Research On Building Settlement Prediction By Chaotic Particle Swarm Optimization Combined Least Squares Support Vector Machine

According to incomplete statistics in China,all types of buildings built in the middle of the 20th century will cause building subsidence due to the changes of years,which will lead to tilting,cracking and other building safety problems.And in these old buildings,there are many buildings that are not allowed to be arbitrarily altered or rebuilt.For these buildings with safety problems,if they are not inspected and repaired in time,it is easy to cause safety accidents.In recent years,with the overall development of the construction industry,the number of urban buildings is also increasing.However,the land supply is also decreasing due to the increasingly reduced layout of urban planning.In order to save land resources,many high-rise buildings have appeared in major cities across the country.Some high-rise buildings will cause different degrees of settlement due to excessive load during construction,and some high-rise buildings will cause different degrees of settlement due to various external reasons such as the change of underground water level and adjacent new buildings during operation.In this case,how to effectively predict the settlement has become the primary problem to be solved in the construction field.(1)The commonly used method of building settlement prediction in China,but there is a problem that the accuracy of prediction parameters is not high.For this reason,this paper proposes to combine particle algorithm and support vector machine algorithm,and optimize the support vector machine prediction by using the advantages of high particle fitness and strong search ability.Based on the relevant parameters,a CPSO-based building settlement prediction model was established.(2)Combined with the measured data of settlement in a certain exhibition center in Shandong,and taking advantage of the phase space reconstruction of chaos theory,the learning samples are obtained through the established phase space structure.Using Matlab R2016b as the platform,libsvm is used for programming extension,and compiled with Microsoft Visual Studio 2013.The single support vector machine prediction model and the optimized support vector machine prediction model are trained.(3)In the computer simulation environment,the mean square error,square sum error and average absolute error are used as subjective and objective evaluation indicators to evaluate the building settlement established by least squares support vector machine,chaotic particle swarm optimization and chaotic particle swarm optimization combined least squares support vector machine.The prediction performance of the prediction model,the experimental results show that the mean square error,square sum error and average absolute error of the building settlement prediction model based on CPSO-LSSVM are better than the traditional model,and the robustness is higher than the prediction model established by other algorithms.It can meet the requirements of building settlement prediction for its accuracy.It is proved that the least squares support vector machine prediction model based on chaotic particle swarm optimization algorithm can track the change trend of building settlement and has high applicability in the construction industry.