Research On Distribution Network Risk Assessment Method Based On Fuzzy Mechanism

Adding renewable energy to the distribution network has become a major goal of China’s energy planning policy.Wind and photovoltaic power,which are representative sources of renewable energy,exhibit significant intermittency and randomness.As renewable energy sources are being integrated into the power grid,the uncertainty in system operation increases,which has made risk assessment of distribution networks a current research focus.Traditional methods rely on expert groups for risk decision-making,but nowadays machine learning methods are mainly used to train models for risk prediction,aiming to ensure safe and efficient operation of the distribution network.Currently,most of the risk assessment methods for distribution networks have the following problems: firstly,the risk assessment of power distribution network is a significant decision,and the ultimate decision-maker should be a human.However,most existing methods only evaluate risks through machine models without a suitable approach to fully consider the evaluation opinions of both humans and machine models.Secondly,the evaluation results of existing machine models are deterministic values that cannot be integrated with human evaluation opinions that have fuzziness.To address these three issues,we proposes a fuzzy prediction model that converts machine predictions into picture fuzzy values for integration with human opinions.Combining with the picture fuzzy mechanism,a complete and reasonable prototype system for distribution network risk assessment is proposed to meet practical needs.The research presented in this paper mainly focuses on three aspects:(1)A fuzzy prediction method that combines fuzzy mechanism with machine learning methods has been designed.This method imitates human thinking patterns and uses machine learning models as a basis to train risk prediction models and fuzzy prediction models through data training.By using the risk values and fuzzy values predicted by the two models,the final prediction result is transformed into picture fuzzy values.In addition,comparative experiments were conducted to verify the feasibility of this method.(2)An improved and novel scoring function for picture fuzzy values has been designed.Firstly,the existing functions were analyzed,which may require the use of precision functions when comparing picture fuzzy values,and have defects in calculation results that contradict basic theory and intuition.Then,a new scoring function was designed based on the degree of influence of positive,neutral,and negative membership degrees on the change in function value.The differences in properties between this function and existing functions were compared,and some properties of this function were analyzed and demonstrated.Finally,a multiple attribute decision making method using this function was designed,and the effectiveness of the designed function in practical applications was compared and verified through three sets of examples.(3)A risk assessment method for distribution networks under picture fuzzy mechanism has been designed and proposed.Firstly,the fuzzy prediction method was used to combine the picture fuzzy mechanism with machine learning methods to obtain a risk warning model with predicted results in the form of picture fuzzy values.After the model warns of risks,experts are notified to conduct risk assessments.By using fuzzy theory to aggregate the picture fuzzy values representing expert opinions and model predicted results,the final risk assessment value is obtained to determine whether there is a decision-making risk.Through simulation experiments using a dataset provided by Guangxi Power Grid,this method was compared with methods that solely use machines or expert groups for evaluation,proving its effectiveness and reliability.