The IEC standard or the International Electrotechnical Commission is a mathematical standard in modeling inverse curve characteristic of overcurrent relay or OCR. Using the IEC standard allows for inaccurate coordination of the relay such as the intersection of the inverse curve, because this standard does not always correspond to various problems in the industry. Accordingly, in this research paper suggests modeling an unconventional curve of OCR with adaptive neuro fuzzy inference system or ANFIS. The propose method is implemented for checking and training data with different kinds of membership functions or MF. Overload current is used as input data and OCR operation time is used as output data in the learning of ANFIS. The example data of OCR curve taken from relay coordination of Hess Indonesia Company. To obtain preliminary data, new data points are created on the R-VCB-09 curve which intersecting with R-VCB-11 curve. The simulation results of ANFIS output are compared with training data to get the optimum OCR curve design; thus, the ANFIS model can be implemented ideally in the industrial power systems. In the simulation result, ANFIS models using 5 trap-type MF results a less error percentage of 7,122204227%.