Introduction to Aspen Inferential Qualities - Developing and Deploying Inferential Soft Sensors for Industrial Processes

Course Id:  APC170   |   Duration:  3.00 day(s)   |   CEUs Awarded:  2.1   |   Level:  Intermediate


Course Objective

Learn how to use Aspen IQmodel to develop linear steady state inferential predictors. Learn how to configure Aspen IQ applications using several different types of models, and how to deploy these applications for on-line use. Learn how to use Aspen Production Control Web Server and Aspen IQview to make on-line tuning changes. Learn how to use the Aspen IQ history files for model development and maintenance. Learn how to use Aspen IQmodel to develop dynamic inferential predictors, and how to deploy these applications.

Course Overview

  • Build a high-performance property estimator
  • Determine when you have sufficient data to generate a valid model
  • Select the appropriate inputs
  • Use Aspen IQmodel to develop a linear steady state inferential predictors
  • Configure Aspen IQ applications using several different types of models and how to deploy these applications for on-line use
  • Use the PCWS (Web Viewer) and Aspen IQview to make on-line tuning changes
  • Use the Aspen IQ history files for model development and maintenance

Benefits

  • Be able to generate an inferential property estimator that will provide property stream specifications
  • Be able to deploy an inferential property estimator that will allow tighter control to property stream specifications

Audience

  • Engineers who are responsible for developing linear inferential property predictors for off-line or on-line use
  • Engineers who will be responsible for designing controller applications that include inferential predictors

Approach

  • Introduction to basic concepts behind  inferential property estimation
  • Hands on workshops build the ability to get the most out of Aspen Inferential Property Estimator Tools

Prerequisites

  • Background in chemical process engineering and/or process operations
  • Some familiarity with Microsoft® Windows operating systems
  • Experience or training in the use of DMCplus (APC101)

Class Schedule

Class Agenda

APC170: Introduction to Aspen Inferential Qualities - Developing and Deploying Inferential Soft Sensors for Industrial Processes

Aspen Inferential Qualities® Introduction

  • Introduce Inferential Property Estimation
  • Provide an overview of Linear PLS Techniques
  • Compare First Principles Models with Regressed Models
  • Introduce the concept behind dynamic Inferential Property Estimation

Aspen IQmodel Overview
  • Discuss how Aspen IQmodel fits into the overall Aspen IQ architecture
  • Discover what Aspen IQmodel can do
  • Differentiate between Steady state versus dynamic inferentials and identify when to use
  • Explain the concept of a step-like approach in building a regression model
  • Demonstrate the steps required to build linear steady state inferentials

Data Specification
  • Set up the model type (inferential and/or sensor validation, dynamic or steady state)
  • Set up the proper dependent variable type to avoid dependent variable data loss
  • Import data through both Aspen IQmodel and the Excel Aspen IQmodel Add-In
  • Set up your ASCII text files in Aspen IQmodel
  • View the imported data
  • Workshop: Perform the Data Specification step in the generation of a typical steady state property estimator

Data Conditioning
  • Use the features of IQModel to condition the input data to detect and handle:
  • Spikes
  • Missing data
  • Bad data
  • Graphically view the input data and use graphical tools to address data problems
  • Review the results of the data conditioning process

Variable Selection
  • Select variables for the inferential
  • Discuss the use of the Genetic Algorithm to determine which independents have the most predictive power for your inferential
  • Learn how to control the variable selection by forcing independent to be used or not to be used
  • Review and discuss the importance of engineering judgment in this process
  • Workshop: Perform the Data Conditioning and Variable Selection step in the generation of a  typical steady state property estimator

Building and Evaluating an Inferential Sensor
  • Build an inferential sensor
  • Specify the model type to be employed
  • Evaluate the quality of your sensor
  • Workshop Build and Evaluate the steady state property estimator generated in earlier workshops

Data Alignment
  • Use IQModel to perform dynamic data alignment
  • Workshop: Perform the dynamic Data Alignment step in the generation of a typical dynamic property estimator

Analyzing Dynamics
  • Use IQModel to analyze the dynamic in the process data
  • Workshop: Analyze Dynamics in the input data to the dynamic property estimator

Introduction to Nonlinear Modeling
  • Review nonlinear modeling options including:
  • Bounded Derivative Networks
  • Nonlinear Fuzzy PLS
  • Monotonic Neural Nets

Deployment Overview
  • Provide an overview of the online deployment process and tools

Introduction to Aspen IQ Config
  • Identify and explain how to use IQconfig
  • Learn how the prediction process fits into the online inferential estimation
  • Configure the subprocesses for an IQ application
  • Validation
  • Input Calculation (IC)
  • Create and maintain a formula library using AspenCalc
  • Workshop: Start configuring a typical property estimator using IQconfig (Step1: IC module)

Prediction Process Parameters
  • Outline how the prediction process fits into the online inferential estimation
  • Discuss the various subprocesses of the prediction process including:
  • Validation
  • Input Calculation (IC)
  • Prediction (PR)
  • Output Calculation (OC)
  • Data Collection (DC)
  • Steady State Detection (SSD)
  • Configure each of the prediction process subprocesses
  • View and modify the prediction process parameters online 
  • Workshop: Continue configuring a typical property estimator using IQconfig (Step2: PR module)

Update Process Parameters
  • Identify and explain the tasks performed by the IQ update process (IQupdate):
  • Lab data collection
  • Lab bias update
  • Analyzer bias update 
  • Workshop: Continue configuring a typical property estimator using IQconfig (Step3: DC, LDC and LBU modules)

Aspen IQ Online Tools: Manage, View, Extract and PCWS
  • Develop a working knowledge of the online tools used to monitor and control inferential sensors:
  • IQ Manage
  • IQ Collect and Extract
  • IQ View
  • Process Control Web Viewer
  • Workshop: Execute the entire process of deploying an online property estimator

Dynamic Prediction and Update Parameters
  • Provide an overview of parameters controlling the dynamic update process  and investigate important subset in more detail
  • Workshop: Configure and deploy a dynamic property estimator

Best Practices and Troubleshooting Tips
  • Provide additional tools and best practices tips to configure, manage and troubleshoot inferentials implemented using AspenIQ
  • Workshop: Configure, implement and tune a dynamic property estimator using simulated live data

Aspen Technology, Inc. awards Continuing Education Units (CEUs) for training classes conducted by our organization. One CEU is granted for every 10 hours of class participation.