Bruxism, a condition commonly seen today, occurs as teeth grinding due to locking of the lower jaw and involuntary movements of the lower jaw muscles especially during sleep. This disorder causes adverse effects in patients' daily life such as neck ache, fatigue in jaw muscles, wear in lower jaw joints and tooth enamel. There are some known methods for the diagnosis of bruxism in the electronic environment. These methods are based on electrocardiography, electroencephalography, microelectromechanical measurements combined with surface electromyography. In this study, a low-cost, wearable and user-friendly system that can be used in the home environment, which does not require additional hardware equipment, has been developed for the use of only MEMS Accelerometers for the diagnosis of bruxism. In this system, by determining the time-frequency features of the microvibration signals obtained from the muscles, the effectiveness of the lower jaw activities can be examined with different machine learning algorithms, and these activities can be distinguished in the most effective way. The designed system is particularly useful for the wearable stimulating electronic systems recently developed for the treatment of bruxism.