IN08
Course Contents
Module 1: Industrial Data and Automation Foundations
- Evolution of process automation: from conventional loops to AI integration
- Characteristics of industrial sensor data: noise, sampling rates and missing values
- Data preprocessing: cleaning, filtering, outlier detection and time alignment
- Feature engineering: creating meaningful indicators from raw process values
- Understanding machine learning concepts for instrument and control data
- Integrating historical data platforms with analytical environments
- Overcoming data quality issues in older process facilities
Module 2: Predictive Maintenance and Anomaly Detection
- Principles of anomaly detection for abnormal process behaviour
- Equipment condition monitoring and failure prediction methods
- Determining remaining useful life (RUL) and maintenance prioritisation
- Supervised vs unsupervised learning models for process diagnostics
- Tracking degradation in rotating equipment, heat exchangers and valves
- Model accuracy, false alarms and performance measures
- Reducing unplanned downtime through early warning indicators
Module 3: Soft Sensors and Intelligent Instrumentation
- Concept of soft sensors for difficult, costly or delayed measurements
- Selecting process variables and engineering features for virtual sensing
- Regression and classification models for real-time estimation
- Hybrid models combining fundamental process chemistry and plant data
- Sensor validation, data reconciliation and instrument fault isolation
- Online model updating, bias correction and performance monitoring
- Building explainability and operator confidence in AI outputs
Module 4: Model Predictive Control and Process Optimisation
- Limitations of conventional PID control in complex, coupled processes
- Managing process dynamics, interactions, dead times and constraints
- System identification and empirical predictive model development
- Defining controlled, manipulated and disturbance variables
- Setting prediction horizons, control horizons and optimisation targets
- Constraint handling and multivariable economic optimization strategies
- MPC testing, commissioning, performance assessment and tuning
Module 5: Deployment, Governance and Sustainable Performance
- Integrating AI models with existing Distributed Control Systems (DCS)
- Comparing edge, on-premise and cloud deployment architectures
- Model lifecycle management: version control, drift and change management
- Human oversight: alarm management, overrides and safe fallback arrangements
- Industrial cybersecurity standards and secure data connectivity protocols
- AI governance, validation requirements and compliance audit documentation
- Developing a site-specific implementation and technology roadmap
