Because of the large number and amount of investment projects in power grid enterprises, investment decision-making is a relatively complex task, which needs to weigh many factors and objective functions and consider how to allocate the limited funds to different construction projects. By introducing the concept of portfolio optimization, this paper takes the maximization of economy, security and sociality as the objective function, considers the constraints of power demand, reliability, enterprise investment capacity and other resources, and constructs optimal decision-making of multi-project portfolio investment for power grid enterprises. Cuckoo search algorithm (CS), as a new heuristic algorithm, has high performance in solving optimization problems. In this paper, stochastic weights are introduced to dynamically optimize cuckoo algorithm, which can further improve the optimization performance of the algorithm. Combining with practical examples and using MATLAB tools, the operability and practicability of the model in multi-project portfolio optimization of power grid are verified.