Substation planning is an important component of urban power networks. In this paper, a new mathematics model for unified substation planning of two voltage levels is presented, which makes the total cost of two voltage levels as objective function and the geographical information as restrictive condition, introduces the penalty function to ensure the uniformity of the load distribution. Bacterial Colony Chemotaxis (BCC) algorithm is one of the new heuristic algorithms based on colony intelligence. In the paper, parameters control strategies of the BCC algorithm are studied, parameter control mean is improved, and the algorithm is applied to the problem of two voltage levels substation planning. Large amount of experiments show that the improved model of the unified substation planning whose objective function value excels the original model has good convergence. The improved BCC algorithm applied to substation planning problem has good performance of global convergence and better convergence speed.