TL;DR
This paper introduces a novel, automated approach for selecting optimal machining schemes for manufacturing features. By building a process knowledge graph and employing an improved cosine similarity metric that considers parameter influence and scheme usage, the method aims to reduce the subjectivity and inefficiency of traditional manual selection. The results show that the proposed method's recommendations closely match established, mature machining schemes.
What Problem Does This Paper Address?
In manufacturing, choosing the right machining scheme for a specific feature (like a hole or a groove) is critical for part quality, production efficiency, and overall cost. Traditionally, this selection relies heavily on the experience and intuition of engineers, making it a subjective, time-consuming, and often inefficient process. This manual approach can lead to suboptimal choices, potentially causing part defects or requiring rework. The paper addresses the need for a more objective, automated, and efficient method for Machining Scheme Selection (MSS) that can leverage existing knowledge and data.
Methods
The core of the proposed method involves two main components:
- Process Knowledge Graph Construction:
- Knowledge Extraction: Natural Language Processing (NLP) techniques are used to extract process knowledge from various sources. This knowledge is structured into ternary groups:
part-feature,feature-attribute, andscheme-resource.
- Knowledge Extraction: Natural Language Processing (NLP) techniques are used to extract process knowledge from various sources. This knowledge is structured into ternary groups:
- Multi-level Graph: These triplets are used to build a multi-level process knowledge graph. This graph acts as a structured repository of manufacturing information, enabling efficient retrieval of available machining schemes for specific features.
- Improved Cosine Similarity Matching:
- Feature Information Model: A model is created to represent feature information, incorporating part properties, basic feature details, and manufacturing information. This model is then encoded into a numerical format.
- Dimensionality Reduction (AHP): The Analytic Hierarchy Process (AHP) is used to evaluate the importance of different influencing factors (parameters) on MSS, reducing the dimensionality of the feature information.
- Influence and Usage Coefficients: The similarity calculation is enhanced by incorporating:
- Influence Coefficient ($\alpha$): Derived from AHP, this quantifies the importance of each process parameter.
- Usage Coefficient ($\beta$): Based on historical data, this reflects the frequency or popularity of a particular machining scheme.
- Novel Similarity Formula: An improved cosine similarity formula is designed, which combines the standard cosine similarity with these influence and usage coefficients to provide a more accurate and context-aware similarity score. The optimal scheme is then selected based on the highest similarity score.
Datasets
The paper does not explicitly mention training on large datasets. Instead, it relies on:
- Extracted Process Knowledge: Information gathered from various sources (production plants, research institutes, online encyclopedias) is used to build the knowledge graph.
- Historical Data: Usage frequency of machining schemes is derived from historical production data to calculate the usage coefficient.
- Typical Shell Part: A specific shell part with six types of hole features is used as a case study for validation. The manufacturing parameters for these features are provided in Table 14.
Key Insights and Ideas
- Bridging the Knowledge Gap: The knowledge graph effectively bridges the gap between raw textual process information and structured, retrievable manufacturing knowledge, making it accessible for automated decision-making.
- Beyond Simple Similarity: The improved cosine similarity, by incorporating parameter influence and scheme usage, moves beyond basic geometric or attribute matching to consider practical manufacturing realities and historical performance. This is crucial for selecting schemes that are not just technically feasible but also efficient and proven.
- AHP for Parameter Weighting: Using AHP to assign weights to influencing factors provides a systematic way to quantify the relative importance of different feature attributes (like size, accuracy, roughness) in the MSS process.
- Practical Relevance: The method's focus on simplicity and ease of use, aiming to be operated by skilled workers, makes it more likely to be adopted in real-world manufacturing environments.
Key Results
- The proposed method successfully retrieves available machining schemes for hole features from a process knowledge graph.
- The improved cosine similarity formula, considering influence and usage coefficients, is used to rank these schemes.
- For a typical shell part with six hole features, the recommended machining schemes by the proposed method show a high degree of overlap (66.67%) with existing mature schemes.
- The average similarity score of the recommended schemes (0.8450) is slightly higher than that of the existing mature schemes (0.8448), indicating the effectiveness and potential superiority of the proposed method.
Limitations and Future Work
- Limitations:
- The paper acknowledges that the current method primarily focuses on hole features and suggests expansion to other feature types (grooves, steps, windows) in future work.
- The accuracy of the method is dependent on the quality and completeness of the extracted process knowledge.
- While the method aims for simplicity, the initial construction and maintenance of the knowledge graph can be resource-intensive.
- Future Work:
- Enhance the knowledge graph by incorporating reasoning capabilities for more sophisticated knowledge retrieval.
- Continuously update and refine the feature information model by adding or removing influence factors to improve comprehensiveness.
- Further optimize the similarity formula to further narrow the gap with mature machining schemes.
- Expand the application of the method to a wider range of manufacturing features.
Summary updated