With the increasing number of temperature sensitive load such as air conditioning, electric heaters has become available, the relationship between temperature and power load is more and more significant and complex. An example is given to illustrate the effect of temperature accumulation effect on load change. On this basis, the influence of temperature on load on the day to be analyzed and its premier two days is also considered, and a modified temperature model considering temperature accumulation effect is proposed. Based on Sigmoid curve and quantization of date type, date serial number, a load-temperature fitting mathematical model is established. The undetermined coefficients in the two models are solved by simulated annealing with the maximum load-temperature similarity and the minimum global fitting error as the objective functions. The actual load and temperature data of some cities in a province are used for testing and verification. The results prove that the load - temperature fitting model proposed can have good fitting accuracy in the whole temperature range, and the fitting error is further reduced after the modified temperature is adopted. Sensitivity curve of temperature to load change has been obtained respectively through the load-temperature fitting model using the actual temperature and the corrected temperature, including the sensitivity of the modified temperature curve for more comprehensive quantitative history two adjacent temperature change on the influence of load, research results for more sophisticated,more accurate quantitative temperature change on the influence of load change provides a sufficient theoretical basis.