Advanced control
AI process control and advanced process control.
For processes where a PID controller is not enough: changing dynamics, long dead times, several interacting variables. We design and deploy control based on a mathematical model and artificial intelligence methods that adapts itself to the process.
When a PID controller is not enough.
Advanced process control makes sense where conventional control reaches its limits. A combustion process changes with fuel quality, a furnace responds with a long delay, CHP output and network temperature affect each other. A controller tuned at start-up then spends most of the year running below optimum, and operators switch to manual. The result is quality fluctuations, higher fuel consumption and unnecessary wear on actuators.
Our engineering team has worked on AI-based control for many years and collaborates with academic institutions on development. We use neural networks, fuzzy logic, expert and neuro-fuzzy systems, genetic algorithms and LQR optimal state control. This includes our own adaptive controller, which identifies the controlled system during operation, builds its mathematical model and continuously recalculates the optimal control parameters without a process engineer having to step in.
This is not a laboratory experiment. We deploy advanced control as a layer on top of existing PLCs and control systems, with clear constraints and the option to switch back to the original control at any time. Typical applications are in energy and district heating, incinerators, metals, the chemical and food industries, and anywhere savings in fuel, energy or raw materials can be measured. We assess the benefit on real plant data, not on promises from a presentation.
The methods we use.
Adaptive self-tuning controllers
The controller continuously identifies the process and recalculates the PID or state controller parameters itself when the process dynamics change.
Neural networks
Process models trained on operating data, prediction of variables that cannot be measured directly, and control of non-linear systems.
Fuzzy and expert systems
The experience of your best operators turned into rules that run the process equally well on every shift.
LQR optimal control
State control of multivariable systems, where variables affect each other and tuning one loop at a time does not work.
Process models and simulation
A mathematical model of the process on which we test the control before it touches the plant. We can also derive the model offline from measured data.
Advanced control directly in the PLC
We deploy the algorithms as library functions and Add-On Instructions for the PLC, on an industrial card in the control system or in a panel-mounted controller.
What the delivery includes.
We supply the algorithm and everything it needs to run reliably in production.
- Analysis of the process and historical data
- System identification and mathematical model of the process
- Control algorithm design and simulation
- Implementation in the PLC, on an industrial card or in a panel-mounted controller
- Integration with the existing control system and visualisation
- Bumpless transfer between advanced and original control
- Mobile development workstation for analysis and modelling on site
- Benefit assessment and training for process engineers
Methods and platforms
How we deploy advanced control.
Data analysis
Using historical data and measurements, we find where control is losing performance and how much room there is for improvement.
Identification and model
From operating data, or short tests on the plant if needed, we build a mathematical model of the process.
Simulation and design
We design the control algorithm and verify it on the model, including constraints and emergency conditions.
Deployment and evaluation
We bring the control online step by step, compare it with the original state and assess the benefit on real data.
Related services.
Instrumentation and control
Turnkey I&C and process control systems: sensors, actuators, control loops, panels and the control system.
Energy management
Optimising electricity and heat generation and consumption, renewables integration, battery storage and energy audits.
Production data collection
Historisation of data from PLCs and meters, high-availability historian, visualisation, reporting and remote monitoring.
Frequently asked questions.
Did not find your answer? Write to us and we will reply within 24 hours.
What is advanced process control (APC)?
Advanced process control is a set of methods that sit above conventional PID control. It uses a mathematical model of the process, prediction and optimisation to keep the plant closer to its optimum. Artificial intelligence helps where the model cannot easily be derived from physics.
How is artificial intelligence used in process control?
Neural networks learn the behaviour of a process from data and predict variables, fuzzy logic and expert systems turn operator experience into rules, and genetic algorithms search for optimal settings. The most practical application is an adaptive controller that tunes itself during operation.
What is a self-tuning controller?
It is a controller that, in automatic mode, continuously identifies the controlled system, builds its mathematical model and uses it to calculate the optimal control parameters. When the process, fuel or load changes, it does not need a process engineer to intervene.
Can advanced control be added to an existing PLC?
Yes. We deploy the algorithms as library functions directly in the PLC, on an industrial card in the control system or as a standalone panel-mounted controller with analogue inputs, outputs and Ethernet. The original control remains as a fallback.
How much does deploying advanced control cost?
It depends on the number of control loops, the quality of the available data, whether additional measurements are needed and the platform the control will run on. The first step is usually a data analysis that shows whether deployment pays off and by how much. We give a fixed price after the technical concept.
Which industries benefit most from advanced control?
Those that control slow, non-linear or interacting processes and where savings can be expressed in fuel, energy or raw materials. Typically energy and district heating, incinerators, metals, chemicals, food and water utilities.
Have a process that is hard to control?
Describe the process and send us trends of the problem variables. We will get back to you within 24 hours with an initial assessment and a free consultation.