基于粒子群萤火虫混合算法的计算机辅助配棉

Computer aided cotton blending based on particle and glowworm swarm hybrid optimization algorithm

为促进棉纺企业精细化管理水平和生产效益的提升,提出一种粒子群萤火虫混合算法的计算机辅助配棉方法。首先,构造混和棉成本最小和混和棉综合质量指标最优的多目标优化函数,建立库存、总重量、原棉种类及质量指标边界等多约束条件,按照线性加权方式将多目标模型转化为单目标模型。其次,对粒子群和萤火虫群进行分阶段初始化,设计一种学习因子动态调整和非线性递减惯性权重策略用以提高粒子群算法的综合寻优能力,并采用自适应移动步长更新萤火虫个体位置。最后,使用粒子群、萤火虫及粒子群萤火虫混合算法对配棉模型进行求解。试验结果表明:3种配棉算法均展现出良好的求解可行性,所得混和棉质量指标的综合绝对误差分别为0.0268、0.0240、0.0281,皆处于较低水平;并且粒子群萤火虫混合算法在成本节约方面更具优势,其配棉的总成本相比粒子群算法、萤火虫算法分别降低了1.20%、2.27%。

To enhance the refined management level and production efficiency of cotton spinning enterprises,a particle and glowworm swarm hybrid optimization algorithm was proposed for computer aided cotton blending.Firstly,a multi-objective optimization function was constructed to minimize the cost of mixed cotton and optimize the comprehensive quality indicators of mixed cotton.Multiple constraints such as inventory,total weight,types,and quality indicator boundary were established.The multi-objective model was transformed into a single-objective model using a linear weighting approach.Secondly,a phased initialization was applied to both particle swarm and glowworm swarm.A dynamic learning factor and a non-linearly decreasing inertia weight strategy were designed to enhance the comprehensive optimization capability of the particle swarm algorithm.An adaptive movement step was employed to update the position of glowworm individuals.Finally,the cotton blending model was solved using a hybrid algorithm of particle swarm,glowworm,and particle-glowworm swarm.The experimental results showed that the three cotton blending algorithms demonstrated better feasibility in solving problems,and the comprehensive absolute errors of the quality index for obtained mixed cotton were 0.0268,0.0240,and 0.0281,respectively,and all at a relatively low level.Moreover,the particle and glowworm swarm hybrid algorithm had more advantages in cost savings.The total cost of cotton blending was reduced by 1.20%and 2.27%compared to the particle swarm algorithm and glowworm algorithm,respectively.