Line loss plays an important role in evaluating the economic operation of a power system. Aiming at the complex relationshipbetween high voltage line loss and various characteristic parameters,considering the influence of gate voltage,gate current and temperatureand humidity on high voltage line loss,we propose a high voltage line loss prediction model based on improved sparrow search algorithm(ISSA) optimized BP neural network ( ISSA-BP) . Firstly,the position update methods of the discoverers and joiners of the sparrow search algorithm?( SSA) are improved through the Levy mutation strategy and the rotation strategy, respectively, and tested on sixbenchmark functions. The results show that the optimization ability of ISSA has been improved. Secondly,the optimal initial weights andoptimal initial thresholds are given to the BP neural network through ISSA,and then the relationship between the characteristic parameters and the line loss rate is fitted. Finally,taking the data of a 500kV high voltage transmission line in Shanxi Province as the researchobject,the prediction effects of five prediction models of BP,GWO-BP,WOA-BP,SSA-BP and ISSA-BP are compared and analyzed.The predicted value is the closest to the actual value,and its RMSE,MAPE,MAE and R2 are 4. 29% ,
3. 67% ,3. 57% and 99. 01% ,respectively,which are the best among various prediction models. Compared with SSA-BP,the RMSE of ISSA-BP decreased by 33. 4% ,MAPE decreased by?
36. 7% ,MAE decreased by 37. 1% ,and R2 increased by 1. 24% . It is showed that ISSA-BP model can accuratelypredict high voltage line loss.