Academia.eduAcademia.edu

Figure 1 – uploaded by Alfonso Paoletti

See full PDFdownloadDownload figure
Fig. 1. Schematic of the mutual relationship between absorbed power and material state

Figure 1 Schematic of the mutual relationship between absorbed power and material state

Related Figures (12)

Fig. 2. Schematic of the phases used for the development of the Machine Learning Model. The developed model was based on the data encoding-decoding process. First (Encoding) power and temperature curves of a certain test are subdivided into non-sequential data records that are used to train the network. Thus, the network is not provided with information con- cerning what is the source of the data. Then, once trained, (Decoding) the network performs the prediction of temperature variation based on
Fig. 2. Schematic of the phases used for the development of the Machine Learning Model. The developed model was based on the data encoding-decoding process. First (Encoding) power and temperature curves of a certain test are subdivided into non-sequential data records that are used to train the network. Thus, the network is not provided with information con- cerning what is the source of the data. Then, once trained, (Decoding) the network performs the prediction of temperature variation based on
Main physical and thermal characteristics of involved materials as provided by materials manufacturers.
Main physical and thermal characteristics of involved materials as provided by materials manufacturers.
Fig. 3. Schematic of the equipment used to produce FAJ joints.
Fig. 3. Schematic of the equipment used to produce FAJ joints.
Fig. 4. Sequence of Data Manipulation Steps.
Fig. 4. Sequence of Data Manipulation Steps.
Fig. 6. Influence of the processing conditions on (a) plunging load; (b) torque; (c) power and (d) temperature. Fig. 5. Schematic of data encoding.
Fig. 6. Influence of the processing conditions on (a) plunging load; (b) torque; (c) power and (d) temperature. Fig. 5. Schematic of data encoding.
Fig. 7. Correspondence between the occurrence of power peak (and valleys) with the onset of temperature knees.
Fig. 7. Correspondence between the occurrence of power peak (and valleys) with the onset of temperature knees.
Fig. 8. Comparison of experimental measurements and simulated temperature trends for some reference plunging loads using different network configurations (a) ANN1; (b) ANN2; (c) ANN3 and (d) ANN4. (@ = 6000 RPM).
Fig. 8. Comparison of experimental measurements and simulated temperature trends for some reference plunging loads using different network configurations (a) ANN1; (b) ANN2; (c) ANN3 and (d) ANN4. (@ = 6000 RPM).
Fig. 9. Comparison of experimental measurements and simulated temperature trends for some reference plunging loads using different network configurations (a) ANN1; (b) ANN2; (c) ANN3 and (d) ANN4. (@ = 8000 RPM).
Fig. 9. Comparison of experimental measurements and simulated temperature trends for some reference plunging loads using different network configurations (a) ANN1; (b) ANN2; (c) ANN3 and (d) ANN4. (@ = 8000 RPM).
Mean and standard deviations of the error calculated on the entire set of tests. Table 2 joints and the process control.
Mean and standard deviations of the error calculated on the entire set of tests. Table 2 joints and the process control.
Fig. 10. Average temperature errors of the different networks (ANN1-ANN4) under processing conditions.
Fig. 10. Average temperature errors of the different networks (ANN1-ANN4) under processing conditions.
Fig. 11. Applications of the developed model. A Machine Learning model based on ANN was developed to predict the temperature variation during friction Assisted Joining of metal- polymer structures. The model takes the absorbed power as input and forecasts temperature trend recursively. The model was trained using processing loads (the plunging load and the torque) and temperature measurements performed employing an IR camera. The entire data processing flow is explained in the paper. The main conclusions are summarized as follows:
Fig. 11. Applications of the developed model. A Machine Learning model based on ANN was developed to predict the temperature variation during friction Assisted Joining of metal- polymer structures. The model takes the absorbed power as input and forecasts temperature trend recursively. The model was trained using processing loads (the plunging load and the torque) and temperature measurements performed employing an IR camera. The entire data processing flow is explained in the paper. The main conclusions are summarized as follows:
Related topics:
Materials ScienceManufacturing Engineering
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved