Statistics Assessment Method Of Aviation Material And Structure Performance

It is crucial for the acknowledgment of the mechanical design allowable values when we design product or select materials. Because aviation materials and structures require excellent performance, their properties commonly use A-basis value and B-basis value to represent. A-basis value is the lowest value of the upper99%of a specified population with95%confidence. B-basis value is the lowest value of the upper90%of a specified population with95%confidence. Because of the constraints of time and money, it will lead to rare experimental datas of mechanical properties of materials for the development of new materials and the improvement of production processes. And the experimental datas will be multiple-sample and have different structure types. However, the traditional methods of data processing are based on large sample.In this paper, the statistics assessment methods of aviation material and structure performance were researched, and the relevant statistical theorys were introduced to the existing calculation methods of basis values of materials and structure performance. And the calculation methods of A-basis value and B-basis value of materials and structure performance were given to adapt them to different data sources and sample sizes. And then, application software for basis values calculating was writed with VC++to achieve the calculation methods in this paper. And the software was used to calculate and analysis the samples of relevant calculation methods to validate the calculation methods of the samples from different batches, different sample sizes and different structure types. Conclude that the calculation methods of basis values in this paper are correct, and the software is useful, convenient and reliable, improving the efficiency of data processing.The experimental datas processed in this paper were small samples mostly. As the bayesian methods make full use of the prior information, they appropriate to the case of small sample for experimental data processing. In this paper, bayesian methods were introduced to the parameter estimation of distribution function which was fitted for the processing of experimental datas. According to relevant calculation examples, bayesian parameter estimation methods of normal distribution model and lognormal distribution model were given in this paper. And the bayesian parameter estimation method for three parameters Weibull distribution model was discussed in this paper.