A Genetic Algorithm-based Approach to Scheduling of Batch Production with Maximum Profit

The optimal scheduling of multi-product batch process is studied and a new mathematics model targeting the maximum profit is proposed, which can be solved by the modified genetic algorithm (MGA) with mixed coding (sequence coding and decimal coding) developed by us. In which, the partially matched cross over (PMX) and reverse mutation are used for the sequence coding, whereas the arithmetic crossover and heteropic mutation are used for the decimal coding. In addition, the relationship between production scale and production cost is analyzed and the maximum profit is always a trade-off of the production scale and production cost. Two examples are solved to demonstrate the effectiveness of the method.

Batch and fed-batch production of butyric acid by Clostridium butyricum ZJUCB

The production of butyric acid by Clostridium butyricum ZJUCB at various pH values was investigated. In order to study the effect of pH on cell growth, butyric acid biosynthesis and reducing sugar consumption, different cultivation pH values ranging from 6.0 to 7.5 were evaluated in 5-L bioreactor. In controlled pH batch fermentation, the optimum pH for cell growth and butyric acid production was 6.5 with a cell yield of 3.65 g/L and butyric acid yield of 12.25 g/L. Based on these results, this study then compared batch and fed-batch fermentation of butyric acid production at pH 6.5. Maximum value (16.74 g/L) of butyric acid concentration was obtained in fed-batch fermentation compared to 12.25 g/L in batch fermentation. It was concluded that culti- vation under fed-batch fermentation mode could enhance butyric acid production significantly (P<0.01) by C. butyricum ZJUCB.

Genetic algorithm for short-term scheduling of make-and-pack batch production process

This paper considers a scheduling problem in industrial make-and-pack batch production process. This process equips with sequence-dependent changeover time, multipurpose storage units with limited capacity, storage time, batch splitting, partial equipment connectivity and transfer time. The objective is to make a production plan to satisfy all constraints while meeting demand requirement of packed products from various product families. This problem is NP-hard and the problem size is exponentially large for a realistic-sized problem. Therefore,we propose a genetic algorithm to handle this problem. Solutions to the problems are represented by chromosomes of product family sequences. These sequences are decoded to assign the resource for producing packed products according to forward assignment strategy and resource selection rules. These techniques greatly reduce unnecessary search space and improve search speed. In addition, design of experiment is carefully utilized to determine appropriate parameter settings. Ant colony optimization and Tabu search are also implemented for comparison. At the end of each heuristics, local search is applied for the packed product sequence to improve makespan. In an experimental analysis, all heuristics show the capability to solve large instances within reasonable computational time. In all problem instances, genetic algorithm averagely outperforms ant colony optimization and Tabu search with slightly longer computational time.

MIIT issued 2017 the first batch of rare earth production control quotas

MIIT issued the notice about 2017 the first batch of rare earth production control quotas recently. According to the notice, overall rare earth mineral production quantity is controlled at 52,500 tons and smelting & separation products at 50,075 tons, while it was 52,500 tons and 45,000 tons, respectively, for 2016. According to the notice, rare earth control quotas shall be allocated to enterprises that meet the national rare earth industry policies and industrial development planning requirements.

Material Flow Optimization for Make-to-Stock Production System

Under the condition of the market competition becom in g more and more drastic, the demands of market take on some new features such as individuation, diversification, small batch, unstableness and quick delivery et c. The Make-to-Stock mode is usually adopted by many enterprises to improve th e balance and stableness of production process. In such enterprises, order batch , production batch and sales batch are the important factors, which affect the s atisfaction of clients, efficiency and benefit of the enterprise. It takes purch ase, production and sales into account respectively when optimizing product batc h in traditional way. However, it ignored the influences of relations between ea ch links of whole system. It is assumed that the consumption and market demand a re continuous process whereas the factual demands are batched when economic batc h is determined. So there exist some deviations between the economic batch deter mined by traditional way and that by integral optimization. Through the integral analysis of Logistics in the production system, we know that from materials are purchased, then manufactured, finally sold, the material changed in appearance and value, it still exist in different links of production system. The amount of materials occupied varies just in different status, from stock status to produc tion status, then to waiting-be-sold status, there is not any substantial chan ge in quantity until they are sold. So we must comprehensively analyze the relat ions among each link based on integral production system, to optimize the materi al batch and cut short production cycle in order to optimize the whole system. In this paper, the production system is taken as a global entity, and in which m aterials variation law and their relations of each link are analyzed; To optimiz e the whole materials flow, a new model of multi-product systems’ economic orde r batch, economic production batch and optimal sale lot multi-product syste ms’ is developed which based on the limit of capitals and stock area.

First Batch of Anhui Brandname Products Recommended by Anhui Provincial People's Government

Product NameMagang brand high quality carbon steel,common low carbon steel-made cold/hotrolling coil rod(without twisting control)Magang brand integral roll steel-madepassenger train wheel(φ915mm) forrailway useJingjing brand electrolytic

APPLICATION OF CONVECTION-DIFFUSION EQUATION TO THEANALYSES OF CONTAMINATION BETWEEN BATCHESIN MULTI-PRODUCTS PIPELINE TRANSPORT

Contamination between batches in multi-products pipeline transport is studied. The influences of convection and diffusion on the contamination are studied in detail. Diffusion equations, which are mainly controlled by convection, are developed under turbulent pipe flow. The diffusion equation is separated into a pure convection equation and a pure diffusion equation which are solved by characteristics method and finite difference method respectively to obtain numerical solutions. The results of numerical computation explain the forming and developing of contamination very well.

基于MOMA的可重入混合流水车间调度问题研究

针对可重入制造系统多具有多品种、大规模、混流生产等特点,构建带批处理机的可重入混合流水车间调度问题(reentrant hybrid flow shop scheduling problem with batch processors,BPRHFSP)模型,提出一种改进的多目标蜉蝣算法(multi-objective mayfly algorithm,MOMA)进行求解。提出了单件加工阶段和批处理阶段的解码规则;设计了基于Logistic混沌映射的反向学习初始化策略、改进的蜉蝣交配和变异策略,提高了算法初始解的质量和局部搜索能力;根据编码规则设计了基于变邻域下降搜索的蜉蝣运动策略,优化了种群方向。通过对不同规模大量测试算例的仿真实验,验证了MOMA相比传统算法求解BP-RHFSP更具有效性和优越性。所提出的模型能够反映生产的基础特征,达到减少最大完工时间、机器负载和碳排放的目的。

基于SMOTE-IKPCA-SeNet深度迁移学习的小批量生产质量预测研究

随着智能制造技术的发展和客户个性化需求的增加,多品种小批量生产方式逐渐成为制造业的主流。面向大批量生产、以统计过程控制为核心的质量管理方式并不适用于小批量生产。针对复杂生产过程存在参数多、非线性和交互作用的问题,提出利用深度迁移学习的方式将历史生产数据作为源域迁移至小样本目标产品数据进行质量预测。首先,通过合成少数类过采样技术(synthetic minority over-sampling technique,SMOTE)和改进的核主成分分析(improved kernel principal component analysis,IKPCA)算法筛选源域和目标域的可迁移特征,这不仅兼顾了特征重要性和可迁移性,还减少了“负迁移”,提高了模型泛化能力;然后,采用结合通道注意力机制的卷积神经网络SeNet构建基于深度迁移学习的质量预测模型。仿真结果表明,随着目标域样本的增加,所提方法的预测准确性明显优于广泛采用的支持向量机建模方法。同时,所提可迁移特征筛选方法显著提高了深度迁移学习的质量预测效果,为复杂的小批量生产过程质量保证提供了新方法。

基于市场需求与可再制造零件数量双重不确定性的再制造供应链批量策略研究

零件的回收再制造并再销对资源节约及减少环境负担具有重要意义。本文的再制造供应链由采购、运输、仓储、制造活动组成,在市场需求与可再制造零件数量双重不确定情景下,构建包含回收件的采购批量、再制造生产批量的决策模型,探索可能存在的最优批量和成本水平以及双重不确定性对生产批量和采购批量的影响。研究表明:分别实现年环境成本、年经济成本、年总成本最低时,均存在唯一最优生产批量和采购批量;市场需求与可再制造零件数量的双重不确定性水平会对最优解产生不同影响,而可再制造零件或新零件处于散存和制成品两种制备阶段的相关成本变化,将会决定企业再制造的不同批量策略。

基于动态成本卷积的复杂产品批产路径优化问题建模与求解研究

具有复杂工艺路线的装备产品,其生产系统设计的待决策变量之间通常相互耦合、相互影响,对产品最终的批产成本产生着影响,构成了较为复杂的组合优化问题。对此,提出一种基于动态成本卷积的复杂产品批产路径优化模型。该模型考虑了复杂产品工艺路线的各节点均可能存在多种生产设计选项,在各选项对应着不同生产投入和批量效率的情况下,以最终交付产品的批产成本最小化为目标输出生产系统设计和成本卷积路径。同时,建立了混合整数规划模型,并通过对模型中非线性成分实施可控精度线性化,将模型换化为可最优求解的线性模型。最后,针对某连续生产行业设计了小、中、大规模实验算例,验证了所提模型的可行性、合理性和求解效率。

批产卫星磁试验流程优化方法

为了在确保验证有效性的同时提高批产卫星的磁试验效率,提出了一种优化批产卫星磁试验流程的方法.建立了与卫星磁试验国际标准相适应的卫星剩磁矩和杂散磁矩模型,揭示了批产卫星各星之间磁矩的相关性,并以此优化和重构了卫星磁试验流程.将优化后的流程应用于3种批产的卫星或部件的磁试验中,结果表明:该方法在整星磁试验中的有效性为85.7%,可将整星磁试验周期压缩92.6%;在部件磁试验中的有效性为100%,可将部件磁试验周期压缩53.8%,为未来批产卫星磁试验的流程优化和标准编制提供了理论参考.

面向智能制革的自动配料系统应用分析

在皮革生产过程中需要使用多种皮革化工材料,精准控制其配比和用量,才能保证成品革的质量和不同批次皮革质量的稳定性。鉴于目前皮革化料的配料环节仍然依赖于传统的粗放式人工配料模式,文章结合制革行业的实际应用场景和功能需求,主要从过程控制层和现场设备层两个层级出发,分析了自动配料系统在制革生产中的应用方案,旨在为制革用自动配料系统的开发和研制提供切实可行的参考依据。

面向批产模式的小卫星构型与总装设计研究

中国小卫星的批量生产和发射大多集中在微纳卫星领域.目前,百千克级以上的小卫星设计制造只是对传统模式的改进优化,从构型及总装的设计上就决定了生产阶段无法实现大量批产.而国外具有代表性的批产型卫星构型并不适应中国单机的技术现状,载荷适应性较差.结合现阶段单机及总装的研制技术状态,提出了批产型小卫星构型和总装设计思路,分别从批发射、模块化、标准化、智能化及低成本5个方面阐述了适应批产模式的小卫星构型设计和总装设计的原则、方法,并在工程实践的基础上提出了发展思路,可供后续同类型小卫星构型与总装设计参考.

小卫星批生产装配一致性控制方法

针对微小卫星批量化生产制造过程,分析了批生产质量管理特点和每组星之间的装配质量一致性影响因素.从工程实践的角度,提出了一种适于卫星系统批生产的装配一致性控制方法.该方法通过建立规范化体系、质量重心前移、提高工艺方法稳定性、固化技术状态、装配流程标准化、实施质量确认制、实行"去型号化"物资管理、引进自动化设备及统筹人员配置等质量控制手段,保证了小卫星批量化生产过程中装配工作的质量一致性.将其应用到卫星批产总装工作后,实践结果表明,卫星装配一致性控制方法能够有效提升批量化生产制造工作质量,大幅提升批次卫星产品的质量一致性,是卫星批量生产研制质量的有力保证.

基于TOPSIS模型对小批量物料的生产安排研究

在企业小批量物料的生产安排中,逐渐出现因事先无法知道物料的实际需求导致生产安排有误等现象。针对这一情况,为帮助企业进行更合理的物料生产,通过分析某工厂2019-2022年的历史数据,基于Topsis算法建立模型,选出需要重点关注的物料,使用时序预测建立关于物料预测模型,通过最小二乘法验证模型的准确性。物料预测为企业解决了小批量物料的生产安排问题,该模型具有一定的推广价值。

小卫星综合电子系统设计

针对小卫星综合电子系统非标准化、互联性能较低、接口关系复杂多变、不易进行软件开发和不利于推广应用的问题,研究了目前国内外小卫星综合电子系统及概念,详细比较了各种综合电子架构的优缺点.并结合小卫星空间使用环境,提出了一种基于3U尺寸OpenVPX标准架构的小卫星综合电子系统方案.该方案有效克服了传统小卫星综合电子系统非标准化和互连性能较差等缺点,可以满足小卫星综合电子系统标准化、软件化、低成本及易于批生产的需求,具有良好的应用和推广价值.

基于改进深度强化学习的注采调控模型研究

提出一种基于改进深度强化学习的注采调控模型。首先,建立以最大化经济效益为目标函数的注采调控强化环境。其次,针对模型参数量大、网络内部协变量位移多等问题,提出一种改进双重深度Q网络的深度强化学习方法,应用批量归一化技术逐层归一化模型的输入数据,增强模型的泛化能力;再通过剪枝模块压缩模型体积,加速网络的训练过程,引入动态ε策略思想提高模型的鲁棒性和稳定性。最后,将所提模型同其他模型进行对比,实验结果表明,所提模型能获得更高、更稳定的平均累计奖励和更快的收敛速度和运行速度。

成品油管网批次在线跟踪技术

成品油管网批次界面跟踪系统可基于实时运行数据实现管内批次位置实时跟踪、到站时间预测等,进而辅助站场精确掌握批次分输及混油下载时间,对于防止油品质量事故、提高站场管理效率具有重要意义。但现有批次跟踪方法基于体积推移算法,易受管内水力热力变化干扰,模型误差较大。提取管网实时流量密度数据,根据密度、阀门状态变化动态识别到站批次,并启动下游管道批次计算;基于油品体积流入、流出守恒定律构建批次界面推移计算模型,并考虑了由于混油、温度压力影响导致的油品体积动态变化,修正批次跟踪模型以补偿体积动态变化造成的批次位置偏差,最终开发了成品油管网批次界面实时跟踪系统。利用某成品油管网历史运行数据进行测试,结果表明,不考虑偏差修正的批次跟踪结果与实际到站时间的平均误差为10.87 min,而考虑温度压力修正模型的平均误差仅为3.8 min,误差降低了65%。该系统可满足管道运行管理的生产需求,有助于提高批次界面跟踪的工作效率,防止油品质量事故。

应用Aspen Batch对年产25吨的美罗培南原料药生产工艺模拟设计

美罗培南是一种新型碳青霉烯类抗生素,具有广阔的市场前景,其生产过程为间歇生产。文章利用Aspen Batch Process Developer 7.2对年产25吨的美罗培南原料药生产工艺流程进行模拟,得到生产过程中的物料衡算结果误差为0.8%,生产时间甘德图表明,生产周期为48小时,并得到该生产过程的公用工程消耗量,对实际的工艺设计具有一定的参考价值。