Soft Sensor in Process Industry: 6 Smart Steps to Build One

Share:
Industrial AI
Soft Sensor in Process Industry: 6 Smart Steps to Build One

Product quality is often known only when a lab sample comes back hours later, long after the process has drifted. A model built from temperatures, pressures and flows can estimate that quality every few seconds instead.

Inferential Measurement Regression Machine Learning Lab Bias Update

Some key variables, such as composition, viscosity or melt index, are expensive or slow to measure. Virtual sensors calculate them from easy measurements that the DCS already collects.

Hello everyone, today we are going to learn what a soft sensor is, which models are used to build one, how to keep it accurate over time, and where it helps most in process plants.
soft sensor

What Is a Soft Sensor?

A soft sensor, also called a virtual sensor or inferential measurement, is a software model that estimates a hard to measure process variable in real time from other measured variables. It combines process data from the process historian with statistical or machine learning methods.

Automation World, citing TrendMiner, notes that most designs rely on five regression methods: linear regression, decision trees, random forests, gradient boosting and neural networks. There is no single best model for every process.

Trend comparing a model estimate with actual measured values
Image credit: Automation World

The soft sensor output becomes a normal DCS tag that operators can trend and alarm. It can also feed control loops, as long as its quality is monitored.

Kadlec, Gabrys and Strandt describe data driven soft sensor models as a major tool for online prediction, process monitoring and sensor fault detection.

Why Plants Use Them

Lab Replacement
Estimate quality between lab samples.
Analyzer Backup
Cover analyzer downtime and faults.
Fouled Sensors
Back up physical sensors in dirty service.
Faster Control
Close loops on quality, not just temperature.
Cost Saving
Avoid expensive online analyzers.
Fault Detection
Compare model and sensor to spot drift.

Automation World highlights estimating lab properties when staff are short, and backing up fouled physical sensors. Both give quick returns.

Estimates in the loop allow quality control through cascade control, where the model output drives a temperature or flow set point.

6 Smart Steps to Build a Soft Sensor

1
Define the Target
Choose the variable and required accuracy.
2
Collect Data
Match process data to lab results with correct time delays.
3
Clean Data
Remove outliers, shutdowns and bad values.
4
Select Inputs
Use process knowledge and correlation.
5
Train and Validate
Test on data the model has never seen.
6
Deploy and Monitor
Track error and update the bias.

Time alignment is the most common pitfall. A lab sample reflects conditions minutes or hours before it was taken.

Input selection benefits from engineering judgement, not only algorithms. Consider sensor linearity and reliability of each input.

Model Types Compared

ModelStrengthWeakness
Linear or PLS regressionSimple, explainableMisses strong nonlinearity
Decision trees and random forestsHandle nonlinearityStep like outputs
Gradient boostingHigh accuracyNeeds careful tuning
Neural networksComplex relationshipsNeeds lots of data, less transparent
First principles hybridUses physicsMore engineering effort

Start with a simple soft sensor and move to complex models only if accuracy demands it. Operators trust models they can understand, and simple models are also easier to maintain when the plant changes.

Hybrid models combine physics with data, a step toward a digital twin.

Keeping the Estimate Accurate

Online EstimateModel runs every few seconds
Lab Result ArrivesCompared with the estimate at sample time
Bias UpdateOffset corrected slowly
Error TrackingAlarm if error grows
RetrainUpdate model after process changes

Bias updating with a filter prevents one bad lab result from jumping the estimate. Large, lasting errors mean the process has changed and the model needs retraining.

Linear Model Example

Estimate = a + b1 × x1 + b2 × x2

Example, column top impurity in percent:
a = 12.4, b1 = minus 0.085 per °C, b2 = 0.9 per bar
Tray temperature x1 = 128 °C, pressure x2 = 1.8 bar
Estimate = 12.4 minus 10.88 + 1.62 = 3.14 percent
Lab result 3.00 percent, error 0.14 percent

Even this simple model can guide operators between lab samples. Confidence limits help them know when to trust it.

Estimate Calculator

Two Input Linear Model
Result
Estimate 3.14, error 0.14

Enter negative coefficients with the minus key in the input box. Replace the values with your own regression results.

Benefits
  • Continuous quality estimates.
  • Lower analyzer costs.
  • Backup for failed instruments.
  • Enables tighter control.
Risks
  • Drift after process changes.
  • Poor data alignment.
  • Overfitted models.
  • Operator mistrust.

Running models at the edge or in the cloud is compared in edge AI vs cloud AI.

Kadlec Review PDF on Data Driven Models

PDF
Data Driven Soft Sensor Models in the Process Industry
Kadlec, Gabrys and Strandt, Computers and Chemical Engineering

Virtual Sensor Video

Soft Sensor FAQ

What is a soft sensor?
It is a software model that estimates a hard to measure variable from other process data. It runs continuously like a normal instrument tag.
Which models are used?
Linear regression, PLS, decision trees, random forests, gradient boosting and neural networks are common. Hybrid physics models are also used.
Can it replace an analyzer?
Sometimes, or it can cover analyzer downtime. Critical or regulated measurements usually still need a real analyzer.
Why is time alignment important?
Lab samples reflect process conditions from minutes or hours before they were analysed. Misaligned data teaches the model the wrong relationships and ruins its accuracy.
How is accuracy maintained?
A slow bias update from lab results corrects small drifts. Large lasting errors trigger model retraining.
Can it be used in control loops?
Yes, often as the outer loop in cascade control. Quality checks must switch control back if the estimate becomes unreliable.
How much data is needed?
Months of data covering normal operating ranges is typical. Complex models such as neural networks need more.

Related Articles

External References

What We Learn Today

  • A soft sensor estimates hard to measure variables from process data.
  • Clean, time aligned data matters more than complex algorithms.
  • Bias updates and monitoring keep estimates trustworthy.
I hope you like above blog. There is no cost associated in sharing the article in your social media. Thanks for reading!! Happy Learning!!

Leave a Reply

Your email address will not be published. Required fields are marked *

Support free learning

Help keep every guide, calculator and tool 100% free

Hosting, software and hours of research go into every article. Even a small contribution helps us publish more in-depth guides for students, technicians and engineers.

  • No paywalls
  • No sign up
  • Free for everyone