Interval Optimization And Neural Network Optimization Of The Key Parameter Error Scheme Of Artillery

Uncertainty is widely existed in the artillery system structure.And then it was significant to consider the uncertainty of artillery structure design for improving the related performance index of artillery and parts process.Interval optimization method was a design method considering parameter uncertainties.Hence the model and algorithm of interval optimization method of mechanical structure were researched in this paper,and then this method was introduced to structure optimization design of artillery.Afterwards,not-dominated solutions artillery key parameters error schemes which were obtained from interval optimization were optimized by using the optimal selection of neural networks.The detail research is as below.Firstly,in light of the uncertainty of structure size,two interval optimization methods were proposed in this paper.An improved interval optimization method considering tolerance was proposed,in which a discretization-based possibility degree of interval model was constructed by combining probability with interval method and the tolerance index with coefficient of tolerance grade was improved.Another three-parameters interval uncertainty optimization method was proposed,in which three-parameters interval was defined with tolerance zone and transformed to the two-parameters interval by the actual constraint of tolerance zone.In addition,an algorithm based on the combination of NSGA-II and test design was proposed.Two-level nest optimization was transformed to single nest optimization of interval optimization by this method,which ensures the accuracy and efficiency of calculation.Secondly,the interval optimization was combined with three structure optimization examples of artillery.The key structure sizes and tolerance of recoil mechanism were optimized synchronously by using the improved interval optimization method considering tolerance.The recoil resistance considering fluid hole size uncertainty was analyzed and optimized by three-parameters interval uncertainty optimization method.Taking the artillery structure key parameters as design variables,the tolerance index was improved.Interval optimization model and algorithm based on approximate model were proposed,and the artillery structure key parameters were further optimized.Thirdly,based on the not-dominated solutions of artillery key parameters error schemes which were obtained from interval optimization,the error schemes were optimized by optimal selection of neural networks.The multy-properties optimization method based on deviation maximization was used to obtain the study samples and verification samples of optimal selection of neural networks.Afterwards,the key technology applied in artillery key parameters error schemes optimization was researched from three aspects and GA-BP optimal selection of neural networks procedures whose property values include both interval number and real number was established.The optimal selection procedures were verified by the results of interval optimization.