Fuzzy classifier of device state basing on MEMS signals in burglar alarm application

The synthesis of fuzzy logic classifier in burglary detector application is presented in the article. The signal source for the system are micro electro-mechanical (MEMS) devices: accelerometer and gyroscope. The use of selected MEMS ICs made the application compact and self-contained. Vibrations and rotational speed signals of protected surface are acquired and processed in computational core of classifier to detect critical events such as intrusion or attempt of it. Preliminary algorithms called direct-DSP transformation into fuzzy logic system is shown. The goal and the motivation of the work is to obtain more general results and better systematization of classifier activity. Fuzzy logic gave possibility of easier sensitivity tuning by changing rules or parameters of membership functions. FL classifier is also a step to create ANFIS structure, where membership functions parameters may be adopted in automated learning procedure. Hardware in general is described. The algorithm synthesis based on the fuzzy logic fundamentals is introduced together with its components such a: linguistic variables, its membership functions sets selection and fuzzy rules. The results are obtained on dedicated laboratory stand, evaluated and combined with existing results.