Research And Implementation On Conceptual Design Algorithms Of Mining And Transportation Equipment Based On Machine Learning

With the increasingly improved structure and function, the conceptual design of mining and transportation equipment is becoming more and more important in their development cycle. The conceptual design of mining and transport equipment not only reflects the innovation level of final products, but also determines 70%-80% of the investment. The defect and error at this stage is difficult to correct in detailed design phase, and it is apt to trigger cost increasing.Traditional conceptual design methods have many shortcomings, such as excessive dependence on expertise, knowledge dispersion and absence of scientific reasoning model. In order to assist designers to make decision,improve work efficiency, and shorten product design period, this paper discussed the intelligent machine learning algorithms and developed a conceptual design system based on WEB which has been used in enterprises,focusing on overall conceptual design parameters of shearer, heading machine and scraper conveyor.The organization and management for conceptual design knowledge and experience of mining and transport equipment can facilitate their sharing and reuse according to their characteristics. CBR can use past successful cases in database to solve similar new problems, which can reduce repeated design labor,and avoid unnecessary redesign. Structured reasoning model based on machine learning algorithms, such as SVR, ELM and GA, can solve problems with innovative solutions. By algorithm improvement and parameter optimization,model reasoning accuracy has been improved and its results are more reliable.To put the reasoning model into engineering practice, conceptual design system for mining and transportation equipment has been developed and delivered to enterprises. The system is developed with B/S application mode,integrated environment VS2010, ASP.NET, SQL2008 and mixed programming of C#&MATLAB. It has perfect Human Machine interaction, and solves the space limitation problem for designers, which means they can telnet the servers and finish design by browser.The system can assist designers to determine the overall parameters of mining and transporting equipment, and can generate design scheme automatically. After inputting shearer cutting height, cutting depth, coal seam pitch and other user attributes, the system can finish similarity comparison of existing products automatically, and output the similar product cases. If there is no similar instance, the system will enable reasoning model based on SVM,ELM and other intelligent machine learning algorithms to output products' overall technical parameters. After manual adjustment, the new instance can be saved to expand the case library. The system has high prediction accuracy and computational efficiency. On standard configuration computer, it can finish parameters reasoning within 3 seconds, and the mean relative error can be controlled within 5%.As Conceptual design system of mining and transportation equipment helped the enterprise integrate their experience and knowledge, it satisfied their requirements of both intranet and remote invocation, improved the design efficiency as well as saved design costs and material costs. It has been verified that the system is easy to use and operates stably and the result is accurate and reliable.