SVM-PSO混合算法的光学电压传感器内电场优化

对光学电压传感器内电场的优化多采用有限元计算结合穷举搜索，无法反映电场分布与传感结构参数之间的非线性映射关系，且计算与搜索费时、效率低，容易落入局部陷阱。提出基于支持向量机(SVM)与粒子群(PSO)的电场优化混合算法，通过PSO优化SVM的参数以建立电光晶体内电场分布的模型，再由PSO对模型求解得到最优电场分布。以介质包裹结构为例，构建包裹介质的介电常数、厚度以及长度与晶体内电场分布之间的非线性映射关系，且求解过程能够有效地避开局部陷阱，优化后晶体内电场均匀度提高了20.8%、训练时间节省了89%。最后通过实验验证了模型的有效性。

Electric field optimization in longitudinal modulated optical voltage sensor based on SVM-PSO algorithm

The electric field optimization in the optical voltage sensor (OVS) is mostly based on the finite element calculation and the exhaustive search, which is unable to map the nonlinear mapping relationship between the electric field distributions and the sensing structure parameters, and it is time-consuming, inefficient, and easy to fall into local traps. Therefore an hybrid algorithm based on SVM (Support Vector Machine) and PSO (Particle Swarm Optimization) is proposed in this paper. The parameters of SVM are optimized by PSO to establish an electric field model for the electro-optic crystal, and then the model is modulated by PSO to achieve the optimal electric field distributions. The medium enwrapping structure is taken as an example and the nonlinear relationship among the dielectric constant, the dimension of the wrapping medium and the electric field distributions in the crystal are established, and meanwhile the solutions can effectively avoid falling into the local traps. The uniformity of electric field distributions is increased by 20.8% and the training time is saved by 89%. Finally, the validity of the model is verified by experiments.