Power load forecasting is particularly important for the planning and operation of the power grid

as it can help power grid companies to make reasonable scheduling of power resources

better balance power resources

and thus improve the economy and stability of the power system.This article proposes a multi-factor collaborative power load forecasting method based on Levenberg Marquardt(L-M)optimized backpropagation(BP)algorithm to address the shortcomings of existing algorithm optimized BP neural networks

such as poor accuracy and low convergence. On the basis of improving the BP algorithm

the L-M algorithm is used to integrate the steepest descent method and Gaussian Newton method to overcome the disadvantages of slow training

easy fall into local optima

and large prediction errors.Trainlm

Traingd

Trainrp

etc.

are used to train the L-M algorithm

considering factors that affect power loads such as temperature

air temperature

and date

and to predict power load in conjunction with cooling and heating loads.Finally

the effectiveness of the proposed method was verified using a residential community as an example.Compared with other methods

the L-M optimized BP algorithm for multi-factor collaborative power load forecasting performs well in solving nonlinear and complex forecasting problems with high accuracy

faster calculation speed

and smaller errors.