
本书着重介绍了武器装备备件的基本概念、需求影响因素、需求预测的程序及武器装备备件保障决策时常用的需求预测方法,主要包括基于相似分析法、模糊综合评判法、灰色评估法的备件品种确定方法,基于仿真、时间序列、灰色模型、支持向量机、故障分析、组合模型的备件需求预测方法。
在人类社会和自然科学问题中,普遍存在预测预报问题,大至宇宙的未 来演化、人类历史发展的趋势,小至销售预测、股市预测、天气预报等。预 测是做决策、规划之前必不可少的重要环节,是科学决策、规划的重要前提。 武器装备是由许多部(组)件组成的,在使用期需要对这些部(组)件 进行修理和更换。为了缩短武器装备的修理间歇时间,增加可正常工作的时 间,对那些易磨损、易故障或因形状复杂、要求高、加工周期长且可能有故 障的部(组)件,储备合理数量的备件,以备随时更换,是提高武器装备可 用度和降低寿命周期费用,保证能按时发挥战斗效能的有效措施。备件是武 器装备维修保障资源中不可缺少的组成部分,是部队成系统、成建制形成作 战能力和保障能力的物质基础,备件供应直接影响装备的战备完好性和寿命 周期费用。现代高技术战争的实践表明,备件保障已成为战争中谋求战场主 动权的有效手段,备件保障能力直接影响武器装备的使用及战备完好性,关 系到军事目标的实现和部队“两成两力”建设。目前,我军建设正处于难得 的战略机遇期,新型武器装备不断入列服役,随着我军武器装备结构原理的 复杂和保障任务的增加,对备件需求保障工作提出了更多、更高的要求。 在备件保障问题中,备件需求预测与优化配置是推进“精确保障”的重 要手段,是实现“保障有力”目标的重要途径,也是备件保障的重点、难点 问题。如果不能科学、合理地确定备件的品种和数量,就将会造成装备保障 经费的浪费,对装备的战备完好性产生非常大的影响,甚至会影响装备的保 障能力,进而影响部队的战斗力,影响部队能否顺利地完成各项任务,关系 到作战的胜负。因此,武器装备备件需求预测与配置优化技术,是实现武器 装备备件供应保障由“规模型”向“精益型”转型的重要手段,是保证备件 武器装备备件需求预测理论与方法 ·IV· 供应规划“优生”和“优育”的有效途径,也是装备综合保障方向研究的重 点和难点内容之一。 现有的预测方法分别从不同的角度分析与解决问题,但运用于武器装备 备件保障时,对问题分析和方法选择缺乏系统的研究,往往预测结果精度不 高,方法可扩展性不强,实际效果达不到预期要求。笔者曾在海军一线部队 从事装备保障工作多年,后来调入院校从事武器装备综合保障方面的教学工 作,同时参与了总部和海军下达的多项备件消耗规律研究、备件消耗标准的 制定、备件需求辅助决策支持系统开发等多项科研项目,对武器装备备件供 应保障工作有较为全面和系统的了解。在装备保障和教学科研工作过程中, 深深体会到备件需求预测与配置优化工作必须做到具体问题具体分析、理论 与实践相结合。为有助于武器装备备件保障决策与管理工作的顺利实施,笔 者参阅有关预测书刊和资料,吸取各行业预测实例的经验和教训,在个人历 年的研究攻关、工作实践和思考,以及在学习工作中陆续发表的备件需求预 测相关的论文,形成了本书的主要内容。 本书在系统分析集成的研究框架下,较为系统地、完整地从理论和应用 方面介绍备件需求预测的理论与方法。第1 章为概述,主要介绍武器装备备 件的基本概念,备件的分类,备件需求影响因素分析和备件需求特点分析等 内容。第2 章为武器装备备件需求预测与配置优化基础,主要介绍武器装备 备件的消耗模式,备件需求预测基本原则、一般程序和方法,以及备件配置 优化指标与方法等内容。第3 章为武器装备备件品种确定方法,主要介绍备 件确定常用的方法及其特点,以及相似分析法、模糊综合评判法和灰色评估 法及其应用等内容。第4 章为基于寿命分布和仿真的武器装备备件需求预测 方法,主要介绍不同寿命分布类型的备件需求预测模型,以及蒙特卡罗仿真、 考虑设备停机的备件需求预测模型、Logistic 回归和马尔可夫过程的备件需 求预测模型及其应用等内容。第5 章为基于时间序列和回归分析的武器装备 备件需求预测方法,主要介绍时间序列法、平稳随机时间序列法和回归分析 法及其应用等内容。第6 章为基于灰色模型的武器装备备件需求预测方法, 主要介绍灰色预测的建模过程,以及含有空缺值序列的GM(1,1)模型、加入 影响因子的灰色模型、灰色-马尔可夫模型的建模过程及其应用等内容。第7 章为基于人工智能的武器装备备件需求预测方法,主要介绍BP 神经网络、 支持向量机回归和组合相关向量机预测方法及其应用。第8 章为基于故障分 析的武器装备备件需求预测方法,主要介绍基于故障预测的备件需求预测程 序,基于信息融合和IMUGM(1,1)故障预测、基于信息融合和IMUGM(1,m,w) 前 言 ·V· 故障预测以及基于故障规律的备件需求预测方法及其应用等内容。第9 章为 基于组合模型的武器装备备件需求预测方法,主要介绍组合模型的建模过 程,基于回归分析和灰色模型、小波变换和GM(1,1)-ARMA 模型、Theil 不 等系数和指数平滑法、GM(1,1)和支持向量机的备件需求预测方法及其应用 等内容。书中所列的数据和曲线,一般基于海军有关单位的保障数据,某些 分析结论可能不适合于所有的武器装备,但是分析方法具有通用性。 本书第1 和3 章由徐廷学编写,第2、6 和8 章由赵建忠编写,第4 和 5 章由叶文编写,第9 章由丛林虎编写,全书由赵建忠等统稿。 在本书编写过程中,得到了许多同事、专家的鼓励、支持和帮助,在此 一并表示感谢。感谢路明、赵建印、邓建球、张旭、刘勇做出大量的审稿工 作,并提出宝贵意见。感谢海军航空大学徐廷学教授及所带领的金天球、王 相飞等硕士、博士们的帮助,没有他们协助建模计算和受益匪浅的讨论,本 书是难以完成的。此外,对编写中参阅的有关书刊、文献的作者致以敬意和 感谢。特别感谢妻子魏炜女士对我工作的理解和支持。 由于作者水平有限,书中难免有错误和不妥之处,真诚期待专家和广大 读者批评和指正。 编著者 2017 年10 月
第1 章 概述 ·················································································· 1 1.1 基本概念 ··········································································· 1 1.2 备件分类 ··········································································· 3 1.2.1 常见的备件分类方法 ···················································· 3 1.2.2 低消耗备件 ································································ 4 1.2.3 基于模糊隶属度的低消耗备件定义与界定 ························· 4 1.3 备件需求影响因素分析 ······················································· 10 1.3.1 备件需求内在影响因素分析 ········································· 11 1.3.2 备件需求外在影响因素分析 ········································· 12 1.3.3 影响因素的量化及规范化处理 ······································ 16 1.3.4 基于粗糙集属性约简的主要影响因素确定 ······················· 20 1.4 备件需求层次性分析 ·························································· 22 1.4.1 备件层次需求分析 ····················································· 22 1.4.2 装备系统层次需求分析 ··············································· 23 1.4.3 维修保障组织需求分析 ··············································· 23 1.5 备件需求特点分析 ····························································· 23 1.5.1 备件需求特征 ··························································· 23 1.5.2 常消耗备件需求特点 ·················································· 24 1.5.3 低消耗备件需求特点 ·················································· 25 第2 章 武器装备备件需求预测与配置优化基础 ··································· 26 2.1 备件消耗模式分析 ····························································· 26 2.2 备件需求预测的基本原则 ···················································· 27 2.3 备件需求预测的一般程序 ···················································· 29 2.4 备件需求预测的方法 ·························································· 31 2.4.1 定性备件需求预测方法 ··············································· 31 2.4.2 定量备件需求预测方法 ··············································· 32 2.4.3 模型复杂性与模型选择 ··············································· 34 第3 章 武器装备备件品种确定方法 ·················································· 36 3.1 概述 ··············································································· 36 3.2 基于相似分析法的武器装备备件品种确定方法 ·························· 42 3.2.1 相似分析法简介 ························································ 42 3.2.2 实例分析 ································································· 44 3.3 基于模糊层次综合评判的备件品种确定方法 ····························· 46 3.3.1 模糊层次综合评判法简介 ············································ 46 3.3.2 模型建立 ································································· 47 3.3.3 示例分析 ································································· 50 3.4 基于D-S 理论和模糊综合评判的备件品种确定方法 ··················· 52 3.4.1 模型建立 ································································· 52 3.4.2 实现步骤 ································································· 54 3.4.3 示例分析 ································································· 57 3.5 基于灰色评估法的武器装备备件品种确定方法 ·························· 61 3.5.1 评价指标体系的建立 ·················································· 61 3.5.2 评价模型的建立 ························································ 63 3.5.3 示例分析 ································································· 65 第4 章 基于寿命分布和仿真的武器装备备件需求预测方法 ···················· 69 4.1 基于寿命分布的武器装备备件需求预测方法 ····························· 69 4.1.1 模型的假设 ······························································ 70 4.1.2 模型的建立 ······························································ 70 4.1.3 示例分析 ································································· 74 4.2 基于蒙特卡罗仿真的备件需求预测方法 ·································· 77 4.2.1 模型的假设 ······························································ 78 4.2.2 参数确定 ································································· 78 4.2.3 蒙特卡罗方法 ··························································· 79 4.2.4 仿真模型的建立 ························································ 79 4.2.5 实例分析 ································································· 81 4.3 基于马尔可夫与蒙特卡罗仿真的备件需求预测方法 ···················· 82 4.3.1 马尔可夫预测模型 ····················································· 82 4.3.2 备件消耗的马尔可夫性分析 ········································· 84 4.3.3 基于马尔可夫与蒙特卡罗仿真预测模型的建立 ················· 85 4.3.4 实例分析 ································································· 88 4.4 基于Logistic 回归和马尔可夫过程的备件需求 预测仿真模型 ······· 89 4.4.1 模型建立 ································································· 90 4.4.2 需求发生概率预测 ····················································· 91 4.4.3 需求数量预测 ··························································· 93 4.4.4 误差分析 ································································· 96 4.4.5 示例分析 ································································· 97 4.5 考虑设备停机的备件需求预测仿真模型 ································· 100 4.5.1 基于最大熵原理与概率加权矩的备件寿命分布确定 ·········· 101 4.5.2 基于保障度的间断工作备件需求预测模型建立 ················ 106 4.5.3 仿真思路及流程规划 ················································· 109 4.5.4 示例分析 ································································ 111 第5 章 基于时间序列和回归分析的武器装备备件需求预测方法 ············· 115 5.1 时间序列法概述 ······························································· 115 5.2 基于时间序列法的备件需求预测 ·········································· 116 5.2.1 算数平均预测法 ······················································· 116 5.2.2 移动平均预测法 ······················································· 117 5.2.3 指数平滑预测法 ······················································· 118 5.2.4 示例分析 ································································ 122 5.3 基于回归分析的备件需求预测方法 ······································· 126 5.3.1 线性回归概述 ·························································· 126 5.3.2 一元线性回归模型 ···················································· 127 5.3.3 多元线性回归模型 ···················································· 129 5.3.4 示例分析 ································································ 131 第6 章 基于灰色模型的武器装备备件需求预测方法 ···························· 133 6.1 灰色预测方法概述 ···························································· 133 6.1.1 灰色系统简介 ·························································· 133 6.1.2 灰色系统理论基础 ···················································· 134 6.1.3 灰色预测方法 ·························································· 136 6.2 基于一般序列GM(1,1)模型的备件需求预测方法 ······················ 137 6.2.1 模型建立 ································································ 137 6.2.2 示例分析 ································································ 141 6.3 基于含有空缺值序列GM(1,1)模型的备件需求预测方法 ········· 144 6.3.1 模型建立 ································································ 144 6.3.2 示例分析 ································································ 145 6.4 基于加入影响因子改进灰色模型的备件需求预测方法 ················ 146 6.4.1 改进灰色预测模型的建立 ··········································· 146 6.4.2 示例分析 ································································ 147 6.5 基于灰色-马尔可夫模型的备件需求预测方法 ··························· 155 6.5.1 经典灰色-马尔可夫预测模型 ······································· 155 6.5.2 改进的灰色-马尔可夫模型 ·········································· 158 6.5.3 示例分析 ································································ 162 第7 章 基于人工智能的武器装备备件需求预测方法 ···························· 166 7.1 基于BP 神经网络的备件需求预测方法 ·································· 166 7.1.1 BP 神经网络基本原理 ··············································· 167 7.1.2 预测模型的建立 ······················································· 169 7.1.3 示例分析 ································································ 170 7.2 基于支持向量机(SVM)的备件需求预测方法 ························ 175 7.2.1 支持向量机概述 ······················································· 176 7.2.2 支持向量机预测原理 ················································· 177 7.2.3 支持向量机预测模型 ················································· 185 7.2.4 示例分析 ································································ 188 7.3 基于复杂时间序列的组合相关向量机备件需求预测方法 ············· 192 7.3.1 相空间重构 ····························································· 192 7.3.2 小波变换基本理论 ···················································· 193 7.3.3 小波函数的选取 ······················································· 198 7.3.4 基于组合相关向量机的预测原理 ································ 199 7.3.5 示例分析 ································································ 206 第8 章 基于故障分析的武器装备备件需求预测方法 ···························· 213 8.1 故障预测的基本原理分析 ··················································· 213 8.2 基于故障预测的备件需求预测系统构建 ································· 215 8.2.1 工作流程 ································································ 216 8.2.2 数据预处理 ····························································· 216 8.2.3 故障预测 ································································ 219 8.2.4 备件需求确定 ·························································· 220 8.3 基于信息融合和IMUGM(1,1)故障预测的备件需求预测方法 ········ 220 8.3.1 问题描述 ································································ 221 8.3.2 IMUGM(1,1)模型的建立 ············································ 221 8.3.3 基于信息融合和IMUGM(1,1)故障预测的备件需求 预测模型构建 ·························································· 224 8.3.4 示例分析 ································································ 225 8.4 基于信息融合和IMUGM(1,m,w)故障预测的备件需求预测建模 ···· 230 8.4.1 问题描述 ································································ 231 8.4.2 IMUGM(1,m,w)模型的建立 ········································· 231 8.4.3 基于信息融合和IMUGM(1,m,w)故障预测的备件需求 预测模型构建 ·························································· 234 8.4.4 示例分析 ································································ 234 8.5 基于故障规律的备件需求预测方法 ······································· 239 8.5.1 备件需求量计算公式 ················································· 239 8.5.2 示例分析 ································································ 242 第9 章 基于组合模型的武器装备备件需求预测方法 ···························· 244 9.1 组合预测方法概述 ···························································· 244 9.1.1 一般组合预测模型 ···················································· 245 9.1.2 最优加权系数的确定 ················································· 246 9.1.3 模型的检验 ····························································· 247 9.2 基于回归分析和灰色模型的备件需求预测方法 ························· 247 9.2.1 单个预测模型建立 ···················································· 247 9.2.2 最优组合预测模型 ···················································· 248 9.2.3 示例分析 ································································ 248 9.3 基于小波变换的备件需求预测方法 ······································· 251 9.3.1 小波变换理论 ·························································· 251 9.3.2 GM(1,1)-ARMA 模型的改进 ········································ 254 9.3.3 WGM -ARMA 模型的构建 ·········································· 264 9.3.4 示例分析 ································································ 265 9.4 基于改进Theil 不等系数的备件需求预测建模 ·························· 270 9.4.1 单项预测模型的建立 ················································· 271 9.4.2 基于Theil 不等系数的IOWHA 算子的组合预测模型构建 ··· 275 9.4.3 基于Theil 不等系数的IOWHA 算子的组合预测模型改进 ··· 278 9.4.4 示例分析 ································································ 282