Research On The Optimization Of Algorithms For Precise Localization Of Rock Rupture Seismic Sources

With the continuous advancement of the deep earth exploration programme,the burial depth of underground projects is constantly moving towards 2000?3000m.Under high ground stress conditions,rock bursts and other engineering disasters are more likely to occur during construction,seriously threatening people's lives and property safety.Microseismic positioning as an effective means of rock burst monitoring and early warning,only ensuring the accuracy of the rupture source positioning can achieve accurate early warning of rock burst.The current research on seismic source localization is mostly focused on the algorithm itself and lacks analysis on the impact of different methods on localization accuracy from the aspect of sensor placement,which is precisely the easiest to be operated in engineering practice.At the same time,many new intelligent algorithms are still at the theoretical stage and have not been applied to the localization of seismic sources.Therefore,in view of the abovementioned problems,this paper analyses the impact of sensor placement on the positioning accuracy of traditional positioning methods and mainstream intelligent positioning algorithms,introducting new intelligent algorithms,and optimizing the new intelligent algorithm.The main research results and conclusions are as follows:(1)Among the traditional localization methods,two representative localization methods are investigated,source localization based on a single wave velocity model using linear solution and source localization based on a sectional wave velocity model using iterative search method in the confidence domain.It is found that the sensor placement method and the number of sensors have the most significant impact on the positioning accuracy,while the sensor layer distance has relatively less impact on the positioning results.When the sensors are arranged in a straight line,neither of the two calculation methods can provide stable and accurate positioning,while the positioning accuracy is significantly improved when the sensors are arranged in a staggered manner.According to the calculation analysis,it is suggested that the average positioning error can be reduced by more than 50% when compared to the in-plane staggered arrangement and double-layer arrangement by using three layers of arrangement or more.As the number of sensors increases,the combination of sensor arrangements selected for positioning increases rapidly,resulting in more possible positioning results.In this case,the abnormal values of positioning results should be eliminated to ensure the accuracy of the final positioning results when averaged.(2)A calculation method based on Snell's law to quickly obtain the minimum travel time is proposed in the rock mass of a single partitioned interface.Based on this,the degree of influence of the sensor arrangement on the localization accuracy of the seismic source in the genetic algorithm,particle swarm algorithm and simulated annealing algorithm is investigated,and it is found that the influence of the sensor arrangement on the localization results can be effectively overcome by using the intelligent algorithm compared with the traditional localization method.Through calculation and analysis,it is recommended that at least five sensors be deployed for locating the seismic source in practical engineering,which can control the average positioning error within 5m.(3)A new intelligent algorithm "Cuckoo algorithm" was introduced to study the localization of seismic sources.It was found that the degree of influence of sensor placement on the localization accuracy of seismic sources was the same as that of genetic algorithm and particle swarm algorithm: number of sensors > sensor placement.However,the cuckoo algorithm is simpler in terms of parameter setting and has a stronger global search capability,and can consistently locate the source with high accuracy by deploying only four sensors.With the same number of iterations,the cuckoo algorithm can control the average positioning error within 1m,which has good prospects for engineering applications.(4)The solution domain is optimised based on the idea of grid partitioning and grid node to time difference residual ordering,so that the newly introduced cuckoo algorithm can achieve accurate localization of earthquake sources in a large area without changing the number of iterations.And the study shows that the optimization method has strong generality and can be applied to any intelligent algorithm,providing a new way for high-precision localization of seismic sources in a large regional range.