Distribution Network Structure Based On Plant Growth Simulation Algorithm

At present our country urban distribution network exists many problems such as capacity lack, unreasonable layout, line loss on the high side, and substandard power quality. Distribution network planning work is the effective mean to solve these problems. This paper adopts comprehensive cost minimum model, this model considers the annual investment cost, depreciation cost, maintenance cost and loss cost of electricity, the constraint conditions include voltage drop, equipment capacity, ratio of power capacity and load for transformer, connectivity, radial network. Which is the most comprehensive model based on economic optimum target. This paper studies various algorithms in distribution network planning application, these algorithms have slow convergence speed , bad search direction, and uncertain parameters, so this paper proposes the Plant Growth Simulation Algorithm and the improved Plant Growth Simulation Algorithm for distribution network planning. Some existing bionic algorithms have fell into local optimal solution due to some parameters are difficult to be determined or the search direction lacks of leading, but the Plant Growth Simulation Algorithm has avoided this problem ,because this algorithm fully considers the intelligent factors in the process of plant growth, and uses a random and directional search mechanism. According to the radial distribution network structure characteristics, this paper uses basic loop thought on the coding of line, takes basic loop set as decision variable, which greatly reduces the dimension of the variables, and avoids the checking process for radial network structure, therefor improves the efficiency of the algorithm. Aiming at inefficient computation in solving large-scale network planning, this paper puts forward the improved algorithm, which puts the growing point scale under the control related with the network scale. This article selects 3 times of planning network line number as the largest scale of the growing point set, when growing point set scale is more than this number, the algorithm sorts the objective function values, reserves the new growing points and some better growing points generated before, and keeps the growing point set scale under the provision, which ensures the diversity and optimality of the growing point. This paper validates the plant growth simulation algorithm with a 10 nodes example, and compares the results with the Genetic Algorithm and Particle Swarm Optimization, the results show that the plant growth algorithm has better direction search mechanism, which can find the optimal solutions with fewer iteration times. This paper validates the improved Plant Growth Simulation Algorithm with a 54 nodes example, and compares the results with the Plant Growth Simulation Algorithm, the results show that the improved Plant Growth Algorithm is much more effective, this effect will be more obvious if the network scale increases.