This paper establishes the multiple linear regression model and the optimizing model, which adjust the composition proportion with the best strategy to improve the structure properties and control the cost by a company. Firstly, in order to get the effective data, we take filtering process method to eliminate withdrawn data. As for the numerous data, we take the pretreatment to revise the data until getting the reasonable data. Then according to the question, we set up a multiple linear regression model. Dealing with the correlation between these chemical elements, we adopt cluster analyses to discuss the result. Secondly, based on the model one, we mainly adopt controlled variable method. At first, we take the content of the chemical elements as dependent variable and the three properties as independent variable. Then we get the functional expression. Taking the important degree of the corresponding independent variables of the regression coefficient into consideration, we find out the coefficients can representative the influential degree. In this way, we could get the influence rule. Thirdly, with the improvement of the model, we set up a partial differential equation to attain the better results.