Models And Algorithms Study For City Emergency Service Facility Location

City is the most centralized place of population flows, commodity flows,capital flows, technology flow and information flow. Since it is affected by many factors such as climate, geology, and social factors, city is facing the challenge of emergencies all the time. Especially with the rapid development and expansion of a city, the occurrence frequencies of all kinds of unexpected events are increasingly, thus the demands for emergency services continue to generate and change, and it is required to strengthening the cities’ emergency management capability to respond to emergencies. Proper layout of the city emergency service facilities is the basis of improving city emergency management capacity. However, the coverage domains of many cities’ existing emergency service facilities have been difficult to meet the emergency requirements for effective emergency management. How to transform original layouts of the city emergency services facilities in old districts and build necessary emergency service facilities for reconstructed new city district while optimizing the layouts of new facilities is urgent to be solved.After analyzing and combing theories of public emergencies and city emergencies management in domestic and foreign, the thesis finds that many existing emergency service facility location models have many problems of addressing the fairness, efficiency and economy of location. Furthermore, the performance of many model solution algorithms still needs to be improved. The thesis analyzes the characteristics of urban emergencies and focuses on the emergency service categories and classification issues. Moreover, this thesis proposes location characteristics of urban emergency services and location basic requirements through combining modern urban development, construction and operation characteristics. Later, it analyzes objectives, principles and influencing factors of the emergency service facility location, and proposes fairness,Efficiency and economic issues of emergency service facility location models construction, which builds foundations for the following construction and solution of location models.Next, the thesis discusses the construction and solution problems of emergency services facility location models. For the fairness of city emergency service facility location, the thesis points out that the fairness can be realized by minimizing the maximum distance between city emergency service facilities and service demand points and constructs a modified P-center model of city emergency service facility location. To solve this model, the thesis designs ASC and ASCFLP algorithms by introducing non-random sampling strategy and spectral clustering thought. Experimental results on many data sets verify that ASC and ASCFLP algorithms have advantages over k-means and NYSC algorithms.The thesis puts forward that the efficiency of location problem can be realized by maximizing the coverage scale of emergency service facilities. It builds a modified maximum coverage model by taking the number of service demand points covered by emergency service facilities as an objective function.It also designs clustering ensemble spectral algorithms, ESC and ESCCFLP, to solve this model. Test results on UCI data sets validate the ESC’s superiority over k-means and KBLANM algorithms. Meanwhile, experimental results on GIS urban demand points show that ESCCFLP algorithm can effectively improve the service efficiency of emergency services facilities, which means that the designed model and proposed solving algorithm can provide auxiliary location decision support for building efficiency emphasizing urban emergency service facilities.On the basis of taking into account of both economy and efficiency principles of city emergency service facility location, the thesis designs a global optimal dual objective decision-making model. Since the spectral clustering algorithm has the advantage of executing multi-way partition and commute time embed has the same idea of clustering as traditional spectral clustering algorithms, the thesis designs commute time embed clustering algorithms(CTEC and FLPCTEC) to solve the above model. Experimental results indicates that CTEC and FLPCTEC algorithms can perform better than k-means andSCCAT algorithms when partitioning areas, and FLPCTEC algorithm can obtain a global optimal solution of the facility location problem.The example verification indicates that the models and the algorithms proposed is feasible and could be applied widely to enhance the fairness,efficiency and economic issues of emergency service facility location.