From multi-attribute ranking to justified choice: a rule-based screening framework (MASINISA) for industrial robot selection
提出MASINISA框架，通过非补偿性筛选与补偿性精炼结合的三阶段流程，在27个工业机器人选型案例中实现100%权重扰动稳定性，将决策空间缩小81%，帮助制造商避免选型失误。
The selection of industrial robots in modern manufacturing is a complex decision-making problem, driven by the wide variety of available models and multiple, often conflicting, evaluation criteria. This paper introduces MASINISA (Multi-Attribute Superiority, Inferiority, and Non-Inferiority Selection Approach), a novel rule-based decision support framework that advances beyond existing dominance-based MADM methods such as ELECTRE, COPRAS, and MABAC. Unlike these approaches, which primarily produce rankings or rely on compensatory aggregation, MASINISA integrates three complementary dominance perspectives (superiority, inferiority, non-inferiority) into a unified choice-oriented pipeline that first applies non-compensatory screening before compensatory refinement. The method relies on a three-dimensional scoring system superiority (S), inferiority (I), and non-inferiority (NI) derived from pairwise comparisons to provide complementary dominance perspectives and enable scale-independent assessments. A structured multi-stage process is implemented: (1) initial non-compensatory filtering of intrinsically poor and score-dominated alternatives, (2) compensatory evaluation of weighted composite scores, and (3) final selection using an exponential goodness function that flexibly balances dominance (S) and robustness (NI) through a tuneable parameter λ. MASINISA’s applicability and effectiveness are demonstrated through an established benchmark case study involving the selection of 27 industrial robots a representative sample of commercially available models evaluated against four conflicting criteria: repeatability (min), cost (min), load capacity (max), and velocity (max). Criteria weights were derived from expert elicitation with industrial automation specialists, ensuring alignment with practical procurement priorities. Benchmark comparisons against TOPSIS, VIKOR, and CARCACS show that MASINISA achieves 100% selection stability under ±10% weight perturbations, outperforming traditional methods, which exhibit rank reversal or sensitivity. Furthermore, MASINISA reduces the decision space by 81% (from 27 to 2 candidates) through systematic filtering, demonstrating both robustness and computational efficiency. By enabling more reliable, transparent, and justifiable robot selection, MASINISA helps manufacturers avoid costly mismatches, reduce downtime, and enhance operational efficiency and safety, directly supporting improved manufacturing performance and return on investment. The operational logic of MASINISA, from pairwise scoring to final parametric selection, is detailed in Section 3 and illustrated step-by-step in the case study.