Progress 11/01/23 to 10/31/24
Outputs
Target Audience:Microwaveable food companies, microwave oven manufacturers, and other researchers in the field. Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?One Ph.D. student, one Master's student, and one Post-doctoral Research associate were trained in developing multiphysics and machine learning models to simulate the microwave heating processes. They were also trained in developing Python-based programs to control the multiport solid-state microwave system for heating experiments. How have the results been disseminated to communities of interest?The results have been published as five peer-reviewed journal articles and presented as five presentations at various conferences/webinars. We also discussed collaboration opportunities with industrial collaborators. What do you plan to do during the next reporting period to accomplish the goals?We will further improve the accuracy of the machine learning model in predicting the frequency-dependent thermal contributions. Then, the machine learning model will be integrated with the previously developed complementary frequency/relative phase strategies and implemented on a real solid-state microwave system. We will continue to interact with industrial collaborators to develop and adopt integrated multiphysics modeling and machine learning modeling platforms further.
Impacts
What was accomplished under these goals?
During Year 4, the team focused on the development and implementation of solid-state microwave processing strategies, as well as reaching out to food industries for potential adoption of the developed Product Development Platform. a. Development of advanced relative phase shifting strategies for smart dual-port solid-state microwave processing The previous modeling results of constructive, destructive, and interferences of dual-port microwaves showed that relative phase greatly impacts the microwave heating performance. We have explored the relative phase control strategies that can improve microwave heating performance. We first evaluated the combined effect of active shielding packaging and relative phase sweeping on heating performance in a dual-port solid-state microwave heating process. Four types of heating strategies that combine two factors (passive and active packaging and fixed and sweeping relative phase) were used in heating a tray of 300 g gellan gel sample for 3 min. The temperature distributions at the top and middle layers of the heated samples were collected and analyzed for heating uniformity index (HUI) and power absorption efficiency (PAE). The results showed that both active shielding package and relative phase sweeping can individually improve the HUI while maintaining high PAE. The packaging factor and relative phase factor showed a significant interaction effect (p = 0.0021) in influencing HUI but not PAE, highlighting the necessity of considering both factors (packaging and relative phase) when optimizing the microwave heating uniformity. The combination of active packaging and sweeping relative phase is a robust heating strategy that delivers the best heating performance that is significantly better than or similar to other combinations of packaging and relative phase strategies. We then developed and evaluated a combined-sweeping & complementary-relative-phase-shifting heating strategy for dual-port microwave heating of foods. The combined-sweeping & complementary-relative-phase-shifting strategy first orderly swept the relative phases between two ports from 0° to 315° at an interval of 45° operated at a frequency of 2.45?GHz and a power level of 200?W for each port to determine relative phase-dependent thermal contributions. Then, relative phases that have complementary thermal contributions to the real-time heating results were determined and used in the heating process. The performance of microwaving a tray of 300?g gellan gel model food for 3?minutes was compared with that of the fixed-relative-phase and orderly-sweeping-relative-phase strategies. The results showed that the complementary strategy delivered significantly more uniform heating results (HUI = 0.27 ± 0.01) than the other strategies. Similar microwave power absorption and maximum and minimum temperatures were observed in the complementary and orderly sweeping heating strategies, which was better than the fixed heating strategy. Overall, the complementary strategy showed superior heating uniformity, followed by the orderly sweeping and then the fixed-relative-phase heating strategy. The complementary relative phase shifting strategy is promising to be incorporated into solid-state microwave systems for smart heating processes. Upon the observation of sinusoidal relative phase-dependent microwave power absorption in dual-port microwave processing, a more efficient predictive complementary heating strategy is under development. b. Identification of turntable function in a solid-state microwave heating process with diverse frequency-shifting strategies applied The turntable has been used in conventional magnetron-based microwave ovens to improve the heating uniformity. Meanwhile, with the emerging solid-state technique, well-designed frequency-shifting strategies also show promise in achieving improved microwave outcomes without a rotating turntable. We used modeling tools to comprehensively assess the turntable function under various conditions, where the affecting factors included vertical and horizontal positions of food, rotation status of the turntable, frequency-shifting strategy, and food configurations. Results illustrated that the elevation of food by the turntable is more critical to power efficiency and heating uniformity, while rotation mainly works to narrow the temperature difference between hot and cold spots. Moreover, the complementary-shifting strategy lessened the rotation function of the turntable. Hence, in a solid-state microwave system embedded with a proper frequency-shifting strategy, a simpler supporting component shall replace the current rotating turntable, streamlining the oven design without compromising the microwave performance. c. Development and implementation of machine learning-guided smart control of solid-state microwave processing From the last report, we have developed a Density-Based Spatial Clustering of Algorithms with Noise (DBSCAN) clustering method to categorize and measure the hotspot information. Based on the findings, we developed a convolutional neural network (CNN) machine learning model to predict the effect of microwave frequency on solid-state microwave heating performance. We first used multiphysics modeling to simulate the effect of microwave frequency on heating results. The combinations of three arbitrary food geometries (round, rectangular, and square) and seven food products (mashed potato, tuna flesh, egg white, fried chicken, lean beef, egg yolk, and turnip) were used in the multiphysics modeling to generate the training datasets for the CNN model. The CNN model used a newly proposed architecture (W-net) that first used frequency-dependent thermal patterns as input layers in one convolution and deconvolutional process (U-net), and then the frequency values were attached to the deconvolutional results to conduct another U-net process. The double U-net ("W-net") process showed better prediction results than the commonly used U-net structure. The hyperparameters of the CNN model were also optimized in the model training process. The trained CNN model was tested on a new food geometry and food product product, showing good predictions on frequency-dependent thermal patterns. This trained machine learning model using multiphysics modeling results was preliminary tested on a real microwave heating process with promising results. To further implement the trained machine learning model on a real microwave system, we are improving the accuracy of the CNN model by incorporating more training datasets. We are collecting more experimental frequency-dependent thermal pattern results for various food products and packages. The experimental patterns will be combined with the modeling results as new training datasets for CNN model refinement. d. Demonstration of Product Development Platform for microwaveable food product design and reach out for potential industrial collaboration We have fine-tuned the interface of the Product Development Platform. We also reached out to industrial collaborators to seek more opportunities for collaboration and technology adoption. Dr. Chen has presented at the "The More You Know" Webinar Series at Kraft Heinz Company. The University of Tennessee and Kraft Heinz Company also signed a non-disclosure agreement. Currently, the teams are discussing more collaboration opportunities on microwave processing technology.
Publications
- Type:
Peer Reviewed Journal Articles
Status:
Published
Year Published:
2024
Citation:
Ghimire, Arjun, and Jiajia Chen. "A combined-sweeping & complementary-relative-phase strategy to improve heating performance in a dual-port solid-state microwave system." Food and Bioproducts Processing 148 (2024): 392-399.
- Type:
Peer Reviewed Journal Articles
Status:
Published
Year Published:
2024
Citation:
Yang, Ran, Zhenbo Wang, and Jiajia Chen. "3-D geometric design of microwaveable food products for optimal heating uniformity based on machine learning-supervised multiphysics models." Food and Bioproducts Processing 147 (2024): 393-405.
- Type:
Peer Reviewed Journal Articles
Status:
Published
Year Published:
2024
Citation:
Ghimire, Arjun, Ran Yang, and Jiajia Chen. "The combined effect of active packaging and relative phase sweeping on microwave heating performance in a dual-port solid-state system." Journal of Microwave Power and Electromagnetic Energy 58.3 (2024): 170-185.
- Type:
Peer Reviewed Journal Articles
Status:
Published
Year Published:
2024
Citation:
Yang, Ran, and Jiajia Chen. "Identification of turntable function in a solid-state microwave heating process with diverse frequency-shifting strategies applied." Innovative Food Science & Emerging Technologies 94 (2024): 103670.
- Type:
Peer Reviewed Journal Articles
Status:
Published
Year Published:
2024
Citation:
Verma, Kartik, et al. "An integrated numerical and analytical model to understand the effect of relative phase in a dual-port solid-state microwave heating process." Journal of Food Engineering 367 (2024): 111869.
|
Progress 11/01/22 to 10/31/23
Outputs
Target Audience:Microwaveable food companies, microwave oven manufacturers, and other researchers in the field. Changes/Problems:We proposed to develop the modeling platforms as interfaced applications without using a COMSOL Multiphysics license. However, the developed applications (without a license) cannot interact with our machine learning algorithms. Then, we used Python language to develop the modeling interface that can interact with both multiphysics modeling and machine learning algorithms; however, a COMSOL license is still needed. Due to the complicated solid-state microwave processing in parameter control, we have delays in working on the Knowledge Generation Platform. We will need to request a non-cost extension to spend more time completing the proposed work of solid-state microwave processing. What opportunities for training and professional development has the project provided?One Ph.D. student, one Master's student, and one Post-doctoral Research associate were trained in developing multiphysics and machine learning models to simulate the microwave heating of food in domestic microwave ovens. They were also trained in developing Python-based programs to control the multiport solid-state microwave system for heating experiments. How have the results been disseminated to communities of interest?The results have been published as three peer-reviewed journal articles and one Master's thesis and presented as one presentation at 2023 IMPI. What do you plan to do during the next reporting period to accomplish the goals?For the domestic oven-related work, we will continue to improve the interface of the Product Development Platform and interact with industrial collaborators for feedback. For the solid-state oven-related work, we will integrate the validated solid-state microwave heating model with the machine learning model to understand interactions between solid-state microwave heating and food products and develop the Knowledge Generation Platform. We will continue to develop algorithms to control solid-state microwave parameters for smart oven development.
Impacts
What was accomplished under these goals?
During Year 3, the team accomplished the following work of the two parallel research directions of magnetron-based and solid-state microwave processing under three objectives: a. Development and validation of a 3-D scanning approach to capture oven geometry for model accuracy improvement that supports both magnetron-based and solid-state microwave processing The accuracy of the multiphysics models is significantly influenced by the microwave oven geometry. To improve the modeling accuracy further, Drs. Chen and Gan developed a 3-D scanning approach to characterize the accurate geometric details of the cavity and incorporate it in the multiphysics modeling of microwave heating. A quantitative approach was also developed to replace the previously often-used qualitative approach to compare the spatial temperature profiles between the simulation and experiments. Results showed that the 3-D scanned approach can accurately incorporate the irregular geometric details of the oven cavity and can improve the prediction accuracy of microwave heating models for future food products and oven development. This new approach supports both domestic and solid-state microwave modeling work. b. Development of an integrated multiphysics modeling and machine learning-based model framework of multi-parameter optimization and preliminary Product Development Platform for microwaveable food product design under magnetron-based microwave processing Dr. Chen and Dr. Wang completed the integrated multiphysics modeling and machine learning-based modeling framework for multi-parameter (3D geometry) optimization. We developed and utilized an online machine learning-supervised multiphysics modeling strategy to optimize multiple geometrical parameters (e.g., top surface width, top surface length, and the ratio of top-to-bottom dimensions) simultaneously for optimal microwave heating uniformity. Initially, the multiphysics model simulated the microwave heating process at one or more given dimensional conditions and evaluated the heating uniformity after heating. The dimensional parameters and heating uniformity results were then used as initial training data for the Gaussian Process Regression and Bayesian optimization to predict the dimensional parameters with potential "best" heating uniformity. The multiphysics modeling was supervised by the machine learning model to conduct more modeling work and generate more paired data of geometry design and heating uniformity to train the machine learning model with an expanded training dataset. The loop of multiphysics modeling and machine learning was run until the stopping criterion was satisfied. The results indicated that the machine learning-supervised optimization strategy is efficient and robust using small (even with only one model) and randomly selected initial training models. Then, a Product Development Platform with a preliminary interface was developed using the Python Language based on the integrated multiphysics modeling and machine learning framework. The platform allows the users to input the types and dimensional ranges of the food geometries and perform an automatic geometry optimization process in the background using the developed multiphysics modeling and machine learning framework. After reaching the predefined optimization criterion (e.g., number of models, desired heating uniformity index), the modeling platform could provide the users with the most promising geometry designs (e.g., ranges of dimensional parameters) and screen out bad designs in a much more efficient way than multiphysics modeling alone. The team is currently working on further improving the interface of the modeling platform for easy use by the food developers. c. Development of multiphysics modeling of solid-state microwave processing and preparation data for machine learning modeling The solid-state microwave heating process is complicated because multiple parameters of multi-source can be controlled separately. Instead of simulating all combinations of parameters, Drs. Chen, Gan, and Fathy worked on developing and validating a simplified modeling approach to improve multiphysics modeling efficiency. We have developed an integrated numerical and analytical approach that allows efficient modeling and a better understanding of multi-mode wave interactions with source phase differences. First, two numerical models simulated the complex electric fields of individual-port microwave heating processes. The simulated nodal electric fields were used as inputs in the analytical model to predict the resultant power dissipation for dual-port heating with arbitrary source phase differences. The integrated approach was validated by comparing the predicted nodal electric field and power densities with numerical modeling results. The phase-dependent nodal and volumetric average power revealed wave-like patterns with great variances. The constructive, destructive, and interference effects were influenced by the source phase differences, emphasizing the importance of proper control of source phases for efficient and uniform heating. We also developed an algorithm to characterize the simulated frequency-dependent thermal heating results, which will be used in the to-be-developed machine learning models. To quantify the nonuniform heating distribution, a Density-Based Spatial Clustering of Algorithms with Noise (DBSCAN) clustering method was used to categorize and measure the hotspot information. We analyzed various aspects, including the size of the clusters, the spread and separation between them, and the locations of the highest temperatures. We are currently using the clustered heat performance in a convolutional neural network (CNN) machine learning model to understand the effect of microwave frequency on solid-state microwave heating, which will then be developed as one component of the Knowledge Generation Platform. d. Experimental evaluation of solid-state microwave processing technology In addition to the modeling work, Dr. Chen also used the experimental approach in developing solid-state microwave processing technology. Using a previously fabricated four-port solid-state microwave system, the dynamic complementary-frequency shifting strategy developed from the last report period was further comprehensively evaluated on five commercial and/or prepared meals with different characteristics, including single-component Pulled Chicken, multicomponent Beef in Gravy, multilayer Lasagna, multicompartment Pulled Chicken & Lasagna, and multicompartment Mashed Potato & Beef in Gravy. Results showed that the dynamic complementary-frequency shifting strategy improved microwave performance on all food products, especially on multicompartment foods. The presence of liquid components or the use of steam-venting packages may negatively affect the thermal profile collection in the dynamic strategy, hindering the algorithm performance, although considerable improvement was still observed. The complementary-frequency shifting strategy is highly promising to be incorporated in future solid-state microwave systems for improving microwave reheating performance. We also experimentally investigated the port interactions and heating performance of four different frequency shifting approaches (in the range of 2.4 and 2.5 GHz, at an interval of 0.01 GHz), i.e., Fixed-frequency without shifting, Synchronized-shifting, Inverse-shifting and Distinct-shifting, in a dual-source microwave system under both stationary and rotatory conditions. Results showed that port interactions were dependent on the microwave frequency and load position, both of which significantly affected the microwave power efficiency. All three shifting strategies significantly improved heating performance compared to Fixed-frequency heating.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2023
Citation:
Verma, K., Nachtrab, J., Dvorak, J., Alley, P., Yang, R., Gan, H., & Chen, J. (2023). 3-D scanned oven geometry improves the modeling accuracy of the solid-state microwave heating process. Journal of Microwave Power and Electromagnetic Energy, 57(4), 247-263.
- Type:
Journal Articles
Status:
Published
Year Published:
2023
Citation:
Yang, R., & Chen, J. (2023). Heating performance of dual-source microwave heating using different frequency shifting strategies in a solid-state system. Food Research International, 113781.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2023
Citation:
Yang, R. & Chen, J. (2023). Integrated multiphysics-modeling and machine-learning approach in optimizing microwavable food product geometry. Presented at the IMPI-57th Annual Microwave Power Symposium, June, Denver, CO, US. Oral presentation.
- Type:
Theses/Dissertations
Status:
Published
Year Published:
2023
Citation:
Verma, K. (2023). Multiphysics modeling to understand microwave-food interactions in a multi-port solid-state microwave system. Master's Thesis. University of Tennessee, Knoxville.
- Type:
Journal Articles
Status:
Published
Year Published:
2023
Citation:
Yang, R., Morgan, M., Fathy, A., Luckett, C., Wang, Z., & Chen, J. (2023). A Comprehensive Evaluation of Microwave Reheating Performance Using Dynamic Complementary-Frequency Shifting Strategy in a Solid-State System. Food and Bioprocess Technology, 16(5), 1061-1075.
|
Progress 11/01/21 to 10/31/22
Outputs
Target Audience:During this period, we interacted with a microwaveable food company, a microwave oven manufacturer, and other researchers in the field. Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?Two Ph.D. students and one Master's student were trained in developing multiphysics and machine learning models to simulate the microwave heating of food in domestic microwave ovens. The students were also trained in developing Python-based programs to control the multiport solid-state microwave system for heating experiments. The students presented theirresearch results at the IMPI (International Microwave Power Institute) conferences and COFE (Conferences of Food Engineering) to develop communication skills. How have the results been disseminated to communities of interest?The results have been published as two peer-reviewed journal articles and one doctoral dissertation and presented as four presentations at two conferences (2022IMPI and 2022 COFE). We also showedthe results to the industrial stakeholders (microwaveable food and oven companies) and got their input for the project improvement. What do you plan to do during the next reporting period to accomplish the goals?For the domestic oven-related work, we will develop the integrated multiphysics modeling and machine learning-based Product Development Platform. We anticipate completing the development of a preliminary APP that combines multiphysics modeling and machine learning for automatically simulating the microwave heating process. For the solid-state oven-related work, we will integrate the validated solid-state microwave heating model with the machine learning model to understand interactions between solid-state microwave heating and food products. We will continue to develop algorithms to control solid-state microwave parameters for smart-oven development.
Impacts
What was accomplished under these goals?
During the second year, the team further developed an integrated multiphysics modeling and machine learning-based model for multi-parameter optimization. The team also developed and validated the models for the solid-state microwave heating process. In addition, the team developed online closed-loop frequency shifting algorithms that could dynamically control the microwave parameters (frequency and power) to achieve better heating and thawing performances. Objective 1: Develop an integrated multiphysics modeling and machine learning-based Product Development Platform to assist current microwaveable food product development. Dr. Chen and Dr. Wang further developed integrated multiphysics modeling and machine learning-based models for multi-parameter (3D geometry) optimization. Three geometric parameters (length, width, and ratio of top to bottom surfaces) were used as targets in the optimization process. The integrated approach was validated by comparing the optimization performance with a parametric sweep approach, which was solely based on multiphysics modeling. The results showed that multi-parameter optimization could be achieved in a more efficient manner using the integrated approach. However, due to the complicated heating process in microwaves, several optimized results can be achieved with a similar heating uniformity index, although a single best result can be predicted. Considering real-life food product development, the integrated approach can identify ranges for food geometric parameters instead of only providing one single best result. Objective 2: Develop an integrated multiphysics modeling and machine learning-based Knowledge Generation Platform to identify general rules of interactions between solid-state microwaves and foods. Drs. Chen and Fathy worked on developing and validating multiphysics models for simulating the microwave heating of foods in the solid-state microwave system. To improve the modeling accuracy, a 3D scanning approach to better characterize the geometry of the microwave oven cavity was developed. The models using 3D scanned geometry showed more accurate prediction on microwave heating patterns than other models using simplified geometry and manually measured geometry. The models were validated at different microwave frequencies and different ports of the microwave system. The validated model will be used in the next step of integrating multiphysics modeling and machine learning studies. In addition to the modeling work, Dr. Chen further fabricated the four-port solid-state microwave system by including A radiometric-capable thermal camera module (Lepton 3.5, 160 ×120) at the top wall of the microwave oven cavity. Using the solid-state microwave system, a dynamic complementary frequency shifting strategy that continuously monitors the thermal profiles of the products in real-time and dynamically updates microwave frequency that has complementary heating patterns was developed. A dynamic microwave defrosting strategy with shifting frequency and adaptive power was developed for thawing frozen food products in the microwave system. Both strategies showed better heating/thawing results than current magnetron-based microwave systems.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2022
Citation:
Yang, R., Chen, J. (2022). Dynamic solid-state microwave defrosting strategy with shifting frequency and adaptive power improves thawing performance. Innovative Food Science & Emerging Technologies. https://doi.org/10.1016/j.ifset.2022.103157
- Type:
Journal Articles
Status:
Published
Year Published:
2022
Citation:
Yang, R., Fathy, A., Morgan, M., Chen, J. (2022). Development of Online Closed-Loop Frequency Shifting Strategies to Improve Heating Performance of Foods in a Solid-State Microwave System. Food Research International, https://doi.org/10.1016/j.foodres.2022.110985
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2022
Citation:
Yang, R., Chen, J. (2022). Reheating performance comparison between a solid-state microwave system using dynamic complementary-frequency shifting and a household microwave oven. presented at 15th CoFE (Conference of Food Engineering), September, Raleigh, NC, US.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2022
Citation:
Yang, R., Chen, J. (2022). Development of solid-state microwave defrosting strategies with adaptive power and shifting frequency. presented at 56th IMPI (International Microwave Power Institute) Symposium, June, Savannah, GA, US.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2022
Citation:
Verma, K., Yang, R., Gan, H., Chen, J. (2022). A multi-port complementary frequency shifting strategy for improving the heating uniformity in a solid-state microwave system, presented at CoFE-22 (Conference of Food Engineering), September, poster presentation.
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2022
Citation:
Verma, K., Yang R., Gan, H, Chen, J. (2022). Characterizing the Effect of Oven Geometry on the Modeling Accuracy of Microwave Heating, presented at IMPI-56 (International Microwave Power Institute) Symposium, June, poster presentation.
- Type:
Theses/Dissertations
Status:
Published
Year Published:
2022
Citation:
Ran Yang. Microwave Frequency Control Algorithms for Use in a Solid-State System to Achieve Improved Heating Performance. University of Tennessee. Doctoral Dissertation. 2022.
|
Progress 11/01/20 to 10/31/21
Outputs
Target Audience:During this period, we interacted with a microwaveable food company, a micrwoave oven manufacture, and other researchers in the field. Changes/Problems:
Nothing Reported
What opportunities for training and professional development has the project provided?One Ph.D. student was trained on the development of multiphysics models and machine learning models to simulate the microwave heating of food in domestic microwave ovens. The Ph.D. student was also trained on designing 3-D printed trays for validating the optimized results and on using thermal image cameras and fiber optic sensors to collect microwave heating performance data for model validation. One Master's student was trained on using COMSOL Multiphysics to develop multiphysics models for simulating the microwave heating of food using a solid-state microwave system. The Ph.D. also presented theresearch results at the IFT and IMPI conferences to develop communication skills. The student won first place for the Graduate Student Presentation Competitionat the IFT Product Development Session and IMPI Symposium. How have the results been disseminated to communities of interest?The results have been published in three peer-reviewed journal articles and presented in two conferences [2021 FIRST (Food Improved by Research, Science, & Technology) IFT annual meeting, and the 2021IMPI (International Microwave Power Institute) Symposium]. We also showedthe results to the industrial stakeholders (microwaveable food and oven companies) and got their inputs for the project improvement. What do you plan to do during the next reporting period to accomplish the goals?For the domestic oven-related work, we will continue to improve the machine learning algorithms for multiple parameters optimization. We anticipate completing the development of an integrated multiphysics modeling-machine learning model for 3D geometry optimization. For the solid-state oven-related work, we will continue the development of the multiphysics-based simulation model and validate the model with experiments.
Impacts
What was accomplished under these goals?
Non-uniform heating of microwaveable foods has caused severe food quality and safety issues. The microwaveable food industry has an immediate challenge to develop food products that can be cooked uniformly in the magnetron-based microwave ovens and an emerging challenge to prepare for the burgeoning solid-state microwave technology. A multi-disciplinary team with expertise in multiphysics modeling, microwave technology, machine learning for optimization, and machine learning for image processing collaborated on developing integrated multiphysics modeling and machine learning-based platforms to equip the food developers for high-quality microwavable foods development. During the first year, the team has developed an integrated multiphysics modeling and machine learning-based model for single parameter optimization and is working on expanding the capability of the integrated model for multi-parameter optimization. The team also has fabricated a four-port solid-state microwave system for developing heating strategies to improve microwave heating performance. Objective 1: Develop an integrated multiphysics modeling and machine learning-based Product Development Platform to assist current microwaveable food product development. Dr. Chen and Dr. Fathy developed finite-element-method-based computer simulation models that incorporate multiphysics of electromagnetics and heat transfer for simulating the microwave heating of foods in domestic ovens. The model was validated by heating a tray of mashed potato samples. Dr. Chen and Dr. Wang developed a Bayesian optimization machine-learning algorithm and integrated it with the multiphysics model. The integrated model was used for optimizing the thickness of microwaveable food products. The integrated approach was validated by comparing the optimization performance with a parametric sweep approach, which was solely based on multiphysics modeling. The results showed that the integrated approach had the capability and robustness to optimize the thickness of different-shape products using different initial training datasets with higher efficiency (45.9% to 62.1% improvement) than the parametric sweep approach. Three rectangular-shape trays with one optimized thickness (1.56 cm) and two non-optimized thicknesses (1.20 and 2.00 cm) were 3-D printed and used in microwave heating experiments, which confirmed the feasibility of the integrated approach in thickness optimization. Currently, this integrated approach is under development for multi-parameter (3D geometry) optimization. Objective 2: Develop an integrated multiphysics modeling and machine learning-based Knowledge Generation Platform to identify general rules of interactions between solid-state microwaves and foods. Dr. Chen and Dr. Fathy fabricated a four-port solid-state microwave system. The microwave system is comprised of an oven cavity (modified from a commercial oven, Panasonic Model NN-SN936W), a microwave generator (PA-2400-2500MHz-200W-4), and four waveguides (CWR340 Centric RF). A customized program was developed to control the operation of the solid-state microwave generator with adjustable power, frequency, and relative phases. This system was used in microwave heating to evaluate the effect of a complementary-frequency strategy on improving the microwave heating of a gellan gel sample. The average temperature and heating uniformity results of the complementary strategy were compared to conventional heating strategies like Fixed-Frequency and Sweeping-Frequency strategies. Typically, the Complementary-Frequency strategy could selectively use specific frequencies in the microwave heating process. These frequencies could be identified based on their associated complementary temperature profile of the top surface of the food product collected by a thermal imaging camera and sequenced in a specific order dictated by a predefined algorithm. The results indicated that both the Sweeping-Frequency and the Complementary-Frequency strategies could uniformly heat food products more than the Fixed-Frequency strategy. Meanwhile, the Complementary-Frequency strategy can trade-off microwave power absorption efficiency and heating uniformity to deliver a higher heating rate than the Sweeping- Frequency strategy does. Dr. Chen and Fathy also worked on developing multiphysics models for simulating the microwave heating of foods in the solid-state microwave system. The model development was completed and the validation work is undergoing.
Publications
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Yang, R., Chen, J.* 2021. Mechanistic and Machine Learning Modeling of Microwave Heating Process in Domestic Ovens: A Review. Foods. https://doi.org/10.3390/foods10092029
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Yang, R., Fathy, A., Morgan, M., Chen, J.* 2021. Development of a Complementary-Frequency Strategy to Improve Microwave Heating of Gellan Gel in a Solid-state System. Journal of Food Engineering. https://doi.org/10.1016/j.jfoodeng.2021.110763
- Type:
Journal Articles
Status:
Published
Year Published:
2021
Citation:
Yang, R., Wang, Z., Chen, J.* 2021. An Integrated Approach of Mechanistic-modeling and Machine-learning for Thickness Optimization of Frozen Microwaveable Foods. Foods. 10(4), 763. https://doi.org/10.3390/foods10040763
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2021
Citation:
Yang, R., Chen, J.* 2021. Development of a Multiphysics Modeling Supported Online Machine-Learning Approach to Optimizing Microwaveable Food Geometry for Better Heating Uniformity. presented at the 2021 FIRST (Food Improved by Research, Science, & Technology) IFT annual meeting, July, online. (Won the Graduate Student Competition in the Product Development session)
- Type:
Conference Papers and Presentations
Status:
Published
Year Published:
2021
Citation:
Yang, R., Chen, J.* 2021. Effect of Dynamic Changing Frequency on the Microwave Heating Uniformity of Food in a Solid-State System, presented at the IMPI (International Microwave Power Institute) Symposium, July, online. (Won the Graduate Student Competition)
|
|