The invention discloses a power grid optimization partitioning method for new energy consumption analysis. The method comprises the following steps: S1, determining a unit sub-region power grid Ui; S2, collecting and making statistics of a new energy historical generated output data sequence and eliminating abnormal data Pre, i(t); S3, calculating accumulated generating capacity Ere, i of all newenergy power stations; S4, collecting historical power load data exceeding Pload, i(t) and removing the abnormal data; S5, calculating accumulated power consumption Eload, i of the power load; S6, calculating matched power generation power Pmatchre, i (t) and matched power utilization power Pmatchload, i(t) at each moment; S7, calculating the total match power generation electric quantity Ematchre,i and the total match power utilization quantity Ematchl, i; S8, calculating a time sequence match degree Kmatch, i; and S9, classifying source-load match degree coefficients Kmatch, i of all unit sub-regions. The optimization method, through long-time statistical data analysis, adopts a key characteristic index of the time sequence match degree to measure the match degree of the generated outputand the power load of the photovoltaic power station on the time sequence, and provides a basic support for formulating planning schemes of site selection and constant volume and the like of distributed photovoltaics.