Development And Application Of Riser Process Recommendation Design System Based On Statistical Feature Recognition Algorithm

The riser process design is an important part of the casting process design.In the traditional riser design,the design process is cumbersome and relies heavily on experience.Using the knowledge-based riser design expert system,there are often problems of lack of knowledge rules and insufficient reasoning ability due to the complexity of the riser design,which leads to the unreasonable design of risers.According to similar castings and their original process design,a reasonable initial riser design solution can be obtained more quickly.Based on the idea of process reuse,this paper proposes a recommended design method for riser technology that draws on the original feasible process.The statistical feature recognition algorithm is used to quickly acquire similar castings in the casting process database,and then the corresponding riser process is obtained.After that the riser process of similar castings is transplanted to the target casting,thereby realizing the rapid design of the reasonable riser process.The main work was shown below.Firstly,to calculate the similarity of the model,a feature recognition algorithm model suitable for casting model was established which includes feature extraction and similarity calculation.In this paper,the statistical feature D2 operator is used to extract the model feature indicator,and the feature model was validated by the basic model and the complex casting model.Then a similarity calculation method was proposed based on the D2 operator to calculate the similarity between the two models and the similarity discrimination effect was verified.Secondly,the system operation efficiency optimization technology was studied.For a single model,a single model feature extraction recommendation point optimization technique was proposed.The proposed experimental solution was used to obtain the recommended point curve as the feature extraction selection criterion,and the model feature extraction time was reduced.For all models,the preprocessing and feature representation storage technique was proposed.The model feature indicator was calculated in batches in advance and stored in a storage structure that is easy to search,which reduces double counting.Two technologies were applied to developed system to reduce the time of a single search and increase the speed of the system.Finally,the riser process recommendation design system was designed and developed.A basic database for all process design plans of a foundry manufacturer in China was established and the verification of system application was performed based on castings such as box parts.The experimental results show that the riser process recommendation design method based on statistical feature recognition algorithm can realize the riser process design with high efficiency.