Effective sales prediction for e-commerce would assist retailers in developing accurate production and inventory control plans, which would further help them to reduce inventory costs and overdue losses. This paper develops a systematic method for e-commerce sales prediction, with a particular focus on predicting the sales of products with short shelf lives. The short-shelf-life product sales prediction problem is poorly addressed in the existing literature. Unlike products with long shelf lives, short-shelf-life products such as fresh milk exhibit significant fluctuations in sales volume and incur high inventory costs. Therefore, accurate prediction is crucial for short-shelf-life products. To solve these issues, a stacking method for prediction is developed based on the integration of GRU and LightGBM. The proposed method not only inherits the ability of the GRU model to capture timing features accurately but also acquires the ability of LightGBM to solve multivariable problems. A case study is applied to examine the accuracy and efficiency of the GRU-LightGBM model. Comparisons among other sales prediction methods such as ARIMA and SVR are also presented. The comparative results show that the GRU-LightGBM model is able to predict the sales of short-shelf-life products with higher accuracy and efficiency. The selected features of the GRU-LightGBM model are also useful due to their interpretability while developing sales strategies.

To gain sustainable development, it is a trend that manufacturing companies are change the value chain from manufacturing-centric to service-centric. Therefore, the capability of the manufacturing service is as significant as the production ability of enterprises, which reflects the supply chain management (SCM), flexible production, production efficiency, and other indicators of the enterprises. It is the first paper to discuss the sustainability of service-oriented manufacturing using bibliometric analysis. It derives a detailed review and future outlook on the development of manufacturing servitization, indicating the research directions for future development, and provides a valuable reference for researchers in related directions. The bibliometric analysis discusses countries or regions, research areas, authors, keywords, institutions, and journals based on the literature data from the Web of Science (WoS). The results show that research on manufacturing services has gradually received attention since its inception and has become popular since 2008. The papers published from 2008 to 2021 account for 77.62%. The USA is the most studied country on this topic, followed by China and the UK. The International Journal of Production Research regarding the most quantity of articles, and Beihang University is the most influential institution in this field. The largest amount of articles published in the area of “business and economics”, amounting to 1565 articles. In recent years, the main research areas included “Industry 4.0”, “cloud manufacturing (CMfg)”, “Internet of Things (IoT)”, “big data” and “services innovation”. Finally, “digital and intelligent manufacturing” and “product-service systems” are potential research directions for the future.

The optimal power flow problem is one of the most widely used problems in power system optimizations, which are multi-modal, non-linear, and constrained optimization problems. Effective constrained optimization methods can be considered in tackling the optimal power flow problems. In this paper, an $ϵ$-constrained method-based adaptive differential evolution is proposed to solve the optimal power flow problems. The $ϵ$-constrained method is improved to tackle the constraints, and a p-best selection method based on the constraint violation is implemented in the adaptive differential evolution. The single and multi-objective optimal power flow problems on the IEEE 30-bus test system are used to verify the effectiveness of the proposed and improved $\epsilon$adaptive differential evolution algorithm. The comparison between state-of-the-art algorithms illustrate the effectiveness of the proposed and improved $\epsilon$adaptive differential evolution algorithm. The proposed algorithm demonstrates improvements in nine out of ten cases.

Vehicle routing problems (VRPs) are challenging problems. Many variants of the VRP have been proposed. However, few studies on VRP have combined robustness and just-in-time (JIT) requirements with uncertainty. To solve the problem, this paper proposes the just-in-time-based robust multiobjective vehicle routing problem with time windows (JIT-RMOVRPTW) for the assembly workshop. Based on the conflict between uncertain time and JIT requirements, a JIT strategy was proposed. To measure the robustness of the solution, a metric was designed as the objective. Afterwards, a two-stage nondominated sorting ant colony algorithm with deep reinforcement learning (NSACOWDRL) was proposed. In stage I, ACO combines with NSGA-III to obtain the Pareto frontier. Based on the model, a pheromone update strategy and a transfer probability formula were designed. DDQN was introduced as a local search algorithm which trains networks through Pareto solutions to participate in probabilistic selection and nondominated sorting. In stage II, the Pareto frontier was quantified in feasibility by Monte Carlo simulation, and tested by diversity-robust selection based on uniformly distributed weights in the solution space to select robust Pareto solutions that take diversity into account. The effectiveness of NSACOWDRL was demonstrated through comparative experiments with other algorithms on instances. The impact of JIT strategy is analyzed and the effect of networks on the NSACOWDRL is further discussed.

To address the limitations in predictive capabilities of thermal error models built from single-source, single-structure data, this paper proposes a thermal error prediction model based on decision-level fusion of multi-source heterogeneous information to enhance prediction accuracy. First, an experimental platform for multi-source heterogeneous information acquisition was constructed to collect thermal error data from different signal sources (multi-source) and different structures (heterogeneous). Next, based on the characteristics of the multi-source and heterogeneous data, relevant features were extracted to construct the feature set. Then, using the feature information set of the multi-source and heterogeneous data, thermal error prediction sub-models were established using Nonlinear Autoregressive models with exogenous inputs (NARX) and Gated Recurrent Units (GRUs) for a vertical machining center spindle. Finally, the entropy weight method was employed to assign the weights for the linear-weighted fusion rule, achieving decision-level fusion of multi-source heterogeneous information to obtain the final prediction result. This result was then compared with experimental results and the prediction results of single-source models. The findings indicate that the proposed thermal error prediction model closely matches the actual results and outperforms the single-source and single-structure data models in terms of Root-Mean-Square Error (RMSE), Coefficient of Determination (R²), and Mean Absolute Error (MAE).