In order to solve the problem that traditional methods can not get effective parameters, which leads to the low accuracy of network security risk quantification model, the optimization of network security risk quantification parameters is studied by numerical simulation method. In view of the host risk calculation, the risk vector is defined according to the host state, and the reasonable weight function is obtained by weighting calculation. The host direct risk value and the indirect risk value are combined to obtain the host risk value. Summing up the sum of the host risk values, calculating the mean value, and obtaining the average risk value of the network at a certain time. Aiming at the complex function in the risk value, a network security risk quantification model is established by introducing delay support vector machine method and described. Aiming at the important parameter penalty factor and kernel function width coefficient in the network security risk quantification model, ant colony algorithm is used to optimize the parameters. The optimal penalty factor is 10 and the optimal kernel function width coefficient is 0.112. The results show that the proposed method can quantify the security risk of host 1 in the network, and the results are consistent with the actual risk of host 1, and other hosts can get the same conclusion. It can be seen that the parameters of the proposed method have excellent performance and high accuracy in network security risk quantification.