Abstract: This research aims to design a maximum power point tracking algorithm which tracks the global maximum among multiple local maximums. Multiple local maxima can be exhibited on the power-current characteristic curve under non-uniform insolation conditions. The research was based on a significant number of data simulations of photovoltaic cell and the discovery that PV panels' P-I curves obeied the following rule. The values of local maxima monotonically decreased with respect to the distance from the global maximum. Moreover, statistical studies had been conducted on the slopes of the curves, distances between two local maxima, etc. Based on these observations and discoveries, the algorithm was thus designed. The algorithm obtained the location of the global maximum of a P-I curve by partially scanning the curve, during which Regional Scanning Mechanism was adopted to enhance the precision and efficiency of the algorithm. In addition, another two simple and fast PO algorithms were compared, their appropriate uses were discovered and be applies smoothly in real practices. The design of this algorithm also ensures reasonable time complexity, efficiency and guarantees that no maxima will be ignored. It is also equipped with an interesting property such that the more the maxima a P-I curve possesses, the faster the algorithm is correspondingly.