Research On Forecasting And Monitoring Methods Of Production Cost With Uncertainty

With the improvement of science and the level of internationalization, the traditional manufacturing industry is facing increasingly severe problems of cost control, which restrict the development of enterprises. Aiming at the problem of the difficulty in accurately estimate the cost, this thesis took research in manufacturing cost estimation of machinery manufacturing enterprises. The main contents are as follows:First, the concept of production cost and the establishment of production cost prediction model. Describing the concept of production costs and establishing prediction model of product’s manufacturing costs based on theories of cost accounting in accountancy, and analyzing the uncertainties of product cost combined with actual situation of machinery manufacturing enterprises. Then this thesis took research in simplifying model, including: 1) Based on historical data and experience, dividing variables into constant variables and independent variables. 2) Adopting Sobol’ method to conduct sensitivity analysis and importance sort on independent variables, making the lower important independent variables fastened to their averages, and finally obtaining the simplified model of product manufacturing prediction.Second, research on product prediction and monitoring methods. Based on the simplified model of product manufacturing prediction, a bottom-up approach was used to predict independent variables and target parameters. Researching on the prediction method of independent variables, using improved gray Markov model to conduct the prediction; then adopting nonparametric kernel density estimation method to carry out the uncertainty quantification of prediction result; finally, propagating the uncertainty by Monte Carlo method to obtain the point estimations and confidence intervals of production cost, and fulfilling the prediction of the model. Elaborating the concept of cost monitoring, describing the connection between monitoring and prediction as well as presenting the design plan of production cost dynamic estimation which is conditional on the monitoring result as feedback.Third, case study on production cost prediction modeling and monitoring. Taking a certain model of electromagnetic brake as an application example. A production cost prediction model was established, and the predicting result was analyzed and validated to prove the practicality and rationality of the model. Then analyzing the supervising and control methods of production cost, illustrating the effects that the monitoring had on the prediction with an example, finally the result validation was conducted.Fourth, software design and development. Including cost valuing and calculation modules in ERP system. And a software tool was developed based on Excel-VBA, which achieved Sobol’ method, nonparametric kernel density estimation method and uncertainty propagation based on Monte-Carlo.Aiming at the problem of low accuracy in production cost estimation under the circumstances of uncertainty, this thesis researched on the uncertainties of production cost in the field of machinery manufacturing, established a production cost prediction and monitoring model and conducted a validation. The result proves that the application of the model increases the precision and efficiency of production cost estimation, and also provides the data basis for pricing and enterprises who want to know the cost details of their product, which is very practical.