From the perspective of supply chain finance empowered by blockchain technology,this paper proposes a new theoretical model of bank run embedding a two-echelon supply chain.Through theoretical derivation and numerical examples,we explain the endogenous mechanism of supply chain finance and the application of blockchain technology affecting bank run risk.Our model demonstrates that supply chain finance and blockchain technology affect bank run risk through banks'loan screening efforts.Blockchain-enabled supply chain finance has an asymmetrical effect on banks'loan screening efforts,which leads to a similar effect on alleviating informa-tion asymmetry between banks and enterprises.Furthermore,there may be an inher-ent conflict between the role of blockchain-enabled supply chain finance in mitigating credit rationing for SMEs and maintaining financial stability.When blockchain tech-nology alleviates the information asymmetry between banks and enterprises,and the probability of high-quality non-core enterprises is at a high level,applying blockchain to supply chain finance will improve the credit availability of non-core enterprises and lead to a higher risk of bank runs.For this reason,the regulatory authorities can resolve this conflict by appropriately tightening capital regulatory constraints.
In this paper, the problem of extended dissipative sliding mode control for discrete-time nonlinear singular systems are addressed. First, the T-S fuzzy polynomial singular systems are constructed to represent nonlinear singular systems with parameter uncertainties. A different delay-dependent sliding mode surface is built based on the dynamic event-triggered mechanism. Then, the sliding mode control law is obtained, which guarantees the reachability of the closed-loop system. The admissibility with extended dissipativity is established by SOS. Finally, a numerical example is given to demonstrate the correctness of the proposed strategy.
Multi-task learning (MTL) improves the performance achieved on each task by exploiting the relevant information between tasks. At present, most of the mainstream deep MTL models are based on hard parameter sharing mechanisms, which can reduce the risk of model overfitting. However, negative knowledge transfer may occur, which hinders the performance improvement achieved for each task. In this paper, for situations when multiple tasks are jointly trained, we propose the adaptive hard parameter sharing method. On the basis of the adaptive hard parameter sharing method, the number of nodes in the network is dynamically updated by setting a continuous gradient difference-based sign threshold and a warm-up training iteration threshold through the relationships between the parameters and the loss function. After each task fully utilizes the shared information, adaptive nodes are used to further optimize each task, reducing the impact of negative migration. By using simulation studies and instance analyses, we demonstrate theoretical proof that the performance of the proposed method is better than that of the competing method.
Zonoid depth, as a well-known ordering tool, has been widely used in multivariate analysis. However, since its depth value vanishes outside the convex hull of the data cloud, it suffers from the so-called `outside problem', which consequently hinders its many practical applications, e.g., classification. In this note, we propose a new class of \emph{$L_q$-norm zonoid depths}, which has no such problem. Examples are also provided to illustrate their contours.
Singular two-point boundary value problems arise in a variety of applied mathematics and physics.It is a classical problem and many researchers have done a lot of research work on this issue.In this paper,we apply cubic B-spline to explore the numerical solutions of a class of singular two-point boundary value problems.The paper method is primarily based on the super convergence in approximating second-order derivative values at the knots by the combination of second-order derivative values of cubic B-spline.The paper proposed cubic B-spline possesses super con-vergence in approximating the first-order derivative/second-order derivative of given function and thus the approximation order of our method reaches fourth order.Some numerical experiments are provided to demonstrate the effectiveness of our method compared to the other existing methods.
For a class of discrete-time Lipschitz nonlinear systems with actuator faults,a reliable preview tracking controller design method based on linear matrix inequality(LMI)is proposed.Firstly,with the aid of the difference method and the state augmentation technique,and fusing the preview information of reference signal and disturbance,an augmented error system is constructed,which transforms the reliable preview tracking control problem into a reliable stabilization problem.The use of the differential mean value theorem overcomes the difficulty of dealing with the difference of nonlinear functions,making the augmented error system be a formal linear parameter varying system.Then,a static output feedback reliable controller is proposed,and a sufficient condition for the asymptotic stability of the closed-loop system is given in the form of LMI.Based on this,the reliable preview controller of the original system is obtained.Finally,a numerical example is given to verify the effectiveness of the proposed design method.
The volatility of commodity market pricing has made it a new risk point for the financial system.However,the question of how risk spillover between the com-modity market,stock,and oil markets varies across time-frequency domains remains unclear.Therefore,this paper uses the time-frequency spillover approach proposed by Baruník and kárehlík to analyze risk spillover between commodity,stock,and oil markets.Also,the paper further discusses the portfolio and risk hedging of commodi-ties in different sectors.First,the paper finds that commodity,stock,and oil markets are relatively well integrated and that the risk spillover rises rapidly during crisis events.And,throughout the sample period,the oil market is the transmitter of risk,while the stock market is the receiver of risk.Second,risk spillovers show heterogene-ity across time-frequency domains,and cross-market risk spillovers are dominated by short-term time-frequency domains.Finally,precious metals and grains are particu-larly well suited to hedge against oil and stock assets.Our results are an important guide to risk management for Investors and Regulators in stock,oil,and commodity markets.
Due to the uncertainty and inconsistency of internal parameters of posi-tion subsystems on the operation side and transmission side of the cold strip rolling mill,the dynamic performance of the hydraulic position system may differ,which of-ten leads to inconsistent thickness of the produced sheets.In this paper,a parameter adaptive synchronous control strategy based on the fully actuated system method is proposed.First,an identification model and a fully actuated system of the hydraulic position system are established.Then,the unknown parameters of the subsystems are identified by using the recursive least squares algorithm based on the identifica-tion model.Subsequently,a parameterized controller is designed for the established fully actuated synchronous system model to achieve the stability of the closed-loop system and the synchronization of the two subsystems.Finally,a simulation result demonstrates the effectiveness of the proposed synchronous control scheme.
In recent years,probabilistic linguistic term sets have become a research hotspot in the recommendation field due to their advantages of being able to accu-rately convey the preferences of decision makers and effectively deal with uncertain information in the decision-making process.However,relative to personalized recom-mendation scenarios,research on probabilistic linguistic term sets in non-personalized recommendation has seldom been involved thus far.This paper combines the fea-tures of probabilistic linguistic term sets with non-personalized recommendation and proposes a recommendation algorithm for non-personalized products based on the horizontal comparison among various probabilistic linguistic term values to enrich the research of non-personalized recommendation algorithms.First,the horizontally comparable relationship between multiattribute products is realized by completing the data of probabilistic linguistic terms.Subsequently,the standardized probabilis-tic linguistic term sets are obtained by using the standardized equation.Based on this,the ranking matrix of non-personalized products is constructed.Finally,with the solving method of eigenvalues and eigenvectors,the general recommendation ranking for non-personalized products is obtained.Taking the movie rating data in ml-latest-small in the MovieLens dataset as an example,effective recommendation results are obtained.Compared with the non-personalized recommendation method based on the vague set,the reliability and scientificity of the algorithm are verified.The purpose of this study is to provide a new idea for the application of probabilistic linguistic term sets in the field of non-personalized recommendation.
It is very important for enterprises to adopt effective service strategy and reasonable service pricing for customers to stabilize market demand.This paper considers two service mechanisms of differentiated pricing with service priority setting and unified pricing without service priority setting,and studies the service mechanism selection and pricing of service provider in view of the waiting aversion of regular customers caused by increased waiting time during service priority setting.The results show that when the market size of priority customers is small and the degree of waiting aversion of regular customers is large,service providers should not set service priority for unified pricing.Otherwise,the service provider should set service priority for differentiated pricing.Moreover,the service provider can further improve the optimal revenue by increasing the service rate during differentiated pricing.
As an important part of urban demand-response transportation,cus-tomized bus network planning directly affects passenger satisfaction and system op-erating costs.In order to solve the morning and evening peak commuting customized bus network design problem of passenger separation and heterogeneous fleets,a mixed integer nonlinear programming model with the goal of minimizing the total operating cost and the passenger travel cost is established to simultaneously optimize vehicle routes,vehicle type selection,travel time and passenger allocation in this paper.To effectively solve the model,an improved adaptive large neighbourhood search algo-rithm(ALNS)based on problem properties is designed respectively.Especially,a new separation destroy operator is proposed to improve the search ability of the algorithm.Finally,the model and algorithm are tested on a large number of randomly generated instances to demonstrate their effectiveness and efficiency.The results show that,when solving small-scale instances,the average difference between the solutions ob-tained by ALNS and the best solutions is only 0.24％and the solving time is all less than 12 s;when solving large-scale instances,compared with the solutions obtained by the large neighbourhood search algorithm and Genetic algorithm in the literature,the average cost saving are respectively 1.20％and 2.27％;and compared with the case without considering passenger separation and heterogeneous fleets,considering passenger separation and heterogeneous fleets reduces 11.46％cost,and improves the resource utilization.
The alternating direction method of multipliers(ADMM)is a simple and effective method for solving separable optimization problems,and related studies have been relatively perfect.However,when the objective function contains the coupled term,the research on the convergence of ADMM algorithm is still in the early stage.In this paper,we propose a new linearied symmetric proximal ADMM for solving non-convex nonseparable optimization,which based on the symmetric ADMM(SADMM)and combined with linearization technique.Under certain conditions,it is proved that the iterative sequence generated by the algorithm is bounded and converges to a critical point of the augmented Lagrangian function.Secondly,the strong conver-gence of the algorithm is proved when the auxiliary function satisfies the Kurdyka-Lojasiewicz(KL)property.Finally,numerical experiments show the effectiveness of the algorithm.
Digital finance is a new financial service model that combines the Internet and digital technology tools with traditional financial service industry;It is essential for promoting industrial transformation and upgrading,assisting the real economy in moving toward high-quality development,and strengthening China's position in the international industrial and value chains.This research examines the impact and transmission path of corporate technology innovation based on data from A-share listed companies in Shanghai and Shenzhen from 2011 to 2018 using both theoretical and empirical analyses.Digital finance has network effect characteristics,according to the development history of digital inclusive finance;additionally,a panel threshold model is set up to estimate the non-linear relationship between digital finance and significant technological innovation of enterprises and define the threshold value.The results demonstrated that the growth of digital finance has a substantial impact on the company's technical innovation.Based on the heterogeneity research,the cen-tral and western regions,which have stricter financial regulations and less developed banking sectors,are the places where digital finance plays a more significant role in fostering corporate technological innovation.The analysis of the mechanism of action reveals that three different types of channels-financial costs,the degree of en-terprise internal control,and the degree of factor market development-are crucial for releasing the dividends of digital financial innovation,opening up a fresh viewpoint on the transmission mechanism.Finally,it is shown that the"marginal impact"of digital finance's incentive effect is non-linearly increasing.The research in this pa-per advances the understanding of the drivers of corporate technological innovation and the effects and mechanisms of digital finance-enabled technological innovation;It has significant ramifications for speeding up the development of China's digital financial infrastructure,enhancing digital factor markets,actively developing an ESG system with Chinese characteristics and a green performance evaluation system,and for constructing a financial regulatory framework.
This paper studies the distributed consensus filtering problem in sensor networks.In the sensor network,there is the shortest path link communication be-tween sensors to obtain information of each other.This paper presents a distributed consensus filtering algorithm with a two-stage filtering structure.At each moment,each sensor node performs local filtering based on its observations,and then sends its estimate to neighbor nodes and receives estimates from neighbor nodes.By employ-ing the optimal matrix-weighted fusion algorithm in the linear unbiased minimum variance sense,each sensor node performs the fusion estimate based on the estimates of its and neighbor nodes.After many times of information exchanges and fusions,the estimation accuracy of all sensor nodes trends to consensus.The proposed algo-rithm needs to calculate cross-covariance matrices between nodes,which brings heavy computational burden.To avoid the calculation of cross-covariance matrices between nodes,a distributed suboptimal consensus filter is also proposed based on the paral-lel covariance intersection(PCI)fusion algorithm.The accuracy and computational amount of the proposed algorithms are compared and analyzed.Simulation results demonstrate the effectiveness of the proposed algorithms.
The secondary supply chain consisting of a manufacturer and a retailer is studied,and the product quality decisions and their effects of the supply chain under three different risk-taking strategies,namely the manufacturer's risk-taking strat-egy,the retailer's risk-taking strategy and the risk-benefit coordination strategy are analyzed through the Stackelberg game.The results show that when the supply inter-ruption loss rate is low,no matter which risk-taking strategy is adopted,the increase in the supply interruption risk loss rate will reduce the quality of the manufacturer's products,affect the retailer's sales price and market demand,and reduce the profit of the manufacturer and the retailer.Manufacturer risk-taking strategies reduce product quality,product prices,market demand,and supply chain member performance.The retailer's risk-taking strategy is conducive to improving product quality and supply chain member performance.The degree to which a risk-benefit coordination strategy can improve product quality and the profitability of supply chain members depends on the bargaining power of manufacturers and retailers,and as the manufacturer's bargaining power increases,the higher the manufacturer's product quality and profit,the lower the retailer's profit,which is the best risk-taking strategy.
In this paper,the event-triggered formation control problem for multi-agent systems is investigated in the case of denial-of-service(DoS)attacks.Firstly,a sliding mode formation control algorithm is constructed based on the states of the multi-agent system,which can effectively overcome external interference and realize time-varying formation task.In order to reduce control consumption,the event-triggered strategy is applied to the formation control of multi-agent systems,and the exclusion of Zeno behavior can be guaranteed by the inter-event time between two successive trigger events have a positive lower bound.Assume that the DoS attacks occur periodically according to time sequence,then an improved trigger mechanism is proposed for the multi-agent systems in the case of DoS attacks.The stability of the control system is proved by Lyapunov function method and an inductive way.Finally,some numerical simulations are presented to illustrate the effectiveness of the proposed event-triggered formation control protocol.
We propose a novel Bayesian Integrated Nested Laplace Approxima-tion(INLA)estimation method for a class of models known as varying coefficient spatial autoregressive models(VCSAM).These models involve a response variable with spatial autocorrelation and a nonparametric component containing varying coef-ficients.Although extensive research has been conducted on VCSAM,there have been limited studies on using Bayesian methods to propose efficient solutions for model es-timation.Motivated by this,the paper introduces a method based on Bayesian-INLA technology to seek an effective alternative to the MCMC method within the VCSAM framework.The varying coefficient component is approximated through a linear com-bination of B-spline basis functions,and the spatial autocorrelation coefficients are estimated using the M-H algorithm.The Bayesian-INLA technique is applied to es-timate the posterior distributions of the varying coefficient function and correlation coefficients.Simulation studies demonstrate that the proposed estimation method exhibits advantages in simultaneously handling spatial dependency effects and spa-tial non-stationarity.Furthermore,empirical analysis of Boston housing price data showcases the superior performance of the proposed estimation method.
To accurately assess the comprehensive quality level of manufacturing small and medium-sized enterprises(SMEs)and address the issues in their production and management,this paper proposes a comprehensive quality assessment method based on combined weighting VIKOR method.The comprehensive quality assessment index system was designed by referencing the requirements of the quality management system and considering the characteristics of manufacturing SMEs.It consists of five dimensions:Quality certification,quality supervision,innovation and development ability,financial quality,and social reputation.The improved information overlap in-dex selection method was used to optimize the indicators.AHP and CRITIC methods were used for weight assignment,and the VIKOR method was applied to calculate the interest ratio value,to achieve comprehensive quality assessment of SMEs.Empirical research was conducted on 119 SMEs in the electronics industry using the compre-hensive quality assessment method proposed in this paper.The research findings indicate a strong consistency between the results obtained from the proposed method and the objective quality development status of the sample enterprises,which verifies the effectiveness of the method proposed in this paper.
The optimization problem of split delivery truck multi-drone collabora-tive routing for emergency supplies under traffic restriction scenario was studied.A model is formulated for truck multi-drone collaborative routing by considering the factors such as the road network conditions in the disaster area,the truck launch-ing/receiving drones en route,drones deliver multiple nodes in a single takeoff,and split delivery.The goal of the model is to minimize the completion time of emer-gency delivery.According to the problem and model characteristics,an improved ant colony algorithm(IACA)is proposed.The experimental results show that the pro-posed approaches can reasonably allocate the delivery tasks of trucks and drones,and scientifically plan the truck multi-drone collaborative routes for split emergency de-livery under regional traffic restriction situation.Truck launching/receiving drone en route can effectively shorten the drone flight distance,reduce the collaboration time between trucks and drones,and shorten the distribution time of emergency delivery under traffic restriction situation.The proposed approaches are feasible,reasonable and effective.
In order to predict joint trajectory with irregular sampling caused by occlusion,an improved(spectral restricted)neural ordinary differential equation net-work(SODEnet)is proposed in this paper.Compared with Gaussian process re-gression and cyclic neural network,SODEnet has advantages in low complexity,high precision,and iteration,which has become an important method for trajectory pre-diction in recent years.First of all,it establishes the human body joint trajectory pre-diction model based on irregular sampling and deduces forward propagation process of SODEnet via Euler method.Secondly,it derives the spectral constraint theorem contributed to Lipschitz condition which formulates SODEnet.Then it designs the loss function with penalty term of spectral norm and introduces the back propagation process via adjoint method.In the meanwhile,the error analysis results are given.Finally,the effectiveness of SODEnet is verified in the wrist joint trajectory prediction experiment.The results show that SODEnet can accurately predict joint trajectory and has faster convergence and smaller prediction error than classical neural ordinary differential equation(NODE)and recurrent neural network(RNN).
