Research On Statistical Modeling Methods For Dimensional Error Of The Hull Block Multi-stage Construction Processes

The hull block consists of hundreds of parts,with a huge size and complex structure.Its construction involves complex processes such as cutting,welding,hoisting,outfitting and so on,including parts processing,sub-assembly welding,hull block welding and other processes.During the multistage manufacturing process,due to the thermal effect of parts cutting or welding,geometric nonlinearity and other complex factors,the dimensional error and form-position error of intermediate products in double-deck block construction processes are constantly generating and evolving.which makes it hard to control the accuracy of the hull block.In the aspect of accuracy management and control of hull block construction,the compensation and allowance of intermediate products in each process are mainly set up based on construction experience,an accurate error management and control system is still not established.Due to the lack of quantitative analysis and compensation calculation,a large number of errors are found and fed back in the later processes,and the errors have to be corrected by compensation procedure like flame and twice cutting.As a result,the cost increases,the efficiency decreases and the accuracy level is largely affected.This paper's main research work is as followed:(1)Analysis of the process of double-deck bottom block constructionTaking B225 double-deck bottom block in a shipyard in Shanghai as an example,the construction processes of the double-deck bottom block is analyzed in detail.On this basis,the standard of dimension error control in each process is elaborated.In the end,taking the internal bottom plate assembly process as the research object,the fluctuation period of the process is extracted by the ensemble empirical mode decomposition,and the initial diagnosis of the error source is realized.(2)Dimensional error statistical process control in single processThe characteristic of small batch production in shipbuilding is the main reason that restricts the application of statistical process control in shipbuilding industry.However,the hull block construction processes include not only small batch production processes such as internal bottom plate assembly,but also large batch production processes such as numerical control plasma cutting.Based on historical data and prior information,the practicability of Bayesian control diagram used in small batch production like internal bottom plates welding was verified.Aiming at mass production processes such as NC plasma cutting process,traditional statistical process control is applied to realize closed-loop control of single process dimensional error combined with process improvement and hypothesis test.(3)Research on the law of error transmission in multi-processIn order to change the current situation of size error compensation based on experience in a shipbuilding company in Shanghai.A support vector regression model based on error separation is established,which not only realizes the prediction of dimensional error transmission and change in multi-process,but also explains and decomposes the composition of dimensional error in each process.On the basis of this,the dimensional error compensation based on experience is optimized to improve the accuracy of hull block construction.The conclusions of this paper can be applied not only to the construction process of double-deck bottom block to instruct the dimensional error control work,but also to the construction process of other types of hull blocks in dimensional error control quantitatively.Furthermore,it has important reference significance for improving the accuracy level of ship construction.