Research On The Methods Of Locating The Hotspots And Estimating The Dose Rate Of UAV Nuclear Radiation Monitoring

When the large-area radioactive contamination is caused by a nuclear and radiation terrorist attack or a major nuclear and radiation safety accident,the unmanned aerial vehicle（UAV）platforms equipped with nuclear radiation detection system can be used for nuclear radiation reconnaissance at low and medium altitude and obtain information on the intensity and distribution of radioactive pollution in the affected area.It is extremely important to accurately determine the state of the accident and its development situation,which is the bases for leaders’command decisions.In the past two decades,UAV platform technology and airborne nuclear radiation detector have developed rapidly.However,there have been relatively few research works and achievements focusing on simulation modeling,hotspot localization algorithms,and dose rate estimation for UAV-based nuclear radiation monitoring systems.These are critical factors determining whether UAV-based nuclear radiation monitoring systems can be"fast and accurate"in their measurements.Therefore,this thesis focuses on the core issues of surface radioactive hotspot localization and surface dose rate estimation in UAV-based nuclear radiation monitoring.From the practical application requirements,various technical means such as numerical simulation,Monte Carlo simulation,and experimental measurement were comprehensively employed to systematically study the search and localization methods for single and multiple radiation sources under different scenarios.Additionally,a UAV-basedγ-radiation monitoring surface dose rate estimation method was established,providing key technical support for promoting the deep application of UAV-based nuclear radiation monitoring technology in military and civilian fields.The main research contents and conclusions of this paper are summarized as follows:（1）Research on the single radioactive source localization methods in large area open environment.Five existing single radioactive source localization methods,such as the trilateral localization method and parameter estimation method,were studied in depth,and methods for improvement were proposed.a)The improved trilateral localization method could provide the probability distribution of radiation source parameters,and the results were stable and reliable.The localization error was less than 0.1 m within a range of 20 m×20 m.b)The optimal sampling range and threshold of the parameter estimation method and cross localization method were studied under various measurement conditions.A combination localization method based on optimal parameters was proposed,which achieved the minimum localization error during the outdoor experiment of UAV-based radioactive source location.Compared with the cross localization method,the accuracy of the proposed method increased by 49.6%,and the time consumption relative to the parameter estimation method was reduced by 71.1%.c)Using the Bayesian estimation method and Markov Chain Monte Carlo method,accurate localization of radiation sources was achieved in open areas.The numerical and MCNP simulation localization results were all less than 1 m when the area size was60 m×60 m.The minimum localization error during the actual UAV search process with an area size of 200 m×250 m was about 3 m.The impact of different factors such as the number and distribution of measurement points and the intensity of radiation sources on the localization results was analyzed.d)Two dynamic planning methods based on Bayesian estimation and maximum likelihood estimation were explored.The former took only 1%of the time required by the latter when the localization results are comparable and thus had a significant advantages in terms of real-time performance.（2）Research on the multi-radiation source localization method in large open areas.To address issues such as the coexisting of multiple radiation sources and uncertain distribution of position and intensity,two localization methods based on nonlinear least squares optimization and response matrix deconvolution were proposed.The former method efficiently and accurately located a typical multi-source radiation field by establishing an objective function and incrementally increasing the number of radiation sources.The latter method discretized the ground area into grids,established a response matrix,and performed deconvolution to obtain the position and activity information of the measured radiation source.When using the inverse square law to construct the response matrix,the position estimation error did not exceed one grid size,and the activity estimation error did not exceed 20%.With further modification of the response matrix using MCNP simulation data,the activity error was reduced to below 5%.（3）Research on the method of the radioactive source localization in shielded environments based on Markov Chain Monte Carlo.For the problem of locating the radioactive source when the shielding exists in the search area,a radioactive source localization algorithm based on the Markov Chain Monte Carlo method was proposed.The algorithm first treated the attenuation coefficient of the shielding material as an unknown parameter and constructed a Markov chain during the random walk process.The radiation source could be located with minimized the computation time through the real-time judgment of whether the Markov chain has converged or reached a stable state.Through simulation data and measured data validation,the proposed method would accurately locate the radiation source in the presence of shielding materials and greatly reduce the computation time while ensuring positioning accuracy.In a laboratory environment of 5.4 m×2.4 m,the localization error was less than 0.1 m,and the computation time was reduced to 10%.（4）Research on the reconstruction of surface radioactive distribution and dose rate estimation based on deconvolution.As regards the reconstruction of surface source radiation distribution and dose rate estimation,an algorithm based on spectral information was proposed.The algorithm accurately reconstructed multiple types and distributions of radioactive nuclides and estimated the dose rate 1 m above the ground by deconvolving the response matrix and introducing energy window-type spectral information to improve the accuracy of the deconvolution results.The feasibility and effectiveness of the method were verified through specific examples of single nuclide(¹³⁷Cs)and multi-nuclide(¹³⁷Cs and ⁶⁰Co)surface sources.The results showed that the cosine similarity between the estimated surface radioactive distribution and the true value were 0.9950 and 0.9965,respectively,demonstrating that the proposed reconstruction algorithm could effectively distinguish between multiple radioactive nuclides and restore their surface radioactive distribution.The radiation source localization method and the deconvolution-based surface radioactive distribution and dose rate estimation method using spectral information in this study can provide key technical support for UAV-based nuclear radiation monitoring systems to quickly determine the surface radioactive material distribution and accurately estimate the surface dose rate,enabling timely nuclear emergency response actions.