The composite sheet layup process involves stacking several layers of a viscoelastic prepreg sheet and curing the laminate to manufacture the component. Demands for automating functional tasks in the composite manufacturing processes have dramatically increased in the past decade. A simulation system representing a digital twin of the composite sheet can aid in the development of such an autonomous system for prepreg sheet layup. While finite element analysis (FEA) is a popular approach for simulating flexible materials, material properties need to be encoded to produce high-fidelity mechanical simulations. We present a methodology to predict material parameters of a thin-shell FEA model based on real-world observations of the deformations of the object. We utilize the model to develop a digital twin of a composite sheet. The method is tested on viscoelastic composite prepreg sheets and fabric materials such as cotton cloth, felt, and canvas. We discuss the implementation and development of a high-speed FEA simulator based on the VegaFEM library. By using our method to identify sheet material parameters, the sheet simulation system is able to predict sheet behavior within 5 cm of average error and have proven its capability for 10 fps real-time sheet simulation.

Precision turning with slow tool servo (STS) plays an increasingly important role in advanced manufacturing nowadays. However, it is difficult to promote machining quality for surfaces with local complex geometric features by the conventional global machining method. Hence, a subregional processing method in STS is proposed. First, the continuous equipotential line is taken to express the local geometric feature. Thus, a potential field is built, where the surface could be divided into subregions. Then, a subregional toolpath with variable feed rates is generated by the field and stitched to ensure the feeding motion stability of X-axis. Finally, the surface is subdivided for variable spindle speed planning, considering the feeding motion stability of Z-axis. It is found that the profile arithmetic average error reduces by 31.58% with the proposed method compared with that with the conventional method and the machining time is shortened by 41.00%. Thus, it is proved that the new processing method effectively promotes machining quality and efficiency.

The five-axis machining of a free-form surface often contains the reversal of a rotary axis’ rotation direction with linear axis synchronized with it. This paper proposes a new machining test to quantitatively evaluate the influence of the reversal of rotation direction on the surface geometry and roughness. In the five-axis machining, the trajectory of tool position and orientation is first given in the workpiece coordinate system (WCS) by the computer-aided manufacturing ( cam ) software, and the computerized numerical control (CNC) system converts it to the machine coordinate system (MCS) to calculate command trajectories. This paper clarifies that the tool path smoothing in the MCS can potentially cause a large contour error because of the dynamic synchronization error of rotary and linear axes. Although some academic works in the literature presented the smoothing in the WCS, many commercial CNC systems still employ the smoothing in the MCS, partly because machine tool users or makers do not clearly see how significant this influence can be on the machining accuracy. The proposed machining test enables a user to quantitatively evaluate it. The machining experiment shows that the geometric error of the finished test piece was as large as 0.16 mm under the conventional smoothing in a commercial CNC system, which can be significantly larger than the influence of other typical geometric errors of a five-axis machine tool. This paper shows, by numerical simulation, that the smoothing in the WCS can completely eliminate this contour error.

Machine tool geometric errors are frequently corrected by populating compensation tables that contain position-dependent offsets to each commanded axis position. While each offset can be determined by directly measuring the individual geometric error at that location, it is often more efficient to compute the compensation using a volumetric error model derived from measurements across the entire workspace. However, interpolation and extrapolation of measurements, once explicit in direct measurement methods, become implicit and obfuscated in the curve fitting process of volumetric error methods. The drive to maximize model accuracy while minimizing measurement sets can lead to significant model errors in workspace regions at or beyond the range of the metrology equipment. In this paper, a novel method of constructing machine tool volumetric error models is presented in which the characteristics of the interpolation and extrapolation errors are constrained. Using a typical five-axis machine tool compensation methodology, a constraint bounding the tool tip modeled error slope is added to the error model identification process. By including this constraint over the entire space, the geometric errors over the interpolation space are still well-identified. Also, the model performance over the extrapolation space is consistent with the behavior of the geometric error model over the interpolation space. The methodology is applied to an industrial five-axis machine tool. In the experimental implementation, for measurements outside of the measured region, an unconstrained model increases the mean residual by 40% while the constrained model reduces the mean residual by 40%.

Thermal deviation induced by ambient temperature changes and heat generated during machine operations influences the accuracy of machine tools. A thermal test is essential to evaluate the influence of thermal deviation. ISO 230-3 provides displacement sensor-based thermal tests for machine tools. This paper proposes a machining test that enables a user to visually, by the naked eye, observe the integrated thermal influence on the tool trajectory's displacement in the direction normal to the test piece surface from the length of the machined slots. The proposed test consists of the machining of the five surfaces to observe the thermal influence of the tool position with respect to the test piece in X-, Y- and Z-directions, as well as the position of two rotary axes with respect to the tool position. The advantages of the proposed test include that it requires no measuring instrument to quantitatively evaluate the thermal error in all directions. And since the thermal influence is evaluated by observing the position where the cutting tool leaves the test piece surface, where the cutting force is zero, the influence of the cutting force on the test results can be ignored. Thermal influences of a five-axis machine tool during the warm-up cycle are investigated by experiment to validate the feasibility of the proposed method. Results show that 150 min is needed for sufficient warm-up for the selected machine tool if permissible tolerance for thermal deviation is 2.5 um for all the errors.

The in-plane torsion test (IPT) is a shear test that has already been successfully used to determine flow curves up to high strains for thin sheets with thicknesses between 0.5 mm and 3.0 mm. In the same way as with other shear tests, the formation of wrinkles is a major challenge in determining flow curves with the IPT, especially when testing ultra-thin sheets with a thickness between 0.1 mm and 0.5 mm. A new method for suppressing wrinkling is introduced, in which the formation of wrinkles is avoided by arranging and gluing single sheets to multi-layered specimens. The influence of the used adhesive on the determination of flow curves is negligible. The proposed method is used to identify flow curves for two materials, the high-strength steel TH620 and the soft steel TS230, used in the packaging industry. The materials are tested in sheet thicknesses between 0.17 mm and 0.6 mm. The determined equivalent plastic strains for the TH620 with a sheet thickness of 0.20 mm could be increased from 0.38 (bulge-test) to over 0.8 with the new method using four-layered specimens.

A large-scale production of carbon nanotubes has been of great interest due to their practical needs, which is limited by the difficulty of producing them with controlled structures and properties. We seek for a surrogate modeling to predict the process yield for a given process configuration under control uncertainties. The predictive power can be used to optimize the process configuration in a closed-loop production system. A challenge in the surrogate modeling is that some process conditions are controlled by other manipulating factors, and the control precision is not high. Therefore, the process conditions vary significantly even under the same setting of the manipulating factors. Due to this variation, the surrogate modeling that directly relates the manipulating factors to the process outcome does not provide a great predictive power on the outcome. At the same time, the model relating the process conditions to the outcome is not appropriate for the prediction purpose because the process conditions cannot be accurately set as planned due to the control uncertainties for a future process run. Motivated by the example, we propose a two-tiered Gaussian process (GP) model, where the bottom tier relates the manipulating factors to the process conditions with control variation, and the top tier relates the process conditions to the outcome. It explicitly models the propagation of the control uncertainty to the outcome through the two modeling tiers. The benefits of the approach over the standard GP approach are illustrated with multiple simulated scenarios and carbon nanotube production processes.

Mirror error compensation is usually used to improve the machining precision of thin-walled parts. However, due to the time-varying cutting condition of thin-walled parts, this zero-order method may result in inadequate error compensation. To cope with this problem, an online first-order error compensation method is proposed for thin-walled parts. With this context, first, the time-varying cutting condition of thin-walled parts is defined with its in-process geometric and physical characteristics. Based on it, a first-order machining error compensation model is constructed. Then, before process starting, the theory geometric and physical characteristics of thin-walled parts are, respectively, obtained with CAM software and structure dynamic modification theory. After process performing, the real geometric characteristic of thin-walled parts is measured, and it is used to calculate the dimension error of thin-walled parts. Next, the error compensated value is evaluated to construct an error compensation surface, which is used to modify the tool center points of next process step. Finally, the machining error is compensated by performing the next process step. Two typical experiments, milling of thin-walled parts with plane- and curved-surface, are used to validate the proposed method, and the experiment results shown that this method can significantly improve the error compensation effect for low-stiffness structure. Compared with the mirror compensation, the final thickness error of thin-walled parts is reduced by 71.4% and 56.2%, respectively, for plane- and curved-surface parts.

The rapidly growing deployment of lithium-ion batteries in electric vehicles is associated with a great waste of natural resource and environmental pollution caused by manufacturing and disposal. Repurposing the retired lithium-ion batteries can extend their useful life, creating environmental and economic benefits. However, the residual capacity of retired lithium-ion batteries is unknown and can be drastically different owing to various working history and calendar life. In this study, we used the incremental capacity (IC) curve to estimate the residual capacity of waste power batteries. First, through experimental means, the parameters of the battery and the IC charging curve are measured. Second, to achieve rapid capacity estimation, a battery capacity estimation method based on the adaptive genetic algorithm-back propagation neural network (AGA-BPNN) is proposed and compared with other classic machine learning methods. The proposed algorithm reduced the error of capacity estimation to 3%. Finally, through the analysis of the IC curve, a method for identifying aging mechanism of large-scale decommissioned batteries is obtained. This research provides effective support for the capacity-based classification of large-scale decommissioned power batteries.

Additive manufacturing (AM) is a type of advanced manufacturing process that enables fast prototyping to realize personalized products in complex shapes. However, quality defects existed in AM products can directly lead to significant failures (e.g., cracking caused by voids) in practice. Thus, various inspection techniques have been investigated to evaluate the quality of AM products, where X-ray computed tomography (CT) serves as one of the most accurate techniques to detect geometric defects (e.g., voids inside an AM product). Taking a selective laser melting (SLM) process as an example, voids can be detected by investigating CT images after the fabrication of products with limited disturbance from noises. However, limited by the sensor size and scanning speed issue, CT is difficult to be used for online (i.e., layer-wise) voids detection, monitoring, and process control to mitigate the defects. As an alternative, optical cameras can provide layer-wise images to support online voids detection. The intricate texture of the layer-wise image restricts the accuracy of void detection in AM products. Therefore, we propose a new method called pyramid ensemble convolutional neural network (PECNN) to efficiently detect voids and predict the texture of CT images using layer-wise optical images. The proposed PECNN can efficiently extract informative features based on the ensemble of the multiscale feature-maps (i.e., image pyramid) from optical images. Unlike deterministic ensemble strategies, this ensemble strategy is optimized by training a neural network in a data-driven manner to learn the fine-grained information from the extracted feature-maps. The merits of the proposed method are illustrated by both simulations and a real case study in a SLM process.

Modern CNC machine tools provide lookup tables to enhance the machine tool’s precision but the generation of table entries can be a demanding task. In this paper, the coefficients of the 25 cubic polynomial functions used to generate the LUTs entries for a five-axis machine tool are obtained by solving a linear system incorporating a Vandermonde expansion of the nominal control jacobian. The necessary volumetric errors within the working volume are predicted from the machine’s geometric errors estimated by the indirect error identification method based on the on-machine touch probing measurement of a reconfigurable uncalibrated master ball artifact (RUMBA). The proposed scheme is applied to a small Mitsubishi M730 CNC machine. Two different error models are used for modeling the erroneous machine tool, one estimating mainly inter-axis errors and the other including numerous intra-axis errors. The table-based compensation is validated through additional on-machine measurements. Experimental tests demonstrate a significant reduction in volumetric errors and in the effective machine error parameters. The LUTs reduce most of the dominant machine error parameters. It is concluded that although being effective in correcting some geometric errors, the generated LUTs cannot compensate for some axis misalignments such as EB(OX)A and EB(OX)Z. The root-mean-square of the translational volumetric errors is improved from 87.3, 75.4, and 71.5 µm down to 24.8, 18.8, and 22.1 µm in the X-, Y-, and Z-directions, respectively.

Rotary axes are the key components for five-axis computerized numerical control machines, while their motions are dramatically influenced by thermal issues. To precisely model the thermal error of rotary axis, a convolutional neural network (CNN) model is developed. To form data sets for the CNN, a laser interferometer is used to measure the angular positioning error at different temperatures and a thermal imager is taken to obtain thermal images of the rotary axis. The measured thermal error is fitted to a sine curve so that training parameters are reduced. And the thermal pixel values of the initial thermal image are subtracted from all the thermal images to consider the incremental thermal effect, so the influence of the initial temperature is negligible. Finally, a deep CNN model with multiple output classifications is designed to complete the data training, verifying and testing. The experimental results show that the prediction accuracy for the parameters is higher than 90%, and the percentage reduction in error is higher than 80%.

This paper presents a novel method for aligning the scanned point clouds of damaged blades with their nominal computer-aided design (CAD) model. To inspect a damaged blade, the blade surface is scanned and the scan data in the form of a point cloud is compared with the nominal CAD model of the blade. To be able to compare the scanned point cloud and the CAD model, they must be brought to a common coordinate system via a registration algorithm. The geometric nonconformity between the scanned damaged blade and its nominal model stemmed from the damaged regions can affect the registration (alignment) outcome. The alignment errors then cause wrong inspection results. To prevent this from happening, the data points from the damaged regions have to be removed from the alignment calculations. The proposed registration method in this work can accurately and automatically eliminate the unreliable scanned data points of the damaged regions from the registration process. The main feature is a correspondence search technique based on the geometric properties of the local neighborhood of points. By combining the average curvature Hausdorff distance and average Euclidean Hausdorff distance, a metric is defined to locally measure the dissimilarities between the scan data and the nominal model and progressively remove the identified unreliable data points of the damaged regions with each iteration of the fine-tuned alignment algorithm. Implementation results have demonstrated that the proposed method is accurate and robust to noise with superior performance in comparison with the existing methods.

Modeling the variation propagation based on the stream of variation (SoV) methodology for multistage machining processes (MMPs) has been investigated intensively in the past two decades; however, little research is conducted on the variation reduction and the existing work fails to be applied to irregular features caused by the machining-induced variation varying with the positions of the contour points on the machined surface. This paper proposes a novel error compensation method for MMPs by modifying the tool path to reduce variation for general features. The method based on differential motion vector (DMV) sets of multiple contour points is presented to represent the deviation of the irregular feature. Then, the conventional stream of variation (SoV) model is further extended to more accurately describe variation propagation for irregular features considering the actual datum-induced variations and the varying machining-induced variations, especially the deformation errors for the low stiffness workpiece. Based on the extended SoV model and error equivalence mechanism, the datum error and fixture error are transformed to the equivalent tool path error. Then, the original tool path is modified through shifting the machine zero points of machine tools with no need for changing the original G code and workpiece setup. A real cutting experiment validates the effectiveness of the proposed error compensation method for MMPs with an average precision improvement of over 60%. The application of the extended SoV model significantly contributes to compensating more complex error sources for MMPs, such as the clamp force, the internal residual stress, etc.

To guarantee the final assembly quality of the large-scale components, the assembly interfaces of large components need to be finish-machined on site. Such assembly interfaces are often in low-stiffness structure and made of difficult-to-cut materials, which makes it hard to fulfill machining tolerance. To solve this issue, a data-driven adaptive machining error analysis and compensation method is proposed based on on-machine measurement. Within this context, an initial definite plane is fitted via an improved robust iterating least-squares plane-fitting method based on the spatial statistical analysis result of machining errors of the key measurement points. Then, the parameters of the definite plane are solved by a simulated annealing-particle swarm optimization (SA-PSO) algorithm to determine the optimal definite plane; it effectively decomposes the machining error into systematic error and process error. To reduce these errors, compensation methods, tool-path adjustment method, and an optimized group of cutting parameters are proposed. The proposed method is validated by a set of cutting tests of an assembly interface of a large-scale aircraft vertical tail. The results indicate that the machining errors are successfully separated, and each type of error has been reduced by the proposed method. A 0.017 mm machining accuracy of the wall-thickness of the assembly interface has been achieved, well fulfilling the requirement of 0.05 mm tolerance.

One major impediment to wider adoption of additive manufacturing (AM) is the presence of larger-than-expected shape deviations between an actual print and the intended design. Since large shape deviations/deformations lead to costly scrap and rework, effective learning from previous prints is critical to improve build accuracy of new products for cost reduction. However, products to be built often differ from the past, posing a significant challenge to achieving learning efficacy. The fundamental issue is how to learn a predictive model from a small set of training shapes to predict the accuracy of a new object. Recently an emerging body of work has attempted to generate parametric models through statistical learning to predict and compensate for shape deviations in AM. However, generating such models for 3D freeform shapes currently requires extensive human intervention. This work takes a completely different path by establishing a random forest model through learning from a small training set. One novelty of this approach is to extract features from training shapes/products represented by triangular meshes, as opposed to point cloud forms. This facilitates fast generation of predictive models for 3D freeform shapes with little human intervention in model specification. A real case study for a fused deposition modeling (FDM) process is conducted to validate model predictions. A practical compensation procedure based on the learned random forest model is also tested for a new part. The overall shape deviation is reduced by 44%, which shows a promising prospect for improving AM print accuracy.

This paper presents a novel thickness profile measuring system that measures double-sided thin pipe wall surfaces in a non-contact, continuous, cosine error-free, and fast manner. The surface metrology tool path was developed to align the displacement sensors always normal to the double-sided surfaces to remove cosine error. A pair of capacitive-type sensors that were placed on the rotary and linear axes simultaneously scans the inner and outer surfaces of thin walls. Because the rotational error of the rotary axis can severely affect the accuracy in thickness profile measurement, such error was initially characterized by a reversal method. It was compensated for along the rotational direction while measuring the measurement target. Two measurement targets (circular and elliptical metal pipe-type thin walls) were prepared to validate the developed measurement method and system. Not only inner and outer surface profiles but also thin-wall thickness profiles were measured simultaneously. Based on the output data, the circularity and wall thickness variation were calculated. The thickness profile results showed a good agreement with those obtained by a contact-type micrometer (1-µm resolution) at every 6-deg interval. The uncertainty budget for this measuring system with metrology tool path planning was estimated at approximately 1.4 µm.

Surface errors due to force-induced tool and workpiece deflections are one of the major errors in multi-axis machining of parts especially with thin-walled structures. Dominant approaches to reduce these surface errors are re-machining the part, feed scheduling, and tool path modification. These methods are time consuming and computationally costly, and they rely on experimental data which is used in cutting force and deflection predictions. The present paper introduces a pure geometrical approach to reduce surface errors drastically by minimizing the engagement lengths of flutes’ cutting edges when a point on the flute’s cutting edge is in contact with the design surface. The total engagement length of the flutes’ cutting edges when one of them generates a contact point on the workpiece surface is formulated and considered as the minimization objective function of an optimization problem. Tilt and lead angles, which define the tool orientation, are the design variables of the optimization problem subjected to constraints based on the geometrical requirements of the ball end milling process. The optimization problem uses the nominal tool path to generate an optimal tool path with adjusted tool orientations. The presented method is computationally inexpensive and does not need any experimentally calibrated coefficients to predict cutting forces because of the pure geometrical nature of the approach. The method is experimentally validated through five-axis ball end milling experiments in which more than 90% surface error reduction is achieved.

Thermo-mechanical models, based on the discretization of the heat transfer and elasticity equations, enable the analysis and optimization of the thermal design of machine tools. This work investigates the thermo-mechanical response of a five-axis precision machine tool to fluctuations of the environmental temperature. To increase the computational efficiency of the thermo-mechanical model, a surrogate model by means of projection-based model order reduction (MOR) is created. This article uses the parametric Krylov Modal Subspace (KMS) method, which enables the evaluation of the thermo-mechanical response of the machine tool for different values of the parameters describing the convective boundary conditions. The thermo-mechanical model is validated comparing the simulated and measured response of the machine tool to a step in the environmental temperature. The validation process uses the global sensitivity analysis (GSA) to determine the convective boundary conditions with the largest impact on the thermally induced deviations. The reduced-order model ensures the computational tractability of the Monte Carlo simulation associated with the sensitivity analysis and parameter identification. The validated thermo-mechanical model is used to investigate the thermo-mechanical design of the machine tool.

Surface profile is one of the foremost aspects to evaluate milling performance. Its generation mechanism is affected by a variety of factors such as tool geometry, runout values, and process parameters and thus still deserves further investigation. This paper aims to propose a unified method to study the surface generation mechanism and to predict the machining quality for variable pitch tools considering runout. First, a Floquet theory based algorithm is extended to analyze the machining stability and output the dynamic responses of the machining system. The resultant trajectories of cutting edges are obtained by kinematic synthesis of system vibrations, tool rotations, and machining feed. Next, both the surface location error (SLE) and the surface roughness are simultaneously extracted from the edge trajectories. A series of cutting tests are performed to validate the prediction results. Some new characteristics of the machined surface profile in terms of form errors and teeth marks are discovered and theoretically explained. Finally, the joint influences of tool geometry, runout values, and process parameters on the surface generation mechanism and quality are analyzed in detail with the proposed method.