As for the fact that random factors or convergence problem in the traditional reactive power compensation

a kind of compensation scheme based on multi knapsack problem is proposed in this paper. Firstly

reactive power deficiency of each substation is calculated

then

the reactive power balance progress is abstracted out as a multi knapsack problem. In that way

the substation with and without extra reactive power is defined as knapsacks and subjects respectively. The optimal solution is obtained by way of genetic algorithm and the reactive power balance result is obtained by way of decoding. Finally

as for the substations still having reactive power deficiency after balance

the reactive power compensation supplementation is made as per the principle of handy regression. In this paper

the multi knapsack combinatorial optimization problem is introduced into reactive power compensation analysis for the first time

and intelligent algorithm is used to increase analysis efficiency of reactive power balance and reduce the workload of reactive planning. It is verified by calculation analysis that this solution is feasible and has reference and application value for the reactive optimization management of grid.