Risk Assessment And Emergency Capability Evaluation Of Power System In Uncertain Environments

With the rapid development of the modern economy,the scale of the power system is increasing,and the network structure is becoming more complex.The stable and reliable operation of the power system is facing increasingly severe challenges.During the operation of a power system,uncertain factors,such as load fluctuations,component failures,and extreme natural disasters,have a profound impact on the safe and stable operation of the power system.Scientific and reasonable risk assessment can effectively identify potential risks of the power system and reduce the possibility of the occurrence of large-scale power outages.On the other hand,because the risk of large-scale power outages in power systems always exists,in addition to improving existing power system risk assessment methods,it is also necessary to evaluate their emergency capability.Regular evaluation of the emergency capability of the power system can promptly find shortcomings in emergency rescue,which is of great significance for reducing the losses caused by disasters.This thesis has conducted discussion and research on the above issues deeply,mainly including the following four aspects:(1)A daily peak load forecasting algorithm is proposed based on dynamic time warping and gated recurrent neural network.Firstly,the daily peak load segment length is determined by the autocorrelation coefficient and human social activity cycle,and the shortest dynamic time warping distance is the target to match the most similar load segment in history.Secondly,the calendar information is encoded using some-hot encoding scheme to expand features effectively.Finally,the gated recurrent neural network is applied to forecast the daily peak load.The simulation results show that the dynamic time warping distance can not only capture the trend of the load curve change,but also find the local information of the load curve.Through comparison and analysis with different matching distances,encoding schemes and forecasting algorithms results,it is verified that the algorithm proposed in this thesis can achieve accurate daily peak load forecasting results.(2)A power system security risk assessment algorithm is proposed based on Latin hypercube sampling and daily peak load forecasting.Firstly,the algorithm proposed in the previous section is applied to forecast the daily peak load.In order to describe the state of the power system accurately,a Markov-based component failure model is constructed and the component state is determined by Latin hypercube sampling in the second place.Based on the results of Latin hypercube sampling,the optimal load reduction and security risk of the power system is calculated.Finally,the relationship between security risk and daily peak load is analyzed.The simulation results show that,compared to Monte Carlo sampling,Latin Hypercube sampling requires fewer samples to achieve the same accuracy.It can be found by observing the fluctuation of power system security risk and daily peak load that their trend of changing is basically the same but the amplitude is different.When the load level is low,there is a "compression" effect on the power system risk changes,and when the load level is high,there is a "stretch" effect.Therefore,accurate load forecasting can determine the risk trend of the system in advance,which is beneficial to the safe and stable operation of the power system.(3)A power system resilience assessment algorithm is proposed considering the identification of important components of the network and the disaster lifecycle.Firstly,the technique for order preference by similarity to an ideal solution(TOPSIS)method is used to identify the importance of nodes and lines by fusing the topological and electrical properties of the power system,and determine the repair order of power system components when resources are insufficient.Secondly,a new framework for resilience assessment including extreme disaster models,component failure models and maintenance models is constructed.Finally,a novel index based on Dy Liacco's security concept is applied to assess power system resilience.The simulation results show that the proposed resilience index overcomes the shortcomings of the traditional indexes that do not consider the duration of the system affected by the disaster and the maintenance time,and can reflect the power system resilience objectively.At the same time,the TOPSIS method can identify the importance of the nodes and lines of the power system effectively,and the maintenance strategy proposed accordingly helps to increase the resilience of the power system.(4)A power system emergency capability evaluation algorithm is proposed based on entropy fuzzy evidence reasoning.According to the risk assessment and emergency management theory,a power system emergency capability evaluation framework is constructed,including 4 primary indexes and 14 secondary indexes.In order to avoid subjective arbitrariness,entropy weight method is applied to obtain the weight of each index.Considering the lack of initial evaluation results,the fuzzy evidence reasoning algorithm is used to evaluate the emergency capacity of the power system.The simulation results show that entropy weight method can identify the most important influencing factors in power system emergency capacity effectively.Compared with other algorithms,the evaluation results obtained by the entropy fuzzy evidence reasoning algorithm have stronger discrimination.