Abstract-to improve to generalization capability of a neural-network ultrasonic technique for gas holdup measurement of a bubble column, a SONIA (Self-Organized Network inspired by Immune Algorithm)ultrasonic technique is proposed. B cell mutation mechanism of SONIA creates hidden units having diverse data characteristics that improve generalization capability ofthe neural network-ultrasonic technique. Experiments on simultaneous measurements of the local gas holdup in au air-water bubble column based on the ultrasound energy attennuation and the velocity change show that the proposed method decreases the approximation error x% The proposed method enables a non-invalsive technique for gas hold measurements in a bubble column.