In view of the order cancellation problem faced by online fresh food retail business, this paper proposes tiered refund decision model aiming to optimize the retailer's revenue, considering consumers' cancellation behaviors with the combination of cost varying characteristics in order fulfillment processes. Based on the classical Hotelling model, we first analyze and depict the consumer behavioral preference on the timing of cancelling orders, then construct the consumer's utility function and establish the retailer's revenue model, and make the optimal tiered refund decision by piecewise optimization based on the non-tiered refund decision. It is found that the retailer's optimal decision of tiered refund can not only improve the retailer's revenue, but also enhance consumers' purchase intention and their market participations by taking advantage of consumers' behavioral preferences. The optimal tiered refund strategy can achieve a “win-win” situation for both retailers and consumers. Finally, we verify the theoretical results by conducting numerical experiments, and provide extensive managerial insights. This research provides decision-making support for online fresh food retailing businesses when developing the refund strategy for order cancellations.
With the increasing scale of large-scale online supermarkets, it has to set new warehouse layouts to stock enormous kinds of goods, e.g., multiple warehouses in one city, multiple warehouses in multiple cities. As a result, a customer's order with multiple types of items is separated into several suborders, which are picked and packed in different warehouses, and then delivered to the customer independently. It will bring about some challenges (high cost, high pollution, high disturbance to customers) for green and healthy development of e-commerce. The package consolidation for split orders among multiple warehouses is an important way to solve this problem, and also an urgent task to improve the performance quality of split order fulfillment. This paper aims at the decision-making problem of package consolidation for split orders in large-scale online supermarket. In order to improve the scientificity and efficiency of online generation of package consolidation schemes for split orders, it focuses on decreasing the difficulty of problem solution, synthetically applying the theories of combination optimization and state-space searching. A two-stage online decision-making method of intelligent optimization for split order package consolidation is proposed to solve three key issues of whether split orders can be consolidated and packaged, which warehouses can be selected as package consolidation centers for each order, and which goods can be consolidated and packaged. In the first stage, the main influencing factors of the split order package consolidation are summarized and the decision rules are designed to decide whether to consolidate. In the second stage, for the split orders that need to be consolidated and packaged, the state-space search algorithm is proposed to generate the alternative package consolidation schemes, and the qualitative control strategies are transformed into control rules and incorporated into the scheme generation process, thereby reducing the solution space. Finally, numerical experiments show that the decision-making method of package consolidation can effectively reduce the fulfillment cost of split orders of online supermarket, and the theoretical results can provide a reference for operational activities of online supermarket or other B2C e-commerce enterprises with serious order splitting situation.
With the development of online retail and the appearance of new retail with the integration of online and offline, the omni-channel operation and especially the collaborative distribution optimization problem have attracted the attention of academics and industry. The drugs operational characteristics under the background of new retail, such as pre-storage, compatibility and timeliness of customer demand, bring new difficulties to the formulation of drug distribution strategy in the new retail. In this paper, the idea of space-time network modeling is adopted to accurately describe the decision-making process of the system and simplify the model structure. According to the two-stage complex decision characteristics of the optimization problem, a comprehensive optimization algorithm based on Lagrangian relaxation method is designed to solve the model. Finally, the effectiveness and efficiency of the algorithm are verified by case studies, and the corresponding management implications are proposed through parameter sensitivity analysis, which provide reference and suggestions for the operation decision of drug collaborative distribution under the background of new retail.
The study focuses on instant delivery service (IDS) for online to offline (O2O) business model in take-out food and new retail industries. The O2O-IDS system is defined as an organization management system consisting of the O2O and IDS platforms that interact with each other. IDS is a new type of logistics service in O2O model. The many-to-many cooperation between O2O and IDS faces disruptions due to many factors, such as 2019 novel coronavirus, which lower the operational efficiency of O2O-IDS system. Therefore, resilience theory is introduced to the O2O-IDS system research based on the idea that reducing vulnerability can improve resilience. An optimization model is constructed through the concepts of key cooperation relationship, loss of system function and performance impact index (PII), etc. The key cooperation relationship was identified through optimizing maximum loss of system function. The model gave the ranking of the cooperation by calculating the PII of the partnership, and the study proposed a reasonable protection strategy according to the budget. In addition, resilience is assessed and quantified by drawing a curve of loss of system function, and under different number of disruptions, the more smoothness the curve is, the better resilient the system would be. The example of O2O-IDS system showed that the optimization strategy proposed in this paper can significantly improve the system's ability to prepare, mitigate and prevent risks.
With applications of the dynamic pickup and delivery problem (DPDP) in new fields, such as ride hailing and take-away food delivery, the DPDP with large scale, strong real-time and strong dynamic characteristics is becoming a hot research topic. Firstly, this paper analyzes the applications and influencing factors of the DPDP, and classifies different applications from the perspectives of delivery mode and delivery entity. Then, a general definition of the DPDP, the solving methods, as well as the evaluation metrics for different dynamic solutions are provided. Focusing on three classical applications (i.e., dynamic dial a ride problem, vehicle scheduling in the ride hailing service, and instant delivery service), this paper analyzes and compares the three applications in various aspects, and surveys different mathematical models as well as solving algorithms. Finally, this paper points out potential future research directions.
In order to improve the service level of pilot scheduling in ports under the complex channel, this paper studies the integrated optimization problem of ship scheduling and pilot scheduling. This paper considers factors such as tide, blocking channel, safe distance, the opposite avoidance of single and two-way channel. Aiming at minimizing ship delay cost and pilot scheduling cost, an integer programming model is established. Combined with the characteristics of the problem, a two-stage variable neighborhood search algorithm is designed to solve the model. The first stage discusses the effect of pilot scheduling on ship delay, and obtains a set of all ship and part pilot scheduling schemes; In the second stage, the complete scheduling scheme of all ships and pilots obtained by traditional scheduling rules is used as the initial solution, and the variable neighborhood search algorithm is used to solve the optimal scheduling scheme among all scheduling schemes. Finally, the effectiveness of the model and algorithm is verified by a series of numerical experiments.
The rapid development of the cruise industry has brought a series of marine environmental problems. In the process of cruise navigation, cruise operation and passengers on board will produce an uncertain amount of waste. Generally, the total amount of waste is far greater than its capacity. It is necessary to dispose of waste in multiple ports of the line, which requires a certain fee to be paid to the port. Therefore, this paper studied a waste disposal problem for cruise ships to decide which port to sign and how much waste should be discharged under the uncertain amount of produced waste. These decisions affect the operating costs of cruise companies, so it is necessary to calculate and analyze scientifically through some optimization models. Based on the theory of system engineering, this paper analyzed the background of the above-mentioned decision-making problems and proposed three models successively by using the theory and tools of mathematical programming. The models include a deterministic model, a stochastic programming model that applied to arbitrary probability distributions of emission parameters, and a three-stage robust optimization model that can cope with the uncertain parameter interval. Then, considering the above problems and the characteristics of the model, this paper designed a Tabu search (TS) and particle swarm optimization (PSO) algorithms to solve large-scale problems in stochastic and robust models, respectively. Extensive experiments in this paper validated the effectiveness and efficiency of the proposed models and algorithms. The decision-making method of cruise waste emissions proposed in this paper has certain application value and guiding significance for the development of the current global green shipping industry.
With the increasing concern of environment pollution caused by the massive usage of fossil fuels, the energy-efficient operation of green container terminals is becoming one of the hottest topics in recent years. This paper investigates the yard cranes with the pre-marshalling tasks, in which the power demand of crane is different when the crane runs in different mode. Considering the peak power constraint, a mixed integer programming model is established to minimize the total energy consumption of all cranes with non-crossing and safety clearance requirements in the storage yard. Two kinds of coding methods are designed based on the idea of discrete bay and continuous bay, then the chromosome is decoded using heuristic rules. And, the crossover and mutation methods are designed for the genetic algorithm (GA) correspondingly. Numerical results show that the method of continuous bay is better than the discrete bay. Then, the effectiveness of designed GA is validated by the results obtained by the comparison between GA and Cplex, particle swarm optimization, artificial bee colony algorithm. In addition, compared with the traditional policy, our model performs well in both of total energy consumption and peak demand. The research findings can guide the port managers to reduce the energy cost effectively without interfering the processing of daily tasks.
With the rapid development of station-based one-way electric-car sharing, the temporal and spatial imbalanced distribution of cars resulted from the directional feature of requirements is becoming increasingly prominent. A transportation problem of station-based one-way shared electric-cars with recharging scheduling is studied. This problem considers the constraints of the remained mileage of cars and the working hours of employees with the objective criteria of total transportation costs. It simultaneously decides the source and destination stations of cars and the assigned employees so that the distribution of cars among stations is balanced and that the cars with low level of batteries are parked at the stations with charging piles. A 0-1 nonlinear programming model is built. In order to solve the model conveniently, aiming at the homogeny of employees, three valid inequalities about single transportation and the total routes are proposed to eliminate symmetry between solutions. The aforementioned mathematical formulations are validated based on randomly generated instances. The results indicate that all the three valid inequalities can help to solve the mathematical model and that the simultaneous application of the three valid inequalities can save about 50\% of solving time for medium- and large- scaled instances.
Signaling product quality through crowdfunding campaign design parameter (including the funding goal and reward price) to potential consumers is one of the most concerned problems for creators. This paper, by incorporating consumer information acquisition behavior, establishes a signaling game model under asymmetry information situation, and characterizes the existence conditions for the H-type creator to signal his product quality by utilizing the campaign design parameter. Furthermore, on this basis, this paper investigates the equilibrium crowdfunding campaign design strategy by comparing the expected profit under separating and pooling equilibrium, and explores the effect of the existence of consumer information acquisition behavior on the equilibrium outcome. We find that:1) when the quality difference between the two types of crowdfunding products is not very large, the H-type creator can signal his high quality by either increasing his funding target, decreasing the reward price, or both simultaneously. Furthermore, the funding goal is a more effective signal device than the reward price; 2) compared with the case where there is no consumer information acquisition behavior, the campaign design parameter under pooling equilibrium is more profitable for the H-type creator when considering this consumer information acquisition behavior. We also explore the effect of some market parameters on the equilibrium crowdfunding campaign design parameter to verify the robustness of our results.
Currently, the domestic and international economic situation has become more complex and changeable, with increasing uncertainties, and the dynamics of credit risk in the financial market has been significantly enhanced. Dynamic credit risk evaluation has become an urgent problem that financial institutions need to solve. To this end, this paper proposes a mixture survival analysis-based dynamic credit scoring method, which consists of three parts. First, constructing mixture survival analysis-based dynamic credit scoring model, including default status discrimination model and default time estimation method, to predict “whether default” and “when to default” for evaluation objects. Then, generating multiple surviv-al status vectors using panel data to characterize the dynamic correlations between credit features and survival time. Finally, based on generated multiple survival status vectors, iteratively estimating model parameters using the expectation maximum algorithm. Experimental research shows that the predictive performance of the proposed method is significantly superior to the single classification-based, ensemble learning-based, and survival analysis-based credit scoring methods.
Service design is a high-risk and high-investment activity. In the complex, volatile and uncertain market environment, how to use scientific theories and methods to design service products and ensure service quality on the basis of reasonable utilization of service resources is one of the keys for service enterprises to obtain competitive advantages. Based on collecting and reviewing the existing literature on service design, this paper first analyzes the relevant concepts and extensions of service design. Then, the relevant literature in the field of service design is summarized from the aspects of service mode, service content, service process and service system from the perspective of management. Finally, this paper also outlines the potential research opportunities of service design under the new information technology environment.
Sharing economy has been one of the emerging and promising research topics in the operation management field, and there has been a growing body of related literature in recent years. The academic notion on sharing economy is first clarified, and the corresponding research scopes are introduced. Then based on the ownership of the shared resource, the business modes of sharing economy are categorized into C2C and B2C modes. Then within each type of business mode, representative platforms (including house-sharing, ride-sharing and bike-sharing platforms) are picked up, in the context of which, literature related to the corresponding business mode is reviewed, especially the pricing problem in the C2C mode and resource management problem in the B2C mode.
Under the background of digital economy and sharing economy, the platform model has received the attention of academia and practice. Value co-creation in platform ecosystems is different from value co-creation in general organization. It is a dynamic process led by platform leaders, and various stakeholders create value through resource integration and interaction. The research shows the characteristics of interdisciplinary integration, and there have few systematic literature reviews from theoretical grounding to research agenda and emerging trends. This article first defines the connotation and characteristics of value co-creation in platform ecosystems, expounds the key factors and process mechanism. On this basis, focusing on existing research in three aspects, including the co-opetition relationship, platform governance, and business model innovation. In addition, combining complex characteristics of platform ecosystems, uncertain environment and digital background, the future research directions as well as possible research questions are discussed.
As a human-centered production system, SERU production system is based on the interrelated socio-technical practices. Fitted organizational culture can provide great support to the implementation of SERU production system. In this paper, we construct a conceptual model to identify the relationships among group culture, SERU production and operational performance. Moderating effects of environmental uncertainty are discussed. We conduct an empirical analysis based on a survey data of 357 Chinese manufacturing firms. The result reveals that group culture has significant positive impacts on SERU production implementation. We conclude that the group culture of a firm is a prerequisite for SERU production system. Environmental uncertainty has a significant moderating effect on the relationship between group culture and SERU production implementation, and the interrelationship between SERU production and operational performance. In particular, environmental uncertainty strengthens the relationship between group culture and multi-skilled worker application, and that of multi-skilled worker application and operational performance. These conclusions enrich the content of SERU production theory and have important implications for SERU practitioners.
In the face of increasing demand uncertainty and fierce market competition, SERU production has emerged and has been widely used in the Japanese electronics assembly manufacturing companies. Some domestic manufacturing and assembly industries are trying to convert the traditional assembly line into SERU production system, which is shorter for LINE-SERU conversion. Some literatures have reported the effect of LINE-SERU conversion in Japanese companies, but the academic community lacks theoretical analysis and benchmarking analysis of the production organization and operation efficiency of SERU production implemented in domestic enterprises. This paper is based on the LINE-SERU conversion example of a domestic medical device assembly company. Through theoretical analysis and data analysis, systematically explored the background, conversion process, influencing factors and production organization form of LINE-SERU conversion in Chinese manufacturing enterprises. Analyzed and compared the improvement effects and reasons from a multi-dimensional perspective. Summarized the differences in SERU system formation, layout and enterprise culture between domestic and Japanese enterprises. The study found that Chinese companies are in the early stages of LINE-SERU conversion and SERU system has its own unique organizational form, which are significantly enhanced in terms of improving production efficiency, shortening the production cycle, reducing the number of work-in-progress, reducing the number of workers and reducing the rate of defective products. However, the improvement effect is different from that of Japanese enterprises. It is suggested that Chinese companies should form a SERU system, optimize unit layout, and cultivate a people-oriented corporate culture based on their human resource characteristics to increase system flexibility and improve efficiency.
The systematic exploration of the enablers and mechanism of SERU operation is of great significance for manufacturers to implement SERU production. This paper constructs a framework of enablers for the effective operation of SERU production from the perspective of theory of swift and even flow (TSEF). Case studies then are carried out in the context of SERU production practice of Hualu Panasonic. The study explores the mechanism of SERU's effective operation and the internal relationship between its various enablers. Based on that, the framework model is testified and corrected. The research shows that swift and even process is helpful for SERU production to respond to high demand volatility. Reducing variability is the leading factor. Factory focus is the basis of the operation. Scientific methods, quality management and bottleneck reduction can support the removing of variability. Workers' competency is a key enabler and integrates with other factors to enable manufactures to achieve a swift and even production flow.
As a new production model, SERU production has been widely adopted by electronic companies such as Canon and Sony and has achieved good results due to its advantages of quick response, good flexibility and high efficiency. Also, SERU production has received extensive attention by academic. The operational management of SERU production system includes two key decision-making processes, i.e., SERU formation and SERU scheduling. Both of them are NP-hard problems. For the simplicity, most of the researches are to decide SERU formation or SERU scheduling. However, the only one decision of SERU formation or SERU scheduling cannot obtain the global optimal solution. To obtain the global optimal solution, joint decision on SERU formation and SERU scheduling are considered. This paper reviews the researches on SERU formation and SERU scheduling from three aspects:the separate decisions on SERU formation and SERU scheduling, the joint decision on SERU formation and SERU scheduling of pure SERU production system, the joint decision on SERU formation and SERU scheduling of hybrid SERU production system. Finally, future researches on the SERU formation and SERU scheduling are classified and provided.
Traditional quality management theories focus on improving the quality of products to meet user needs. However, the arrival of the internet-of-things era has given new connotation to quality management. Nowadays, enterprises pay more attention to continuous improvement of user experience in order to increase the stickiness of customers. As a result, new challenges have been brought to traditional quality management theories. In this background, this paper pioneers the construction of a product life-cycle quality management system based on the internet-of-things. With the product life-cycle as the main line, we elaborate on the underlying technical architecture and management collaborative architecture of the system. Furthermore, from the perspective of theoretical methods, business practices, current challenges, and future research outlook, we discuss the implementation of product life-cycle quality management based on the internet of things. Finally, we hope that this study will not only promote the theoretical innovation of quality management, but also provide a reference for the practice and model innovation of quality management industry based on the internet of things.
In the development of emerging e-commerce, the collection, utilization, openness and sharing of user data has reached an unprecedented level, which also brings great challenges to personal privacy security. In order to resolve the contradiction between user data usage and personal privacy protection, this study proposes a data privacy preservation approach that aims at minimizing the information loss, based on k-anonymity principle. Firstly, a new record sorting algorithm is proposed based on the features of attribute domains. Secondly, the privacy protection process is transformed into an optimal allocation problem between the entire records and the available anonymization function candidates, thus an optimization model is constructed to minimize the total information loss generated by assigning different functions to anonymize every record. To reduce the computation time, a heuristic method is developed to solve the optimal allocation model and implement anonymization for each record. A numerical study is conducted on three real-world datasets of different scales, by comparing with the most advanced privacy protection methods in existing research to demonstrate the effectiveness of the proposed approach. The results show that this approach can produce maximum data utility and has superior computational efficiency over the benchmarks, under the same level of privacy protection. This research provides theoretical and technical innovations for privacy preserving user data in emerging e-commerce, and offers an effective solution for privacy protection in large data applications.
As governments at all levels in China have started to promote mandatory garbage classification, in order to meet the standardization and automated garbage classification in all aspects of classification and recycling need a fine-grained image classification model suitable for cloud deployment with high accuracy and low latency. This article takes advantage of deep transfer learning to establish an end-to-end transfer learning network architecture GANet (garbage neural network). Aiming at the challenges of category confusion and background interference in garbage classification, this paper proposes a new pixel-level spatial attention mechanism PSATT (pixel-level spatial attention). In order to overcome the challenges of multi-class and sample imbalance, this paper proposes a label smoothing regularization loss function. In order to improve convergence speed, model stability and generalization, this paper proposes a stepped OneCycle learning rate control method, and gives a combined use strategy combining Rectified Adam (RAdam) optimization method and stochastic weight averaging. Experiments used the training data which are marked by the Shenzhen garbage classification standard and provided by the “Huawei cloud artificial intelligence competition · garbage classification challenge cup”, and verified the significant effect of GANet in the garbage classification problem, and won the national second prize (2nd place). At the same time, the proposed PSATT mechanism is superior to the comparison methods with improvement on different backbone network architectures, and has good versatility. The GANet architecture, PSATT mechanism and training strategies proposed in this paper not only have important engineering reference value, but also have good academic value.
The class imbalance learning widely occurs in classification tasks in the research field of data mining, such as manufacturing quality conditions, medical diagnosis, financial service, etc. The synthetic minority over-sampling technique (SMOTE) is a common technique to deal with imbalanced datasets, which can be enhanced using the framework of the boosting algorithm. However, this strategy can easily result in the lack of diversity of the base classifiers in the ensemble learning system. On this account, a boosting learning algorithm integrated Gaussian process smote oversampling is proposed to solve the imbalance learning problem, namely Gaussian-based smote in boosting (GSMOTEBoost). In order to improve the robustness of the classification system, the proposed GSMOTEBoost algorithm is developed using the framework of AdaBoost, in which a smote oversampling technology based on Gaussian process is used to increase the diversity of the base classifiers for each iteration. To verify the effectiveness of our algorithm, we develop the experiments on twenty datasets selected from the KEEL repository with these well-known imbalance learning algorithms. The G-mean, F-measure and AUC are considered as the assessment metrics and the hypothesis testing is used to analyze the experimental results. The obtained results, supported by the proper statistical analysis, indicate that the proposed GSMOTEBoost significantly outperforms the comparison methods.
The performance of decision making by artificial intelligence has exceeded the capability of the human being in many specific domains. Countries like China and the USA have promulgated artificial intelligence development strategies and action plans to encourage the applications of artificial intelligence. In the artificial intelligence decision-making process, the inherent black-box algorithms and opaque system information lead to highly correct but incomprehensible results, which hinder the further development of artificial intelligence. For the commercialization and popularization of artificial intelligence, the need for explainability of intelligent decision-making is becoming more and more urgent. It is necessary to study the transformation of black-box decision-making into a transparent process and establish trust between humans and machines. From the perspective of system application and decision beneficiaries, this paper focuses on the domestic and foreign-related research on four aspects:The basic concepts of explainable artificial intelligence decision-making, explanation methods of black-box models, applications of explanation methods in high-risk domains and explanation methods evaluation. Meanwhile, we state insights into future research and development trends.