订单型服装企业流水线设备配置比例的预测及生产计划的优化
|
摘要
随着市场竞争的日趋激烈以及市场需求的日趋多样化，越来越多的企业选择多品种小批量这种生产方式来快速的适应市场的变化。在当今的市场中，企业的竞争主要取决于产品的供货周期、质量和生产成本。先进的管理是实现在上述目标的一个重要手段。调度是生产管理的核心内容和关键技术，其任务是在企业有限的资源约束下，保证所选定的生产目标最优。科学的生产计划的安排方案，对于控制企业的在制品库存，提高产品交货满足率，缩短产品供货周期和提高企业生产率起着至关重要的作用。
本文首先介绍了订单型服装企业的特点，指出柔性加工将是服装加工企业的发展方向，智能技术在其中要起到重要作用也是发展的趋势，通过对目前比较热门的人工神经网络技术和遗传算法的介绍，针对它们的特点和缺陷，本文提出了一种以人工神经网络和遗传算法为基础的BP-GA智能混合算法，并通过实验证明了这种算法的有效性。同时利用BP-GA智能混合算法对订单型服装生产企业的设备配置比例进行了预测研究，旨在通过合理利用设备来提高设备利用率、降低生产成本，其中网络的输入参数是通过对服装各部件的分解，找到服装各部件的共同特点以及分析这些特征与设备配置之间的关系来确定的，输出函数是来自专家的设备配置比例方案，隐含层结点数是通过实验比较而得到。最后对预测模型进行了仿真，仿真结果令人满意，证明了这一方法的可行性。
随着我国市场经济的发展，高产值已不再是企业追求的主要目标，而准时生产准时交货的(Just in time，JIT)生产则成为企业在市场中取胜的必备条件，因而体现JIT思想，与交货期有关的提前／拖期问题成了新的研究热点。本文简要介绍了目前生产计划提前／拖期问题的研究状况，建立了一个该问题的数学模型，在此基础上，应用遗传算法，对交货期窗口下的提前／拖期问题进行了研究，并在计算机上得以仿真实现。仿真结果表明，该算法不受问题规模的限制，对于解决大规模复杂的问题更显示出其优越性，为MRPII与JIT思想在车间作业计划上的结合提供了有利工具。
Becoming more diversified with fierceness and market demand day by day of the market competition, More and more enterprise choose many variety short run the mode of production come fast adaptation change of market. In the current market, the competitions of enterprises depend on the stock cycle, quality and production cost of the products mainly. Advanced management is realizing in an importance means of above-mentioned goals. The dispatcher is the key content of the production management and key technology, its task is under the circumstances that limited resource of enterprise are restrained, Guarantee that the selected production goal is optimum. The arrangement scheme of the scientific production schedule, for controlling the material in process of enterprises stock, Improve the products and deliver the satisfying rate, shorten the products and supply cycle and raise enterprise's productivity to play an essential role.
This text has introduced order type the characteristics of clothing enterprises at first, it is the developing direction of the processing enterprise of the clothing to point out flexibility and process, And it is a development trend too that intelligent technology should play a important role among them, Pass to hot artificial, To their characteristic and defect, this text has proposed that a kind of BP-GA intelligence based on artificial neural network and genetic algorithm mixes algorithms, And has proved the validity of this kind of algorithm through the experiment. Utilize BP-GA intelligence mix between algorithm and order type
clothing equipment allocation ratio of manufacturing enterprise predict that study at the same time, Aim at it through making use of equipment to raise the utilization rate of equipment and installations, reduce the production cost rationally, Among them the introduction parameter of the network is through resolving to every part of the clothing, Find clothing every common characteristic and analyse characteristic and relation of device layout these come and confirm of part, Output function from expert equipment allocation ratio scheme, imply layer form and count and receive through experiment. To predicting that models carried on emulation, the emulation result is satisfactory, proved the feasibility of this method finally.
With the development of market economy of our country, the high output value has already no longer been the main goal that enterprises pursue, And produce on time delivering goods on time(Just In Time,JIT) produce and become enterprise win essential condition among market, Whether therefore it reflect JIT thought it is the related to delivery date ana aneaa or time /tow tnere aren't issues oi issue, wnetner tnis text production schedule the ahead of time /last issues of the research states of issue, Have set up a mathematics model of this question, on this basis, use the hereditary algorithm, Whether it is for delivery date the windows under it is the ahead of time /tow because there aren't issues of issue, and must use in computer there aren't emulation. Emulation result indicates , this algorithm is not restricted by scale of question , demonstrate his superiority even more in solving the problem that extensive and complicated, Have offered the favorable tool for MRPII and combination in the production pla
n of the workshop of JIT thought.
本文首先介绍了订单型服装企业的特点，指出柔性加工将是服装加工企业的发展方向，智能技术在其中要起到重要作用也是发展的趋势，通过对目前比较热门的人工神经网络技术和遗传算法的介绍，针对它们的特点和缺陷，本文提出了一种以人工神经网络和遗传算法为基础的BP-GA智能混合算法，并通过实验证明了这种算法的有效性。同时利用BP-GA智能混合算法对订单型服装生产企业的设备配置比例进行了预测研究，旨在通过合理利用设备来提高设备利用率、降低生产成本，其中网络的输入参数是通过对服装各部件的分解，找到服装各部件的共同特点以及分析这些特征与设备配置之间的关系来确定的，输出函数是来自专家的设备配置比例方案，隐含层结点数是通过实验比较而得到。最后对预测模型进行了仿真，仿真结果令人满意，证明了这一方法的可行性。
随着我国市场经济的发展，高产值已不再是企业追求的主要目标，而准时生产准时交货的(Just in time，JIT)生产则成为企业在市场中取胜的必备条件，因而体现JIT思想，与交货期有关的提前／拖期问题成了新的研究热点。本文简要介绍了目前生产计划提前／拖期问题的研究状况，建立了一个该问题的数学模型，在此基础上，应用遗传算法，对交货期窗口下的提前／拖期问题进行了研究，并在计算机上得以仿真实现。仿真结果表明，该算法不受问题规模的限制，对于解决大规模复杂的问题更显示出其优越性，为MRPII与JIT思想在车间作业计划上的结合提供了有利工具。
Becoming more diversified with fierceness and market demand day by day of the market competition, More and more enterprise choose many variety short run the mode of production come fast adaptation change of market. In the current market, the competitions of enterprises depend on the stock cycle, quality and production cost of the products mainly. Advanced management is realizing in an importance means of above-mentioned goals. The dispatcher is the key content of the production management and key technology, its task is under the circumstances that limited resource of enterprise are restrained, Guarantee that the selected production goal is optimum. The arrangement scheme of the scientific production schedule, for controlling the material in process of enterprises stock, Improve the products and deliver the satisfying rate, shorten the products and supply cycle and raise enterprise's productivity to play an essential role.
This text has introduced order type the characteristics of clothing enterprises at first, it is the developing direction of the processing enterprise of the clothing to point out flexibility and process, And it is a development trend too that intelligent technology should play a important role among them, Pass to hot artificial, To their characteristic and defect, this text has proposed that a kind of BP-GA intelligence based on artificial neural network and genetic algorithm mixes algorithms, And has proved the validity of this kind of algorithm through the experiment. Utilize BP-GA intelligence mix between algorithm and order type
clothing equipment allocation ratio of manufacturing enterprise predict that study at the same time, Aim at it through making use of equipment to raise the utilization rate of equipment and installations, reduce the production cost rationally, Among them the introduction parameter of the network is through resolving to every part of the clothing, Find clothing every common characteristic and analyse characteristic and relation of device layout these come and confirm of part, Output function from expert equipment allocation ratio scheme, imply layer form and count and receive through experiment. To predicting that models carried on emulation, the emulation result is satisfactory, proved the feasibility of this method finally.
With the development of market economy of our country, the high output value has already no longer been the main goal that enterprises pursue, And produce on time delivering goods on time(Just In Time,JIT) produce and become enterprise win essential condition among market, Whether therefore it reflect JIT thought it is the related to delivery date ana aneaa or time /tow tnere aren't issues oi issue, wnetner tnis text production schedule the ahead of time /last issues of the research states of issue, Have set up a mathematics model of this question, on this basis, use the hereditary algorithm, Whether it is for delivery date the windows under it is the ahead of time /tow because there aren't issues of issue, and must use in computer there aren't emulation. Emulation result indicates , this algorithm is not restricted by scale of question , demonstrate his superiority even more in solving the problem that extensive and complicated, Have offered the favorable tool for MRPII and combination in the production pla
n of the workshop of JIT thought.
引文
[1]许庆瑞、王爱民、张友仁 生产管理 1988．9 高等教育出版社 北京 第113页
[2]闻新等 Matlab神经网络应用设计 科学出版社 北京 2002
[3]刘永红 神经网络理论的发展与前沿问题 信息与控制 1999．28(1)31-46
[4]越胜、王保中、康立山 遗传算法设计中的赋值函数和节点分类计算机科学 1998．25(3)
[5]陈世福 人工智能与知识工程 南京大学出版社 南京 1997
[6]张文修、梁怡 遗传算法的数学基础西安交通大学出版社 西安 2000第3页
[7]王小平 曹立明 遗传算法—理论、应用与软件实现 西安交通大学出版社 西安 第4页
[8]陈颖琪、余松、赵奕进化计算和人工神经网络的结合 信息与控制 1999．29(8)31-33
[9]李书全、赵良英、史智兴、寇纪松、李敏强 一种防止遗传算法在成熟前收敛的有效算法 系统工程理论与实践 1999．5 72-77
[10]Vittorio M. Genetic Evolution of the Topplogy and Weight Distribution of Neural Networks. IEEE Trans on Neural Networks 1994.5(1)
[11]艾丽蓉、何华灿 遗传算法综述 计算机应用研究 1999．41-8
[12]杨恒辉、戴韧、陈康民 遗传算法中参数选取及其在数据全局寻优中的应用 计算机与现代化 2003 2第3页
[13]Holland J H. Adaptation in Natural and Artificial Systems MIT Press 1992
[14]刘国华、包宏、李文超 基于WWW的人工神经网络遗传算法系统的实现 计算机工程 28 2第33页
[15]Richard P Lippmann An Introduction to Computing with Neural Nets[J] IEEE ASSP Magazine 1987 4-22
[16]焦李成 神经网络的应用于实现 西安电子科技大学出版社 西安 1993
[17]梁化楼、戴贵亮 人工神经网络与遗传算法的结合：进展与展望电子学报 1995(10) 194-200
[18]徐荣、徐用懋、兰鸿森 多层前向网络的研究——遗传算法BP算法和结构优化策略 自动化学报 1997 23(1)43-49
[19]穆文全、廖晓峰、虞厩邦 基于遗传算法和BP算法的多层感知机杂交训练算法 电子科学学报 1993 21(11)1-6
[20]熊凌、赵明旺 基于遗传算法的BP网络全局收敛的混台智能学习算法 武汉科技大学学报(自然科学版) 2000 23 (2)183页
[21]王万良 基于遗传算法的车间调度方法研究 同济大学博士学位论文 54-60
[22]胡守仁 神经网络导论 国防科技大学出版社 北京 1993．1
[23]Schradoph N,et al Dynamic Parameter Encoding for Genetic Algorithms Machine Learning 1992 9(1) 9-12
[24]曹德武、赵枚 BP算法的改进及其在非线性转子建模中的应用数值计算与计算机应用 2000 2 121-126
[25]焦李成、保铮 同伦BP网络理论与算法 电子科学学报 1993 21(11) 1-6
[26]蔡自兴、徐光有 人工智能及其应用 清华大学出版社 北京 1996第324页
[27]娄天顺、于卫 基于MATLAB的系统分析与设计 西安电子科技大学出版社 西安 1998
[28]许东、吴铮 基于Matlab6.x的系统分析与设计—神经网络西安电子科技大学出版社 西安 2002
[29]罗四维 人工神经网络建造 中国铁道出版社 1998 54-67
[30]耿兆强、邹益仁 基于遗传算法的作业车间模糊调度问题的研究 计算机集成制造系统-CINN 2002 8第8卷第8期．
[31]吴悦、汪定伟．模糊交货期窗口下极大化顾客满意数的最优调度．系统工程理论方法应用 1999 8(1)．
[32]TADAM I H, MASUDA T. Two scheduling problems with fuzzy due dates. Fuzzy Sets and Systems 1992 46 (3) :339-347.
[33]战德臣、陈伟、王忠杰 基于启发式规则的混合遗传算法及其在生产计划中的应用 计算机工程及应用 2003 8第215页．
[34]LAN H ,ROSS P ,CORNE D. A promising genetic algorithm ap2proach to job - shop scheduling, rescheduling and open - shopscheduling problems [A] . Proceedings of the Fifth International Conference on Genetic Algorithms [C] . San Manto : Morgan Kaufmann, 1993. 375 - 382.
[35]车间生产调度问题研究、机械工程学报 2002 5 97-98．
[36]吴悦、汪定伟 单机作业下极大化顾客满意数的提前／拖期调度问题 系统工程理论方法应用 1998 7(3) 45-49页．
[37]黄德才、经玲、杨晚年一个基于JIT的FMS零件排序问题的模型及解法 计算机集成制造系统—CIMS 1997 3(6)45～48．
[38]李秀、刘文煌、姜惩宇、王宁生 遗传算法在车间批量生产计划问题中的应用 南京航空航天大学学报 2001 33(6)544-54(5
[39]李秀 集成环境下连续式生产模式车间管理控制器的研究 南京航空航天大学学位论文 32-40
[40]陈国良、王煦法、庄镇泉遗传算法以及应用 人民邮电出版社 北京 1996
[41]金毅 生产计划的模糊方法与应用山西人民出版社 西安 1997 1-7
[2]闻新等 Matlab神经网络应用设计 科学出版社 北京 2002
[3]刘永红 神经网络理论的发展与前沿问题 信息与控制 1999．28(1)31-46
[4]越胜、王保中、康立山 遗传算法设计中的赋值函数和节点分类计算机科学 1998．25(3)
[5]陈世福 人工智能与知识工程 南京大学出版社 南京 1997
[6]张文修、梁怡 遗传算法的数学基础西安交通大学出版社 西安 2000第3页
[7]王小平 曹立明 遗传算法—理论、应用与软件实现 西安交通大学出版社 西安 第4页
[8]陈颖琪、余松、赵奕进化计算和人工神经网络的结合 信息与控制 1999．29(8)31-33
[9]李书全、赵良英、史智兴、寇纪松、李敏强 一种防止遗传算法在成熟前收敛的有效算法 系统工程理论与实践 1999．5 72-77
[10]Vittorio M. Genetic Evolution of the Topplogy and Weight Distribution of Neural Networks. IEEE Trans on Neural Networks 1994.5(1)
[11]艾丽蓉、何华灿 遗传算法综述 计算机应用研究 1999．41-8
[12]杨恒辉、戴韧、陈康民 遗传算法中参数选取及其在数据全局寻优中的应用 计算机与现代化 2003 2第3页
[13]Holland J H. Adaptation in Natural and Artificial Systems MIT Press 1992
[14]刘国华、包宏、李文超 基于WWW的人工神经网络遗传算法系统的实现 计算机工程 28 2第33页
[15]Richard P Lippmann An Introduction to Computing with Neural Nets[J] IEEE ASSP Magazine 1987 4-22
[16]焦李成 神经网络的应用于实现 西安电子科技大学出版社 西安 1993
[17]梁化楼、戴贵亮 人工神经网络与遗传算法的结合：进展与展望电子学报 1995(10) 194-200
[18]徐荣、徐用懋、兰鸿森 多层前向网络的研究——遗传算法BP算法和结构优化策略 自动化学报 1997 23(1)43-49
[19]穆文全、廖晓峰、虞厩邦 基于遗传算法和BP算法的多层感知机杂交训练算法 电子科学学报 1993 21(11)1-6
[20]熊凌、赵明旺 基于遗传算法的BP网络全局收敛的混台智能学习算法 武汉科技大学学报(自然科学版) 2000 23 (2)183页
[21]王万良 基于遗传算法的车间调度方法研究 同济大学博士学位论文 54-60
[22]胡守仁 神经网络导论 国防科技大学出版社 北京 1993．1
[23]Schradoph N,et al Dynamic Parameter Encoding for Genetic Algorithms Machine Learning 1992 9(1) 9-12
[24]曹德武、赵枚 BP算法的改进及其在非线性转子建模中的应用数值计算与计算机应用 2000 2 121-126
[25]焦李成、保铮 同伦BP网络理论与算法 电子科学学报 1993 21(11) 1-6
[26]蔡自兴、徐光有 人工智能及其应用 清华大学出版社 北京 1996第324页
[27]娄天顺、于卫 基于MATLAB的系统分析与设计 西安电子科技大学出版社 西安 1998
[28]许东、吴铮 基于Matlab6.x的系统分析与设计—神经网络西安电子科技大学出版社 西安 2002
[29]罗四维 人工神经网络建造 中国铁道出版社 1998 54-67
[30]耿兆强、邹益仁 基于遗传算法的作业车间模糊调度问题的研究 计算机集成制造系统-CINN 2002 8第8卷第8期．
[31]吴悦、汪定伟．模糊交货期窗口下极大化顾客满意数的最优调度．系统工程理论方法应用 1999 8(1)．
[32]TADAM I H, MASUDA T. Two scheduling problems with fuzzy due dates. Fuzzy Sets and Systems 1992 46 (3) :339-347.
[33]战德臣、陈伟、王忠杰 基于启发式规则的混合遗传算法及其在生产计划中的应用 计算机工程及应用 2003 8第215页．
[34]LAN H ,ROSS P ,CORNE D. A promising genetic algorithm ap2proach to job - shop scheduling, rescheduling and open - shopscheduling problems [A] . Proceedings of the Fifth International Conference on Genetic Algorithms [C] . San Manto : Morgan Kaufmann, 1993. 375 - 382.
[35]车间生产调度问题研究、机械工程学报 2002 5 97-98．
[36]吴悦、汪定伟 单机作业下极大化顾客满意数的提前／拖期调度问题 系统工程理论方法应用 1998 7(3) 45-49页．
[37]黄德才、经玲、杨晚年一个基于JIT的FMS零件排序问题的模型及解法 计算机集成制造系统—CIMS 1997 3(6)45～48．
[38]李秀、刘文煌、姜惩宇、王宁生 遗传算法在车间批量生产计划问题中的应用 南京航空航天大学学报 2001 33(6)544-54(5
[39]李秀 集成环境下连续式生产模式车间管理控制器的研究 南京航空航天大学学位论文 32-40
[40]陈国良、王煦法、庄镇泉遗传算法以及应用 人民邮电出版社 北京 1996
[41]金毅 生产计划的模糊方法与应用山西人民出版社 西安 1997 1-7