Machine learning control explained

Machine learning control (MLC) is a subfield of machine learning, intelligent control, and control theory which aims to solve optimal control problems with machine learning methods. Key applications are complex nonlinear systems for which linear control theory methods are not applicable.

Types of problems and tasks

Four types of problems are commonly encountered:

• Control parameter identification: MLC translates to a parameter identification^[1] if the structure of the control law is given but the parameters are unknown. One example is the genetic algorithm for optimizing coefficients of a PID controller^[2] or discrete-time optimal control.^[3]
• Control design as regression problem of the first kind: MLC approximates a general nonlinear mapping from sensor signals to actuation commands, if the sensor signals and the optimal actuation command are known for every state. One example is the computation of sensor feedback from a known full state feedback. Neural networks are commonly used for such tasks.^[4]
• Control design as regression problem of the second kind: MLC may also identify arbitrary nonlinear control laws which minimize the cost function of the plant. In this case, neither a model, the control law structure, nor the optimizing actuation command needs to be known. The optimization is only based on the control performance (cost function) as measured in the plant. Genetic programming is a powerful regression technique for this purpose.^[5]
• Reinforcement learning control: The control law may be continually updated over measured performance changes (rewards) using reinforcement learning.^[6]

Applications

MLC has been successfully appliedto many nonlinear control problems,exploring unknown and often unexpected actuation mechanisms. Example applications include:

• spacecraft attitude control,^[7]
• thermal control of buildings,^[8]
• feedback control of turbulence,^[9]
• and remotely operated underwater vehicles.^[10]

Many more engineering MLC application are summarized in the review article of PJ Fleming & RC Purshouse (2002).^[11]

As is the case for all general nonlinear methods,MLC does not guarantee convergence, optimality, or robustness for a range of operating conditions.

See also

• Reinforcement learning

Further reading

• Dimitris C Dracopoulos (August 1997) "Evolutionary Learning Algorithms for Neural Adaptive Control", Springer. .
• Thomas Duriez, Steven L. Brunton & Bernd R. Noack (November 2016) "Machine Learning Control - Taming Nonlinear Dynamics and Turbulence", Springer. .

Notes and References

1. Thomas Bäck & Hans-Paul Schwefel (Spring 1993) "An overview of evolutionary algorithms for parameter optimization", Journal of Evolutionary Computation (MIT Press), vol. 1, no. 1, pp. 1-23
2. N. Benard, J. Pons-Prats, J. Periaux, G. Bugeda, J.-P. Bonnet & E. Moreau, (2015) "Multi-Input Genetic Algorithm for Experimental Optimization of the Reattachment Downstream of a Backward-Facing Step with Surface Plasma Actuator", Paper AIAA 2015-2957 at 46th AIAA Plasmadynamics and Lasers Conference, Dallas, TX, USA, pp. 1-23.
3. Zbigniew Michalewicz, Cezary Z. Janikow & Jacek B. Krawczyk (July 1992) "A modified genetic algorithm for optimal control problems", [Computers & Mathematics with Applications], vol. 23, no 12, pp. 83-94.
4. C. Lee, J. Kim, D. Babcock & R. Goodman (1997) "Application of neural networks to turbulence control for drag reduction", Physics of Fluids, vol. 6, no. 9, pp. 1740-1747
5. D. C. Dracopoulos & S. Kent (December 1997) "Genetic programming for prediction and control", Neural Computing & Applications (Springer), vol. 6, no. 4, pp. 214-228.
6. Andrew G. Barto (December 1994) "Reinforcement learning control", Current Opinion in Neurobiology, vol. 6, no. 4, pp. 888–893
7. Dimitris. C. Dracopoulos & Antonia. J. Jones (1994) Neuro-genetic adaptive attitude control, Neural Computing & Applications (Springer), vol. 2, no. 4, pp. 183-204.
8. Jonathan A. Wright, Heather A. Loosemore & Raziyeh Farmani (2002) "Optimization of building thermal design and control by multi-criterion genetic algorithm, [Energy and Buildings], vol. 34, no. 9, pp. 959-972.
9. Steven J. Brunton & Bernd R. Noack (2015) Closed-loop turbulence control: Progress and challenges, Applied Mechanics Reviews, vol. 67, no. 5, article 050801, pp. 1-48.
10. J. Javadi-Moghaddam, & A. Bagheri (2010 "An adaptive neuro-fuzzy sliding mode based genetic algorithm control system for under water remotely operated vehicle", Expert Systems with Applications, vol. 37 no. 1, pp. 647-660.
11. Peter J. Fleming, R. C. Purshouse (2002 "Evolutionary algorithms in control systems engineering: a survey"Control Engineering Practice, vol. 10, no. 11, pp. 1223-1241