Operation and maintenance (O&M) events resulting from environmental factors (e.g., precipitation, temperature, seasonality, and unexpected weather conditions) are among the primary sources of operating costs and downtime in run-of-river small hydropower plants (SHPs). This paper presents a data-driven methodology for predicting such long events using machine learning models trained on historical power production, weather radar, and forecast data. Case studies on two Slovenian SHPs with different structural designs and levels of automation demonstrate how environmental features—such as day of year, rain duration, cumulative amount of rain, and rolling precipitation sums—can be used to forecast long events or shutdowns. The proposed approach integrates probabilistic classification outputs with threshold-consistency smoothing to reduce noise and stabilize predictions. Several algorithms were tested—including Logistic Regression, Support Vector Machine (SVM), Random Forest, Gradient Boosting, and k-Nearest Neighbors (k-NN)—across varying feature combinations for O&M model development, with cross-validation ensuring robust evaluation. The models achieved an F1-score of up to 0.58 in SHP1 (k-NN), showing strong seasonality dependence, and up to 0.68 in SHP2 (Gradient Boosting). For SHP1, the best model (k-NN) correctly detected 36 long events, while 15 were misclassified as no events and 38 false alarms were produced. For SHP2, the best model (Gradient Boosting) correctly detected 69 long events, misclassified 23 as no events, and produced 42 false alarms. The findings highlight that probabilistic machine learning-based forecasting can effectively support predictive O&M planning, particularly for manually operated or service-operated SHPs.

Driven by technological advancements and the increasing need for efficiency and customization, the manufacturing industry is shifting towards Shared Manufacturing. This strategy enhances global production flexibility and resource utilization by enabling diverse entities to collaboratively engage in distributed manufacturing activities. Expanded resource sharing across industries and society, along with the redefinition of manufacturing resources as marketable services, creates a complex, interconnected production network that demands autonomous production control for effective and flexible management. This study proposes an architectural design for Autonomous Production Units within a blockchain-based Shared Manufacturing system. The design enhances autonomy, allowing independent management and optimization of manufacturing processes while integrating with the global market. The architecture includes two key decision-making submodules: the Business Decisions Submodule, which handles operational activities and interactions with the blockchain network, and the Manufacturing Decisions Submodule, which oversees physical manufacturing processes. The concept is implemented and tested on a case study, demonstrating the Autonomous Production Unit’s capability to autonomously execute the entire manufacturing process, from negotiation to manufacturing service execution, while also handling malfunctions.

Part-scale modeling of the temperature field in metal powder bed additive manufacturing (AM) is critical for predicting mechanical properties of the AM-ed parts. Track-by-track heat transfer analysis is impractical due to the extensive number of layers and the intricate design of scan strategies for the heat source, particularly in the fabrication of specimen clusters or parts with complex geometry, where multiple regions in the powder bed are manufactured simultaneously. Many part-scale modeling approaches only focus on the thermal behavior of a single part without considering the thermal interaction from the surrounding parts to reduce computational cost. However, experimental observations have revealed that the temperature distribution along the building direction can vary among samples with identical local geometries. This discrepancy can be attributed to the heating effects from neighboring samples. In this study, we propose an integrated part-scale modeling framework that combines layer-wise equivalent heat flux attribution with layer-wise element activation. Before the layer-wise attribution, we justify the equivalent heat flux of individual layers through high-fidelity track-scale simulations. Unlike traditional heat transfer analysis for single parts, our analysis incorporates heat conduction effects through the powder bed between different fusion zones. The temperature data obtained from each equivalent layer using our approach shows consistency when compared to the experimental observations. This research presents an efficient, physically grounded method for modeling the thermal behavior of large AM specimen clusters, enhancing our understanding of temperature field evolution in AM and supporting the design of optimized scanning path strategies for large samples.

Femtosecond laser micromachining (FL mu M) has emerged as a transformative technology for precision microfabrication, offering minimal thermal damage and exceptional resolution across diverse materials, including metals, semiconductors, polymers, and ceramics. Despite its advantages, FL mu M faces challenges in maintaining consistent quality and efficiency, especially in industrial applications. This study investigates the integration of acoustic emission (AE) monitoring with FL mu M for in-situ quality assessment and process optimization. AE signals, captured at high frequencies (up to 1.5 MHz), are analysed using physics-based methods such as RMS, MARSE, and STFT, as well as machine learning (ML)-based approaches. The results reveal strong correlations between AE signal characteristics and laser parameters such as pulse energy, scanning speed and focal position. They also show the possibility to detect critical material removal regimes, including ablation and melting. Feature importance analysis using ML techniques can identify process-specific frequency bands-350, 469, 547, and 664 kHz-which are particularly relevant for detecting process instabilities and ensuring quality control. By combining AE monitoring with advanced signal analysis, this approach demonstrates a scalable, non-invasive solution for improving the precision and reliability of FL mu M, with potential applicability across a wide range of materials and microfabrication processes.

This paper presents a novel approach to thermal process control in the food industry, specifically targeting the pasteurization and cooking of soft-boiled eggs. The unique challenge of this process lies in the precise temperature control required, as pasteurization and cooking must occur within a narrow temperature range. Traditional control methods, such as fuzzy logic controllers, have proven insufficient due to their limitations in handling varying loads and environmental conditions. To address these challenges, we propose the integration of robust reinforcement learning (RL) techniques, particularly the utilization of the Deep Q-Network (DQN) algorithm. Our approach involves training an RL agent in a simulated environment to manage the thermal process with high accuracy. The RL-based system adapts to different heat capacities, initial conditions, and environmental variations, demonstrating superior performance over traditional methods. Experimental results indicate that the RL-based controller significantly improves temperature regulation accuracy, ensuring consistent pasteurization and cooking quality. This study opens new avenues for the application of artificial intelligence in industrial food processing, highlighting the potential for RL algorithms to enhance process control and efficiency.

Download This Paper Open PDF in Browser Add Paper to My Library Share: Permalink Using these links will ensure access to this page indefinitely Copy URL Enhanced Large-Scale Modeling of Additive Manufacturing: Layer-Wise Equivalent Heat Flux Attribution for Thermal Interaction Analysis Across Multiple Fabrications 22 Pages Posted: 27 Feb 2024 See all articles by Fan ChenFan ChenNorthwestern UniversityJian CaoNorthwestern UniversityKozjek DominikNorthwestern UniversityConor PorterNorthwestern University Abstract Part-scale modeling of the temperature field in metal powder bed additive manufacturing (AM) is critical for predicting mechanical properties of the AM-ed parts. Track-by-track heat transfer analysis is impractical due to the extensive number of layers and the intricate design of scan strategies for the heat source, particularly in the fabrication of specimen clusters or parts with complex geometry, where multiple regions in the powder bed are manufactured simultaneously. Many part-scale modeling approaches only focus on the thermal behavior of a single part without considering the thermal interaction from the surrounding parts to reduce computational cost. However, experimental observations have revealed that the temperature distribution along the building direction can vary among samples with identical local geometries. This discrepancy can be attributed to the heating effects from neighboring samples. In this study, we propose an integrated part-scale modeling framework that combines layer-wise equivalent heat flux attribution with layer-wise element activation. Before the layer-wise attribution, we justify the equivalent heat flux of individual layers through high-fidelity track-scale simulations. Unlike traditional heat transfer analysis for single parts, our analysis incorporates heat conduction effects through the powder bed between different fusion zones. The temperature data obtained from each equivalent layer using our approach shows consistency when compared to the experimental observations. Overall, the layer-wise equivalent heat flux attribution method offers a sound physical basis and excellent computational efficiency. This research offers a more precise and streamlined approach for modeling large AM specimen clusters and contributes to the advancement of our understanding of the temperature field evolution in AM, and the work enables intelligent scanning path strategy design for scalable additive manufacturing. Keywords: Additive manufacturing, Laser powder bed fusion, part-scale thermal simulation, temperature history prediction, Melt pool temperature Suggested Citation: Suggested Citation Chen, Fan and Cao, Jian and Dominik, Kozjek and Porter, Conor, Enhanced Large-Scale Modeling of Additive Manufacturing: Layer-Wise Equivalent Heat Flux Attribution for Thermal Interaction Analysis Across Multiple Fabrications. Available at SSRN: https://ssrn.com/abstract=4741149 Fan Chen Northwestern University ( email ) 2001 Sheridan RoadEvanston, IL 60208United States Jian Cao (Contact Author) Northwestern University ( email ) 2001 Sheridan RoadEvanston, IL 60208United States Kozjek Dominik Northwestern University ( email ) 2001 Sheridan RoadEvanston, IL 60208United States Conor Porter Northwestern University ( email ) 2001 Sheridan RoadEvanston, IL 60208United States Download This Paper Open PDF in Browser Do you have negative results from your research you’d like to share? Submit Negative Results Paper statistics Downloads 0 Abstract Views 8 32 References PlumX Metrics Feedback Feedback to SSRN Feedback (required) Email (required) Submit If you need immediate assistance, call 877-SSRNHelp (877 777 6435) in the United States, or +1 212 448 2500 outside of the United States, 8:30AM to 6:00PM U.S. Eastern, Monday - Friday.

The pressing problem of laser powder bed fusion of metals (PBF-LB/M) is the inability to predict and mitigate undesired process conditions in a build. This study aims to develop an efficient and geometric-agnostic prediction tool for the variations of the average meltpool temperature of each layer given a laser scanning path. The approach is a data-driven model based on defined physics-based features and measurement data from a two-color coaxial photodiode system. Although the coaxial photodiode system cannot measure absolute meltpool temperature, it can measure relative changes in the meltpool temperature if the process is performed in the shallow or no-keyhole regimes, the focus of this study. Once trained, the model can predict meltpool temperature variations before the start of the build process for a part geometry which had not yet been printed. The results show that variations in the layer average meltpool temperature can be predicted with the average coefficient of determination (r2) score of 0.65. Furthermore, the method shows the significance of considering the effect of surrounding parts on the average meltpool temperature profile of an observed part. For part geometries investigated in this study, the model shows the information from at least 40 previous layers is needed to enable sufficient and stabilized prediction performance.

An adaptive laser power control strategy for Selective Laser Melting (SLM) has been developed using data from a co-axial photodiode monitoring system with 200 KHz temporal resolution. A supervised machine learning based algorithm outputs variable laser power along the scanning path based on mechanistic features. The approach was implemented on a commercial machine and demonstrated an average 12 % reduction in porosity size and 65 % reduction in the standard deviation of porosity size measured by X-Ray Computed Tomography (CT) compared to parts built with constant laser power. This approach is scalable and its precalculated nature is compatible with regulatory concerns. (c) 2024 CIRP. Published by Elsevier Ltd. All rights reserved.

This paper proposes a transfer learning approach using graph neural networks (GNN) for predicting the forming force during double-sided incremental forming (DSIF) processes. In order to address the geometry complexity of DSIF parts, a GNN-based model was proposed to aggregate surface geometric information of DSIF parts and toolpaths. Furthermore, a transfer learning method was adopted to improve the prediction. The model was pre-trained on a dataset of previously formed DSIF parts with varying geometries. To address material and machine variations, the model was further trained on the initial few layers of the observed part for calibration and subsequently predicted the forming force in the vertical direction relative to the part's coordinate system for the rest of the layers of the observed part. The performance of our proposed approach was evaluated using experimental datasets from two different machines and different input materials, demonstrating the generality and effectiveness of the approach in forming force prediction.

Coaxial photodiode monitoring sensors provide a digital signature at the melt pool level in laser powder bed fusion (L-PBF), facilitating faster part qualification. However, current signatures are plagued by significant noise and signal variation and show counterintuitive trends such as a decrease in measured temperature near overhang surfaces. This paper investigates the behavior of coaxial photodiode-based melt pool monitoring (PD-MPM) thermal measurements at overhang regions by conducting a combination of experiments and multiphysics simulations. High-speed in situ synchrotron X-ray imaging is coupled with a coaxial photodiode system to enable comparison of the observed melt pool phenomena and monitoring signals during double-track AlSi10Mg experiments. A multiphysics model is developed to simulate the melt pool dynamics and concurrent sensor signals throughout the process. We propose a surrogate model that clarifies the correlation between sensor signals and melt pool temperature. Both experimental and simulation results emphasize the significant impacts of laser energy and keyhole formation on solid-liquid interface discontinuities and PD-MPM signals. Results reveal that a boiling region with a relatively smaller projected area, as near an overhang, can lead to a decrease in the measured melt pool temperature, even when the true peak temperature remains constant, meaning that PD-MPM temperature measurements cannot be used to estimate absolute melt pool temperature. Consequently, in overhang regions, interaction between the melt pool and underlying powder results in an abruptly deforming melt pool and a pronounced decrease in both individual photodiode intensities and overall measured melt pool temperature. This work illustrates how to correctly interpret the coaxial photodiode monitoring signals in L-PBF by uncovering the intricate dynamics within the melt pool, particularly in challenging overhang regions, and pave the foundation for data-driven part qualification.

With the development of additive manufacturing (AM) processes, a variety of methods and technologies for process monitoring and post-process inspection have become available. These, together with modern computerized machines, generate large amounts of data about the process. Hence, efficient analysis methods for extracting useful information and knowledge from these datasets are needed. This paper presents a method for automated data fusion of machine input toolpath commands, on-axis intensity monitoring, and Computed Tomography (CT) scan measurement data in laser metal powder-bed fusion AM. The iterative closest point (ICP) point-clouds registration method is used for fine spatial alignments of (1) on-axis intensity measurements with machine input toolpath commands and (2) machine input toolpath and CT scan. The input-toolpath point cloud is selected as the reference point cloud. We further propose a post-processing method that can correct part of the wrong connections between measurements and toolpath points as results of ICP registration. The proposed method is suitable for merging data where there is no temporal synchronization between the machine commands and the monitoring system, and where the actual laser scan path may deviate from the machine input toolpath commands, e.g., due to dynamics constraints of scanning galvo mirror system. The method is designed to handle dynamic changes in input process parameters such as dynamic changes in laser power and scan speed, which lays a solid foundation for advanced AM process control, qualification and certification. A practical solution for organizing and storing the fused data, which allows for efficient analysis, is also presented. The results show that the method can spatially align all three different types of data and find the corresponding input toolpath command for each measurement point. The alignment of on-axis measurements and machine input toolpath commands has been improved by using a clustering algorithm. The main limitation is that, for estimating the improvement and error of spatial alignment, the average difference between the closest points of two point clouds is used, but this metric does not give an exact value of an actual registration error. The estimation of connectivity error is based on a visual comparison of registration results.

A major source of flaws and defects in Laser Powder Bed Fusion stems from the creation of pores during the build process. In-process mechanisms that produce pores, such as keyholing and lack of fusion, and secondary processes that can reduce porosity, such as hot isostatic pressing, have been extensively studied. Many of these works also suggest that if pores are detectable then it could be possible to develop a schema to "heal" the defects in-situ. However, the effects of a subsequent laser pass on pores generated in prior layers have not been fully studied due to the experimental difficulties. It is therefore important to understand if these interaction mechanisms have the potential to "heal" and enable control methods that select laser parameters to target areas of suspected porosity. This paper identifies and qualitatively analyzes pore healing mechanisms in L-PBF using high speed X-ray imaging. Several key phenomena affecting keyhole and melt pool interactions with existing pores are identified in the X-ray images and the effects of these phenomena are discussed. Finally, several areas for future work are presented, including the need for robust quantitative analysis of these phenomena.

Geometrical features and toolpath sequence are two important factors that cause process condition variations, such as variations in the meltpool temperature or meltpool size, that might lead to undesired material properties in the laser powder bed fusion (LPBF) process. Due to the high dynamics and complex physics of the LPBF process, it is difficult to predict variations in process conditions with simulations alone. Advances in measurement technology and computational technologies open up new possibilities for smart manufacturing. In this paper, a data-driven method to predict intra-layer variations in the processing conditions that source from the toolpath sequence and part geometry is presented. The approach is demonstrated using two-color on-axis pyrometer measurements. Three demonstration cases are presented in which it is demonstrated (1) how the trained predictive model can be used as a filter to ease the interpretation of process variations and discover patterns related to toolpath and part geometry, and (2) how to generate predictions that can be used for feedforward control, i.e., for adjusting laser power or scanning speed along the toolpath using a meltpool temperature prediction model generated based on on-axis measurements. Results show that the developed prediction model is able to meaningfully predict process variations resulted from toolpath sequence and geometry. Predictions are aligned with the results from the related work of others and for the case of 180° laser path turnarounds in our high-speed X-ray imaging experiments. The potential issues related to the current maturity status of the process and measuring equipment that could in practice affect the performance of the proposed solutions are also discussed.

Cost estimation is critical for effective decision making in engineering projects. However, it is often hampered by a lack of sufficient data. For this, data imputation techniques can be used to estimate missing costs based on statistical estimates or analogies with historical data. However, these techniques are often limited because they do not consider the existing knowledge of experts. In this paper, a novel cognitive data imputation technique is proposed for cost estimation that uses explanatory interactive machine learning to integrate and improve human knowledge. Through a case study in maintenance cost estimation the effectiveness of the approach is demonstrated.

A co-axial photodiode monitoring system with high temporal resolution has been integrated into a proven test bench enabling synchronized side-view high speed X-ray imaging of melt pool dynamics and top-view spectral emission characterization of the melt pool. This setup enables direct observation of melt pool phenomena and correlation between the two monitoring systems which can be directly scaled to commercial systems. The work demonstrates a 92% detection rate in keyhole collapse phenomena related to defect generation in SLM. Furthermore, the impact of gas flow on monitoring signals is studied to understand the fundamental importance of gas flow in commercial systems.

In this paper an IoT application of LoRa is presented. The application is related to the so-called precision beekeeping. This term is associated with the monitoring of many variables within a hive and in its vicinity, which can assist the beekeeper at all the activities that have to be done in beekeeping practice. In order to achieve that kind of functionality the so called wireless sensor network has to be realized. First an overview of such technologies is given. Our decision was to take LoRa. The core of the system is an Arduino microcontroller with the addition of the LoRa shield. In the paper a process of communication between master and slave unit is described. In order to demonstrate the applicability, the slave unit has an additional temperature sensor attached. The system was verified by a successful temperature measurement lasting several days

Uncontrolled process variability, stemming from geometry, machine, or parameter variation, can lead to metallurgical defects such as keyhole porosity and lack of fusion as well as geometrical defects such as increased surface roughness or increased deformation in the produced parts. This lack of control is a pressing problem in laser powder bed fusion (L-PBF) processes. One way to reduce this variability is to use model-based predictive control. Process parameters such as laser power and scan speed can be adjusted during the process based on in situ measurements of process conditions such as melt pool size or temperature. In this paper, a predictive model that is an essential element in a larger predictive control ecosystem for L-PBF is developed and tested. The proposed machine learning-based regression model is trained using high-resolution co-axial melt pool temperature measurements from the previous layers. The machine learning model can predict the melt pool temperatures along the toolpath for the next layer assuming processing parameters remain the same. The paper describes the development of the machine learning-based prediction model and presents the guidelines for the design and selection of the features in feature vector. The estimation of the prediction performance based on real physical data is presented followed by suggestions of future work. The main limitation of the current approach is the relatively high computational cost. Some guidelines for implementation and possible improvements are given in the discussion of results.

Autonomous mobile robots (AMRs) are increasingly used in modern intralogistics systems as complexity and performance requirements become more stringent. One way to increase performance is to improve the operation and cooperation of multiple robots in their shared environment. The paper addresses these problems with a method for off-line route planning and on-line route execution. In the proposed approach, pre-computation of routes for frequent pick-up and drop-off locations limits the movements of AMRs to avoid conflict situations between them. The paper proposes a reinforcement learning approach where an agent builds the routes on a given layout while being rewarded according to different criteria based on the desired characteristics of the system. The results show that the proposed approach performs better in terms of throughput and reliability than the commonly used shortest-path-based approach for a large number of AMRs operating in the system. The use of the proposed approach is recommended when the need for high throughput requires the operation of a relatively large number of AMRs in relation to the size of the space in which the robots operate.

Research in the area of robotic systems has greatly benefited from the use of simulation models. Recent approaches allow the transfer of developed algorithms from simulation to reality (sim-to-real) and increasingly accurate representations of real systems as simulation models (real-to-sim). The paper presents an architecture based on open software that supports simultaneous experiments on real robots and their simulation models. Two illustrative examples are shown: a digital twin of an industrial robot and a sim-to-real transfer in an autonomous mobile robot system. The possibilities of future research on the interaction between robotic systems and their simulation models are discussed.