面向复杂纯滞后系统的智能控制系统研究  

Research on Intelligent Control Systems for Complex Pure Hysteresis Systems

作　　者：黄磊 HUANG Lei(Department of Machanical and Electrical Engineering,Henan Information Engineering School,Zhengzhou 450011,China)

机构地区：[1]河南信息工程学校机电工程系,河南郑州450011

出　　处：《自动化仪表》2022年第9期30-34,39,共6页Process Automation Instrumentation

摘　　要：工业场景下,大规模多点输入集中输出的纯滞后系统通常会带来成本控制器布署高、自适应控制运算能力不足、规模柔性和集控能力差、各系统输出偏差大等问题。为了适应多个纯滞后系统的耦合输出要求,同时兼顾快速稳定收敛的控制目标,设计了一套可以支持工业大数据应用的智能控制系统。创新性地提出了一种基于大数据流计算实现的自适应增量式比例积分微分(PID)控制方法。该方法采用数据采集和处理分离的分层结构,可以满足大规模复杂纯滞后系统的低成本集控和系统数量变化的要求。经验证,自适应增量式PID模型通过滑动窗口流计算,能快速实现不同系统不同输入下的同速控制,确保多系统输出的一致性。该系统支持机器学习算法的快速嵌入,为后续实现基于历史数据批处理反馈的工业智能提供了可能。In industrial scenarios, pure hysteresis systems with large-scale multi-point inputs and centralized outputs usually bring about problems such as high-cost controller deployment, insufficient adaptive control computing capability, poor scale flexibility and centralized control capability, and large output deviation of each system. In order to be able to accommodate the coupled output requirements of multiple pure hysteresis systems while taking into account the control objective of fast and stable convergence, an intelligent control system that can support industrial big data applications is designed. An innovative adaptive incremental proportional integral differential(PID) control method based on the computational implementation of big data streams is proposed. The method adopts a hierarchical structure with data acquisition and processing separation, which can meet the requirements of low-cost centralized control of large-scale complex pure hysteresis systems and system quantity variation. It is verified that the adaptive incremental PID model can quickly realize the same speed control under different inputs of different systems and ensure the consistency of multi-system output by sliding window flow calculation. The system supports rapid embedding of machine learning algorithms, providing the possibility of subsequent implementation of industrial intelligence based on batch feedback of historical data.

关 键 词：工业物联网 电气控制系统 工业大数据 纯滞后 耦合输出 智能控制 比例积分微分控制 流计算 

分 类 号：TH166[机械工程—机械制造及自动化]