TL;DR
This paper introduces a novel multi-attribute decision-making (MADM) method for evaluating manufacturing component suppliers, specifically incorporating "green production" indicators alongside traditional economic ones. The core innovation is a new aggregation operator that can handle complex interdependencies between attributes, even when attributes are grouped into partitions, offering a more nuanced and flexible approach to supplier selection.
What Problem Does This Paper Address?
In today's manufacturing landscape, especially within the Industry 4.0 framework, companies are increasingly pressured to adopt sustainable practices. This extends to their supply chains, where suppliers play a crucial role in a company's environmental performance. Traditional supplier evaluation methods often focus heavily on economic factors like price and delivery time, neglecting crucial "green" aspects such as environmental certification, emissions, and resource consumption. Furthermore, existing MADM methods often assume attributes are independent or only consider pairwise correlations. This paper addresses the need for a more comprehensive supplier evaluation system that integrates green production indicators and can effectively model the complex relationships between these attributes, even when they can be logically grouped.
Methods
The authors propose a new multi-attribute decision-making (MADM) method that leverages Probabilistic Linguistic Sets (PLS) to represent expert evaluations, allowing for more expressive and qualitative data representation. The key methodological contribution is the development of the Probabilistic Linguistic Weighted Partitioned Maclaurin Symmetric Mean (PLWPMSM) operator. This operator is designed to:
- Handle Partitioned Attributes: It can group attributes into distinct categories where attributes within a category are correlated, but categories themselves are independent. This is a significant advancement over methods that assume global correlation or no correlation.
- Capture Attribute Correlations: It extends the Maclaurin Symmetric Mean (MSM) operator to account for interrelationships between multiple attributes within a partition.
- Incorporate Weights: The "Weighted" aspect allows for assigning different importance levels to attributes.
- Flexibility: The operator is designed to be flexible, with parameters that can be adjusted to represent different decision-maker preferences (e.g., risk-neutral vs. risk-averse) and can degenerate into simpler operators (like PLPM or PLPBM) under specific parameter settings.
The decision-making process involves:
- Standardizing attribute values: Converting cost attributes to benefit attributes.
- Classifying attributes: Grouping attributes based on their interrelationships.
- Aggregating evaluation information: Using the proposed PLWPMSM operator.
- Ranking alternatives: Calculating scores for each supplier and ranking them.
Datasets
The paper uses a case study involving the evaluation of four component suppliers for a manufacturing company. The evaluation data, presented in Table 1, consists of probabilistic linguistic assessments for six attributes across these four suppliers. The paper also references examples from other studies (e.g., hospital selection, investment projects) to demonstrate the method's applicability and compare its performance. No new benchmark datasets were introduced.
Key Insights and Ideas
- Green Production is Crucial: The paper strongly advocates for the inclusion of green production indicators in supplier evaluation, highlighting its growing importance for corporate sustainability and competitiveness.
- Attribute Grouping Matters: The ability to partition attributes and model correlations within partitions but independence between them is a significant step towards more realistic decision-making models. This acknowledges that not all attributes are related in the same way.
- Flexibility in Decision-Making: The PLWPMSM operator's adjustable parameters allow it to adapt to various decision-maker preferences and risk attitudes, making it a versatile tool.
- Beyond Pairwise Correlations: While previous methods addressed pairwise correlations, the PLWPMSM operator can handle correlations among multiple attributes simultaneously within a defined group.
Key Results
The proposed method was applied to a case study of evaluating four component suppliers. The results demonstrated that the PLWPMSM operator could effectively rank the suppliers. The paper also conducted sensitivity analysis by varying the parameters $\kappa$ (representing the order of the Maclaurin symmetric mean) and $\kappa_1, \kappa_2$ (representing the partition structure). While the specific scores varied with parameter changes, the overall ranking of the best alternative remained consistent in the primary case study, indicating robustness. Comparisons with existing methods like PLWA and PLMSM showed that the proposed method provided more nuanced results, especially when attributes were partitioned, highlighting its superior applicability and generality.
Limitations and Future Work
- Expert Opinion Bias: The authors acknowledge that the evaluation information relies on expert opinions, which may be subject to cognitive bias.
- Focus on Cooperative Suppliers: The study primarily focused on evaluating existing cooperative suppliers, and the dynamic adjustment of subsequent orders was not considered.
- Future Work Suggestions:
- Quantifying green production indicators, specifically carbon emission reduction.
- Combining supplier evaluation with order share allocation for dynamic decision-making.
- Applying the method to other MADM problems like online product recommendation and charging pile site planning.
Summary updated