Fault prognosis is the key technique of PHM(Prognostic and Health Management). Aimed at the difficult problems of parameters selection and setting in fault prognosis of electronic equipment, the paper proposes a new method of parameters selection and setting in fault prognosis. Firstly, the parameter set which can reflect fault state of electronic equipment is extracted by using the multi-signal flow graphs model. By introducing the relative risk of biostatistics, the optimized parameters of fault prognosis are selected by using the data-driven method. The prognostic parameter threshold in the confidence level of 95% is got by building statistical model trend graph. The experiment validates that the optimized prognostic parameters can be quickly and effectively selected and set by the proposed method. Simultaneously the complicated analyses of fault mode, fault state, fault diagnosing bases and the influence pf subjective factors are avoided.