Core competence is a critical factor for a firm’s sustained competitive advantage, yet there remains considerable debate regarding its measurement and the validity of its indicators. Using annual reports from China’s A-share listed companies between 2012 and 2023,this study construct a novels analytical framework and measurement indicator for corporate core competence by employing a sentence-dictionary similarity machine learning algorithm, combined with named entity recognition (NER) and an intensity lexicon. The validity of the proposed indicator is examined from the perspectives of traditional resources, human resources, brand resources, operational capability, innovation capability, management and control capability, industrial chain collaboration capability, and overall core competitiveness. The results show that the indicators effectively capture a firm’s core competence. Building on this, we examine the application of these indicators in three contexts: Debt financing, equity value, and uncertainty shocks. The results reveal that: 1) The core competence indicator significantly negatively correlates with debt financing costs, effectively signaling a firm’s credit risk; 2) The indicator predicts stock excess returns and market value, helping guide medium- and long-term investment decisions; 3) The indicator demonstrates a firm’s“immunity” to uncertainty shocks, such as the COVID-19 pandemic. This study presents a new methodological framework for measuring core competence, providing practical implications for investor and regulatory decision-making, as well as contributing to the development of world-class enterprises in China.
Since 2000, China has witnessed large-scale construction of transportation infrastructure alongside the rapid development of the digital economy. The interplay between these two factors has substantially influenced the formation and deepening of a unified national market. However, few studies have investigated them in conjunction. Drawing on industrial and commercial enterprise registration data, this study estimates the scale of the digital economy across cities; Meanwhile, employing a pixel-level assessment of traffic quality, it constructs an intercity travel-time matrix to evaluate the effects of transportation infrastructure and the digital economy on the unified national market. The findings reveal that: 1) The spatiotemporal compression driven by transportation infrastructure construction significantly facilitates the flow of goods, people, and information across cities, effectively mitigating market segmentation. 2) Reductions in road transportation time exert a greater positive impact on market integration than do reductions in rail transportation time. From a temporal and spatial perspective, the contribution of shortened rail freight and road transportation time to market integration has gradually weakened in eastern urban agglomerations, whereas no such diminishing trend is observed in their western counterparts. 3) The digital economy and transportation infrastructure exhibit a synergistic effect in jointly shaping a unified national market, primarily through deep integration that promotes the flow of goods and information, dismantles administrative barriers, and reduces transaction costs. This study contributes to an enhanced understanding of transportation infrastructure and the digital economy, offering insights for building a high-quality, unified national market.
To address the profound changes unseen in a century, China has proposed a new development paradigm featuring “dual circulation”, with the domestic economy as the mainstay and the domestic and international economies reinforcing each other. This paper constructs a dynamic and static CGE (computable general equilibrium) model system based on China’s multi-regional input-output table. The dynamic CGE model is used to simulate the impacts of population growth, tax adjustments, and effective investment on the construction of the dual circulation pattern and related economic variables. Based on the simulation results, an appropriate dual circulation proportion coefficient is selected, and the comprehensive impact of its synergistic effects with technological level and tax changes on industrial development is analyzed through the static CGE model. The research results show that population, tax rate, and effective investment have differentiated impacts on the dual circulation proportion coefficient. Specifically, the natural growth rate of population and tax rate hikes have relatively limited effects in promoting dual circulation, while increased effective investment significantly drives the expansion of the dual circulation proportion coefficient. Based on this, to ensure stable economic operation and promote industrial development, the dual circulation proportion coefficient should be maintained within a range of 1–2 times, while ensuring that technological progress is coordinated with the development direction of dual circulation. Dynamic simulations indicate that moderately increasing the tax rate (5%–15%) can accelerate industrial transformation and upgrading; when the dual circulation proportion coefficient is 1.2 times, the conditions for industrial development are most favorable. It is worth noting that in constructing the dual circulation pattern, one should avoid blindly pursuing excessive expansion of the proportion of domestic circulation to prevent negative impacts on the economic system. Through rigorous theoretical modeling and numerical simulations, this paper provides quantitative evidence and policy implications for the transformation of traditional development models, the scientific construction of the new dual circulation development paradigm, and industrial upgrading.
The introduction of the Green Financial Systems Opinions in 2016 (hereinafter referred to as the “Opinions”) has positioned China as the world’s first economy to establish a comprehensive green financial policy system. In this crucial period of economic restructuring and development mode transformation, it is worth exploring how the green finance system can optimize capital market resource allocation and assist enterprises in green transition. Drawing upon data from A-share listed companies between 2011 and 2021, this paper empirically examines the impact of implementing the Opinions on total factor productivity (TFP) of green enterprises, along with its underlying mechanisms, using the difference-in-differences method. The findings demonstrate a significant improvement in TFP for green companies after the Opinions, particularly among state-owned enterprises and enterprises in the three economic circles. Mechanism analysis reveals that the relaxation of financing constraints and expansion of credit scale serve as primary channels for empowering financial support towards green transition. Additionally, the issuance of green bonds and disclosure of environmental reports exhibit effects on enhancing TFP. There are substantial differences in TFP among enterprises based on their disclosure practices regarding whether and how they disclose (non-disclosure, qualitative disclosure, or quantitative disclosure of) environmental governance and performance. This study provides some useful insights into evaluating micro-economic benefits of policies related to green finance and further promotion.
With the rapid development of the platform economy, review manipulation has become an unethical means for platform sellers to obtain web traffic. Integrating the supply-side traffic competition perspective and demand-side consumer defense perspective, we study the spillover effect of review manipulation on competitors with different competitive relationships (direct vs. indirect) and positions (top-ranking vs. non-top-ranking). The results show that: 1) Seller’s review manipulation has a significant spillover effect on its direct competitors. A 1% increase in the intensity of review manipulation, corresponding to a 1% increase in the proportion of fake positive reviews, results in an average 0.039% increase in the sales for the top-ranking direct competitors, while sales for the non-top-ranking direct competitors decrease by an average of 0.026%. 2) Seller’s review manipulation has no significant spillover effect on its indirect competitors. 3) Brand strength and product price weaken the positive spillover effect of review manipulation on top-ranking direct competitors and strengthen the negative spillover effect on non-top-ranking direct competitors. The conclusions challenge the intuition that sellers’ review manipulation could “hurt the others to benefit oneself” and reveal that review manipulation by direct competitors may actually benefit the top-ranking sellers on the platform. This paper provides implications for the governance of platform misconduct as well as the high-quality development of the platform economy.
Better utilizing the financial engine to help farmers increase their income and become rich is the necessary way to achieve comprehensive rural revitalization. This paper takes the selection of credit towns in the Guiding Opinions of the People’s Bank of China on Promoting the Construction of Rural Credit System (PBC [2009] No. 129) as a shock. Based on rich and various economic and geography data sources, we utilize spatial geographical discontinuity regression design to examine the impact of the rural credit system, on the income level of rural residents and the mechanism. We find that the construction of the credit system increases the per capita disposable income of rural residents by 11.18%. Mechanism analysis shows that the construction of the rural credit system mainly promotes the establishment of farmers’ professional cooperatives and drives employment by providing credit support, thereby increasing farmers’ income. Further heterogeneity analysis finds that the income-increasing effect of rural credit system construction mainly affects areas with higher levels of human capital, younger population structures, and richer financial resources. Our research conclusions show that effectively stimulating the leading role of professional cooperatives is a key way to achieve common income growth, which provides specific policy implications for how finance can help rural revitalization.
The establishment of the carbon emission trading market enriches the set of strategies enterprises can employ to address carbon risks. Could this lead to optimized decision-making, reduced risks, and lower financing costs for enterprises? To clarify the profound impact of carbon trading on the financing of micro-enterprises, this paper constructs a difference-in-difference model to examine the relationship between carbon trading and equity financing costs. The results reveal that, compared to enterprises not included in the carbon trading list, those participating in carbon trading experience a significant reduction in equity financing costs, while those removed from the list see a significant increase in equity financing costs. Mechanism tests show that enterprises participating in carbon trading have smaller bid-ask spreads, lower idiosyncratic volatility, less financing constraints, higher cash holdings, higher ESG scores, and a lower likelihood of being sued. This suggests that carbon trading reduces equity financing costs by lowering cash flow risks, information risks, and reputation risks. Further analysis indicates that the higher the marketization degree of carbon trading, the greater the reduction in equity financing costs. The findings of this paper highlight the logic of how carbon trading optimizes corporate decision-making, reduces risks, and ultimately lowers equity financing costs through market mechanisms. This expands the relevant research on the impact of carbon trading markets on micro-enterprises and provides empirical evidence support for the construction of a unified national carbon market.
To describe the underlying mechanism of jump clustering in stock prices, based on the extrapolation theory of behavioral finance, this paper constructs an asset pricing model which can capture the characteristics of jump clustering in stock prices under the continuous-time consumption asset pricing framework, and explores the effects of investors’ extrapolation bias of stock price jump on market equilibrium and asset prices. This paper finds that the extrapolation bias of stock price jump can lead to the jump clustering in the same direction. Moreover, compared with the traditional consumption asset pricing model with rational investors, the model in this paper can generate higher excess returns and greater stock price volatility, which provides an explanation for the jump clustering, equity premium puzzle and excessive volatility puzzle in a unified framework. Besides, the model calibration and simulation analysis results based on China Securities Index (CSI) 300 stock index show that the extrapolation bias of stock price jump makes the model better match the statistical characteristics of financial assets in the real market, and effectively describe the characteristics of jump clustering in real stock returns.
With the development of information technology and the rise of e-commerce, pharmaceutical B2B e-commerce as an emerging business model has gradually occupied an important position in the healthcare field. To optimize inventory management between pharmaceutical e-commerce platforms and primary healthcare institutions and to improve the responsiveness of the pharmaceutical supply chain, this study explores an optimization scheme for horizontal inventory sharing between pharmaceutical e-commerce platforms and primary healthcare institutions under a transshipment strategy. Based on the Stackelberg game theory, a two-level supply chain model consists of a single B2B pharmaceutical e-commerce platform, multiple heterogeneous primary healthcare institutions, and various types of drugs. This study analyzes the pricing, ordering decisions, and influencing factors of pharmaceutical supply chain members with and without a horizontal inventory sharing strategy. Additionally, a case study based on data from 1Yao.com is conducted to investigate the impact of changes in the transshipment cost-sharing ratio on supply chain decisions. The study finds that without horizontal inventory sharing, pharmaceutical supply chain members typically respond to demand fluctuations by raising prices or increasing order quantities, which is not conducive to improving the overall supply chain efficiency. Implementing a horizontal inventory-sharing strategy can achieve Pareto improvement in the supply chain and maximize overall supply chain profits while enhancing patients’ healthcare experience. Moreover, in different market environments and cost allocations, the transshipment cost-sharing ratio can significantly influence supply chain decisions. Pharmaceutical supply chain members can negotiate the transshipment cost-sharing ratio to determine drug order quantities and wholesale prices, thereby enhancing their own profits and achieving a win-win situation.
The rise of the platform economy has catalyzed embedded innovation practices between platform enterprises and technology-based startups. A robust governance mechanism plays a crucial role in enhancing collaboration efficiency among ecosystem members, thus ensuring the effective functioning of the platform ecosystem. From the perspective of embedded innovation, this study constructs a two-stage dynamic evolutionary game model involving platform enterprises, technology-based startups, and incumbent firms. It systematically analyzes the behavioral stability conditions of each actor across different stages and, through numerical simulations, examines the impact of initial intentions, hindering factors, and driving forces on the realization of innovation collaboration. The study finds that: 1) Initial willingness to participate constitutes a foundational variable in the evolution of strategic behavior, and the selection of strategically aligned partners is a critical prerequisite for initiating and advancing innovation collaborations; 2) The revenue embedding factors exert a dual regulatory effect. The minimum threshold of ecological return required by each system participant directly shapes strategic choices, while any participant’s adjustment in aggregate input redefines the feasible range of benefit-sharing coefficients, thereby influencing the overall stability of cooperation; 3) The organizational embedding factors demonstrate stage-specific characteristics. During the resource-embedded stage, strong relational control intentions tend to hinder cooperation; in contrast, during the ecological-embedded stage, stronger network connections enhance the willingness to cooperate among system members. Notably, platform enterprises, as central nodes within the network, play a pivotal role in stabilizing innovation collaborations; 4) Normative and guiding factors jointly function as dual drivers of behavioral governance. The former deters opportunistic or non-cooperative behaviors by reinforcing breach penalties and reducing the payoff of such behaviors, whereas the latter stimulates members’ innovation motivation through enhanced incentives. However, when the intensity of such incentives exceeds a reasonable threshold, it may ultimately lead to the failure of innovation activities within the ecosystem. Overall, the study has important implications for the governance strategies of platform enterprises and technology-based startups engaged in embedded-innovation practices.
To analyze the implementation effect of the progressive carbon tax policy under different remanufacturing modes, a supply chain game model consisting of original manufacturers and remanufacturers is constructed based on the progressive carbon tax policy, and the impacts of the progressive carbon tax policy on the market competition, consumer surplus, and environmental benefits of the three remanufacturing modes are comparatively analyzed, to establish the optimal remanufacturing mode under the progressive carbon tax, and to achieve the optimization of the economic and environmental benefits at the same time. The main conclusions of the study are: 1) An increase in the progressive carbon tax rate will increase the retail price per unit of new products and per unit of remanufactured products; however, it will also increase the recycling rate of used products and the sales volume of remanufactured products, while decreasing the market sales volume of new products. 2) The independent remanufacturing model can achieve a higher recycling rate of used products and consumer welfare, and significantly reduce the environmental impacts, so the independent remanufacturing model is optimal when the intensity of intellectual property rights protection is weak. 3) The outsourcing remanufacturing model shows better resource recycling efficiency and environmental benefits than the licensed remanufacturing model; therefore, the outsourcing remanufacturing model is optimal when the intensity of intellectual property protection is high.
In the fierce market competition, manufacturers’ investment in supplier quality improvement not only enhances overall product quality but also exerts a decisive influence on corporate revenue. This paper constructs a supply chain consisting of one original equipment manufacturer (OEM) and two suppliers. Using a differential game model, this study systematically investigates how environmental risk, reputational risk, and their dual combination affect the product quality of supply chain members and the OEM’s optimal investment strategies for supplier quality improvement. The results show that: 1) Both environmental and reputational risks faced by the OEM impose negative impacts on the product quality of the OEM and its upstream suppliers. Such risks directly undermine the OEM’s product quality and market sales, which in turn reduces the order volume and revenue of upstream suppliers. To sustain profitability, suppliers tend to lower their product quality standards. 2) The OEM’s investment strategies under single or dual risk scenarios are determined by the OEM’s investment efficiency in supplier quality improvement and the efficiency of the investee supplier in achieving product quality. When the cost coefficient of achieving product quality by the supplier is less than a certain threshold, indicating higher efficiency in achieving product quality, the OEM does not need to invest. When this efficiency is low, simultaneously, when the cost coefficient of OEM’s investment in improving supplier product quality is less than a certain threshold, indicating higher efficiency of OEM’s investment in improving supplier product quality, then investing in supplier quality improvement benefits the OEM’s profit growth; otherwise, not investing is its optimal strategy. 3) The cost coefficient threshold for OEM investment increases monotonically with environmental risk but decreases monotonically with reputational risk. This means that an increase in environmental risk will provide an opportunity for OEMs with lower efficiency in product improvement to profit through supplier investment, while an increase in reputational risk will hinder such OEMs from achieving profit growth through investment.
This paper develops a noncooperative-cooperative biform game model to analyze the relationship between production competition and carbon emission reduction (CER) technology cooperation between two manufacturers with different CER efficiency. By integrating noncooperative and cooperative games, the optimal production strategy, cooperation strategy, and profits for two manufacturers are solved. We examine two scenarios of simultaneous production competition and CER technology cooperation: 1) The manufacturer with lower CER efficiency (manufacturer L) and the manufacturer with higher CER efficiency (manufacturer H) compete in production and collaborates with a technology provider through a technology transfer contract (PT model); 2) both two manufacturers compete in production and cooperate in CER through cost-sharing (MT model). The effects of factors such as the CER efficiency and carbon price on manufacturers’ optimal strategies, profits, and carbon emissions are explored. Finally, we derive conditions under which the manufacturer with lower CER efficiency would choose to engage in different CER cooperation models. The key findings are as follows: 1) The higher the carbon price, the higher (lower) the profit of manufacturer L (H) under the PT model, and the lower (higher) the profit under the MT model. 2) When the CER cost coefficient and carbon price are low, the CER levels of both manufacturers are higher in the PT model compared to the MT model. 3) Manufacturer L has the lowest carbon emissions and highest profits in the PT model, while manufacturer H has the lowest emissions in the PT model and the highest profits in the MT model. This study provides a unified analytical framework for solving the coexistence of production competition and CER technology cooperation. The findings offer valuable insights for decision-making in manufacturing enterprises and for selecting appropriate CER cooperation mode.
With increasingly serious imbalance of global water resources supply and demand, water resources cooperation in transboundary rivers has become an important strategy for resolving water resources conflict among riparian countries and maintaining regional peace and stability. In order to analyze the influence of random factors and population size on the evolution of water resources cooperation in transboundary rivers, this paper constructs a stochastic evolutionary game model of water resources cooperation in transboundary rivers based on the Moran process. The benefits and population size conditions under which the reciprocal strategy becomes the dominant strategy under the strong selection scenario of expected benefit dominance and the weak selection scenario of random factors dominance are obtained respectively. The influence of each parameter on the strategy evolution of riparian countries under the strong and weak selection scenarios is compared and analyzed by numerical simulation. The results show that strategic options of riparian countries mainly depend on the cost-benefit factors, random factors, and the number of riparian countries. The conditions for the reciprocal strategy to become an evolutionary stable strategy are more stringent under the strong selection scenario than under the weak selection scenario. Costs of the tough strategy, synergistic benefits, regulatory probability of basin management agencies, and reputation loss are positively correlated with riparian countries to choose the reciprocal strategy, while costs of the reciprocal strategy, persuasion costs, and water resources benefits are negatively correlated with riparian countries to choose the reciprocal strategy. The study can provide a reference for promoting water resources cooperation in transboundary rivers.
In recent years, the frequent occurrence of extreme weather and other risks has posed significant challenges to the stable supply of water resources in water diversion projects. With the deep integration of information technology and water diversion projects, leveraging the advantages of digital technology to enhance resilience has become a critical issue for the sustainable development of water diversion projects in China. This paper constructs a research model on the impact of water resource scheduling digitalization on the resilience of sustainable supply chains in water diversion projects from the perspectives of single and multiple configurations. Using 425valid data samples, an empirical analysis was conducted employing structural equation modeling (SEM) and fuzzy-set qualitative comparative analysis (fsQCA) methods. The paper found that: 1) Value co-creation is a key mediating variable in the impact of water resource scheduling digitalization on the resilience of sustainable supply chains in water diversion projects, with environmental dynamism positively moderating the relationship between water resource scheduling digitalization and value co-creation. 2) Neither water resource scheduling digitalization nor value co-creation alone can achieve a high level of resilience in the sustainable supply chain of water diversion projects. There exist multiple pathways to generate conditions for high absorptive capacity, high responsiveness, and high transformative capacity. The paper systematically explores the mechanisms by which water resource scheduling digitalization affects the resilience of sustainable supply chains in water diversion projects, aiming to provide theoretical guidance and managerial insights for enhancing the resilience of sustainable supply chains in water diversion projects.
Motivated by the fact that bidders’ decision-making is generally reference-dependent and that salient prices such as the buy-it-now price and the reserve price often serve as important reference anchors, this paper moves beyond the traditional assumption of exogenous reference points. It introduces a generalized reference point jointly determined by the reserve price and the buy-it-now price and constructs a buy-it-now auction model. Theoretical analysis characterizes the general regularities of bidding behavior and pricing strategies, while numerical analysis under a linear specification reveals the underlying economic mechanisms. The results show that the equilibrium bidding strategy is monotonically increasing in the bidder’s private valuation and in the reference point, and decreasing in the bidding cost. Under high valuations or a high reference point, bidders are more inclined to accept the buy-it-now price. The seller’s optimal strategy is to coordinate the reserve price and the buy-it-now price to guide bidders’ reference expectations, with the core requirement that the buy-it-now price is accepted with positive probability in equilibrium. Numerical results indicate that the reserve price exhibits different patterns of variation with respect to the effect intensity across various weight levels, whereas the buy-it-now price declines as the effect intensity and the weight increase. Additionally, an increase in the reference point raises the expected revenues of the seller, the auctioneer, and the overall system, as well as the bidder’s expected payment, whereas a decrease has the opposite effect.
To address the joint optimization problem of multi-robot task allocation and path planning in robotic mobile fulfillment systems, this paper proposes a novel hierarchical multi-agent deep reinforcement learning approach, which effectively resolves the challenges of dimensional explosion and dynamic coupling in the joint optimization. In the method, each robot is modeled as a dual-layer agent: The upper layer handles task selection while the lower layer manages path planning. This hierarchical design mitigates the curse of dimensionality in joint optimization and enhances efficiency through inter-layer collaboration. Then a two-layer Markov decision process model is established, which incorporates task progress weight factors along with collision and congestion penalties into the reward function to guide the agents to learn optimal policies according to dynamic environmental changes. Finally, an improved hierarchical multi-agent advantage actor-critic algorithm is proposed to solve the model. Experimental results demonstrate that the proposed method outperforms baseline algorithms in multiple key metrics, including average return, total picking time, and total picking path length, with its advantage becoming more pronounced as the problem scale increases.
A class of travel decision-making problems with the consideration of network characteristics are explored, where travelers encounter simultaneously the negative impact of road congestion and the positive impact of social interactions when making choices. Based on the route choice scenario, a logit-based discrete choice model with social interactions is constructed. Analytical results reveal that social interactions can lead to the existence of multiple stochastic user equilibria, even in simple networks with two parallel routes. Further, the range of social interaction strengths that ensure a unique equilibrium under heterogeneous routes is provided. For general traffic networks, the stochastic user equilibrium assignment problem is transformed into an equivalent nonlinear mathematical programming model, which can be resolved using the method of successive averages (MSA). Numerical examples demonstrate that the impact of social interactions on total system travel time is non-monotonic, with potential to reduce congestion, particularly when the intensity of social interactions is moderate.
With the in-depth promotion of the rural revitalization strategy and the continuous development of rural distributed energy system (RDES), the participation of RDES in the external energy market competition has become increasingly common and intense. Aiming at the problem of multi-system co-operation to participate in the external energy market game and benefit allocation, this paper proposes a hybrid game optimization model for rural distributed energy system group (RDESG) considering electricity-carbon-biomass trading. Firstly, the RDES operation structure and equipment model are constructed, and multiple inter-RDES electricity-carbon-biomass P2P trading paths are designed. Secondly, a Stackelberg game decision model is constructed between rural distributed energy system group and a higher-level energy network, in which the EN aims at revenue maximization, considering energy transmission and energy sales price constraints, and the RDESG aims at operation cost minimization, considering inter-system energy trading constraints and equipment operation constraints. Then, an asymmetric Nash bargaining model for the lower cooperative game is constructed by combining the marginal contribution rate and the multi-energy sharing contribution rate of electricity, carbon and biomass. Finally, based on the KKT condition, the Stackelberg game is converted into a single-layer mixed-integer linear programming problem by using the Big-M method and strong dyadic theory, and the cooperative game is solved by combining the alternating direction multiplier method (ADMM). The results show that the hybrid game optimization model proposed in this paper improves the utilization rate of energy and the environmental benefits, and realizes the interconnection, energy sharing and win-win situation among RDESs.
In high-quality process monitoring with the time between events following exponential distribution, there is an issue of inaccurately inferring controlled parameters due to the insufficient historical sample sizes. To address this issue, this paper proposes a process monitoring scheme based on Bayesian transfer learning. It models the time between events in high-quality processes using source and target domains that follow an exponential distribution. The correlation between two similar processes is represented through the joint prior distribution of variable parameters. Bayesian inference and generalized hypergeometric functions are utilized to derive the closed forms of the posterior distribution, posterior expectation, posterior variance, and posterior cumulative distribution function under the in-control state of the target process, thereby providing an efficient online real-time monitoring control chart for phase II. Finally, we employ numerical simulations and case studies, using the average time to signal (ATS) metric to verify that our proposed exponential distribution-based Bayesian transfer learning scheme can effectively leverage source domain data to enhance the monitoring performance of the target process.
The analysis of feedback loop structure is a key factor in generating development strategies for complex systems, and the polarity of the feedback loop is an important basis for determining the direction of strategies, but the polarity transition of the causal chain directly affects the judgment of the polarity of the feedback loop in the system; And the analysis of policy simulation is an important basis for whether and how system development strategies are implemented. How to integrate the advantages of feedback loop structure and policy simulation analysis to generate management strategies with feasibility and high accuracy is an important issue to be required further research. In response to the above issues, this article proposes the key causal chain feedback loop structure and simulation analysis strategy generation method: Firstly, the polarity transition situation of the causal chain and feedback loop related to the strategy factors are analyzed through numerical simulation calculation of system operating parameter values; Secondly, based on this, the feedback loop structure is used to deeply analyze the system behavior and generate management strategies for system development; Finally, through the simulation analysis of future benefits of strategy implementation, management measures to promote system development are proposed. Taking the high-quality development of the rural ecological agriculture industry chain as an example, the application research of this method is conducted. Three management strategies are proposed from deep integration of the industry chain, green production, and diversified high-quality agricultural product supply. The feasibility of this method in determining management strategies of dynamic complex system are verified. The new key causal chain feedback loop structure and simulation analysis strategy generation method can generate highly implementable and reliable management strategies for the development of dynamic complex systems.
Accurate forecasting of consumer demand is essential for understanding demand shifts and optimizing corporate resource allocation. However, existing models fail to effectively reveal the dynamically implicit synergistic relationships between corporates (i.e., the evolving correlation in inter-corporate business interactions) and their combined effect with the explicit spatial structure (i.e., the geospatial distance of corporates). This combination can effectively capture the interplay between the relatively static geographic distances of corporates under the spatial agglomeration effects and the evolving correlations in inter-corporate business interactions in the real business environment, which is critically important for the service industry characterized by diverse and rapidly changing consumer demands. In this paper, we propose the SF-STAGCN (sequential feature based spatial-temporal attention graph convolutional network) model, which dynamically adjusts relationship weights by the spatial-temporal attention mechanism to capture the similarity of the inter-corporate business characteristics over time; it combines with the graph convolutional neural network to provide insight into the joint role of both implicit synergistic relationships and explicit spatial structures. Our experimental results demonstrate that the SF-STAGCN model surpasses existing baseline models. The dynamic features, extracted from time-varying online reviews, enhance the representation of the dynamic implicit synergistic relationships among corporates, which in turn, significantly improve the predictive accuracy of the model. This study provides theoretical support for the service industry to accurately analyze consumer demand and optimize business strategies.
As one of the most important energy resources in China, coal is affected by various market factors, resulting in highly random fluctuations and non-stationarity of coal price time series data. In order to depict the above characteristics, this paper puts forward a coal price forecasting framework, which considers multiple multi-scale characteristics to drive two-stage parameter optimization, and predicts the Bohai-rim steam-coal price index. Firstly, multivariate decomposition algorithm is used to extract and reconstruct multi-scale sub-sequence features. Secondly, the time series model, artificial intelligence and deep learning technology are utilized to predict the reconstructed high, medium and low frequencies, respectively. Finally, based on “decomposition and integration” strategy and combining intelligent algorithm, dynamic optimization of artificial intelligence model parameters and combined weights of reconstructed sequences is carried out to achieve the prediction of coal price index. The results show that the proposed framework has good performance and robustness, which is better than the benchmark model, and can be used for short-term prediction of coal price.
The frequent occurrence of global forest fires highlights the importance of forest fire prediction in disaster management, in which the prediction of forest fire spread trend is a very key link. In this paper, based on McArthur fire spread model, a directed graph network reflecting the real fire spread trend is effectively constructed, and a fusion multi-index link prediction model IMI-LP is proposed to predict the forest fire spread trend. Based on the dynamic change trend of fire and the static properties between nodes, the model fused and learned the feature vectors containing multiple dynamic and static indicators, and realized the high-precision prediction of extreme long-distance forest fire spread events. The empirical analysis based on the historical fire data of Australia in November 2022 shows that compared with the traditional link prediction algorithm, the IMI-LP model has shown better performance in practical applications, with the AUC value on the test set reaching 99.97% and the average accuracy AP reaching 86.65%. It provides a powerful decision support for the allocation of forest fire emergency resources.
With global warming, Arctic route has become a significant shipping route in recent years. However, the short opening history of Arctic route causes the limited historical data on Arctic voyage, which makes great challenges for long-term forecasting of the voyage. Besides, the highly volatile characteristic of the data makes the forecasting more difficult. To tackle the issue, we propose a modified data-fusion method:Decomposition-recombination-fusion-forecasting (DRFF). This framework of first decomposition and then fusion can reduce the difficulty of model construction and improve the accuracy of fusion prediction. In this framework, the STL method is used to decompose each dataset (the dataset of Arctic voyage or that of Arctic Sea ice area). Then, combine two trend components and forecast the trend term of the voyage with slope fusion and prediction algorithm. Simultaneously, put together two seasonal components and predict the seasonal term of the voyage with an LSTM-CNN-DNN fusion forecasting method. Furthermore, the residual component of the voyage is forecasted with the gray wave model. Finally, the three forecasting results are summed to obtain overall predictions for the voyage. Experimental results show that the MAPE of DRFF model in 12-step-ahead forecasting is only 6.08%. Compared with previous fusion forecasting methods, our model has higher accuracy and reliability.
