Obtaining Stress Spectrum And Predicting Fatigue Life Of Tower Crane Based On Integrated Learning Algorithm

With the rapid development of China’s market economy,the lifting machinery industry has entered a period of rapid development.At the same time,safety accidents inevitably occur in lifting machinery,and the frequent occurrence of these accidents seriously affects people’s life and property safety.According to statistics,in the national special equipment safety accidents,the incidence rate of lifting machinery accidents is 26.36%,and the mortality rate is 30.30%,ranking second.The occurrence of these accidents is mostly due to fatigue failure of the tower crane structure,so it is particularly important to analyze the fatigue life of the tower crane.The foundation of fatigue life analysis is the acquisition of tower crane load spectrum.Currently,there are two main methods for obtaining tower crane load spectrum: on-site measurement and finite element analysis.The data obtained from on-site measurement is more realistic,but it will consume a lot of manpower and material resources.Although the finite element analysis method is relatively fast,the authenticity of the data needs to be improved.This article proposes a method of ‘finite element analysis + fixed sample database + integrated learning algorithm + on-site measurement’ to quickly obtain load spectra.The method uses finite element analysis technology to obtain equivalent stress data of tower cranes,and establishes a fixed sample database based on the optimized load step size and position movement step size of the trolley lifting.The integrated learning algorithm is used to construct a tower crane equivalent stress prediction model,Based on the on-site measured working conditions data,quickly predict the time stress curve of the dangerous point of the tower crane,and finally build a life prediction platform to quickly obtain the fatigue life of the tower crane.The specific research work is as follows:(1)Taking a tower crane in service as the research object,the finite element model of the tower crane is established by using Py Ansys,a secondary development product of Ansys.And its statics and dynamic analysis is carried out to determine the location of the dangerous point of the tower crane and obtain the equivalent stress data of the dangerous point location.(2)Three classical algorithms of ensemble learning are analyzed and compared,and random forest is selected as the algorithm to build the equivalent stress prediction model.By establishing equivalent stress prediction models for different lifting load steps and trolley position movement steps,it is determined that the lifting load step size is 0.1 t and the trolley position movement step size is 100 mm.And establish a fixed sample database based on this step size data.(3)Establish an equivalent stress prediction model for all hazardous points of the tower crane using data from a fixed sample database.Incremental learning is adopted to improve the prediction accuracy of the equivalent force prediction model for a certain danger point when its prediction accuracy is lower than that of other danger points.Based on the on-site measured operating data,the time stress curve of the tower crane can be quickly obtained using an equivalent force prediction model.(4)Due to the high cycle fatigue type of tower cranes,the nominal stress method is used to estimate their fatigue life.Build a life prediction platform using Py Qt5 to quickly obtain the time stress curve and fatigue life of the tower crane.