SSA-CNN-LSTM Fusion of Multi-Source Heterogeneous Data for Order Performance Evaluation

With the continuous development of the flexible supply chain in the manufacturing industry of the Industrial Internet of Things and the widespread promotion of the order-oriented production mode, the quantity and types of data involved in order performance evaluation tasks are constantly increasing. The applicability of traditional evaluation methods is significantly weakened, leading to additional investment of manpower and time resources. In response to this issue, this paper proposes a SSA-CNN-LSTM multi-source heterogeneous data fusion model aimed at achieving precise order performance evaluation. By integrating and learning data of different structures from various sources, the model fully explores the correlation of data features to obtain precise fusion results, thereby enabling the evaluation of order performance. Simulation experiments conducted on a dataset from a certain intelligent collection customization company demonstrate that the RMSE, MAE, and MAPE of the SSA-CNN-LSTM model results are reduced by 58.71%, 62.94%, and 63.29% respectively compared to the LSTM model, validating the superior accuracy of the proposed model. It also indicates that the model proposed in this study provides new ideas and methods for completing performance evaluation tasks, offering reliable basis and reference for enterprise decision-making, and enriching the research content of the deep learning multi-source heterogeneous data fusion field.