The invention discloses an ABC strategy method suitable for stored grain digital supervision. The method comprises the steps of 1, determining a sampling range and a temperature measuring point of stored grain storage, and determining a temperature measuring sensor position in the sampling range and a grain temperature tested by a temperature measuring sensor; 2, building a fitting function F(i,j,k,d) of the temperature measuring sensor position and the grain temperature, wherein F(i,j,k,d)=A(i,j,k,d)j2+B(i,j,k,d)j+C(i,j,k,d); 3, calculating a deviation of the fitting function in the spatial position of the temperature measuring sensor and coefficients A(i,j,k,d), B(i,j,k,d) and C(i,j,k,d) of the fitting function when the deviation reaches a minimum value, and calculating a parameter D(i,j,k,d)and a parameter F(i,j,k,d); 4, judging and analyzing the coefficients and the parameters; and 5, summarizing abnormal conditions occurring in the step 4, calculating an abnormal proportion of a certain section in a granary and an abnormal proportion of the whole granary, and performing output.