中国科学院数学与系统科学研究院期刊网

23 July 2025, Volume 45 Issue 7
    

  • Select all
    |
  • Qi LIU, Junyi HUANG, Gengzhong FENG, Shouyang WANG
    Systems Engineering - Theory & Practice. 2025, 45(7): 2101-2123. https://doi.org/10.12011/SETP2023-2891
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    In the digital economy era, data has emerged as a new factor of production. However, pervasive data quality issues pose significant challenges to releasing the value of data elements and may potentially become "grey rhinos" for digital economy development. Currently, the field of data science is advancing rapidly, highlighting the pressing need for further consolidation and summarization of research related to data quality. This is essential to effectively support the practice of data quality management and the establishment of reliable data circulation. This paper takes a systematic approach to explore the trajectory of data quality research. By employing a synthesis of diverse methodologies, we conduct a comprehensive review of relevant literature from domestic and international sources during the past 30 years. Our review reveals a logical progression in data quality research, characterized by the interconnected stages of "connotation-theory-method-application". Building upon this, we develop a framework for data quality research. Subsequently, we provide a retrospective summary encompassing the data quality connotation and dimensions, theoretical foundation development, assessment and optimization methods, and influencing factors and value effects. Finally, we explore trends in the development of data quality research and offers insights into future directions.
  • Shuxian LI, Xiaochuan PANG, Jiali MA, Shuhua XIAO, Shushang ZHU
    Systems Engineering - Theory & Practice. 2025, 45(7): 2124-2144. https://doi.org/10.12011/SETP2023-2010
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Using detailed data of interest-bearing debts from more than 3{,}000 local financing platforms and the 2021 annual reports from 60 major banks in China, this paper evaluates the systemic risk of the banking system potentially caused by local government debts in terms of total debt volume, economic industry and economic region, respectively. The results of stress testing show that: 1) Financing platform loans in the leasing and business service industry have the largest risk exposure among all industries and are closely connected to the real estate industry. An increase in default rate in the leasing and business service industry alone can trigger systemic risk in the banking system. 2) In terms of the regional comparison, the implicit debt of local governments exhibits higher default rates in the west and higher risk exposures in the east. The systemic risk of the banking system presents a noticeable "high in the east and low in the west" pattern under the same default rate. Additionally, the safe interval of default rates for implicit debts is narrower in the east compared to the mid-west. 3) At present, either defaults in local implicit debt (financing platform loan) or liquidity crisis triggered by explicit debt (government bond) is unlikely to cause the systemic risk in the banking system. However, combining with the risk contagion from banking networks, they can jointly cause significant losses to the banking system.
  • Jianfei WANG, Cuiqing JIANG, Yong DING, Yingfeng LI
    Systems Engineering - Theory & Practice. 2025, 45(7): 2145-2162. https://doi.org/10.12011/SETP2023-2403
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    With the development of digitalization, networking and intelligence in social and economic activities, the connections between firms are becoming close, and the impact of related risks on the financial distress of firms is increasing. Existing research usually uses social network analysis methods to quantify topological structures and related risk impacts, but those methods are not applicable to heterogeneous networks containing different types of entities and relations. In particular, the related risk paths are long and the quantification of higher-order related risks propagated through indirect paths faces challenges. To end this, we design a framework for predicting financial distress of firms by incorporating higher-order related risk features. In the framework, we propose an unsupervised heterogeneous graph representation learning model to construct higher-order related risk features and develop an explainable method to mine higher-order related risk paths. Experimental evaluations demonstrate the superior predictive power of the unsupervised heterogeneous graph representation learning model over benchmark methods for financial distress prediction. In addition, the experimental results show that there are two types of higher-order related risk paths that help predict the financial distress of firms.
  • Zhen YU, Chenxi LI, Yuankun LI
    Systems Engineering - Theory & Practice. 2025, 45(7): 2163-2187. https://doi.org/10.12011/SETP2024-0689
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Under the orientation of high-quality development, the incentive problem for green transformation of Chinese enterprises urgently needs to be solved. From the perspective of regional trade agreements (RTAs), we construct a theoretical model to analyze the impact mechanism of "Open Environmental Regulations" on the green technology progress of enterprises in developing countries, and conduct an empirical study taking Chinese enterprises as an example. First, we develop a model of enterprise green innovation decision-making in an open economy and find that RTAs' environmental provisions influence the direction of enterprise technological progress through two channels: The technological incentive effect and product structure effect. Then, using the text of RTAs, country-specific bilateral trade data, and import and export data from Chinese customs, we construct the "environmental provisions embeddedness" of enterprises and employ a panel fixed-effects model to empirically test the effect of "Open Environmental Regulations" on green technological progress in Chinese enterprises. Our study finds that embedded environmental provisions significantly promote substantive green innovation in enterprises but have no significant impact on strategic green innovation. Mechanism analysis shows that RTAs' environmental provisions drive enterprises to increase R&D investment and green transformation of product structures, thereby promoting green technological progress. The green innovation effect of environmental provisions is more evident in non-heavy polluting industries, enterprises with low financing constraints, enterprises with high social attention, and enterprises with a high degree of overseas market diversification. Further research finds that some market-based environmental regulation tools amplify the green innovation effect of environmental provisions, indicating the necessity of coordinating internal and external environmental policies in the process of institutional openness. These conclusions provide new theoretical perspectives and empirical evidence for China to promote high-quality economic development and high-level environmental protection through high-level openness. Our study finds that embedded environmental provisions significantly promote substantive green innovation in enterprises but have no significant impact on strategic green innovation. Mechanism analysis shows that RTAs' environmental provisions drive enterprises to increase R&D investment and green transformation of product structures, thereby promoting green technological progress. The green innovation effect of environmental provisions is more evident in non-heavy polluting industries, enterprises with low financing constraints, enterprises with high social attention, and enterprises with a high degree of overseas market diversification. Further research finds that some market-based environmental regulation tools amplify the green innovation effect of environmental provisions, indicating the necessity of coordinating internal and external environmental policies in the process of institutional openness. These conclusions provide new theoretical perspectives and empirical evidence for China to promote high-quality economic development and high-level environmental protection through high-level openness.
  • Guiyu LI, Shuming WANG, Hongbo DUAN
    Systems Engineering - Theory & Practice. 2025, 45(7): 2188-2201. https://doi.org/10.12011/SETP2023-2370
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Climate policy is the key to addressing climate change and realizing carbon neutrality. The assessment of climate policy is filled with uncertainties affecting economy, climate and energy through parameter uncertainty and model uncertainty. We develop a robust assessment framework based on the expected output from IAM for climate policies performance by incorporating the integrate assessment models (IAMs) and distributionlly robust optimization to address the impacts of model uncertainty and joint parameters uncertainties simultaneously on the assessment of climate policy. Besides, Wasserstein ambiguity set is utilized to demonstrate information of uncertain parameters. The results show that: 1) The stricter mitigation efforts, the less impacts of uncertainty on climate performance but the larger economic costs. 2) Uncertainty requires stronger mitigation efforts for the realization of carbon neutrality. 3) The worst-case distribution from the expected output of IAM has less effects on the expected net present value and expected global warming but brings great effects on the risk of future abrupt climate damage. 4) The temperature increase performance of climate policies is relatively robust across IAMs. Our work is the first study that develops a framework for assessing climate policies on the realization of carbon neutrality under uncertainty by integrating distributionally robust optimization and general climate-economic model, which provides a conservative and robust assessment for climate policy.
  • Jianzhong XIAO, Yang WEN, Jiachao PENG, Xiangyi LU
    Systems Engineering - Theory & Practice. 2025, 45(7): 2202-2225. https://doi.org/10.12011/SETP2023-2499
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Climate change and its associated risks have become a significant concern for sustainable development in society. Specifically, the stranding of fossil fuel assets has emerged as a potential factor affecting long-term investments by companies and the stability of financial markets. To understand the impact of asset stranding risk on investor decision-making, this study examines data from Chinese energy-intensive companies listed on the A-share market from 1998 to 2021. The findings of the study are as follows: 1) The risk of stranding fossil fuel assets in Chinese enterprises exhibits an upward trend, particularly after the introduction of the "dual carbon" goal. 2) There is a positive correlation between the stranding risk of fossil fuel assets and stock returns, indicating a premium associated with fossil fuel assets in the Chinese capital market. 3) The risk of asset stranding can be transmitted through the capital market, influencing investors' asset allocation and risk decisions. However, there are variations in decision-making among rational investors. 4) Investors mitigate the risk of stranding assets through various internal and external channels, including green finance, ESG performance, and business transformation. ESG and green finance play a crucial role in guiding investors to adjust their investment portfolios and mitigate the impact of asset stranding risk. This study offers insights for energy-intensive enterprises in China on how to proactively pursue low-carbon transformation and address the risks associated with asset stranding. Additionally, it provides new evidence for investors to reevaluate their involvement with high-carbon enterprises and make informed decisions regarding corporate climate governance.
  • Yue CAO, Ruibo ZHOU, Tianxiao GUO
    Systems Engineering - Theory & Practice. 2025, 45(7): 2226-2244. https://doi.org/10.12011/SETP2024-0027
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Preventing and resolving financial risks and promoting enterprises to "transform from virtual to real" is an inevitable requirement for China's manufacturing industry to realize high-quality development and implement economic structural transformation. Based on the research perspective of intelligent manufacturing, this paper takes the pilot demonstration project of intelligent manufacturing implemented by the Ministry of Industry and Information Technology as a quasi-natural experiment, and adopts a multi-period double-difference research method to systematically investigate whether intelligent manufacturing can effectively promote enterprises to "transfer from virtual to real". The results of the study show that the implementation of intelligent manufacturing significantly reduces the financialization level of enterprises by suppressing the profit-seeking and reservoir motives of enterprises, and the above results are more obvious in enterprises that are not state-owned, have higher minimum wage standards in their respective regions, adopt diversified development modes, and are farther away from the banks and enterprises. The analysis of economic consequences reveals that the implementation of intelligent manufacturing can not only promote the enterprise "de-virtualization", but also promote the enterprise "to the real", and alleviate the crowding out effect of financialization on the enterprise's real investment. This paper further expands the research literature on intelligent manufacturing and enterprise financialization by examining the impact of the implementation of intelligent manufacturing on enterprise financialization, which is of reference value for promoting the transformation and upgrading of manufacturing enterprises and realizing high-quality development.
  • Weihua LIU, Siyu WANG, Zhicheng ZHOU
    Systems Engineering - Theory & Practice. 2025, 45(7): 2245-2263. https://doi.org/10.12011/SETP2023-2605
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    The industry 4.0 technology makes the cooperation of manufacturing enterprises and logistics enterprises more convenient, accelerating the integration of the two industries. Logistics enterprises or manufacturing logistics departments join in the procurement and production of manufacturers, so as to achieve the deep strategic cooperation and "you have me, I have you". However, it is not clear how Industry 4.0 technology affect the integration of the two industries. This study uses qualitative comparative analysis (QCA) and multiple regression, and proposes the influence path of Industry 4.0 technologies including big data, Internet of Things, artificial intelligence, blockchain, augmented reality and 5G on the integration of the two industries from both qualitative and quantitative perspectives, based on the empirical research on Typical Cases of Deep Integration and Innovation Development of Logistics and Manufacturing Industry (2021), and finds that: From the overall path, enterprises can integrate from the innovation, cost and efficiency, such as using the dual factors of process and assets to drive innovation, using the three factors of mode, process and assets to drive cost optimization, and using technology to achieve efficiency optimization. From the perspective of technology, the single application of blockchain, augmented reality technologies with low popularity can improve enterprise efficiency, while more mature technologies require overlapping applications, such as the Internet of Things, big data, artificial intelligence, 5G. Smaller enterprises apply industry 4.0 technology more obvious effects for the integration, but the Internet of Things and big data show opposite characteristics for different enterprises, when the enterprises use them, larger manufacturing enterprises have more significant effects, while smaller logistics enterprises are more obvious.
  • Zhi LIU, Chenyu MING, Xiaoxue ZHENG, Bengang GONG
    Systems Engineering - Theory & Practice. 2025, 45(7): 2264-2281. https://doi.org/10.12011/SETP2023-2466
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    The upcoming "mandatory + voluntary" carbon market mechanism, which allows carbon cap-and-trade mechanisms to coexist with nationally certified emission reduction trading, is a new mechanism that has not been thoroughly studied, which will promote the formation of a new "carbon complementary supply chain" mode of operation. That is, supply chain members promote internal cooperation and complementary advantages of resources through internal trading of carbon emission rights to achieve high-quality, low-cost energy saving and emission reduction. Therefore, under the new mechanism of "mandatory + voluntary", the study of carbon complementary energy supply chain cooperation mode selection and benefit distribution is conducive to further enhance the operational efficiency and synergistic effect of the supply chain, which is of great theoretical significance for promoting the low-carbon transformation of China's energy industry. This paper designs five cooperation models for the carbon complementary energy supply chain composed of emission reduction enterprises, emission limited enterprises and the consumer side, and comprehensively uses non-cooperative and cooperative game theory methods to explore the choice of cooperation models and the distribution of benefits under the "mandatory + voluntary" mechanism and renewable energy subsidy policy. The results of the study show that the cooperation model of the alliance between emission limited enterprises and emission reduction enterprises makes the supply side more competitive relative to the consumption side, with the lowest total quantity of energy products and the highest wholesale price, while the model of the alliance between emission reduction enterprises and the consumption side is just the opposite. Under the premise of the same subsidy level, the cooperation model of the grand alliance eliminates the influence of the double marginal effect, with the highest total profit, which is the optimal cooperation model. The weighted revenue allocation scheme proposed based on the cooperation game model can ensure the stability of the grand alliance under the different subsidy levels, and it has a certain degree of superiority.
  • Qi ZHANG, Jingxian CHEN, Liang LIANG
    Systems Engineering - Theory & Practice. 2025, 45(7): 2282-2295. https://doi.org/10.12011/SETP2023-2630
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    More and more manufacturers are not only selling products through traditional retail channels, but also establishing channels directly targeting consumers to sell products. Faced with the threat of direct channels, retailers can respond to channel competition by switching to external suppliers. Considering the dual channel management problem of manufacturers when retailers hold external choices, a secondary supply chain consisting of one manufacturer and two retailers holding different external choices is constructed. The wholesale price decisions of manufacturers and the switching supplier decisions of retailers are explored, and the impact of outside options by retailers on the manufacturer's optimal channel strategy is analyzed. Research has shown that manufacturers' wholesale price decisions are constrained by retailers' external choices, but adopting customized wholesale price strategies can weaken the constraints of retailers' external choices; Abandoning some or retaining all retailers may become the optimal channel strategy for manufacturers, and abandoning some retailers when the efficiency of direct sales channels is low can improve the profitability of manufacturers; When retailers have limited external selection advantages, closing inefficient direct sales channels may become the optimal channel strategy for manufacturers.
  • Junfei DING, Xujin PU, Xiqiang XIA
    Systems Engineering - Theory & Practice. 2025, 45(7): 2296-2308. https://doi.org/10.12011/SETP2023-2680
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    By considering the inability to independently manufacture and remanufacture core components, suppliers are introduced to provide core components to downstream firms. This paper develops a supply chain model consisting of one supplier, one manufacturer and one remanufacturer. In the supply chain, both the manufacturer and the remanufacturer purchase key components from the supplier to produce final products. We examine the outsourcing and authorization remanufacturing modes. Subsequently, we analyze and compare the equilibrium solutions and corresponding consumer surplus, environmental impact and social welfare under the two remanufacturing modes considering the supplier-led and the manufacturer-led scenarios. The results show that, under the supplier-led scenario, the supplier keeps the procurement prices of core components under the two remanufacturing modes unchanged. Under the manufacturer-led scenario, the supplier changes the procurement price for the remanufacturer; in any scenario, compared to the authorization remanufacturing mode, the outsourcing remanufacturing mode leads to higher collection rate, supply chain profit, consumer surplus, and social welfare, but lower environmental impact. Therefore, the outsourcing remanufacturing mode should be adopted between the manufacturer and the remanufacturer; however, under the outsourcing remanufacturing mode, when the collection scale coefficient is relatively large, the manufacturer should be encouraged to be the leader in the supply chain; on the contrary, the supplier should be the leader in the supply chain.
  • Yehu YUAN, Duanduan WU
    Systems Engineering - Theory & Practice. 2025, 45(7): 2309-2326. https://doi.org/10.12011/SETP2023-2465
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    The digital transformation of enterprises has changed from internal business, organization and business model to digital synergy of production factors and organizational relationship in the upper and lower reaches of supply chain. Based on the data-to-information-to-knowledge-to-wisdom model (DIKW) of data value chain, this paper constructs a multi-stage model of enterprise digital transformation, and uses the data of Chinese A-share listed companies from 2007 to 2021, this paper examines the impact of digital transformation on supply chain resilience. The study found that digital transformation can significantly improve supply chain resilience, with different effects at different stages. In mechanism testing, digital transformation promotes supply chain resilience by promoting information and knowledge spillover in supply chain. Further heterogeneity analysis found that when the enterprise is located in the eastern region, the economic policy uncertainty is high and the industry competition degree is strong, as well as the large-scale, high-tech industry and non-manufacturing industry, for enterprises with low supply chain integration, the impact of digital transformation on supply chain resilience is more significant. Moreover, the digital transformation has the diffusion effect on the upstream and downstream enterprises of the supply chain, which can improve the digital transformation degree and value creation level of the upstream and downstream enterprises. The research results reveal the impact and mechanism of enterprise digital transformation on the resilience of supply chain, provide a new idea for building a resilient supply chain system, and promoting coordinated development of supply chain.
  • Chaoqun YI, Yimin YANG
    Systems Engineering - Theory & Practice. 2025, 45(7): 2327-2339. https://doi.org/10.12011/SETP2023-2536
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Consider a green supply chain composed of a manufacturer and a retailer. We construct two decision models for the retailer: not adopting AI technology and adopting AI technology. Compared to not adopting AI technology, retailers can more accurately predict potential demand when adopting AI technology. The retailer's AI technology adoption strategies and their impact on the green supply chain, environmental benefits, and social welfare are studied. Our paper shows that the retailer's estimations of potential demand and the green input cost coefficient are important factors affecting their AI technology adoption strategies. Specifically, when the retailer underestimates potential market demand, it prefers to adopt AI technology as the estimation bias increases, and as the green input cost coefficient decreases, the retailer prefers to adopt AI technology. On the contrary, when the retailer overestimates potential market demand, it should not adopt AI technology when the green input cost coefficient is low, and when the green input cost coefficient is higher than the given threshold, the retailer should adopt AI technology. Meanwhile, we find that the adoption of AI technology may harm the profits and overall social welfare of the supply chain. Further research shows that both the unit cost of retailer's application of AI technology and prediction accuracy of AI technology can affect the retailer's AI technology adoption strategies.
  • Yuyan WANG, Xiaozhen ZHANG
    Systems Engineering - Theory & Practice. 2025, 45(7): 2340-2357. https://doi.org/10.12011/SETP2023-2142
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    In the context of technology patent licensing, the game models for the supply chain of new energy vehicles have been established based on four management strategies: no introduction, individual introduction by suppliers/manufacturers, and mutual introduction of professional managers. The aim is to comprehensively explore the impact of introducing professional managers on innovation decision-making within companies. Additionally, we have investigated the question of whether supply chain members should introduce professional managers from various perspectives. The study reveals a negative correlation between the innovation level of companies and incentive levels. Enterprises that introduce professional managers experience an increase in profits as their self-set incentive levels increase, but a decrease as the incentive levels set by the other party increase. From the standpoint of enhancing innovation levels and expanding market demand, the decision to have both supply chain members introduce professional managers is considered the most prudent choice. However, a comprehensive consideration of various interests indicates that both parties may opt for two extreme scenarios: Either not introducing professional managers at all or introducing them mutually. From the standpoint of enhancing innovation levels and expanding market demand, the decision to have both supply chain members introduce professional managers is considered the most prudent choice. But, a comprehensive consideration of various interests indicates that both parties may opt for two extreme scenarios: Either not introducing professional managers at all or introducing them mutually. However, the profit optimization of supply chain members can be realized by designing benefit-cost-sharing contract under the mutual introduction of professional managers. This indicates that when all supply chain members introduce professional managers, setting up a proper coordination mechanism is helpful to promote enterprise innovation and maintain the stable operation of supply chain.
  • Xin SUI, Wenqiang DAI
    Systems Engineering - Theory & Practice. 2025, 45(7): 2358-2371. https://doi.org/10.12011/SETP2024-0020
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    In the process of guaranteed delivery of display advertising, the publisher needs to design allocation strategies in advance based on contracts signed with advertisers, delivering impressions to various campaigns according to specified proportions. However, the existence of multi-level demand contracts in reality and the significant uncertainty in impression supply present substantial challenges to this process. To address this issue, this paper proposes a distributionally robust chance-constrained programming model, where the determination of impression demands is regarded as a decision-making process. Adopting a conservative approximation approach, the model is transformed into an easy-to-solve mixed-integer second-order cone programming (MISOCP) model. Besides, considering the problem in which the risk level is a decision variable, a feasible MISOCP model is also designed. Extensive out-of-sample testing, compared with the benchmark model, validates the robust performance of the proposed model and solution method.
  • Jingpeng WANG, Xiaomiao LIN, Pengpeng XIE, Pengfei WANG, Peng LIU
    Systems Engineering - Theory & Practice. 2025, 45(7): 2372-2384. https://doi.org/10.12011/SETP2024-0401
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    To mitigate the high incidence of order cancellations by passengers in ride-sourcing systems and enhance platform operational efficiency, this study adopts a data-driven approach within the theoretical framework of "predict then optimize". By integrating predictive methodologies from data science with operations research optimization techniques, this research analyzes the complex dispatching challenges in ride-sourcing systems, specifically considering passenger order cancellations. The study reveals that: i) The predictive-then-optimization framework effectively simplifies the platform's dispatching optimization problem with consideration of passenger's order cancellations into a linear programming model, which significantly improving the solvability of the model and reducing the difficulty of theoretical analysis; ii) Employing real data, machine learning models can effectively predict whether passengers will cancel orders, thus avoiding the limitations of assumptions inherent in mathematical modeling of passenger decision-making processes; iii) Compared to dispatching models that do not consider passenger's order cancellation behavior, the model proposed in this paper can effectively improve the revenue of the ride-sharing platform. Numerical experiments indicate that as the supply-demand ratio (drivers/passengers) increases, the solutions of the dispatching strategies that consider passenger's order cancellation behavior and those that do not gradually converge; compared to orders with short or long travel distances, orders with medium travel distances contribute more significantly to the platform's revenue; compared to cost-priority and profit-priority strategies, the dispatching strategy that accounts for passenger's order cancellation behavior achieves higher revenue and can effectively reduce the total waiting time of passengers, among other benefits. This paper provides a modeling approach and solution method for the optimization of ride-sharing platform's dispatching considering passenger's order cancellation behavior, offering theoretical reference for the improvement of dispatching strategies.
  • Xingyu DAI, Qunwei WANG
    Systems Engineering - Theory & Practice. 2025, 45(7): 2385-2404. https://doi.org/10.12011/SETP2023-2471
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    The complex computational rules associated with multivariate interval-valued time series (ITS) pose challenges for conventional point-valued methods. This paper treats multivariate interval-valued variable as elements in a high-dimensional random set space and employs the principles of decomposition-ensemble forecasting to construct a MITSSD-FC-IVAR model. Firstly, this paper defines a unit directional discrete weighted inner product and provides formulas for the operations of multivariate ITS distance and variance. Second, this paper defines and calculates the spectral characteristics of multivariate ITS, introducing the multivariate ITS spectral decomposition (MITSSD) method to decompose the original multivariate ITS into multiple sub-sequences with specific frequency components. Finally, to accommodate the frequency domain properties of ITS, this paper builds a Fourier coefficient-interval vector autoregressive (FC-IVAR) model to forecast the sub-sequences of multivariate ITS and ensemble them into the original multivariate ITS predictions. Empirical findings using multivariate energy assets and weather conditions data show that the model proposed in this paper outperforms various benchmark models.
  • Xiaoying CHEN, Jianjun WANG, Shijuan YANG, Suying ZHANG
    Systems Engineering - Theory & Practice. 2025, 45(7): 2405-2417. https://doi.org/10.12011/SETP2023-2444
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    In robust parameter design, the quality of data and model parameter uncertainty often affect the modeling and the acquisition of optimal parameters. In addition, batch effects need to be considered in the actual product manufacturing process. To address the above problems, a robust method has been proposed that can handle batch effects in products or processes and is insensitive to outliers. This method is a multi-response modeling and optimization approach based on Bayesian quantile mixture models. The quantile mixture model is first incorporated into a Bayesian framework. Bayes' theorem is then used to derive the posterior distributions of the model parameters and the Markov chain Monte Carlo algorithm is used to calculate the estimates of the parameters. Next, the multivariate expected loss function is constructed. Finally, the expected quality loss function is minimized to find the optimal parameter settings and the Euclidean distance of the predicted responses between batches is calculated. The simulation example and 3D printing example show that the proposed method is resistant to outliers and reduces the impact of model parameter uncertainty on modeling and optimization.
  • Guodong WANG, Xiaoyang LI
    Systems Engineering - Theory & Practice. 2025, 45(7): 2418-2432. https://doi.org/10.12011/SETP2023-2570
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Sequential accelerated life tests (SALT) can update prior information of model coefficients via collecting lifetime data under different stress levels, and then get better experimental strategy. SALT has been applied in industries successfully, but this method has its drawback. If the model is unknown, then the analysis results may deviate from the true values. To address above-mentioned problem, we propose a SALT for planning and analyzing accelerated lifetime experiment considering model uncertainty. First, we get lifetime data from high stress level, and update the prior information of model coefficients; next, the SALT plan is optimized based on the prior information of candidate models under the criterion that incorporates both asymptotic prediction variance and squared bias; finally, a Bayesian model averaging (BMA) framework is used to derive the posterior model and the posterior distribution for the low lifetime quantile. The proposed method can get the optimal solution of SALT through update prior information sequentially. Furthermore, the method is robust. It can avoid the uncertainty of results caused by a single specified distribution model.
  • Xiaohui HUANG, Xijin TANG
    Systems Engineering - Theory & Practice. 2025, 45(7): 2433-2446. https://doi.org/10.12011/SETP2023-2042
    Abstract ( ) Download PDF ( )   Knowledge map   Save
    Understanding the architecture and genesis mechanisms of echo chambers is important for dismantling the information cocoon, removing filter bubbles, and mitigating the phenomena of information constriction and group polarization in online social media. For this purpose, this study has constructed a novel model for opinion interaction networks, incorporating two indicators for the distribution of user and neighbor stances, and has developed an effect function that can accurately describe the state of echo chambers in the network in terms of distinct stances. To further explore the neighbor effect on the stance of users in the echo chamber, a network causal inference model has been established, and a Dose-Response function is used to quantify the neighbor effects on the stance of users. In the empirical analysis, this study utilizes two datasets about Russia-Ukraine discussions. The experimental results demonstrate that the constructed echo chamber effect function is effective in quantifying the state of the echo chamber in the opinion interaction network. Furthermore, the neighbor effect in the echo chamber not only strengthens the stance of users with similar opinions but also induces users with conflicting opinions to align with the group stance, thereby intensifying the echo chamber architecture.