SCSAI PLM Rule Template System - Complete Summary

System Overview

This system automatically analyzes and extracts from 7,706 AML files in the SCIOT directory, constructing a complete SCSAI PLM rule template and prompt system.

Industry Validation: The system has been validated with 10GB of real process data in the high-end/machinery industry (high-end equipment customer, zero data leakage, 7×24h stable operation) and is being replicated in the phosphorus chemical industry. The BOM/process/equipment/quality management models for high-end/machinery are highly isomorphic to those in phosphorus chemicals, allowing direct reuse of the underlying object models.

Data Scale

AML File Statistics

Generated Data

Core Business Systems

1. Project Management

• ItemType Count: 21
• Core Objects: Project, Activity2, Activity, Task
• Function: Project planning, WBS breakdown, task assignment, Gantt chart, resource management
• Key attributes:
• Project: name, description, date_start_sched, date_due_sched, status, project_manager
• Activity2: name, duration, percent_complete, assigned_to, source_id, cpi, spi

2. Change Management

• ItemType count: 56
• Core objects: ECR, ECN, Change Controlled Item, Affected Item, Express ECO
• Function: Engineering change request, change notice, express change, affected item management
• Process: ECR creation → Affected item identification → Approval → ECN creation → Implementation

3. Part Management

• ItemType count: 15
• Core objects: Part, CAD, Manufacturer Part, Vendor Part, BOM
• Function: Part, BOM, CAD model, manufacturer, supplier management

4. Document Management

• ItemType count: 15
• Core objects: Document, File, Document Book
• Function: Technical documentation, drawings, file management

5. Quality Management

• ItemType count: 11
• Core objects: NCR, Quality Corrective Action, Quality Issue
• Function: Non-conformance, corrective action, quality issue management

Database table structure

Core tables

1. sciot_item_types - Basic information of object classes
2. sciot_properties - Property Definitions
3. sciot_relationships - Relationship Types
4. sciot_rules - Auto-generated Rules
5. sciot_templates - Auto-generated Templates
6. sciot_prompts - Auto-generated Prompts

Business System Tables

1. sciot_business_systems - Business System Definitions
2. sciot_business_objects - Business System Objects
3. sciot_business_prompts - Business System Prompts

Extension Tables

1. sciot_item_types_full - Full Item Type Information
2. sciot_properties_full - Full Property Information
3. sciot_methods_full - Method Information
4. sciot_forms_full - Form Information
5. sciot_form_fields - Form Fields
6. sciot_itemtype_associations - Item Type Associations

API Endpoints

Auto-generated Data API

Business System API

Generated Files

Rule Template Files

• generated/sciot-rules-auto.json - 3489 rules
• generated/sciot-templates-auto.json - 298 templates
• generated/sciot-prompts-auto.json - 298 prompts
• generated/sciot-index-auto.json - Index

Business System Files

• generated/project_management_template.json - Project management template
• generated/change_management_template.json - Change management template
• generated/business_system_prompts.json - Business system prompts

Analysis Scripts

• aml_analyzer_v2.py - AML analyzer
• analyze_core_systems.py - Core business system analyzer
• generate_business_prompts.py - Business prompt generator
• scan_all_aml.py - AML file scanner
• comprehensive_aml_analyzer.py - Comprehensive AML analyzer

Data Flow

SCIOT目录 (7706 AML文件)
    ↓
aml_analyzer_v2.py (自动分析)
    ↓
sciot_import.db (规则模板表 + 业务系统表)
    ↓
API端点 (随时取出)
    ↓
组装大模型输入 → LLMGenerate → 按规则模板组装AML → 写入SCSAI

业务系统提示词示例

项目管理提示词

你是SCSAI PLM系统的项目管理专家助手...

核心对象类型:
1. Project (项目) - 项目基本信息、时间计划、资源分配
2. Activity2 (WBS任务) - 任务、里程碑、摘要任务
3. Activity (工作流活动) - 审批活动

业务规则:
- 日期格式必须为 YYYY-MM-DD
- 百分比字段必须为 0-100 的整数
- 里程碑任务的工期应为0
- 父任务的完成百分比由子任务自动计算

AML创建模板:
[完整的AML模板代码]

变更管理提示词

你是SCSAI PLM系统的变更管理专家助手...

核心对象类型:
1. ECR (工程变更请求)
2. ECN (工程变更通知)
3. Affected Item (受影响项)

变更流程规则:
1. ECR必须先创建并审批通过，才能创建ECN
2. ECN必须关联到已批准的ECR
3. 受影响项必须在ECR阶段识别
4. 变更实施前必须完成影响分析

使用方式

1. 查询对象类模板

const response = await fetch('/api/aml/auto-full/Project');
const data = await response.json();
// data包含: template, rules, prompt, summary

2. 查询业务系统提示词

const response = await fetch('/api/aml/business-prompt/project_management');
const prompt = await response.json();
// 使用prompt.system_prompt作为LLM系统提示词

3. 重新生成数据

const response = await fetch('/api/aml/rebuild-auto', {method: 'POST'});

系统特点

1. 全自动分析 - 从7706个AML文件自动提取元数据
2. 业务系统级 - 不仅分析单个对象类，还构建完整业务系统
3. 完整关联 - 包含ItemType + Properties + Methods + Forms + Relationships
4. 规则驱动 - 自动生成3489条数据验证和业务规则
5. 提示词系统 - 为每个业务系统生成专业的LLM提示词
6. AML模板 - 提供可直接使用的AML创建模板
7. 数据库存储 - 所有数据存入数据库，随时可取
8. API接口 - 提供完整的REST API供前端调用

后续优化方向

1. 添加更多业务系统（需求管理、供应商管理等）
2. 完善Method和Form的关联分析
3. 添加Workflow和LifeCycle的详细分析
4. 生成更复杂的业务场景模板
5. 添加权限和访问控制分析
6. 优化提示词以支持更多LLM模型

生成时间: 2024年

Data Source: SCIOT directory (7706 AML files)

System Version: v2.0