Abstract:

This paper presents a model‐based approach to detect termites from their head banging acoustic signals, and is derived from Bayesian probability theory. The termite head banging is the loudest and most diagnostic sound that termites make, and can be utilized for termite detection. The laser Doppler vibrometry system is used to obtain the termite head‐banging signals from infested wood. An algorithm based on Bayesian probability theory is developed to detect termites’ presence. The atomic model that represents the termites’ data is the sum of decaying sinusoidal signals. First the model selection is performed that tells us about the number of vibration frequency components present in the data under observation. Once the correct model is known, then the vibration frequency that corresponds to termites’ head banging frequency is determined. The calculations are performed using the Markov chain Monte Carlo method. Monte Carlo integration is then used to approximate the marginal posterior probabilities for all the parameters, including the number of exponentials and whether a constant offset is present. The performance of this algorithm is evaluated by testing it on experimental data, and the results obtained reveal the excellent performance of the algorithm.