主管:中国科学院
主办:中国优选法统筹法与经济数学研究会
   中国科学院科技战略咨询研究院

中国管理科学 ›› 2026, Vol. 34 ›› Issue (2): 164-175.doi: 10.16381/j.cnki.issn1003-207x.2023.1065cstr: 32146.14.j.cnki.issn1003-207x.2023.1065

• • 上一篇    下一篇

基于TSO-LS-SVM模型的电煤库存风险评价研究

陈云峰1,2, 于雪1(), 刘吉成1, 马旭颖1, 朱玺瑞1   

  1. 1.华北电力大学经济与管理学院,北京 102206
    2.国家能源投资集团有限责任公司,北京 100010
  • 收稿日期:2023-06-25 修回日期:2023-09-05 出版日期:2026-02-25 发布日期:2026-02-04
  • 通讯作者: 于雪 E-mail:shenwenjieo@qq.com
  • 基金资助:
    国家自然科学基金项目(71771085)

Research on Electric Coal Inventory Risk Assessment Based on TSO-LS-SVM Model

Yunfeng Chen1,2, Xue Yu1(), Jicheng Liu1, Xuying Ma1, Xirui Zhu1   

  1. 1.College of Economic and Management,North China Electric Power University,Beijing 102206,China
    2.National Energy Investment Group Co. ,Ltd. ,Beijing 100010,China
  • Received:2023-06-25 Revised:2023-09-05 Online:2026-02-25 Published:2026-02-04
  • Contact: Xue Yu E-mail:shenwenjieo@qq.com

摘要:

为提高电煤企业库存风险评估的准确度和效率,本文提出一种金枪鱼群优化算法与最小二乘支持向量机(TSO-LS-SVM)的风险组合评价模型。首先,该方法利用金枪鱼群优化算法(tuna swarm optimization algorithm,TSO)实现了最小二乘法(least squares,LS)和支持向量机模型(support vector machine,SVM)的参数设置优化。其次,通过算例分析验证了所提TSO-LS-SVM模型在电煤库存风险评价中的适用性。再次,通过对比金枪鱼群优化算法、鲸鱼优化算法(whale optimization algorithm,WOA)和粒子群优化算法(particle swarm optimization,PSO)验证了本文所提方法的优越性。结果显示,TSO-LS-SVM模型收敛速度快,准确率更高,均方误差更小,在电煤库存风险评价中表现最优。最后,通过灵敏性分析从煤炭损耗、政策机遇、设施建设、员工素养和信息传导5个角度提出了风险管控策略,为电煤企业提高库存风险管控水平提供了参考。

关键词: 电煤库存风险, 风险评价, 支持向量机, 金枪鱼群优化算法

Abstract:

In the current international pattern characterized by significant changes, adaptations and reorganization, energy issues are no longer simply matters of supply or development, but a comprehensive consideration involving national security and international strategy. However, the recent years' local disparities in energy supply and demand, coupled with electricity shortages in China's energy market, have laid bare the deficiencies in China's coal management. At present, these shortcomings primarily manifest in three key areas: firstly, as a large thermal power generation country, China's electric coal supply risk prevention awareness and response ability is insufficient; secondly, the electric coal stockpile reserve mechanism is not perfect; thirdly, there is a deficiency in suitable risk assessment models. Therefore, to improve China's risk control level in electric coal risk control, especially in electric coal inventory, in addition to strengthening the risk prevention awareness, appropriate risk assessment models should also be constructed and the mechanism for electric coal inventory reserve should be improved. Therefore, the thermal power plant is taken as the research object in this paper, and the risk assessment indexes are established considers the risk factors affecting the management of electric coal inventory from three levels of macro-environment (policy, economy, natural environment and market), meso-environment (information transmission and enterprise cooperation) and micro-environment (business operation and coal yard condition). On this basis, by analyzing the advantages and disadvantages of the SVM assessment model, the regularization parameter λ is introduced to prevent the overfitting problem, and the tuna swarm optimization algorithm(TSO)is used to optimize the model regularization parameterλ, the kernel function parameter σ and other related parameters, and thus the risk combination evaluation model of the tuna swarm optimization algorithm and the least-squares support vector machine (TSO-LS-SVM) is proposed. Finally, the proposed model is applied to 200 electric coal enterprises in China to analyze the examples in the context of reality, and the specific research results are as follows. (1)The indicators are selected from four aspects considering the digitalization background: information effectiveness, information sharing degree, information symmetry and vulnerability of information transmission mechanism, and a combination of quantitative and qualitative methods is used to construct the inventory risk assessment index system of electric coal enterprises. (2) The applicability of the proposed TSO-LS-SVM model in the assessment of electric coal inventory risk is verified through case analysis. The superiority of the proposed method is verified by comparing TSO, whale optimization algorithm (WOA) and particle swarm optimization (PSO), for it has faster convergence speed, higher accuracy, smaller mean square error, and the best performance in the assessment of coal inventory risk. (3) Sensitivity analysis is conducted on risk factors, and it is found that the top five risk indicators are coal loss, policy opportunities, facility construction, employee quality and information transmission, providing scientific guidance and basis for the proposal of electric coal risk control strategy. Based on the conclusions above, certain management insights are provided for enhancing the risk management and control level of electric coal enterprises, improving the electric coal stockpile reserve mechanism, and ensuring the stable supply of energy.

Key words: electric coal inventory risk, risk assessment, support vector machine(SVM), tuna swarm optimization algorithm(TSO)

中图分类号: