Research On Data Mining Method And Application Of Digital Prototype Design Performance Evaluation For Bag Filter

With the rapid development of modern computer technology,digital prototyping technology has been widely used in aircraft,automobile and other manufacturing industries,which makes the product design driven by simulation become a mainstream.In the process of modeling and running of complex products,digital simulation technology will produce large scale,high dimension and complex interrelation simulation data,how to make full use of the hidden knowledge and rules in simulation data to guide the design and optimization of products is a difficult problem to be solved at present.However,data mining technology can reveal new relationships,patterns and trends by analyzing a large amount of data.So data mining technology is an effective method to solve this kind of problems.In recent years,with the development of data-driven artificial intelligence technology,Deep learning is widely used in data mining because of its strong ability of data analysis and learning.In this paper,the deep belief network was improved and used to predict the performance of digital prototype of bag filter.The specific contents of the study are as follows:(1)The working principle of bag filter was analyzed.Based on that,the key technology and performance of a self developed bag filter prototype were studied.From the perspective of data mining technology,the simulation data of digital prototype of bag filter was analyzed,the target of data mining was determined.(2)The learning algorithm of the deep belief network(DBN)model was analyzed.When the DBN model was predicted in nonlinear systems,it is difficult to find the global optimum and the learning speed is slow because of the fixed learning rate in the model construction,an improved DBN prediction model was proposed.The momentum learning rate was introduced into the unsupervised pretraining stage of DBN,and the Restricted Boltzmann Machine network was improved to improve the accuracy of feature extraction and the anti-oscillation ability of parameters in the training process.At the same time,the conjugate gradient method was embedded into the DBN fine-tuning stage to accelerate the learning speed.The performance prediction model of the digital prototype of the bag filter based on the improved DBN was constructed,and the optimal network structure of the model was determined by the experimental method.A prediction model with high prediction accuracy was established.(3)The performance prediction platform of digital prototype of bag filter was built.The function module of digital prototype simulation and data mining for product oriented bag filter was developed and integrated.The platform consists of five modules: digital prototype model reconstruction module,numerical simulation module,data preprocessing module,data mining module,performance prediction module.The feasibility and effectiveness of the platform were proved by predicting the working performance of 192 series bag filter.The results of research and application show that the improved DBN prediction model is not only fast but also accurate.When the model is used to predict the working performance of the digital prototype of the bag filter,not only can the index value of the evaluation performance be obtained quickly and accurately,What is more important is that the designer can evaluate the influence of different design schemes on their working performance and determine the optimal structure of the dust collector.