带钢热连轧过程具有工序众多、工况频变、多系统耦合等特点，这给对其运行工况进行有效的监测和评估带来一定困难。常见的方法是从数据出发对生产过程数据进行特征提取，并利用提取的特征构建监测和评估模型，当该过程生产产品不断切换导致工况不断变化时，所提取的特征往往不足以反映该过程特性，进而导致工况评估错误。基于以上问题，本书创新性地设计了基于共性-个性特征建模的热连轧过程特征分析框架，并提出了多种特征建模方法。同时，基于所构建的特征模型，实现热连轧过程工况的评估，并进行应用验证。本书所包含的主要内容有：（1）带钢热连轧过程简介，分析热连轧过程的特点，并以数据为基础，展示热连轧过程多工况运行特性；（2）带钢热连轧过程常见的故障及异常的工况，分析该过程运行工况特点、异常工况的特性及导致的原因；（3）热连轧过程故障注入系统，设计开发热连轧精轧过程故障注入系统，实现对不同异常工况的仿真分析；（4）提出热连轧过程特征建模框架及工况评估方法，构建基于共性-个性特征建模的特征分析框架，利用空间分析、三维张量建模、深度置信网络等多种分析方法实现基于共性-个性特征建模的分析方法，同时提出相应的异常工况评估方法；（5）原型系统方法验证，利用实际的热连轧数据验证方法的有效性，构建验证原型系统，利用实验室实时数据验证方法的实用性。

1 Introduction2 Brief Introduction to the Hot Strip Milling Process2．1 Introduction2．2 Description of the Hot Strip Milling Process2．3 Data Characteristics of the Hot Strip Milling Process2．4 Multi-scenario Hot Spots in the Hot Strip Milling Process2．5 Analysis of Control and Operation Conditions in the Finishing Milling Process3．5 Conclusion3 Milling Process Fault Injection System3．1 Common Faults in the Milling Process3．2 Fault Injection Model for the Milling Process3．3 Milling Process Fault Injection System and Implementation3．4 Presentation of Several Types of Fault Data3．5 Conclusion4 Commonality-Individuality Feature Modeling Framework4．1 Introduction4．2 Common and Individual Features in the Hot Strip Milling Process4．3 Commonality-Individuality Feature Modeling Framework4．4 Feature Modeling and Condition Assessment4．5 Conclusion5 Feature Modeling Method Based on Commonality-Individuality PLS5．1 Introduction5．2 Brief Introduction to PCA and PLS Algorithms5．3 Commonality-Individuality PCA Algorithm5．4 Commonality-Individuality PLS Algorithm5．5 Results of Commonality-Individuality Feature Modeling5．6 Conclusion6 Feature Modeling Method Based on Commonality-Individuality Subspace6．1 Introduction6．2 Description of Commonality-Individuality Subspace6．3 Extraction of Commonality-Individuality Features6．4 Migration of Common Subspace6．5 Results of Commonality-Individuality Subspace Modeling6．6 Conclusion7 Feature Modeling Method Based on Tensor Decomposition7．1 Introduction7．2 Brief Introduction to Tensor Decomposition Methods7．3 Extraction of Common Features by Tensor Decomposition7．4 Modeling of Individual Features by Tensor Decomposition7．5 Summary and Comparison of Tensor Decomposition Methods7．6 Conclusion8 Feature Modeling Method Based on Deep Belief Networks8．1 Introduction8．2 Description of Deep Belief Network Method8．3 Feature Modeling Based on Deep Belief Networks8．4 Network Construction and Parameter Learning8．5 Conclusion9 Condition Assessment Based on Linear Commonality-Individuality Feature Modeling9．1 Introduction9．2 Results of Condition Assessment Based on Commonality-Individuality PCA9．3 Results of Condition Assessment Based on Commonality-Individuality Subspace9．4 Results of Condition Assessment Based on Tensor Decomposition9．5 Conclusion10 Condition Assessment Based on Commonality-Individuality Deep Belief Networks10．1 Introduction10．2 Training of Abnormal Condition Models10．3 Presentation of Abnormal Condition Assessment10．4 Conclusion11 Prototype System for Hot Strip Milling Process Condition Assessment11．1 Introduction11．2 Hardware Framework of the Prototype System11．3 Software Framework of the Prototype System11．4 Data Replay and Presentation of the Prototype System11．5 Application of Condition Assessment in the Prototype System11．6 ConclusionReferences