Tool wear prediction is of critical importance to maintain the desired part quality and improve productivity. Inspired by the successful application of deep learning in many condition monitoring tasks. A novel modeling framework is presented, which includes multiple stacked sparse autoencoders and a nonlinear regression function for tool wear prediction. Multiple stacked sparse autoencoders consists of two main structures: one model is designed with multi-dimensional stacked sparse autoencoders, which can learn more features from different feature-domains in the raw vibration signal; and another single-dimensional stacked sparse autoencoders is used for feature fusion and deeper features learning. And a modified loss function is applied that improves the learning ability. In addition, due to the good properties of tool wear process in nonstationarity and complex nonlinear, a nonlinear regression function is utilized to enhance the progressive tool wear prediction tasks. A dataset from a real manufacturing process is used to evaluate the performance of the proposed modeling framework. Experimental results show that tool wear can be predicted accurately and stably by the proposed tool wear predictive model, which outperforms the already developed methods.

A high-speed reciprocating operating mechanism with fast running and frequent stopping produces a certain vibration due to inertia. The inherent vibration of this mechanism becomes a key factor affecting positioning accuracy and time. However, the existing vibration identification method is suitable only for static or low-speed structure operations. Therefore, a method of identification for operational modal parameters based on strain response is proposed. This method combines the strain mode with the operational modal analysis method, both of which are verified under the static and operational states, respectively. First, the theoretical background of the identification method is derived. Next, the proposed approach is used to identify the displacement mode shapes and the strain mode shapes of the beam structure with modal parameters that are obtained by experimental modal analysis. These results show that the beam structure presents different displacement mode shapes and strain mode shapes under different constraints. Finally, the displacement mode shapes of the sorting arm are obtained by the method under operation. The recognition results are compared with the computational modal analysis, and the two mode shapes are found to be highly consistent, which verifies that the method is equally reliable during operation. This method expands the application scope of the classic experimental modal analysis and reduces the error because of the additional mass of the actual sensor.

Production system flexibility has become an important factor to meet the needs of diversified markets in an ever-changing and turbulent market environment. Thus, this paper addresses the problems of flexibility evaluation for decision making by evaluation method driven by digital twin, which models, predicts, and evaluates the flexibility of a given production system design scheme (PSDS). Based on the definition and characteristic analysis of flexibility in the production system design phase, and combined with data modelling (static function model (SFM) and dynamic prediction model (DPM)) and data analysis (AHP-based and simulation-based analysis), the proposed framework evaluates system flexibility from the static and dynamic levels, helps enterprises quickly obtain the system’s comprehensive flexibility evaluation index, realizing the rapid screening of the production system design scheme (PSDS). The successful application of the evaluation framework in case enterprise has fully proved the feasibility and effectiveness of the method.