不同元启发式算法在热管换热器结构优化中的对比研究OA
Comparative study of different metaheuristic algorithms in optimizing the structure of heat pipe heat exchangers
[目的]热管换热器在燃煤电厂具有节能降耗的作用,但如何在制造材料最少情况下获得最优效率是一项工程化应用难题.目前针对热管换热器的结构优化,需要突破算法选型无据、参数设置依赖经验及理论解难以落地的问题,找出优化过程中的参数敏感性规律并建立"理论-工程"评估新视角.[方法]本文以某90 t/h CFB锅炉的空气预热器所选用的气-气式热管换热器为研究对象,构建以降低金属用量为目标的多约束优化模型,选取带约束遗传算法(constrained genetic algorithm,CGA)、粒子群优化算法(particle swarm optimization,PSO)和基于线性种群规模缩减的成功历史自适应差分进化算法(linear population size reduction success-history based adaptive differential evolution,L-SHADE)3种算法,在相同条件下进行系统对比.创新性地量化分析种群规模、迭代次数及随机性对性能的耦合影响,并引入"理论最优解向工程标准规格取整"的后评估机制,模拟真实制造约束.[结果]研究表明,3种算法均存在最优参数区间.CGA算法虽可有效处理约束,但收敛速度不及PSO算法和L-SHADE算法;PSO在保证解质量前提下计算效率最高.将理论连续解按制造标准取整后,CGA、PSO和L-SHADE算法计算的金属用量分别为0.592、0.586和0.588 m³,较初始设计分别降低24.30%、25.06%和24.81%.尽管算法在理论空间表现各异,但经工程规格取整后,3种最终方案趋于一致,揭示了理想模型与实际约束间的性能收敛规律.[结论]虽最终工程方案相近,但PSO凭借卓越效率与稳定性成为首选.本文所述方法显著提升了优化设计的工程适用性.
[Objective]The heat pipe heat exchanger has the effect of saving energy and reducing consumption in coal-fired power plants,but how to obtain the optimal efficiency with the least manufacturing materials is an engineering application problem.At present,for the structural optimization of heat pipe heat exchanger,it is necessary to break through the problems of no basis for algorithm selection,dependence on experience for parameter setting and difficulty in landing theoretical solutions,find out the parameter sensitivity law in the optimization process and establish a new perspective of"theory-engineering"evaluation.[Methods]In this paper,the gas-gas heat pipe heat exchanger selected for the air preheater of a 90 t/h CFB boiler is taken as the research object,and a multi-constraint optimization model with the goal of reducing the metal consumption is constructed.Three algorithms,constrained genetic algorithm(CGA),particle swarm optimization(PSO)and linear population size reduction success-history based adaptive differential evolution(L-SHADE),are selected.The system comparison is carried out under the same conditions.The coupling effects of population size,iteration times and randomness on performance are quantitatively analyzed innovatively,and the post-evaluation mechanism of'rounding the theoretical optimal solution to the engineering standard specification'is first proposed to simulate the real manufacturing constraints.[Results]Studies have shown that all three algorithms have optimal parameter intervals.Although CGA algorithm can effectively deal with constraints,its convergence speed is not as fast as PSO algorithm and L-SHADE algorithm.PSO has the highest computational efficiency under the premise of ensuring the quality of the solution.After rounding the theoretical continuous solution according to the manufacturing standard,the metal consumptions calculated by CGA,PSO and L-SHADE algorithms are 0.592 m3,0.586 m3 and 0.588 m3,respectively,which are 24.30%,25.06%and 24.81%lower than the initial design.Although the performance of the algorithm in the theoretical space is different,the three final schemes tend to be consistent after the engineering specifications are rounded,which reveals the performance convergence between the ideal model and the actual constraints.[Conclusion]Although the final engineering schemes are similar,PSO has become the first choice due to its excellent efficiency and stability.The method described in this paper significantly improves the engineering applicability of the optimization design.
赵洋;黄林滨;谭锐;朱旭初;马凯;王跃童
国能南京电力试验研究有限公司,江苏 南京 210023国能南京电力试验研究有限公司,江苏 南京 210023||华中科技大学能源与动力工程学院,湖北 武汉 430074国能南京电力试验研究有限公司,江苏 南京 210023国能南京电力试验研究有限公司,江苏 南京 210023河北省低碳高效发电技术重点实验室(华北电力大学),河北 保定 071003河北省低碳高效发电技术重点实验室(华北电力大学),河北 保定 071003
能源科技
热管换热器结构参数优化元启发式算法L-SHADE
heat pipe heat exchangerstructural parameter optimizationmetaheuristic algorithmsL-SHADE
《电力科技与环保》 2026 (4)
544-553,10
低碳智能燃煤发电与超净排放全国重点实验室开放课题项目(D2023FK084)国家能源投资集团有限责任公司科技创新项目(GJNY-23-68)
评论