Construction Height Measurement Method Based On Kalman BP Network

High suspended platform has been widely used in tall buildings construction.Due to its characteristics of flexible and convenient operation,loading and unloading easily,gondola has been applied in construction works such as curtain wall construction and maintenance,thermal insulation construction,doors and windows installation and so on.Gondola has been a common non-permanent machine in buildings construction,and there have been more than 1.5m units in our country.Because of its unique friction lifting method,rope broken/dumping/falling and other construction accidents are easy to happen,constant online monitoring to gondola by advanced Internet of things(IoT)has been a new trend to solve safety accidents in gondola construction.The construction height is an important data for monitoring system.The purpose of this study is how to collect and get the construction height data.There are two usual way to get height data in production practices.The height value can be obtained by converting the pressure data obtained by pressure sensors,but not keep long-term effect because of the change of the pressure data.The other way is to use the GPS algorithm to derive the height data of the gondola,which is easily affected by the occlusion of the signal,and interference factors in construction site.Therefore the big error for height value is easy to generate if one of the two methods is used due to the various uncertainties of the error factors.Based on the properties of measurement error,in this thesis Kalman filtering model and BP neural network,as well as the statistical theory of data are all used to process the construction height measure data of gondola.Detailed research works are as follows:(1)The pressure and the GPS height measurement principle were analyzed,and the errors of each height measurement method in practical application were studied.(2)Based on principles of mathematics and statistics,measurement data was processed.The gross datas in the measurement results were excluded by the Bessel equation to reduce the error.(3)Based on Kalman filtering theory and BP neural network theory,Kalman BP neural network error correction model was established,which was used to process the height data to reduce the error.The optimization of Kalman filtering parameters and BP neural network structure parameters was proposed.(4)According to the characteristics of error correction and the actual construction conditions of the powered suspended platform,the experiment was designed to test and verify the Kalman BP neural network error correction model.The pressure/GPS fusion altimetry method was improved to reduce the error existing in the process.The theoretical analysis and the error correction were proved to be helpful in improving the stability and accuracy of the measurement data.