While meeting the needs of network traffic, dynamic deployment of VNF is the key to cost optimization of NFV system. In addition to considering server resource capacity and dynamic user traffic, the dynamic expansion strategy also needs to strike a balance between the operating costs of VNF, the cost of maintaining VNF instances, the cost of VNF deployment, and the additional cost of servers placing new VNF instances. In this paper, the problem is modeled as a 0-1 integer linear programming model with quadratic terms, and a new linear programming method is used to solve the problem. The algorithm reduces the complexity of the algorithm to a great extent by reducing the number of variables in the constraint. Through the experimental simulation, we prove the effectiveness of the algorithm and reduce the cost of VNF-FG scaling.