Research On The Decision Making Of The Startup Sequence Of Generators After Blackout Based On Monte Carlo Terr Search Algorithm

Blackout will not only threaten the safety of people’s lives and property,but also cause social unrest and even endanger the stability of the country.Practice has shown that,due to the influence of various unpredictable factors in reality,such as natural disasters,man-made damage,and hidden dangers of the system itself,the power system blackout cannot be completely avoided.Therefore,it is necessary to conduct research on the restoration of the power system after a blackout.Among the various problems of system restoration after a blackout,the decision making of the startup sequence of generators is the core issue.Its purpose is to determine the startup sequence of the non-black-start generators in the power system and the corresponding power restoration path to ensure that the grid is gradually restored to stability state,lay the foundation for subsequent large-scale load restoration.The restoration plan formulated offline based on the preset scene can provide the optimal or nearly optimal restoration strategy.As it is difficult to consider the impact of unpredictable factors in practice,it may not be possible to effectively guide the decision making of the startup sequence of generators based on the restoration plan alone.It is necessary to make an online decision on the startup sequence of generators based on real-time data,which complements the offline plan to ensure the smooth restoration of the generators.This paper focuses on the decision making problem of the startup sequence of generators,and on the basis of existing research results,uses artificial intelligence technology to study the offline and online decision-making methods of the startup sequence of generators after a blackout.Considering the application scenarios of whether the real-time status information of the system can be obtained,the offline decision-making method and the online decision-making method of the startup sequence of generators are studied.The main contents of the paper are as follows:(1)The optimization model of the decision making of the startup sequence of generators is studied.By analyzing the power output characteristics of the black-start generators and the non-black-start generators,the output power expressions of generators in different restoration phases are obtained.Based on this,an optimization model for the decision making of the startup sequence of generators is constructed.Among them,the objective function is to maximize the generating capacity of generators.(2)Proposed an offline decision-making method for the startup sequence of generators based on Monte Carlo tree search algorithm and Dijkstra algorithm.Analyzed the mutual influence between the startup sequence decision and the power supply path restoration decision.In the decision-making process,both were considered,the Monte Carlo tree search algorithm was used to determine the restoration sequence of the generator set,and Dijkstra algorithm was introduced in the Monte Carlo simulation process.Proposes an offline decision making method for the startup sequence of generators.The IEEE 39 bus test system and a local actual power grid in a certain province are used as examples for simulation tests.The results verify the feasibility of the proposed offline decision-making method.(3)An online decision-making method for the startup sequence of generators based on improved Monte Carlo tree search algorithm is proposed.This paper proposes the robustness constraints of the restoration plan to limit the risk of the generator restoration delay caused by the uncertain factors in the actual restoration,and constructs an online decision-making model based on the previous foundation.An improved method of Monte Carlo tree search algorithm is proposed,which improves and optimizes the upper confidence interval,default strategy,backpropagation,etc.,and increases the calculation speed under the premise of ensuring the accuracy requirements to adapt to the characteristics of online applications.The IEEE 39 bus test system is used for simulation testing,and the results show the effectiveness of the proposed online decision-making method for unit recovery sequence,and in the case of unpredictable failures during the restoration process,the subsequent restoration strategy can be quickly optimized and adjusted.