In this paper, a current-mode design methodology to implement a set of fuzzy linguistic hedge circuits is proposed. The so-called fuzzy linguistic hedge is a fuzzy operation applied to adjust the membership function of a fuzzy set. The fuzzy membership function of control variable and the control rules are very important in a fuzzy logic controller because they dominate the control strategies. If the control results fail to meet the system requirements, the control objective can still be achieved by adjusting the membership function of the fuzzy set or the control rules. Moreover, the adjustment effect of the control strategies through the modifications of the fuzzy membership function is the same as that of the system control rules. In this paper, we propose a set of fuzzy linguistic hedge circuits, including absolutely, very, much more, more, plus minus, more or less, slightly, and contrast intensification, which has been fabricated in 0.8 μm CMOS process. Experimental results show that the average error of the circuits is within 1% of the full scale current. Under the power supply voltage of 3.3 V, the operating dynamic range is 50 μA. Furthermore, these circuits still work well even when the power supply voltage is down to 2.5 V. In addition, in real world application, we can incorporate a membership function generator, a fuzzification unit, a multi-input maximum/minimum circuit, and a defuzzification unit with the linguistic hedge used to modify the membership function in order to develop a real-time adaptive fuzzy logic controller.
Read full abstract