磷矿选矿厂人因可靠性分析及预防对策研究

Human Reliability Analysis and Preventive Countermeasures in Phosphate Ore Processing Plants

摘要: 为应对磷矿选矿厂人因失误高发问题及提升传统HRA方法的行业适用性，本研究构建了融合改进人为因素分类系统(HFICS-MP)与贝叶斯网络(BN)的综合分析框架。基于200起事故数据，采用HFICS-MP提炼出36项专属风险因素，并建立四层级结构。卡方检验识别了层级内/间非线性路径，BN量化分析揭示：近端行为层的违章操作(C31)是核心因素，发生频率81.0%，后验影响力最大；发生概率67.0%的组织层的安全教育培训不力(A23)，与发生概率58.5%的风险管控缺失(A31)为深层诱因；发生概率45.0%的监督层的制度执行失效(B41)，与防护设备监管缺失(B31)，发生概率41.0%为关键短板。案例反演验证了“组织缺陷→监督失效→近端行为→事故”的传导机制及跨层级影响(如监督层的制度执行失效→物体打击)，敏感性分析表明控制人力资源配置缺陷、防护设备监管缺失、生产环境恶劣、违章操作等关键因素可显著降低机械伤害概率。本研究为磷矿选矿厂提供了人因失效根源定位、动态评估与分级防控的量化方法论。

Abstract: To address the high frequency of human errors in phosphate ore processing plants and enhance the industry-specific applicability of traditional Human Reliability Analysis (HRA) methods, this study constructs an integrated analysis framework combining an improved Human Factors Classification System (HFICS-MP) with a Bayesian Network (BN). Based on 200 accident cases, the HFICS-MP framework was used to refine 36 unique risk factors and establish a four-layer structure. Chi-square tests identified non-linear transmission paths within and between layers. BN quantification revealed that Violation Operations (C31) in the proximal behavior layer was the core factor, with the highest occurrence frequency (81.0%) and the greatest posterior influence. In the organizational layer, Inadequate Safety Training (A23, occurrence frequency: 67.0%) and Deficient Risk Management (A31, occurrence frequency: 58.5%) were deep-rooted contributors. Within the supervision layer, Failure in Procedure Execution (B41, occurrence frequency: 45.0%) and Lack of Protective Equipment Oversight (B31, occurrence frequency: 41.0%) were key weaknesses. Case back-calculation validated the “organizational deficiencies → supervisory failures → proximal behaviors → accident” transmission mechanism and cross-layer influences. Sensitivity analysis demonstrated that controlling key factors such as Inadequate Human Resource Allocation (A12), Lack of Protective Equipment Oversight (B31), Interference from the Operational Environment (C11), and Violation Operations (C31) could significantly reduce the probability of mechanical injuries. This study provides a quantitative methodology for root cause identification, dynamic assessment, and hierarchical prevention and control of human failure in phosphate ore processing plants.