CAJ | 학술논문

为了求解某些类型的复杂函数优化问题，基于 SEIV 传染病模型提出了一种新型函数优化算法，即SEIV算法。在该算法中，假设某个生态系统由若干个人和动物个体组成；每个人和动物个体均由若干个特征来表征。该生态系统存在一种传染病在人与动物之间传染，其传染规律为动物传给人或动物传给动物，这种传染病攻击的是个体的部分特征。每个染病个体均经历易感、暴露、接种或发病等阶段。个体的体质强弱是通过该个体的某些特征的暴露、某些特征的接种、某些特征的发病与某些特征的易感等情况综合决定的。依据SEIV传染病模型的疾病传播规律构造出了相关演化算子，其中E-E、V-V和I-I算子能传递强壮个体的特征信息，使得虚弱个体能向好的方向发展；S-E和S-S算子能使异类或同类（仅指动物）个体之间交换信息；S-V、V-S、E-I和E-V算子能使个体获得其他同类个体的平均特征信息，从而降低了个体陷入局部最优解的概率；S-S算子能使个体的活跃度提高，从而扩大搜索范围。体质强壮的个体能继续生长，而体质虚弱的个体则停止生长，从而确保该算法具有全局收敛性。结果表明，本算法对求解某些复杂函数优化问题具有较高的适应性和收敛速度。
위료구해모사류형적복잡함수우화문제，기우 SEIV 전염병모형제출료일충신형함수우화산법，즉SEIV산법。재해산법중，가설모개생태계통유약간개인화동물개체조성；매개인화동물개체균유약간개특정래표정。해생태계통존재일충전염병재인여동물지간전염，기전염규률위동물전급인혹동물전급동물，저충전염병공격적시개체적부분특정。매개염병개체균경력역감、폭로、접충혹발병등계단。개체적체질강약시통과해개체적모사특정적폭로、모사특정적접충、모사특정적발병여모사특정적역감등정황종합결정적。의거SEIV전염병모형적질병전파규률구조출료상관연화산자，기중E-E、V-V화I-I산자능전체강장개체적특정신식，사득허약개체능향호적방향발전；S-E화S-S산자능사이류혹동류（부지동물）개체지간교환신식；S-V、V-S、E-I화E-V산자능사개체획득기타동류개체적평균특정신식，종이강저료개체함입국부최우해적개솔；S-S산자능사개체적활약도제고，종이확대수색범위。체질강장적개체능계속생장，이체질허약적개체칙정지생장，종이학보해산법구유전국수렴성。결과표명，본산법대구해모사복잡함수우화문제구유교고적괄응성화수렴속도。
In order to solve some types of complicated function optimization problems,this paper constructed a SEIV algorithm based on the SEIV epidemic model.The algorithm supposed that some human and animal individuals exist in an ecosystem;it each characterized individual by a number of features;an infectious disease exists in the ecosystem and infects among individu-als,the rule of infection was animal individuals infect human individuals or animal individuals were infected each other,the dis-ease attacked a part of features of an individual.Each infected individual passed through such stages as suspected,exposed,in-fected or vaccinated.It decided the individual physique strength of an individual synthetically by the exposure,vaccination,in-fection and susceptibility of certain features.The transmitting rules of the infectious disease in the SEIV epidemic model were used to constructed evolving operators in which the E-E,V-V and I-I operator were used to transfer feature information from some strong individuals to an weak individual so as to make the week individual grows better;the S-E and S-S operator were used to transfer feature information between heterogeneous or homogeneous (only for animal)individuals;the S-V,V-S,E-I,E-V oper-ator were used to ensure an individual to obtain average feature information from other homogeneous individuals so as to reduce probability that the individual droped into local optimum solutions;the S-S operator was used to expands an individual’s search scope by increasing its activity.The individuals with strong physique could continue to grow,while the individuals with weak physique stop growing,this could ensure the algorithm to globally converge.Results show that the algorithm has characteristics of strong search capability and high adaptability for some types of complicated functions optimization problems.