A Novel Approach for Mining Machining Process Decision Knowledge Based on Knowledge Constraint Combined with Water Wave Optimization Algorithm

Xinzheng Xu, Zhicheng Huang, Lihong Qiao, Yongqiang Wan, C. L. Philip Chen, Zhujia LiApplied Sciences 2026 · DOI

TL;DR

This paper introduces a novel approach for discovering "Process Decision Knowledge" (PDK) in manufacturing, crucial for intelligent Computer-Aided Process Planning (CAPP) systems. By combining domain-specific "knowledge constraints" with the "Water Wave Optimization" (WWO) algorithm, they achieve over 90% accuracy in extracting these rules from complex manufacturing data, overcoming limitations of traditional methods.

What Problem Does This Paper Address?

Intelligent CAPP systems aim to automate and improve manufacturing process planning. A key component of these systems is "Process Decision Knowledge" (PDK), which essentially represents the "rules of thumb" or experience-based knowledge that guides decisions like tool selection or parameter settings. However, discovering this PDK from vast, high-dimensional manufacturing data is challenging. Traditional methods often struggle with accuracy due to implicit, experience-dependent decisions and the sheer complexity of the data. This leads to a scarcity of practical, intelligent CAPP systems. The authors aim to bridge this gap by developing a more accurate and robust method for PDK discovery.

Methods

The core of the proposed method lies in two main components:

  1. Knowledge Constraint-Based Association Discrimination Model: This is the novel aspect. Instead of relying solely on statistical measures (like support and confidence) from the data, this model incorporates prior knowledge from process planners. This prior knowledge is mathematically represented as a "semantic association matrix," which captures the relationships between different process data items (e.g., how "machining feature" relates to "machining method"). This matrix is then used to define a "fitness function" that evaluates the validity of potential PDK rules. This ensures that discovered rules are not just statistically frequent but also semantically meaningful and aligned with expert understanding.
  2. Water Wave Optimization (WWO) Algorithm: This nature-inspired metaheuristic algorithm is employed as the search strategy. WWO is chosen for its effectiveness in solving complex combinatorial optimization problems, particularly its ability to balance global exploration and local exploitation. It uses "water wave operators" (propagation, refraction, breaking) to efficiently search the vast space of possible rule combinations for valid PDK.

The overall framework involves:

  • Data Sample Construction: Using OCR and Named Entity Recognition (NER) to convert unstructured process planning documents into structured, tabular data samples.
  • Knowledge Mining: Applying the WWO algorithm with the knowledge constraint-based fitness function to discover PDK from these structured samples.

Datasets

The study utilizes machining process planning documents from a general parts manufacturing plant. These documents were collected and organized into four categories:

  • Shaft parts (527 records)
  • Gear parts (513 records)
  • Housing parts (535 records)
  • Plate parts (568 records)

The data was pre-processed into structured, two-dimensional tables for the PDK mining process. No new benchmark datasets were introduced.

Key Insights and Ideas

  • Prior Knowledge is Key: The most significant insight is that simply relying on statistical patterns in data is insufficient for accurate PDK discovery. Incorporating expert domain knowledge, even in a structured matrix form, dramatically improves the quality and relevance of the discovered rules.
  • Semantic Meaning Matters: The paper emphasizes that PDK isn't just about co-occurrence of data items; it's about the semantic relationships between them. The semantic association matrix is a clever way to quantify this.
  • WWO is Well-Suited for PDK: The Water Wave Optimization algorithm's ability to balance exploration and exploitation, along with its operators, makes it a strong candidate for navigating the complex, discrete search space of PDK.
  • Accuracy Over Raw Quantity: While traditional methods might find more rules, the proposed approach prioritizes correct and usable rules, achieving high accuracy even if the total number of rules discovered is slightly lower.

Key Results

  • High Accuracy: The proposed method achieved an accuracy rate of approximately 90% in PDK mining.
  • Improved Ktotal: The WWO-based strategy achieved the highest mean Ktotal (number of correct rules) of 78.4 compared to PSO (40.4), GA (49.7), and BSO (76.8).
  • Efficient Mining Speed (Kv): WWO also demonstrated the highest mean Kv (mining speed) of 3.179 pcs/s, indicating efficient discovery of correct rules.
  • Comparison with DT-PDKM: The proposed method outperformed the Decision-Tree-Based Process Decision Knowledge Mining (DT-PDKM) method across all metrics (Ktotal, Acc, and Kv). For instance, DT-PDKM's Ktotal ranged from 41-52, significantly lower than the proposed method's average of 78.4.
  • Parameter Optimization: The study identified optimal parameters for WWO, showing that a minimum fitness threshold (minF) around 0.4-0.5 balances Ktotal and Acc, while specific values for attenuation coefficient (α=0.2), maximum wavelength (Amax=P/3), and maximum wave height (hmax=7) yield the highest Ktotal.

Limitations and Future Work

  • Manual Matrix Construction: The current method relies on manually constructing the semantic association matrix. This is manageable for domains with limited semantic labels but becomes challenging for more complex domains.
  • Domain Specificity: The reliance on domain-specific expert knowledge for the matrix means that transferring the method to a new domain requires reconstructing this matrix, limiting its immediate transferability and generalizability.

Future work directions include:

  • Automated Semantic Association Learning: Developing mechanisms to automatically learn semantic associations across multiple domains to improve generalization and reduce manual effort.
  • Hybrid Data-Knowledge-Driven Framework: Integrating data-driven methods to discover implicit relations where explicit semantic associations are unavailable, creating a more scalable and applicable framework.
  • Adaptive Knowledge Base Updates: Developing mechanisms to handle the timeliness decay and obsolescence of knowledge in process knowledge bases due to evolving manufacturing technologies and processes. This would involve adaptive updating mechanisms for PDK.

Summary updated

Abstract

Knowledge discovery constitutes a vital component in building intelligent CAPP systems, and the effective discovery of process knowledge has become a prominent research focus within intelligent manufacturing. Process decision knowledge is a type of knowledge that reflects the relations between process data items, represented in the form of production rules. However, PDK discovery faces low accuracy challenges from complex high-dimensional manufacturing data and implicit experience-dependent process decisions. This paper proposed a PDK mining framework that combines knowledge constraint and the water wave optimization algorithm. This approach formulated prior knowledge mathematically using an association discriminant matrix and embedded this representation into the knowledge mining model, thus equipping the algorithmic framework with the ability to discover PDK accurately. The WWO is utilized to search within the sample space for combinations of process data items that constitute valid knowledge. In contrast to traditional association rule mining algorithms that lack accuracy and template-based methods that are inherently rigid, the proposed approach provides a robust solution by achieving over 90% correctness in PDK mining. It also serves as a demonstration and offers insights for mining similar rule-based knowledge in other fields.
BibTeX
@inproceedings{xu2026novel,
  author = {Xinzheng Xu and Zhicheng Huang and Lihong Qiao and Yongqiang Wan and C. L. Philip Chen and Zhujia Li},
  title = {{A Novel Approach for Mining Machining Process Decision Knowledge Based on Knowledge Constraint Combined with Water Wave Optimization Algorithm}},
  booktitle = {Applied Sciences},
  year = {2026},
  doi = {10.3390/app16062806},
}

Loading related content...

Figure