Probabilistic power flow based on Improved Approximate Bayesian method

An improved approximate Bayesian combined with Markov Chain Monte Carlo method for solving probabilistic power flow (PPF) is proposed for the stochastic and intermittent nature of scenic power generation. When the prior distribution is not close to the posterior distribution, the acceptance rate of the samples is relatively low and the convergence rate is slow. The Delayed Rejection (DR) algorithm in the algorithm can improve the acceptance rate of the samples, speed up the convergence speed, and make the maximum posterior distribution closer to the true posterior distribution. Compared with the traditional approximate Bayesian calculation, the power flow results of IEEE 6-node system show that the Approximate Bayesian Calculation with Delayed Rejection has higher fitting accuracy and faster convergence speed, which can effectively reduce the posterior probability of state variables solving difficulty.