Abstract: This study develops a real-time compaction quality evaluation model to improve the low accuracy and robustness of the existing compaction quality evaluation models. This real-time model adopts a newly fuzzy logic based on the kernel method (KM) and adaptive chaotic bacteria foraging algorithm (AC-BFA). Its input parameters consist of the physical parameters of compacted material, such as the amplitudes of the first harmonic and fundamental wave decomposed from the vibration signal, the characteristic parameters of material source, and the rolling parameters of construction process. Engineering application shows that compared with the commonly used compaction quality evaluation models, the real-time model is more accurate and superior showing stronger robustness in noise data tests and abnormal data tests, and it can achieve real-time evaluation of compaction quality when embedded into a compaction quality real-time monitoring system.

Key words: compaction quality, kernel method, adaptive chaotic bacteria foraging algorithm, fuzzy logic, real-time evaluation.