生產(chǎn)計劃調(diào)度崗位職責(zé)-wenkub
2024-10-25 01 本頁面
【正文】 式運行的：仿真系統(tǒng)僅以啟動時確定下來的靜態(tài)數(shù)據(jù)集作為輸入，仿真運行過程中不能動態(tài)地向系統(tǒng)注入數(shù)據(jù)。各單元模塊的功能如下1．控制單元：控制仿真模型運行，抽取模型的狀態(tài)特征數(shù)據(jù)與注入數(shù)據(jù)對比分析，據(jù)此調(diào)整模型和分辨率；控制實際生產(chǎn)線運行，根據(jù)仿真結(jié)果數(shù)據(jù)反饋生產(chǎn)線調(diào)度優(yōu)化方案；控制數(shù)據(jù)采集，根據(jù)仿真結(jié)果數(shù)據(jù)產(chǎn)生數(shù)據(jù)采集策略，反饋數(shù)據(jù)采集模塊執(zhí)行。3．人機(jī)接口單元：顯示仿真結(jié)果，支持用戶對仿真策略進(jìn)行調(diào)整和控制，支持用戶根據(jù)仿真結(jié)果對生產(chǎn)線運行進(jìn)行控制，支持用戶根據(jù)仿真結(jié)果對數(shù)據(jù)采集策略進(jìn)行選擇控制。其中，調(diào)度模塊包括相互協(xié)同的兩個層次：上層模塊是生產(chǎn)計劃調(diào)度器，采用全局優(yōu)化的方法，利用群體智能蟻群算法進(jìn)行尋優(yōu)，可產(chǎn)生一個靜態(tài)的調(diào)度計劃，其尋優(yōu)時間長的問題可以通過多Agent建模的分布式計算能力得到解決。整個過程是一個動態(tài)反饋過程。(二)動態(tài)數(shù)據(jù)驅(qū)動仿真控制單元由中心推理機(jī)和輔助功能Agent組成，共同構(gòu)成一個調(diào)度決策支持系統(tǒng)。短期優(yōu)化目標(biāo)包括：最大化生產(chǎn)量、最大化WIP移動步數(shù)、最小訂單交貨延遲率、降低加工周期、降低加工周期方差、降低WIP水平等。同時，它也在不斷地動態(tài)調(diào)整，保持與生產(chǎn)線當(dāng)前的實際生產(chǎn)情況一致，完成這個任務(wù)需要中心推理機(jī)的協(xié)同，如基于案例庫的推理等。與傳統(tǒng)仿真不同，動態(tài)數(shù)據(jù)驅(qū)動仿真是一種與生產(chǎn)線生產(chǎn)過程并行的仿真方法。同時，將調(diào)度因素反饋給數(shù)據(jù)采集策略Agent，由后者按一定的策略完成下一步的生產(chǎn)線數(shù)據(jù)采集工作。使用戶可以實時地控制仿真的整個過程，并利用仿真結(jié)果指導(dǎo)生產(chǎn)和數(shù)據(jù)采集過程。其中，仿真模型根據(jù)動態(tài)注入的生產(chǎn)數(shù)據(jù)完成自適應(yīng)調(diào)整是整個DDDAS的核心，本文采用分層優(yōu)化的思想生成調(diào)度方案，同時達(dá)到全局和局部的優(yōu)化目標(biāo)。一方面，通過真實系統(tǒng)運行過程中產(chǎn)生的參數(shù)同步地對仿真系統(tǒng)進(jìn)行調(diào)整，可以大大提高仿真的準(zhǔn)確性、時效性、智能化；另一方面，通過仿真提供的數(shù)據(jù)同步地為真實系統(tǒng)地運行提供決策支持，這些將大大地擴(kuò)展仿真系統(tǒng)的應(yīng)用能力。因此近年來傾向于利用人工智能的原理和技術(shù)進(jìn)行搜索，尋找復(fù)雜問題的較優(yōu)解，特別是以效仿生物處理模式以獲得智能信息處理功能的遺傳算法研究最為深入?！禔 Hybrid Genetic Algorithm for Job Shop Scheduling Problem to Minimize Makespan》 作者：Lin Liu, Yugeng XiIn this paper, we present a hybrid genetic algorithm for the job shop scheduling problem to mimize to improve GA performance is a critical issue when using a GA to solve optimization general way focuses on tuning its parameters such as population size, crossover rate and mutation , if all parameters have attained the useful bounds, the expected improvement is often not worth the efforts of finding even better potential improvements can be only explored by modifying the size of search set of active schedules is usually large and includes a lot of schedules with relatively large idle times on machines, and thus with relatively large idle times on machines, and thus with poor performance in terms of proposed algorithm used the idea of hybrid scheduler to reduce the search space as well as the putational search space can be reduced or increased by controlling the upper bound of idle times allowed on the parameters of the hyubrid scheduler are unlikely to be determined appropriately in advance, we search better values of them in the hybrid GA to Gas in literatures, a chromosome includes not only genes representing the relative priorities of all operations but also genes representing the parameters to determine the upper bound of idle times permitted on a given machine before scheduling an random keys representation is used to encode a element of the chromosome is a real number of [0,1].During the schedule generation phase, the SPV rule is used to convert a real number vector into a job repetition on the hybrid scheduler, a chromosome is decoded into a feasible , a local search is executed in the neighborhood determined by the critical active chain to improve the performance of the schedule generated in the schedule generation In the 2section, we present the formulation of job shop scheduling problem to minimize the 3 section, we describe the proposed hybrid genetic algorithm
醫(yī)療健康相關(guān)推薦
鄂ICP備17016276號-1