Probability distribution shape and skew: (a) positive skew; (b) zero skew; (c) negative skew. 

Probability distribution shape and skew: (a) positive skew; (b) zero skew; (c) negative skew. 

Source publication
Article
Full-text available
Polymers are being used in a wide range of Additive Manufacturing (AM) applications and have been shown to have tremendous potential for producing complex, individually customized parts. In order to improve part quality, it is essential to identify and monitor the process malfunctions of polymer-based AM. The present work endeavored to develop an a...

Context in source publication

Context 1
... a negative skew, the tail on the left side of the probability density function is longer or fatter than the right side. On the contrary, a positive skew indicates that the tail on the right side is longer or fatter than the left side (Figure 3). Generally speaking, the probability density distribution of AE is an abnormal distribution, i.e., the skewness is not zero. ...

Citations

... Wu et al. 22 developed a method of acoustic emission (AE)-based monitoring that uses datadriven methods to identify the normal, out-of-material, and plugged nozzle states of the FFF process. Yang et al. 24 used an AE sensor to monitor filament breakage and the instantaneous skewness of the sensing data to predict it. Kim et al. 25 proposed an AE sensor and three accelerometers to diagnose loosened bolts in the nozzle head during the FFF process by using a support vector machine (SVM) model. ...
Article
Full-text available
3D printing has exhibited significant potential in outer space and medical implants. To use this technology in the specific high-value scenarios, 3D-printed parts need to satisfy quality-related requirements. In this article, the influence of the filament feeder operating states of 3D printer on the compressive properties of 3D-printed parts is studied in the fused filament fabrication process. A machine learning approach, back-propagation neural network with a genetic algorithm (GA-BPNN) optimized by k-fold cross-validation, is proposed to monitor the operating states and predict the compressive properties. Vibration and current sensors are used in situ to monitor the operating states of the filament feeder, and a set of features are extracted and selected from raw sensor data in time and frequency domains. Results show that the operating states of the filament feeder significantly affected the compressive properties of the fabricated samples, the operating states were accurately recognized with 96.3% rate, and compressive properties were successfully predicted by the GA-BPNN. This proposed method has the potential for use in industrial applications after 3D printing without requiring any further quality control.
... An increasingly important role is also covered by in-process monitoring of machine health and part quality. Multiple in-process sensing solutions are being explored, for example based on accelerometers [16,17], thermocouples [13,18,19] pressure sensors [8,18], load cells [20], axes encoders [12], acoustic emission sensors [21], optical and infrared cameras [22][23][24][25][26], in some cases combined to leverage the advantages of multi-sensor data fusion [20,27]. The use of digital twins to support in-process FFF data analysis [28] and the adoption of machine learning [29][30][31] are also demonstrating significant potential to improve in-process monitoring of FFF. ...
Article
A solution for in-process monitoring of part warpage in fused filament fabrication is presented, based on real-time measurement of the upwards-oriented repulsive force acting on the extruder during deposition. The force signal is processed with the help of simulation software implementing a mechanical model which incorporates the stiffness of the machine and the compliance of the part as it is fabricated. The model acts as a digital twin, where compliance is constantly updated while part geometry evolves through the creation of new layers. The incorporation of stiffness/compliance simulation helps the monitoring system in the interpretation of the force signal, leading to a better isolation of warpage events. The proposed monitoring solution is validated through the implementation of a sensorised hardware prototype and the execution of experiments involving the fabrication of test parts. The results indicate warpage detection performance with 92.9% accuracy, 91.5% specificity and 95.7% sensitivity, and demonstrate that it is possible to use a relatively affordable, quick and non-destructive sensing solution (that is, measuring the repulsive force acting on the extruder) as an effective means to detect part warpage. However, the developed solution highlights the challenge of developing accurate digital twins for the prediction of part compliance, essential for the correct interpretation of the force signal.
... The most common reason of filament breaking is non-uniform filament composition that cannot withstand the necessary dragging force. Yang et al. identified filament fracture with acoustic emission sensors, they demonstrated that when the filament is broken acoustic emission distribution changes as compared unbroken filament [21]. Two quantitative parameters were used to quantify variations in signal: relative similarity and instantaneous skewness. ...
Article
Full-text available
Recent improvements in the fused filament fabrication (FFF) technique have extended its usage for the manufacturing of finished products in a variety of engineering fields. However, some challenges are to be overcome including moderate mechanical, thermal, and physical properties of finished products. This paper reviews ways of improving these properties during three phases of manufacturing namely pre-processing phase, in-situ/on-site processing phase, and post-processing phase. A systematic review of these three phases is not much explored in the additive manufacturing process and hence there is a lot of scope for innovation in improving the quality of the finished product. Results of the review are finally concluded and it is found the approaches presented here in this paper resulted in increasing the performance of the fabrication process and allow for the fabrication of products with better mechanical, thermal and physical capabilities.
... The most common reason of filament breaking is non-uniform filament composition that cannot withstand the necessary dragging force. Yang et al. identified filament fracture with acoustic emission sensors, they demonstrated that when the filament is broken acoustic emission distribution changes as compared unbroken filament [21]. Two quantitative parameters were used to quantify variations in signal: relative similarity and instantaneous skewness. ...
Conference Paper
Full-text available
Recent improvements in the fused filament fabrication (FFF) technique have extended its usage for the manufacturing of finished products in a variety of engineering fields. However, some challenges are to be overcome including moderate mechanical, thermal, and physical properties of finished products. This paper reviews ways of improving these properties during three phases of manufacturing namely preprocessing phase, in-situ/on-site processing phase, and post-processing phase. A systematic review of these three phases is not much explored in the additive manufacturing process and hence there is a lot of scope for innovation in improving the quality of the finished product. Results of the review are finally concluded and it is found the approaches presented here in this paper resulted in increasing the performance of the fabrication process and allow for the fabrication of products with better mechanical, thermal and physical capabilities.
... Yang et al. detected filament breakage with AE sensors, showing that the distribution of the AE signal changed after filament breakage occurred. The signal changes are quantified using two measurable indicators: instantaneous skewness and relative similarity [41]. Between these indicators, the relative similarity is shown to outperform instantaneous skewness. ...
... Filament property varies from type to type, so identifying material 6. AE sensor mounts on the extruder shell for detecting filament breakage [41]. ...
... Amplitude spectrum of AE signals under different manufacturing conditions[41]. ...
Article
Real-time monitoring of the additive manufacturing process offers the promise of guaranteeing product quality and increasing the efficiency of the printing process. This paper summarizes research results for the in situ monitoring of the printing process for the fused filament fabrication process. To have a systematic and comprehensive summary, different methods, devices, and achievements in a range of monitoring systems for 3D printing are described. Sensor types and devices used in the literature for printer health-state monitoring and printing process product quality monitoring are summarized. Discussion of current and future research directions concludes the review.
... Various solutions regarding ME process monitoring have been proposed in literature. Acoustic emission technique is employed in order to detect filament blockage due to increased pressure in the hot end [21,22]. The same issue was detected through vibration obtained by an accelerometer that is mounted near the nozzle orifice [23]. ...
Article
Full-text available
Material Extrusion (ME) is one of the established processes in Additive Manufacturing (AM) realm. The process involves deposition of adjacent material strands that are bonded together in order to create successive layers that are stacked one upon the previous until the three-dimensional part is fabricated. One of the key factors for increased performance and part quality is the extrusion of strands that are uniformly deposited according to the designed trajectory, without defects and with minimal deviation from the intended strand shape and dimension. This objective becomes more challenging when ME subsystems must be optimally synchronized, in order to perform material deposition at higher rates, given the timely need to improve ME process efficiency. In this study, eight factors that control material flow and deposition from kinematic, rheological, and thermal aspects are investigated via a fraction factorial design of experiments. Machine vision methods are employed to evaluate factor contribution to the fabrication of simple linear segments, and to effectively measure the free form shape of the resulting strands. Analysis of the experimental results reveals the mechanisms that trigger Material Deposition Discontinuity (MDD) during part fabrication. In addition, Strand Width Deviation (SWD) along the linear path is minimized with special attention to the undesirable material overfills at the start and the end points of the strand. Two separate regression models are generated, describing the relationship between the investigated factors and the MDD and SWD criteria. The models are utilized as roadmaps towards continuous and uniform deposition in indicative case studies at normal (4.80 mm3/s) and increased (7.80 mm3/s) melt flow rate. Compared with non-optimized runs at similar flow rates, continuous and more uniform deposition is achieved for both normal and increased rates as indicated by a corresponding reduction of 34% and 19% of the measured strand width deviation.
... Wu et al. [6] have proposed using acoustic emission (AE) sensors to identify the printer failure using a hidden semi-Markov model. Along the same lines, Li et al. [7] have proposed the use of AE monitoring to identify warping induced print distortions while Yang et al. [8] demonstrated the use of AE monitoring for real-time filament breakage sensing. Kousiatza and Karalekas [9] used fiber Bragg grating (FBG) sensors for in-situ strain field monitoring and to develop temperature profiles during an FDM print. ...
Article
Full-text available
Fused Deposition Modeling (FDM) is a highly versatile additive manufacturing method for 3D printing thermoplastic-based components at small as well as larger production scales. By combining the filament with fibers from other materials including wood, metal, glass, and carbon, the method can easily be adapted to print complex parts using a variety of materials. However, despite its popularity, online print quality and machine monitoring continue to remain a challenge. Here, we present the preliminary results from our efforts on using cheap off-the-shelf sensors in combination with discrete wavelet transform analysis to identify the differences in the vibroacoustic signals measured near the print area during successful and failed first layer filament deposition on the build plate. A failure in creating a strong first layer bond between the extruded filament and the build plate always results in a print failure and is one of most common print issues occurring in FDM printing. By controlling the extruder and build plate temperatures, we control the filament – build plate bond strength while measuring the generated vibroacoustic signals using a PVDF piezo sensor. The measured signals are analyzed using a discrete wavelet transform to partition the signal energy into different energy levels. For the cases studied, we find that the relevant noticeable differences can be observed in specific energy levels during good and bad bond formation. Reconstructing the signal using these energy contributions provides a time domain representation of signal differences under different conditions. The obtained results demonstrate that a cheap and easy to implement method can be developed using PVDF sensors in combination with a wavelet-based signal analysis approach.
... Owing to its excellent ability to identify events at the initial fault stage, AE has been widely used in mechanical structure monitoring [5], civil engineering [6], machining condition monitoring [7], machine failure detection [8], material testing [9], and other fields. It has also proved to be a promising technique in some emerging fields, such as additive manufacturing (AM) process monitoring [10,11]. (2  10 6 samples  2 bytes  10 s). ...
Article
One of the most important issues arising in the use of acoustic emission (AE) for nondestructive process monitoring is the accurate identification of potential process malfunctions to avoid premature failure. In some cases, the AE signals from malfunction sources are relatively weak, with high levels of background noises. Thus, these signals could easily be submerged, making it rather difficult to separate them. Therefore, it is of critically importance to find a solution to the problem of weak emission source identification and obtain a correct representation of original waveforms. The present work proposes a new feature representation method based on similarity of probability distributions from raw AE waveforms. The Bhattacharyya coefficient is used for this purpose. A standard procedure for calculating similarity is formulated. Both an instantaneous similarity and a relative similarity are defined. The influences of the choice of some key parameters are discussed in detail. Tests on filament breakage detection in an additive manufacturing process reveal the feasibility and effectiveness of the proposed method. This method is believed to be appropriate when the target malfunction emission signal amplitude is less than the environmental emission signals generated by other stationary sources, and threshold methods fail to perform properly. It could also be used as an alternative feature representation method for AE signals in other fields.
... Accelerometers, thermocouples, infrared temperature sensor, video borescope were used to monitor the quality of 3D printed parts, and most optimal parameters of feed/flow ratio, extruder temperature, and layer height were recommended for better dimensional accuracy and surface roughness [28,29]. Acoustic emission monitoring technique was used to detect such 3D printing process errors as semiblocked extruder, completely blocked extruder, and run out of the material [30], and filament breakage [31]. Orientation, motion, hygrometry, temperature, and vibration sensors were utilised to track printing and not printing conditions in FDM [32]. ...
Article
Full-text available
3D printing and particularly fused deposition modelling (FDM) is widely used for prototyping and fabricating low-cost customised parts. However, present fused deposition modelling 3D printers have limited nozzle condition monitoring techniques to minimize nozzle clogging errors. Nozzle clogging is one of the significant process errors in fused deposition modelling 3D printers, and it affects the quality of prototyped parts in terms of mechanical properties and geometrical accuracy. This paper proposes a dynamic model for current-based nozzle condition monitoring in fused deposition modelling, which is briefly described as follows. First, all the process forces in filament extrusion of the fused deposition modelling were identified and derived theoretically, and theoretical equations of the feed rolling forces and flow-through-nozzle forces were derived. In addition, the effect of the nozzle clogging on the current of extruding motor were identified. Second, based on the proposed dynamic model, current-based nozzle condition monitoring method was proposed. Next, sets of experiments on FDM machine using polylactic acid (PLA) material were carried out to verify the proposed theoretical model, and the results were analysed and evaluated. Findings of the present study indicate that nozzle clogging in FDM 3D printing can be monitored by sensing the current of the filament extruding motor. The proposed model can be used efficiently for monitoring nozzle clogging conditions in fused deposition modelling 3D printers as it is based on the fundamental process modelling.
... Features in the time and frequency extracted by multiple sensor including thermocouples, infrared temperature sensors and accelerometers, was analyzed by six different machine learning algorithms. Yang et al. [12] used an AE sensor for filament breakage monitoring and found that the instantaneous skewness could be used as a preliminary indicator for filament breakage. Filament jam is also a common failure in FFF machines [20]. ...
... Several methods are currently under development for reliable in-situ monitoring based on the sensors for defect detection and location [8][9][10][11][12][13][14][15][16]. The sensors applied for in-situ monitoring during the FFF process contain acoustic emission (AE) [8], accelerometer sensors [10], infrared temperature sensors [11], fiber Bragg grating (FBG) sensors [15], visual camera [16], and more. ...
... The sensors applied for in-situ monitoring during the FFF process contain acoustic emission (AE) [8], accelerometer sensors [10], infrared temperature sensors [11], fiber Bragg grating (FBG) sensors [15], visual camera [16], and more. According to previous studies, the objectives of in-situ monitoring and diagnosis of the FFF process can be divided into two groups: one for the states of the FFF machine [8][9][10][11][12][13], and the other for the quality of the building parts [14][15][16][17][18][19]. ...
Article
Full-text available
Fused filament fabrication (FFF) is one of the most widely used additive manufacturing (AM) technologies and it has great potential in fabricating prototypes with complex geometry. For high quality manufacturing, monitoring the products in real time is as important as maintaining the FFF machine in the normal state. This paper introduces an approach that is based on the vibration sensors and data-driven methods for in-situ monitoring and diagnosing the FFF process. The least squares support vector machine (LS-SVM) algorithm has been applied for identifying the normal and filament jam states of the FFF machine, besides fault diagnosis in real time. The identification accuracy for the case studies explored here using LS-SVM is greater than 90%. Furthermore, to ensure the product quality during the FFF process, the back-propagation neural network (BPNN) algorithm has been used to monitor and diagnose the quality defects, as well as the warpage and material stack caused by abnormal leakage for the products in-situ. The diagnosis accuracy for the case studies explored here using BPNN is greater than 95%. Results from the experiments show that the proposed approach can accurately recognize the machine failures and quality defects during the FFF process, thus effectively assuring the product quality.