| Recently,to tackle the issue of engineering water shortage and rural drinking water safety in the Karst area,the state has largely invested in water conservancy projects.Furthermore,the proper resettlement of immigrants has become crucial to the success or failure of the project because the improper handling of resettlement problems causes serious social contradictions.When the selection of resettlement sites is unreasonable,it can easily force immigrants into poverty.Therefore,appropriate usage of the rich data resources from various units and departments and implementation of advanced technology and theoretical methods to realize a more scientific and reasonable selection of reservoir resettlement sites is an urgent issue that needs to be answered.Due to the background characteristics of the Karst mountainous areas,such as a fragile ecological environment,inconvenient transportation,and a highly fragmented land type distribution,the problems regarding the reservoir resettlement site selection in this area mainly include four aspects:(1)Site selection evaluation system and model level: quantify the multi-factor impact factors,and establish the evaluation index system and mathematical model for selecting reservoir resettlement site in the Karst areas.(2)Basic data level: collecting data on low-grade rural roads in Karst mountainous areas is difficult.As an important factor affecting the resettlement site selection,the lack of data seriously impacts the results of resettlement site selection.Thus,finding efficient and accurate methods for obtaining low-grade roads in Karst mountainous areas is an urgent problem to be solved for the selection of water reservoir resettlement.(3)Algorithm implementation level: the existing swarm intelligence algorithms heavily rely on setting the initial population;therefore,using an optimization method with an initial population setting suitable for the resettlement location of reservoirs in Karst mountainous areas is necessary.(4)Application level of the algorithm: due to the different sizes of the reservoir,inundation range,and resettlement scale,the applicability of each algorithm in reservoir projects of different sizes requires verification.In this study,we investigated reservoir resettlement using remote sensing multisource images and swarm intelligence modeling analysis to solve the reservoir resettlement site selection problem in the Karst mountainous areas.Our main contributions are as follows:(1)We proposed a mathematical model suitable for reservoir resettlement site selection in the Karst mountainous region to combat the problem of Karst quantitative evaluation of resettlement site selection of reservoirs in mountainous areas.Using the specificity of reservoir resettlement in Karst mountainous areas,we devised an evaluation index system suitable for reservoir resettlement site selection in Karst mountainous areas.Subsequently,we used the analytic hierarchy process to calculate the weight of influencing factors,establish the objective function,formulate constraints,and construct the mathematical model of resettlement location.This fills the gap of the lack of an evaluation system and location model in the resettlement reservoir locations in the Karst region.(2)By analyzing the collected data,we observe complex terrains in Karst areas with several curved and narrow rural roads.Moreover,the road data that significantly impact the resettlement site selection has some complications,such as the lack of lowgrade rural roads and the untimely updating of rural road data.Another important factor affecting the resettlement site selection is the traffic factor.Thus,we proposed an algorithm considering segmentation extraction and network optimization to extract the road network,which resolves the problem of neglecting many small sections of roads in the optimization process when the traditional two-step method is used to extract roads from Synthetic Aperture Radarimages.We employed the Bayesian framework to transform the road network extraction problem into the joint reasoning of piecewise-extracted likelihood value and network optimization priority.Subsequently,based on the optimization of a multi-scale linear feature detector(MLFD)and Beamlet network,a conditional random field(CRF)was used for joint reasoning,and the road was extracted by fully considering the characteristics of curved and narrow roads in the Karst mountainous areas.This method efficiently generated the road data required for the experimental area and provides a data guarantee for the algorithm implementation in Chapters 5 and 6.(3)We proposed an optimization method of initial population setting using a spatial clustering algorithm,which solves the problem of the algorithm falling into local optimization too early when the initial population setting of reservoir resettlement location is unreasonable.Combined with the characteristics of fragmentation of land class distribution in the Karst mountainous areas,the spatial entities were clustered and analyzed using “land class aggregation.” Then,the“internal difference” was conducted for each land class using the area and weight to obtain a reasonable initial population to improve the accuracy of the swarm intelligence algorithm.Next,considering five commonly used swarm intelligence algorithms as examples,we optimized parameter setting adjustment,particle representativeness(particle swarm optimization algorithm),dynamic update adjustment(Gray Wolf algorithm),pheromone update(ant colony algorithm),and adaptive adjustment probability strategy(cuckoo algorithm).Finally,we conducted an experimental test considering the Jiayan water control project.After the optimization,the location results and algorithm performance of the five algorithms were improved,and the feasibility of the optimization algorithm was verified.(4)The impact of scale differences on the results of reservoir resettlement site selection was verified,and we discussed the applicability of different evaluation indexes and optimization algorithms on different scales,respectively.From the influencing factors of the scale of hydraulic engineering,the reservoir resettlement location problem of different scales was transformed into the location problem of different scales.Furthermore,we adjusted the objective function and constraints in the location model proposed in Chapter 3.Through experimental analysis,we verified that different swarm intelligence algorithms have different applications on different scales,which provides a reference for the accurate application of swarm intelligence algorithms in the resettlement location of hydraulic engineering. |
PDF Full Text Request