We experimentally demonstrate the suppression of self-excited thermoacoustic oscillations in a Sondhauss tube via genetic programming (GP). As a data-driven control strategy, GP optimizes both the control-law structure and its parameters based on a predefined cost function. We find that GP closed-loop control suppresses thermoacoustic limit-cycle oscillations through synchronous quenching, in contrast to the asynchronous quenching mechanism observed in conventionally controlled self-excited oscillators. Benchmark tests reveal that GP closed-loop control outperforms both GP open-loop and classic open-loop control, achieving a greater amplitude reduction while consuming an order of magnitude less actuation power. We then show that a Van der Pol oscillator model, driven by the optimal GP control law, faithfully reproduces the experimental dynamics, including synchronous quenching. These findings establish GP as an effective data-driven strategy for active control of thermoacoustic instabilities in propulsion and energy applications.

Close

PrevNext

View figure in article

• Acoustic wave phenomena
• Flow control
• Flow instability
• Instability control
• Nonlinear dynamics in fluids
• Thermoacoustic effect
• Genetic algorithms
• Machine learning models