4251 ×10−32.5521 ×10−34.8768 ×10−34.4650 ×10−4MPA 2.4546 ×10−32.6859 ×10−34.1413 ×10−32.9001 ×10−4NMPA 2.6906 ×10−35.3223 ×10−31.2407 ×10−22.6504 ×10−3QQLMPA 2.6027 ×10−32.6581 ×10−28.7399 ×10−22.8684 ×10−2PSO_snono 2.4251 ×10−32.0777 ×10−22.7425 ×10−16.8904 ×10−2RDPSO 2.5684 ×10−38.3559 ×10−34.8559 ×10−21.1885 ×10−2LMHHO 1.1013 ×10−22.0279 ×10−13.1883 ×10−11.1563 ×10−1SHADE 2.4251 ×10−31.1486 ×10−22.7425 ×10−14.9628 ×10−2注:黑体数据表示最佳结果。 -
[1] FARAMARZI A, HEIDARINEJAD M, MIRJALILI S, et al. Marine predators algorithm: a nature-inspired metaheuristic[J]. Expert Systems with Applications, 2020, 152: 113377. [2] HOLLAND J H. Genetic algorithms[J]. Scientific American, 1992, 267(1): 66-73. [3] KENNEDY J, EBERHART R. Particle swarm optimization[C]// Proceedings of the International Conference on Neural Networks. Piscataway: IEEE Press, 2002: 1942-1948. [4] YANG X S, DEB S. Cuckoo search: recent advances and applications[J]. Neural Computing and Applications, 2014, 24(1): 169-174. [5] RAI R, DHAL K G, DAS A, et al. An inclusive survey on marine predators algorithm: variants and applications[J]. Archives of Computational Methods in Engineering, 2023, 30(5): 3133-3172. [6] HELMI A M, AL-QANESS M A A, DAHOU A, et al. Human activity recognition using marine predators algorithm with deep learning[J]. Future Generation Computer Systems, 2023, 142(C): 340-350. [7] 马驰, 曾国辉, 黄勃, 等. 融合混沌对立和分组学习的海洋捕食者算法[J]. 计算机工程与应用, 2022, 58(22): 271-283.MA C, ZENG G H, HUANG B, et al. Marine predator algorithm based on chaotic opposition learning and group learning[J]. Computer Engineering and Applications, 2022, 58(22): 271-283(in Chinese). [8] 张磊, 刘升, 高文欣, 等. 精英反向黄金正弦海洋捕食者算法[J]. 计算机工程与科学, 2023, 45(2): 355-362.ZHANG L, LIU S, GAO W X, et al. An elite opposition-based golden sine marine predator algorithm[J]. Computer Engineering & Science, 2023, 45(2): 355-362(in Chinese). [9] SHAHEEN M A M, YOUSRI D, FATHY A, et al. A novel application of improved marine predators algorithm and particle swarm optimization for solving the ORPD problem[J]. Energies, 2020, 13(21): 5679. [10] AYDEMIR S B. Enhanced marine predator algorithm for global optimization and engineering design problems[J]. Advances in Engineering Software, 2023, 184: 103517. [11] KUSUMA P D, ADIPUTRA D. Hybrid marine predator algorithm and hide object game optimization[J]. Engineering Letters, 2023, 31(1): 262-270. [12] HAMZACEBI C, KUTAY F. Continuous functions minimization by dynamic random search technique[J]. Applied Mathematical Modelling, 2007, 31(10): 2189-2198. [13] 吴建辉, 王博华, 张小刚, 等. 基于模式搜索法的云模型粒子群算法[J]. 控制与决策, 2017, 32(11): 2076-2080.WU J H, WANG B H, ZHANG X G, et al. Cloud model particle swarm optimization algorithm based on pattern search method[J]. Control and Decision, 2017, 32(11): 2076-2080(in Chinese). [14] TIZHOOSH H R. Opposition-based reinforcement learning[J]. Journal of Advanced Computational Intelligence and Intelligent Informatics, 2006, 10(4): 578-585. [15] RAHNAMAYAN S, JESUTHASAN J, BOURENNANI F, et al. Computing opposition by involving entire population[C]//Proceedings of the IEEE Congress on Evolutionary Computation. Piscataway: IEEE Press, 2014: 1800-1807. [16] ZHU G P, KWONG S. Gbest-guided artificial bee colony algorithm for numerical function optimization[J]. Applied Mathematics and Computation, 2010, 217(7): 3166-3173. [17] YANG X S. Firefly algorithms for multimodal optimization[C]//Proceedings of the International Symposium on Stochastic Algorithms. Berlin: Springer, 2009: 169-178. [18] QIN A K, HUANG V L, SUGANTHAN P N. Differential evolution algorithm with strategy adaptation for global numerical optimization[J]. IEEE Transactions on Evolutionary Computation, 2009, 13(2): 398-417. [19] LIANG J J, QIN A K, SUGANTHAN P N, et al. Comprehensive learning particle swarm optimizer for global optimization of multimodal functions[J]. IEEE Transactions on Evolutionary Computation, 2006, 10(3): 281-295. [20] ZHAO S R, WU Y L, TAN S, et al. QQLMPA: a quasi-opposition learning and Q-learning based marine predators algorithm[J]. Expert Systems with Applications, 2023, 213: 119246. [21] SADIQ A S, DEHKORDI A A, MIRJALILI S, et al. Nonlinear marine predator algorithm: a cost-effective optimizer for fair power allocation in NOMA-VLC-B5G networks[J]. Expert Systems with Applications, 2022, 203: 117395. [22] SUN J, PALADE V, WU X J, et al. Solving the power economic dispatch problem with generator constraints by random drift particle swarm optimization[J]. IEEE Transactions on Industrial Informatics, 2014, 10(1): 222-232. [23] MENG Z Y, ZHONG Y X, MAO G J, et al. PSO-sono: a novel PSO variant for single-objective numerical optimization[J]. Information Sciences, 2022, 586: 176-191. [24] SUN X P, JIANG L S, SHEN Y, et al. Success history-based adaptive differential evolution using turning-based mutation[J]. Mathematics, 2020, 8(9): 1565. [25] HUSSAIN K, ZHU W, SALLEH M N M. Long-term memory Harris’ hawk optimization for high dimensional and optimal power flow problems[J]. IEEE Access, 2019, 7: 147596-147616. [26] JIANG L L, MASKELL D L, PATRA J C. Parameter estimation of solar cells and modules using an improved adaptive differential evolution algorithm[J]. Applied Energy, 2013, 112: 185-193. [27] YU K J, LIANG J J, QU B Y, et al. Parameters identification of photovoltaic models using an improved JAYA optimization algorithm[J]. Energy Conversion and Management, 2017, 150: 742-753. -


下载: