【摘要】 针对人工生态系统算法易限于局部最优、全局探索能力差等缺陷，提出一种改进人工生态系统优化算法(Improved Artificial Ecosystem-based Optimization Algorithm, IAEO)。利用Hammersley点集初始化，使个体分布更加均匀；采用非线性递减及混沌序列来提高算法的探索和开发能力；加入爆炸操作和高斯变异来提高算法跳出局部最优的能力，在四个基准函数的仿真结果表明寻优能力有较大提高。利用多层极限学习机对数据进行特征提取，在有监督部分利用混合核极限学习机进行分类。利用IAEO优化混合核函数的核参数、正则化系数和比例系数，并在标准数据集上进行性能验证。将该方法应用于船舶柴油机故障诊断，该方法有效提高了故障诊断的准确性和稳定性并表现出较好的泛化性能。更多还原

【Abstract】 The artificial ecosystem-based optimization algorithm is easy to fall into local optimum, weak global exploration ability. Therefore, this paper proposes an improved artificial ecosystem-based optimization algorithm(IAEO). Hammersley point set was used to establish the initial population, which made the individual distribution more uniform. The nonlinear decreasing strategy was adopted and chaotic sequence was added to improve the exploration and exploitation ability of the algorithm. The explosion operation and Gaussian variation were added to help the algorithm to jump out of local optimum. The simulation results on four benchmark functions showed a significant improvement in the optimization ability. The multi-layer extreme learning machine was used for feature extraction of data, and the hybrid kernel extreme learning machine was used for classification. The IAEO was used to optimize the kernel parameters, regularization coefficients, and proportional coefficients of mixed kernel functions. The proposed network was evaluated on some benchmark datasets. The proposed method was applied to fault diagnosis of marine diesel engine. The results show that the proposed method effectively improves the diagnostic accuracy and stability, and demonstrates good generalization performance.更多还原