Anchor Rod Corrosion Prediction Based on PSO-LSSVM

According to the corrosion data of the anchor rod of a transmission line tower in western Hubei, the correlation between soil factor and rod corrosion was analyzed by grey relational degree algorithm. The key parameters of least squares support vector machine (LSSVM) were optimized by particle swarm algorithm (PSO), and the data were processed by the grey correlation degree weights. The PSO-LSSVM prediction model and the PSO-LSSVM prediction model considering the grey correlation degree weights were established separately. The example showed that the mean square error of the PSO-LSSVM model training set was decreased by 15.3% and the mean square error of the prediction set was decreased by 35.71% compared with the LSSVM model. When the weight of grey correlation degree was considered, the mean square error of the training set and the predictive set in the PSO-LSSVM prediction model decreased further, and the value reduced 24.59% and 20% respectively. The PSO-LSSVM was used to predict the corrosion of anchor rods with relatively good precision, and the PSO-LSSVM prediction model considering the grey correlation degree weights was more accurate.