metal casting process merging with artificial intelligence
TRANSCRIPT
Prof. Baode Sun
孙宝德教授
Dean of School of Material Science &
Engineering, SJTU
上海交通大学材料学院院长,博士生导师
Metal Casting Process
– Merging with Artificial Intelligence
铸造—— 走向智能
铸造-走向智能
2019年6月20日
孙宝德
Metal Casting Process – Merging with Artifcial Intelligence
上海市先进高温材料及其精密成形重点实验室
Shanghai Key Lab of Advanced High-Temperature Materials and Precision Forming
目录
CO
NT
EN
TS
1. 铸造瓶颈问题The key Problem of Foundry
2. 走向智能铸造Towards Intelligent Casting
3. 数字化铸造是基础Digital Casting is the Foundation
4. 智能化铸造探索Exploration of Intelligent Casting
目录
CO
NT
EN
TS
1. 铸造瓶颈问题The key Problem of Foundry
2. 走向智能铸造Towards Intelligent Casting
3. 数字化铸造是基础Digital Casting is the Foundation
4. 智能化铸造探索Exploration of Intelligent Casting
5
1.国际铸造产业发展状况与面临的问题
China:
49.4 million tons
India:
12.5 million tons
America:
9.6 million
tons
Europe:
13 million tons
Brazil:
2.2 million
tons2017年,全球铸件产量近1.1亿吨
global castings production reached nearly 110 million tons.
1. Development and problems of international foundry industry
铸造产业 是“工业之母”
≈年消耗1亿吨标准煤,铸造能量的利用率17%,能源与资源消耗多,碳排放高。The annual consumption of 100 million tons of standard coal, casting energy utilization rate is 17%, energy and
resources consumption, high carbon emissions.
≈3亿吨金属熔化量300 million tons of melt
The foundry industry is
"foundation of Industry"
6
2.Development and problems of China's foundry industry
45%
11%
9%
5%4%
5%
2%
2%
3% 2%11%
China
India
U.S.A
Japan
Russia
Germany
Brizil
Korea
France
Italy
Other
世界铸造大国
平均生产能力仅为德国的20%
全国 26000家,99%企业
年产万吨以下
现代化铸造企业<10%
2017年
2.中国铸造产业发展状况与面临的问题
China produces 45% of casting products in the world, but compared with the developed countries, the
quality of casting products in China is still far behind the developed countries. The average production
capacity in China is only as large as 20% of Germany’s. There are about 26,000 foundries in China, and
99% of them have the capacity of less than 10,000 tons/year, and only 10% of them are modernized.
7
3.全球熔模铸造行业市场现状与发展趋势
0.00%
1.00%
2.00%
3.00%
4.00%
5.00%
6.00%
0
20
40
60
80
100
120
140
160
2011 2012 2013 2014 2015 2016 2017 2018
2011-2018年全球熔模铸造市场及规模增长率
(单位:亿美元,%)
市场规模(亿美元) 同比增长(%)
航空领域达56亿美元
航空
39%
其他
19%
汽车
14%
医疗
17%
机械11%
全球熔模铸造市场应用分布
3. Current situation and development trend of global investment casting market
为满足熔模铸件“精、大、强”的品质要求,应发展的关键和支柱技术?The market of aviation sector is about $5.6 billion. How to meet the needs of the demanding from aviation
sector?
4.Current problems and the technical issues in foundry industry
2003-2018年中国铸件产量(万吨)和增长率(%)China's castings output (10,000 tons) and growth rate (%) from 2003 to 2018
中国铸造总量在4900万吨,因为铸造缺陷造成的损失约600亿。The total amount of foundry in China is 49 million tons, and the loss caused by casting defects is about 60 billion Yuan.
疏松Shrinkage
夹杂物inclusion
冷隔Cold lap
尺寸超差out of tolerance欠浇misrun
漏钢Breakout
4.目前铸造行业存在的问题及技术瓶颈
9
❑工艺出品率低 Low process yield
❑合格率低 Low qualified rate
❑精度低 Low dimensional accuracy
❑凝固缺陷多 Solidification defects
❑生产线孤岛多 Production islands
铸造走向何方?Where is casting going?
5.目前铸造存在的理论问题及技术瓶颈
铸造技术瓶颈The key Technology issue
❑形核与生长 Nucleation and growth
❑凝固中的解析解 Analytical Solution
❑缺陷形成准则 Defect Formation
❑微观组织Microstructure
❑相场法 Phase field method
❑模数法Modulus method
凝固理论发展Development of solidification theory
理论指导
实践验证
5. Current theoretical problems and the technical issues in foundry
theoretical guidance
practical verification
目录
CO
NT
EN
TS
1. 铸造瓶颈问题The key Problem of Foundry
2. 走向智能铸造Towards Intelligent Casting
3. 数字化铸造基础Digital Casting is the Foundation
4. 智能化铸造探索Exploration of Intelligent Casting
1.智能化与数字化改变传统铸造
1. Artificial Intelligence and Digitalization are Changing Traditional Casting
人工智能已经在面部识别,自动驾驶,甚至在传统认为程序难以
逾越的围棋领域大放异彩。Artificial Intelligence has been successful in facial
recognition, autonomous driving, and it even
beats top GO players that was not considered
possible many years ago.
铸造过程可以视作一个由多种因素影响的黑箱过程,通过分析大
量数据以及恰当的AI算法能掌握这个黑箱的规律,并且将规律用
于自我学习指导工艺设计。Casting process could be considered as a ‘black
box’ process. By analyzing big data and implement
proper AI algorithm, it is possible to master the
essence of the ‘black box’, and in return the casting
process can be optimized.
AlphaGo Zero 在“无(人类)监督下”完成学习与演进并超越人类。
Complete Learning and Evolution without Supervision and Beyond Human Beings
Fully Connected Layers
人工智能辅助理性设计铸造工艺
大数据处理 输入参数vs.材料性能的数据轨迹
机器学习深度学习
掌握规律
材料性能
黑箱过程
2.智能铸造系统
浇铸系统如冒口,浇道设计对于铸件性能影响巨大。通过人工智能技术合理优化设计浇铸系统,既避免疏松等缺陷,又节省材料。Casting system including riser and runner has tremendous impact on the properties. AI is implemented to
optimize casting system to avoid casting defects, and save materials.
2. Intelligent Casting System
Artificial Intelligence Assisted Rational Design for Foundry Technology
❑ 凝固组织
❑ 凝固缺陷
❑ 力学性能
❑ 出品率
➢ 化学成分
➢ 铸造工艺
➢ 环境噪音
➢ 结构形状
模具/产品设计Mold design
铸造工艺参数Process
parameters
铸件性能/缺陷Casting defects
and properties
生产现场数据采集
AI Cast云端服务
人工智能算法
Algorithm
数据库Database
浇冒口智能设计Optimize casting
system
产线质量控制Quality control
铸造工艺优化Optimize
casting process
缺陷自动识别Defects
recognition
铸造企业接入云端服务器
2.智能铸造系统
2. Intelligent Casting System
铸造企业通过传感器与5G网络接入到AI cast
云端系统提供服务,云端系统获得的企业数据又进一步提高了算法的准确性。Foundries have access to AI
Cast via sensors and 5G
network and get services.
In return, the algorithm
can be improved by data
from foundries.
3.智能铸造装备与技术
“智能铸造”是数字化、信息化与铸造生产高度融合的产物。智能铸造系统是具有学习能力的大数据知识库,自我规划、自我改善。"Intelligent casting" is a product of high integration of
digitalization, informatization and casting production. Intelligent casting system is a large data knowledge base with
learning ability, self-planning and self-improvement.
虚拟铸造
Virtual casting
快速造型
Fast moulding
机器人应用
Robotics
缺陷识别
Defect Recognition智能铸造装备与技
术
3. Intelligent Casting Equipment and Technology
物理空间Physical space虚拟空间 Virtual space
机理模型 数学模型
应用验证
分析目标
经验知识
工艺参数优化
装备产线可靠性
产线质量分析
迭代验证
4.铸造数字孪生技术
4. Foundry Digital Twin Technology
By making full use of the data of physical model, sensor update and operation history, the simulation process of multi-discipline,
multi-physical quantity, multi-scale and Multi-probability are integrated, and the mapping is completed in virtual space, thus
reflecting the corresponding life cycle process of physical equipment.
Analytical objectives
Experiential knowledge
Process optimization
Reliability of Equipment
Quality Analysis
Iterative verification
目录
CO
NT
EN
TS
1. 铸造瓶颈问题Bottleneck Problem of Foundry
2. 走向智能铸造Towards Intelligent Casting
3. 数字化铸造基础Digital Casting is the Foundation
4. 智能化铸造探索Exploration of Intelligent Casting
铸造系统数字化
产品数字化
工艺数字化
CAD:UG/Catia/PRO-E
CAE:Nastran/Ansys
PDM: Team-Center
ERP: SAP、Enovia
装备数字化
控制:NC,CNC,DNC
系统:MC,FMC,FMS
涉及装备与产品
的几何、力学行
为的耦合!
CAPP、DFX?
影响产品性能影响制造效率
影响制造效率影响制造质量
数字化铸造技术体系:铸件表达数字化、铸造装备数字化、铸造工艺数字化、铸造系统数字化。
1.数字化铸造技术体系
Digital casting technology system: casting expression digitalization, casting equipment digitalization, casting
technology digitalization, casting system digitalization.
1. Digital Casting Technology System
2.虚拟铸造技术
全过程误差流误差分析软件平台全流程位移场仿真计算平台
全流程虚拟集成计算与误差调控技术是铸件的尺寸精度控制有效路径。The whole process integrated calculation and error control technology is the digitized path of dimensional
accuracy control of castings.
2. Virtual Casting Technology
Simulation platform for displacement field for whole processSoftware platform for error flow and analysis in the whole process
全流程误差流数字化调控 3.数字化建模技术
◆ 数字化建模设计与稳健控制技
术:基于容差控制系数,构造
各阶段容差约束条件,并考虑
各阶段制造难度,多尺寸的综
合精度为目标,对各阶段进行
容差设计。
3. Digital Modeling Technology
◆ Tolerance design and robust control
technology: Based on tolerance control
coefficients, tolerance constraints are
constructed for each stage, and
tolerance design is carried out for each
stage taking into account the
manufacturing difficulty of each stage
and the comprehensive accuracy of
multi-dimension as the goal.
4.集成计算材料工程
微观组织计算 显微缩松预测 力学性能预测宏观模拟
ICME跨层次多尺度计算可构建高温合金铸件成分-组织-结构-性能-工艺的关系
通过了铸件成型流场,温度场,凝固缺陷与凝固组织的多尺度计算,构建了工艺-组织-性能关联关系(凝固参数-二次枝晶间距-显微缩松-强度与寿命),复杂铸件10个特征位置进行了验证,预测与实际晶粒尺寸平均相差8.45%。Through the multi-scale calculation of casting flow field, temperature field, solidification defects and
solidification structure, the process-structure-performance relationship (solidification parameters-secondary dendrite
spacing-micro shrinkage-strength and life) was constructed.
特征位置验证
4. Integrated Computing Materials Engineering
Ten characteristic positions of complex castings were verified, and the average difference between predicted and actual
grain size was 8.45%.multi-scale calculation of ICME can construct the relationship between composition, structure,
performance and process of superalloy castings.
基于SLS增材制造技术,制造周期较传统工艺缩短6倍以上 。增材制造造型设备的数字化与智能化是实
现铸造过程智能化的关键技术。
5.装备数字化快速铸造技术
5. Equipment Digital Rapid Casting Technology
The digitalization and intellectualization of moulding equipment is the key to realize the
digitalization control of casting process. Based on SLS augmented material manufacturing
technology, short cycle and high efficiency casting forming. The manufacturing cycle is
more than 6 times shorter than that of traditional process.
目录
CONTENTS
1. 铸造瓶颈问题The key Problem of Foundry
2. 走向智能铸造Towards Intelligent Casting
3. 数字化铸造基础Digital Casting is the Foundation
4. 智能化铸造探索Exploration of Intelligent Casting
30年代中期 Chvorinov 凝固模数
40年代末Caine J.B 凝固速率比法
60年代末 R.Woldawer铸件形态参数
1982年 Mcadams J. Q参数法
1995年周尧和提出周界商法
2000年高尚书 三次方程解析法
2001年赵清和Δ参数法冒口设计
经验驱动的模型
1.浇冒系统智能优化技术
1. Intelligent optimization of gating system
Experience-driven model
riser design
人工智能算法冒口优化
冒口形状几何拓扑优化➢Riser optimization with artificial intelligence algorithm
➢Topology optimization for riser shape
冒口尺寸数据流自动优化计算流程与框架
智能铸造-冒口数字化智能化设计1.浇冒系统智能优化技术
1. Intelligent optimization of gating system
Software flow
data flow
Framework for automatic optimization of riser size data flow
自动高通量计算
智能算法自动优化寻优
典型环套环铸件
Automatic high-throughput
calculation
Automatic optimization of
intelligent algorithm
Typical ring to ring
castings
采用数据挖掘算法对铸件缺陷位置和工艺出品率进行多目标寻优。铸件工艺出品率提升了14.49%,经模拟与浇注实验验证后,
铸件工艺出品率相差0.78%,算法优化精度可靠。Multi-objective optimization of casting defect location and process yield was
carried out by using data mining algorithm. The casting process yield has been increased by 14.49%. The simulation and pouring
experiments show that the difference of casting process yield is 0.78%. The optimization accuracy of the algorithm is reliable.
智能铸造-冒口数字化智能化设计1.浇冒系统智能优化技术
1. Intelligent optimization of gating system
特征铸件实验验证
Experimental verification of typical castings
模拟验证simulation verification
模数法冒口riser with modulus method
优化算法冒口riser optimization algorithm
智能铸造工艺设计系统
浇注系统模具设计 制壳参数 收缩补偿率
工艺设计模型
浇注参数
铸造工艺分析 铸件结构参数化
浇注系统参数化 补缩系统参数化
Typical Structural Database
2.智能铸造工艺设计软件系统
2. Intelligent Casting Process Design Software System
Process design model
Analysis of casting process and structural parameterization
Parameterization of gating system
gating system
mold design
shell parameters contraction compensation rate
pouring parameters
超级平台模拟计算
上传铸件模型文件
上传计算参数文件 下载模拟计算优化报告
无需购买软件即可在线使用本软件计算、分析服务。并在计算结束后,下载标准化文本和动画报告。This software can be used online
without purchasing. At the end of the calculation, the standardized text and animation report are downloaded.
3.智能铸造云平台
3. Intelligent Foundry Cloud Platform
Upload calculation parameter file
Upload Casting Model File
Submission
model
On-line
Computing
On-line
optimization
Super Platform Simulation
Download simulation and
optimization report
大数据技术体系逐渐成熟 智能铸造工厂是大势所趋决策支撑
数据基础
传统铸造走向智能铸造已成为必然。It is inevitable for traditional casting to move towards
intelligent casting
结论与展望
Big data technology system is maturing Intelligent foundry is the trendData base
Decision support
CONCLUSION AND OUTLOOK
Thank you!