AiDOPWarehouse/doc/_bid_gen_figs.py

# -*- coding: utf-8 -*-
"""生成投标技术方案的总体架构类配图（8 张）。"""

import sys

sys.path.insert(0, __file__.rsplit("\\", 1)[0] if "\\" in __file__ else ".")
sys.stdout.reconfigure(encoding="utf-8", errors="replace")

from _bid_fig_kit import C, Fig, row


# ==================================================================== 图1
def fig_otd_blueprint():
    f = Fig("Ai-DOP 制造业数据智能运营平台 OTD 端到端总体业务蓝图",
            w=13.6, h=10.4,
            sub="以销售订单为主线，贯通 S0～S9 全流程；下层数据中台完成「采—存—治—用」，上层实现指标测量、智能诊断与改善闭环")

    # --- S0 运营建模（基座）
    f.panel(3, 90, 94, 9.0, "S0　运营建模（主数据与流程标准化）", tone="gray")
    s0 = row(6, 87.4, 20.5, 4.6, 4, 2.6,
             ["客户/供应商主数据", "物料主数据\n（18位统一编码）",
              "工艺主数据\n（标准工序库·工艺路线）", "订单模板与流程SOP"], f,
             tone="gray", fs=8.4)

    # --- OTD 业务主线
    f.panel(3, 79, 94, 22.0, "OTD 业务主线（订单接收 → 产品交付）", tone="blue")
    r1 = row(6, 75.6, 13.6, 5.0, 6, 2.2,
             ["销售订单接收\nS1", "订单评审\n交期评估 S1", "交付承诺与\n全流程跟踪 S1",
              "主生产计划\n产能平衡 S2", "作业计划下达\nS2", "物料需求计划\nMRP S3"], f,
             tone="blue", fs=8.4)
    r2 = row(6, 66.4, 13.6, 5.0, 6, 2.2,
             ["采购申请\n采购订单 S3", "供应商交货\n执行 S4", "来料检验IQC\n入库 S5",
              "生产执行\n过程检验 S6", "成品入库\nFQC 检验 S7", "销售发货\n交付客户 S1/S7"], f,
             tone="green", fs=8.4)
    f.chain(r1)
    f.chain(r2)
    wrap_y = 68.5
    f.arrow(r1[5]["b"], (r1[5]["c"][0], wrap_y), style="-")
    f.arrow((r1[5]["c"][0], wrap_y), (r2[0]["c"][0], wrap_y), style="-")
    f.arrow((r2[0]["c"][0], wrap_y), r2[0]["t"])
    f.note(6, 58.8, "变更管理：节点进度滞后 10% / 30% 或确认无法按期交付 → 自动预警 → 交付变更审批 → 与客户协商新交期并同步订单状态",
           fs=8.0, color="#404040")

    # --- S8 异常监控
    f.panel(3, 55.5, 94, 9.5, "S8　全流程异常监控", tone="orange")
    s8 = row(6, 52.4, 20.5, 5.0, 4, 2.6,
             ["异常识别与提报\n（系统自动 / 人工）", "分级响应\n一般·重要·紧急（≤1h）",
              "处理闭环\n措施与结果录入", "AI 智能体集群\n7×24 感知与预警驱动"], f,
             tone="orange", fs=8.4)
    f.chain(s8)

    # --- S9 运营绩效
    f.panel(3, 44.5, 94, 11.0, "S9　运营绩效指标测量与智能运营", tone="purple")
    s9 = row(6, 41.0, 16.2, 5.4, 5, 2.5,
             ["KPI 定义与日批计算\nOTD·OEE·交付率·合格率·周转",
              "九宫格智慧运营看板\n部门看板 · L1–L4 下钻",
              "运营问题诊断\n7 维根因溯源",
              "改善闭环\n建档·派单·跟踪·复盘",
              "ChatBI 智能报表\n自然语言问数"], f, tone="purple", fs=8.2)
    f.chain(s9[:4])
    f.arrow(s9[1]["b"], (s9[4]["c"][0], s9[1]["b"][1] - 1.6), style="-", dashed=True)
    f.arrow((s9[4]["c"][0], s9[1]["b"][1] - 1.6), s9[4]["b"], dashed=True)

    # --- 数据中台
    f.panel(3, 32.5, 94, 11.5, "数据中台（采 · 存 · 治 · 用）", tone="teal")
    dt = row(6, 28.8, 16.2, 5.0, 5, 2.5,
             ["STG 贴源层\n原样落地", "STD 标准层\n清洗·编码统一", "DWD 明细宽表\n主题化关联",
              "DWS / KPI 指标层\n日批聚合", "应用消费\n看板·诊断·ChatBI"], f,
             tone="teal", fs=8.2)
    f.chain(dt)
    f.note(6, 22.2, "数据治理：字段标准化、空值过滤、异常值标记、编码统一；数据质量准确率 ≥ 98%，主数据与单据状态同步延迟 ≤ 5 分钟",
           fs=8.0, color="#404040")

    # --- 系统集成
    f.panel(3, 20.0, 94, 17.5,
            "系统集成层（多源异构接入采集 ↑ 入站　·　业务结果 ↓ 出站回写）", tone="red")
    src = row(6, 16.6, 13.6, 4.4, 6, 2.2,
              ["ERP", "MES", "QMS / IQC", "SRM / 供应商门户", "WMS / 仓储", "CRM · TMS · OA"], f,
              tone="red", fs=8.6)
    f.note(6, 10.0,
           "接入方式：数据库直连 · HTTP-API · WebService · FTP 文件 · MQ 消息 · 物联网网关（6 种）　｜　"
           "同步模式：定时同步 + 实时同步",
           fs=8.2, color="#404040")
    f.note(6, 6.6,
           "出站回写：统一 Outbox 事务表 → API 推送 → 回执核对 → 失败重试 / 断点续传，不重复入库、不丢数据",
           fs=8.2, color="#404040")
    f.note(6, 3.2,
           "移动端应用：生产执行 · 仓储物流 · 质量管理 · 协同决策（订单进度、供应商协同、决策看板与下钻）",
           fs=8.2, color="#404040")

    return f.save("fig_otd_blueprint")


# ==================================================================== 图2
def fig_application():
    f = Fig("应用架构", w=13.0, h=9.2,
            sub="五层分域架构：接入展现 — 业务应用 — 领域服务 — 数据服务 — 集成适配，横向贯穿安全与运维两大支撑体系")

    f.panel(3, 88, 78, 11.0, "① 接入与展现层", tone="blue")
    row(6, 84.6, 16.6, 5.2, 4, 2.6,
        ["PC 门户\nVue3 + Element Plus", "管理层大屏\n九宫格 / 部门看板",
         "移动端应用\n生产·仓储·质量·决策", "ChatBI 对话入口\n浮窗 / 独立页"], f,
        tone="blue", fs=8.4)

    f.panel(3, 75.5, 78, 15.0, "② 业务应用层（S0～S9）", tone="green")
    row(6, 72.1, 10.6, 4.8, 6, 1.9,
        ["S0 运营建模", "S1 产销协同", "S2 制造协同", "S3 供应协同",
         "S4 采购执行", "S5 物料仓储"], f, tone="green", fs=8.2)
    row(6, 65.7, 10.6, 4.8, 6, 1.9,
        ["S6 生产执行", "S7 成品仓储", "S8 异常监控", "S9 指标看板",
         "运营诊断", "改善闭环"], f, tone="green", fs=8.2)

    f.panel(3, 59.0, 78, 14.0, "③ 领域服务层（可复用业务能力）", tone="purple")
    row(6, 55.6, 13.6, 5.0, 5, 2.2,
        ["订单与交付\n评审·承诺·跟踪", "计划与排程\nMPS·MRP·产能",
         "采购与供应\n申请·订单·交货", "质量与追溯\nIQC·IPQC·FQC",
         "指标与诊断\nKPI·根因·改善"], f, tone="purple", fs=8.2)
    row(6, 49.1, 13.6, 4.0, 5, 2.2,
        ["工作流与审批", "消息与预警", "报表与导出", "任务调度", "多租户与权限"], f,
        tone="purple", fs=8.2)

    f.panel(3, 43.5, 78, 12.5, "④ 数据服务层（数据中台）", tone="teal")
    row(6, 40.1, 13.6, 5.0, 5, 2.2,
        ["贴源 STG", "标准 STD", "明细宽表 DWD", "指标 DWS / KPI", "主数据 MDM"], f,
        tone="teal", fs=8.4)
    f.note(6, 33.4, "数据治理：标准化 · 去重 · 空值过滤 · 异常标记 · 编码统一 · 血缘与质量稽核",
           fs=8.2, color="#404040")

    f.panel(3, 29.5, 78, 14.0, "⑤ 集成适配层", tone="red")
    row(6, 26.1, 10.6, 4.6, 6, 1.9,
        ["DB 直连", "HTTP-API", "WebService", "FTP 文件", "MQ 消息", "IoT 网关"], f,
        tone="red", fs=8.4)
    row(6, 20.2, 16.6, 4.6, 4, 2.6,
        ["可视化通道配置", "定时 / 实时调度", "同步监控与告警", "Outbox 出站回写"], f,
        tone="red", fs=8.4)

    f.panel(3, 14.0, 78, 13.0, "外部业务系统", tone="gray")
    row(6, 10.6, 10.6, 4.6, 6, 1.9,
        ["ERP", "MES", "QMS", "SRM", "WMS", "CRM / TMS / OA"], f,
        tone="gray", fs=8.6)
    f.note(6, 3.6, "工业互联网平台：OAuth2.0 单点登录（SSO），员工一次登录即可访问 Ai-DOP",
           fs=8.4, color="#404040")

    # 右侧纵向支撑体系
    f.box(83, 88, 6.5, 87.0, "安\n全\n体\n系\n\n数\n据\n·\n应\n用\n·\n网\n络\n·\n审\n计",
          tone="blue_d", fs=8.8)
    f.box(90.5, 88, 6.5, 87.0, "运\n维\n体\n系\n\n监\n控\n·\n日\n志\n·\n备\n份\n·\n扩\n容",
          tone="gold", fs=8.8)
    f.crop(0)
    return f.save("fig_application")


# ==================================================================== 图3
def fig_data():
    f = Fig("数据架构：数据中台「采—存—治—用」", w=13.4, h=7.4,
            sub="统一数据底座支撑 KPI 测量、智能诊断、BI 可视化与 AI 应用；主数据「一处修改、全系统同步」")

    f.panel(3, 90, 17, 56, "采（多源采集）", tone="red")
    row(5, 85.5, 13, 6.0, 1, 0, ["ERP\n主数据·订单·库存"], f, tone="red", fs=8.4)
    row(5, 78.0, 13, 6.0, 1, 0, ["MES\n工单·工序·设备"], f, tone="red", fs=8.4)
    row(5, 70.5, 13, 6.0, 1, 0, ["QMS\nIQC·IPQC·FQC"], f, tone="red", fs=8.4)
    row(5, 63.0, 13, 6.0, 1, 0, ["SRM / WMS\n交货·出入库"], f, tone="red", fs=8.4)
    row(5, 55.5, 13, 6.0, 1, 0, ["CRM·TMS·OA·IoT"], f, tone="red", fs=8.4)
    f.note(5, 46.0, "直连 / API /\nWebService /\nFTP / MQ /\nIoT 网关", fs=8.2)

    f.panel(22, 90, 34, 56, "存 + 治（分层建模与治理）", tone="teal")
    a = f.box(24.5, 85.5, 29, 6.0, "STG 贴源层　原样落地 · 不做业务改写", tone="teal", fs=8.6)
    b = f.box(24.5, 76.5, 29, 6.0, "STD 标准层　字段标准化 · 编码统一 · 空值过滤 · 异常标记",
              tone="teal", fs=8.0)
    c = f.box(24.5, 67.5, 29, 6.0, "DWD 明细宽表　按订单主线主题化关联", tone="teal", fs=8.6)
    d = f.box(24.5, 58.5, 29, 6.0, "DWS / KPI 指标层　日批聚合 · L1–L4 分层指标",
              tone="teal", fs=8.4)
    for x, y in zip([a, b, c], [b, c, d]):
        f.arrow(x["b"], y["t"])
    f.obox(24.5, 49.5, 14, 6.0, "MDM 主数据\n客户·供应商·物料·工艺", tone="teal", fs=8.0)
    f.obox(39.5, 49.5, 14, 6.0, "元数据 · 血缘\n质量稽核规则", tone="teal", fs=8.0)
    f.note(24.5, 40.5, "质量目标：数据准确率 ≥ 98%；主数据与单据状态同步延迟 ≤ 5 分钟（异常 ≤ 15 分钟）",
           fs=8.2, color="#404040")

    f.panel(57, 90, 40, 56, "用（数据消费）", tone="purple")
    u1 = f.box(59.5, 85.5, 16.5, 6.0, "九宫格智慧运营看板", tone="purple", fs=8.6)
    u2 = f.box(78, 85.5, 16.5, 6.0, "部门看板 / 自助下钻", tone="purple", fs=8.6)
    u3 = f.box(59.5, 76.5, 16.5, 6.0, "运营问题诊断", tone="purple", fs=8.6)
    u4 = f.box(78, 76.5, 16.5, 6.0, "改善闭环与效果验证", tone="purple", fs=8.4)
    u5 = f.box(59.5, 67.5, 16.5, 6.0, "ChatBI 智能报表", tone="purple", fs=8.6)
    u6 = f.box(78, 67.5, 16.5, 6.0, "S8 异常预警与 AI 智能体", tone="purple", fs=8.0)
    f.box(59.5, 58.5, 35, 6.0, "Outbox 出站回写　→　ERP / MES 等业务系统", tone="orange", fs=8.6)
    f.note(59.5, 49.0,
           "KPI 示例口径：\n"
           "· 产销协同　OTD 交付率 = 按时交付订单数 / 总订单数 × 100%\n"
           "· 生产执行　OEE = 可用率 × 表现率 × 质量率 × 100%\n"
           "· 采购供应　供应商交付率 = 按时交付批数 / 总采购批数 × 100%\n"
           "· 质量　　　成品合格率 = 合格成品数 / 总生产数 × 100%\n"
           "· 库存周转　库存周转天数 = 库存数量 / 日均用量",
           fs=8.0, color="#404040", va="top")

    f.arrow((20.0, 60), (24.0, 60), lw=1.6)
    f.arrow((54.0, 60), (58.5, 60), lw=1.6)
    f.crop(30)
    return f.save("fig_data")


# ==================================================================== 图4
def fig_tech():
    f = Fig("技术架构", w=13.0, h=7.6,
            sub="主流开源技术栈 + 容器化部署；前后端分离、服务可水平扩展、支持信创环境适配")

    f.panel(3, 91, 94, 12.5, "客户端", tone="blue")
    row(6, 87.6, 21.5, 5.4, 4, 2.4,
        ["浏览器\nChrome / Edge / 国产浏览器", "移动端\n企业微信 / 钉钉 / H5",
         "大屏终端\n看板一体机", "开放 API 调用方"], f, tone="blue", fs=8.2)
    f.note(6, 80.6, "HTTPS / TLS 1.2+", fs=8.2)

    f.panel(3, 77.5, 94, 8.0, "接入层", tone="gray")
    row(6, 74.6, 21.5, 4.6, 4, 2.4,
        ["Nginx 反向代理 · 负载均衡", "API 网关 · 统一鉴权限流",
         "OAuth2.0 / JWT · SSO", "静态资源 CDN / 缓存"], f, tone="gray", fs=8.2)

    f.panel(3, 68.5, 94, 15.0, "应用层", tone="green")
    row(6, 65.2, 13.6, 5.4, 6, 2.2,
        ["前端\nVue 3 + TypeScript\nElement Plus · ECharts",
         "后端\n.NET 8 + Admin.NET\n分层架构",
         "ORM\nSqlSugar\n多库适配",
         "工作流\n审批流引擎",
         "任务调度\n日批 / 定时同步",
         "集成执行器\nDB Pull · API Push"], f, tone="green", fs=7.8)
    row(6, 57.4, 21.5, 4.6, 4, 2.4,
        ["多租户隔离", "RBAC 权限与数据权限", "操作 / 登录审计日志", "AI 能力接入（LLM）"], f,
        tone="green", fs=8.2)

    f.panel(3, 52.0, 94, 13.5, "数据层", tone="teal")
    row(6, 48.6, 17.4, 5.4, 5, 2.4,
        ["MySQL 8\n业务库", "MySQL 8\n数据中台库（MDP）",
         "Redis\n缓存 · 会话 · 分布式锁", "对象存储 / 文件服务\n附件·标签·报表",
         "SQL Server / 达梦\n源库只读适配"], f, tone="teal", fs=8.0)
    f.note(6, 41.6, "读写分离与索引优化；大表分区与归档；备份策略：全量日备 + 增量，异地留存",
           fs=8.2, color="#404040")

    f.panel(3, 38.5, 94, 12.0, "运行与部署", tone="orange")
    row(6, 35.2, 17.4, 5.4, 5, 2.4,
        ["Docker 容器化", "Docker Compose /\nK8s 编排", "多环境\n开发·测试·UAT·生产",
         "CI/CD 流水线", "灰度发布与回滚"], f, tone="orange", fs=8.0)
    f.note(6, 28.2, "可用性目标：全年 ≥ 99%；关键组件高可用部署；重大故障力争 48 小时内恢复",
           fs=8.2, color="#404040")

    f.panel(3, 25.0, 94, 14.5, "监控 · 安全 · 运维", tone="red")
    row(6, 21.6, 13.6, 5.0, 6, 2.2,
        ["应用与接口监控", "同步任务监控告警", "集中日志检索",
         "数据加密与脱敏", "漏洞扫描与补丁", "备份恢复演练"], f, tone="red", fs=8.0)
    f.note(6, 14.6,
           "性能目标：常规页面与查询 ≤ 30 秒，复杂查询 / 多图联动 ≤ 60 秒，接口成功率 ≥ 99%；"
           "约 1000 人在线、400 人并发；超大导出与 AI 任务按异步任务处理",
           fs=8.2, color="#404040")
    f.crop(7)
    return f.save("fig_tech")


# ==================================================================== 图5
def fig_integration():
    f = Fig("系统集成与接口架构", w=13.2, h=7.2,
            sub="6 种接入方式 × 定时/实时双模式；可视化配置新增一套第三方系统 ≤ 4 小时，无需大量编码")

    f.panel(3, 90, 21, 74, "源系统", tone="gray")
    names = ["ERP\n主数据·订单·库存·财务", "MES\n工单·工序·设备·报工",
             "QMS\n来料·过程·成品检验", "SRM / 供应商门户\n交货计划·发货单",
             "WMS\n出入库·库存·盘点", "CRM · TMS · OA · IoT"]
    srcs = []
    for i, t in enumerate(names):
        srcs.append(f.box(5.5, 85.5 - i * 11.5, 16, 7.6, t, tone="gray", fs=7.8))

    f.panel(26, 90, 30, 74, "集成通道（采集与治理）", tone="red")
    ways = f.box(28.5, 85.5, 25, 12.0,
                 "接入方式（6 种）\n数据库直连 · HTTP-API · WebService\nFTP 文件 · MQ 消息 · 物联网网关",
                 tone="red", fs=8.2)
    mode = f.box(28.5, 71.0, 25, 8.0, "同步模式\n定时同步（夜间批量） · 实时同步", tone="red", fs=8.4)
    cfg = f.box(28.5, 60.5, 25, 8.0, "可视化通道配置\n数据源 · 实体映射 · 调度策略", tone="red", fs=8.2)
    cln = f.box(28.5, 50.0, 25, 9.5,
                "清洗与标准化\n字段标准化 · 空值过滤\n异常值标记 · 编码统一", tone="red", fs=8.2)
    mon = f.box(28.5, 37.5, 25, 9.5,
                "监控与容错\n同步日志 · 异常告警\n断点续传 · 重试去重", tone="red", fs=8.2)
    for a, b in zip([ways, mode, cfg, cln], [mode, cfg, cln, mon]):
        f.arrow(a["b"], b["t"])
    for s in srcs:
        f.arrow(s["r"], (27.6, s["r"][1]), lw=0.9)

    f.panel(58, 90, 18, 74, "统一数据底座", tone="teal")
    st = [f.box(60, 85.5, 14, 8.0, "STG 贴源", tone="teal", fs=8.8),
          f.box(60, 74.5, 14, 8.0, "STD 标准", tone="teal", fs=8.8),
          f.box(60, 63.5, 14, 8.0, "DWD 宽表", tone="teal", fs=8.8),
          f.box(60, 52.5, 14, 8.0, "DWS / KPI", tone="teal", fs=8.8)]
    for a, b in zip(st, st[1:]):
        f.arrow(a["b"], b["t"])
    f.arrow(mon["r"], (59.2, mon["r"][1]), lw=1.4)
    f.box(60, 40.0, 14, 9.5, "mdp_outbox\n出站事务表", tone="orange", fs=8.4)

    f.panel(78, 90, 19, 74, "上层应用 / 回写", tone="purple")
    ups = ["九宫格看板", "部门看板与下钻", "运营问题诊断", "改善闭环",
           "ChatBI 智能报表", "S8 异常预警"]
    for i, t in enumerate(ups):
        f.box(80, 85.5 - i * 8.0, 15, 6.0, t, tone="purple", fs=8.4)
    f.arrow(st[3]["r"], (79.2, st[3]["r"][1]), lw=1.4)
    ob = f.box(80, 37.0, 15, 8.5, "API 推送回写\nERP / MES 回执核对", tone="orange", fs=8.2)
    f.arrow((74.2, 35.2), ob["l"], lw=1.4)

    f.note(4, 13.5,
           "性能与适配指标：实时同步业务数据变更推送时延 ≤ 2–5 秒；夜间批量 1000 万条明细传输 + 清洗转换 ≤ 30 分钟；"
           "网络中断恢复后自动断点续传，不重复入库、不丢失数据；\n"
           "适配主流国产及通用 ERP / MES / WMS；凭据加密托管，前端不明文展示密钥；"
           "外部系统超时或不可用导致的延迟不计入本平台性能指标。",
           fs=8.4, color="#404040", va="top")
    f.crop(2)
    return f.save("fig_integration")


# ==================================================================== 图6
def fig_diagnosis():
    f = Fig("智能诊断与分析架构", w=13.2, h=7.0,
            sub="以销售订单为主线贯通全链路；实时类问题识别 ≤ 3 分钟，周期类问题每日定时分析；支持 7 个维度根因下钻")

    f.panel(3, 90, 94, 15.0, "① 端到端数据链路（以销售订单为主线）", tone="blue")
    ch = row(5.5, 86.6, 10.6, 5.6, 8, 1.4,
             ["需求预测", "销售订单", "采购计划", "来料入库",
              "车间生产", "半成品流转", "成品入库", "发货配送"], f, tone="blue", fs=8.2)
    f.chain(ch)
    f.note(5.5, 78.6, "断点识别：任一节点缺失前序单据、状态未流转或时间倒挂，即判定为链路断点",
           fs=8.2, color="#404040")

    f.panel(3, 74.0, 46, 26.0, "② 异常识别（规则 + 模型）", tone="orange")
    ex = ["断点\n单据链中断", "滞后\n节点超时效", "损耗\n物料/工时异常",
          "供需失衡\n计划与产能错配", "库存积压\n呆滞与周转恶化", "交付延期\nOTD 未达标",
          "产能浪费\n设备与人效低", "物流低效\n发运与配送滞后"]
    for i, t in enumerate(ex):
        f.box(5.5 + (i % 4) * 10.6, 70.6 - (i // 4) * 8.6, 9.4, 6.6, t,
              tone="orange", fs=7.8)
    f.note(5.5, 53.0,
           "实时类（缺料 · 生产停滞 · 发货延迟）识别延迟 ≤ 3 分钟\n"
           "周期类（库存积压 · 供需失衡 · 产能利用率）每日定时执行分析",
           fs=8.2, color="#404040", va="top")

    f.panel(51, 74.0, 46, 26.0, "③ 根因溯源（7 维下钻）", tone="purple")
    dims = ["订单", "物料", "供应商", "产线", "仓库", "时间段", "人员"]
    for i, t in enumerate(dims):
        f.box(53.5 + (i % 4) * 10.6, 70.6 - (i // 4) * 8.6, 9.4, 6.6, t,
              tone="purple", fs=9.4)
    f.box(85.3, 62.0, 9.4, 6.6, "多维关联\n算法", tone="purple", fs=8.0)
    f.note(53.5, 53.0,
           "从指标 → 明细单据 → 责任维度逐级下钻，每条结论均可追溯到订单主线或指标编码，保留证据快照",
           fs=8.2, color="#404040", va="top")

    f.panel(3, 46.0, 94, 16.0, "④ 诊断输出与改善闭环", tone="green")
    out = row(5.5, 42.6, 16.6, 6.6, 5, 2.6,
              ["运营诊断报告\n结论·根因·证据", "改善策略建议\n可落地措施清单",
               "问题建档\n生成改善任务", "整改派单与跟踪\n责任人·行动项·进度",
               "效果量化复盘\n基线 vs 当前 KPI"], f, tone="green", fs=8.0)
    f.chain(out)
    f.note(5.5, 33.0,
           "闭环规则：改善单必须关联诊断问题或指标编码，保证端到端可追溯；效果验证须保留基线值、目标值与验证结论；"
           "复盘通过后沉淀为标准化管控规则并回写运营建模（S0）。",
           fs=8.2, color="#404040", va="top")

    f.panel(3, 26.0, 94, 12.0, "⑤ 支撑能力", tone="teal")
    row(5.5, 22.6, 16.6, 5.6, 5, 2.6,
        ["统一数据抽取\nSTG→STD→DWD", "指标计算引擎\n日批 + 触发式",
         "阈值与规则库\n可配置", "AI 智能体集群\n7×24 感知预警",
         "ChatBI\n自然语言追问"], f, tone="teal", fs=8.0)
    f.crop(11)
    return f.save("fig_diagnosis")


# ==================================================================== 图7
def fig_kanban():
    f = Fig("九宫格智慧运营看板与指标下钻结构", w=12.4, h=7.4,
            sub="管理层一屏总览 → 部门看板 → 指标下钻 → 明细证据 → 诊断改善，五级贯通")

    f.panel(3, 90, 45, 72, "九宫格智慧运营看板（管理层大屏）", tone="blue")
    grid = [("S1 产销协同", "OTD 交付率"), ("S2 制造协同", "计划达成率"),
            ("S3 供应协同", "物料齐套率"), ("S4 采购执行", "供应商交付率"),
            ("S5 物料仓储", "库存周转天数"), ("S6 生产执行", "设备 OEE"),
            ("S7 成品仓储", "成品合格率"), ("S8 异常监控", "异常闭环率"),
            ("S9 运营指标", "综合运营指数")]
    for i, (m, k) in enumerate(grid):
        f.box(5.5 + (i % 3) * 13.8, 85.5 - (i // 3) * 18.0, 12.6, 15.0,
              f"{m}\n\n{k}\n\n● 红 / 黄 / 绿", tone="blue", fs=8.2)
    f.note(5.5, 30.5,
           "红黄绿状态由指标阈值判定；点击任一格子进入对应部门看板或模块详情看板；\n"
           "顶栏统一展示异常预警摘要与待办改善任务数。",
           fs=8.2, color="#404040", va="top")

    f.panel(50, 90, 22, 72, "部门看板", tone="green")
    dept = [("生产部", "工单进度\n设备 OEE\n工序瓶颈分析"),
            ("采购部", "供应商交付排行\n物料齐套率\n逾期订单"),
            ("质量部", "各环节不合格率\n成品与半成品不良率")]
    for i, (d, k) in enumerate(dept):
        f.box(52.5, 85.5 - i * 19.5, 17, 16.5, f"{d}\n\n{k}", tone="green", fs=8.2)
    f.note(52.5, 30.5, "按角色与数据权限\n分发；空数据友好提示", fs=8.2, va="top")

    f.panel(74, 90, 23, 72, "指标下钻与自助分析", tone="purple")
    lv = [("L1", "模块级核心 KPI"), ("L2", "维度分解指标"),
          ("L3", "过程 / 工序级指标"), ("L4", "单据 · 批次明细")]
    boxes = []
    for i, (l, t) in enumerate(lv):
        boxes.append(f.box(76.5, 85.5 - i * 11.0, 18, 8.0, f"{l}　{t}",
                           tone="purple", fs=8.6))
    for a, b in zip(boxes, boxes[1:]):
        f.arrow(a["b"], b["t"])
    f.box(76.5, 41.5, 18, 8.0, "进入运营问题诊断", tone="orange", fs=8.8)
    f.arrow(boxes[3]["b"], (85.5, 41.5))
    f.note(76.5, 30.5,
           "支持按时间、产品、客户、\n供应商等条件自助筛选；\n"
           "示例：成品合格率 → 某工序\n不合格明细。",
           fs=8.2, va="top")
    f.crop(14)
    return f.save("fig_kanban")


# ==================================================================== 图8
def fig_closedloop():
    f = Fig("异常发现 → 诊断 → 改善 → 验证 闭环管理逻辑", w=12.6, h=6.6,
            sub="PDCA 闭环：任何一次异常都必须走完「发现—定位—整改—验证—固化」，未验证通过不得关闭")

    steps = [
        ("① 发现\n异常识别与提报", "orange",
         "系统自动触发 / 人工提报\n实时类 ≤ 3 分钟"),
        ("② 定级\n分级响应", "orange",
         "一般 · 重要 · 紧急\n紧急异常 ≤ 1 小时响应"),
        ("③ 诊断\n根因溯源", "purple",
         "7 维下钻取证\n输出运营诊断报告"),
        ("④ 建档\n改善任务生成", "green",
         "关联指标与诊断结论\n明确目标值与期限"),
        ("⑤ 派单\n责任人与行动项", "green",
         "复用平台审批流\n进入责任人待办"),
        ("⑥ 执行\n过程跟踪", "green",
         "行动项状态更新\n进度可视、超期提醒"),
        ("⑦ 验证\n效果量化复盘", "blue",
         "基线值 vs 当前 KPI\n未达标退回继续改善"),
        ("⑧ 固化\n标准规则沉淀", "blue",
         "回写 S0 运营建模\n形成 SOP 与管控规则"),
    ]
    xs = [4, 28, 52, 76]
    ys = [82, 50]
    nodes = []
    for i, (t, tone, desc) in enumerate(steps):
        x, y = xs[i % 4], ys[i // 4]
        n = f.box(x, y, 20, 13.0, t, tone=tone, fs=9.6)
        f.note(x + 10, y - 16.0, desc, fs=8.2, ha="center", color="#404040")
        nodes.append(n)

    for i in range(3):
        f.arrow(nodes[i]["r"], nodes[i + 1]["l"], lw=1.4)
        f.arrow(nodes[i + 4]["r"], nodes[i + 5]["l"], lw=1.4)
    f.arrow(nodes[3]["b"], (nodes[3]["c"][0], 62.0), style="-", lw=1.4)
    f.arrow((nodes[3]["c"][0], 62.0), (nodes[4]["c"][0], 62.0), style="-", lw=1.4)
    f.arrow((nodes[4]["c"][0], 62.0), nodes[4]["t"], lw=1.4)

    # 回环：沿左侧外缘回到 ①，避免与 ④→⑤ 折线共用同一列
    lx = 1.6
    f.arrow(nodes[7]["b"], (nodes[7]["c"][0], 26.0), style="-", lw=1.4, dashed=True)
    f.arrow((nodes[7]["c"][0], 26.0), (lx, 26.0), style="-", lw=1.4, dashed=True)
    f.arrow((lx, 26.0), (lx, 88.5), style="-", lw=1.4, dashed=True)
    f.arrow((lx, 88.5), (nodes[0]["c"][0], 88.5), style="-", lw=1.4, dashed=True)
    f.arrow((nodes[0]["c"][0], 88.5), nodes[0]["t"], lw=1.4, dashed=True)
    f.note(52, 23.0, "持续治理：规则固化后回到监控环节，形成螺旋上升的运营改善循环",
           fs=8.6, ha="center", color="#404040")

    f.panel(4, 18.0, 92, 13.0, "闭环刚性约束", tone="gray")
    f.note(6, 14.0,
           "· 改善单必须关联诊断问题或指标编码，端到端可追溯　　· 派单复用平台审批流，不另建孤立任务系统\n"
           "· 效果验证须保留基线值、目标值与验证结论　　　　　　· 验证未通过的改善单不得关闭，自动退回继续改善\n"
           "· 所有异常与改善记录留痕，支持按周期统计闭环率与平均闭环时长",
           fs=8.4, color="#404040", va="top")
    f.crop(2)
    return f.save("fig_closedloop")


if __name__ == "__main__":
    print("生成技术方案配图：")
    fig_otd_blueprint()
    fig_application()
    fig_data()
    fig_tech()
    fig_integration()
    fig_diagnosis()
    fig_kanban()
    fig_closedloop()
    print("完成。")