Energy Flow Network Research Articles

Streamlining latent spaces in machine learning using moment pooling

Many machine learning applications involve learning a latent representation of data, which is often high-dimensional and difficult to directly interpret. In this work, we propose “moment pooling,” a natural extension of deep sets networks which drastically decreases the latent space dimensionality of these networks while maintaining or even improving performance. Moment pooling generalizes the summation in deep sets to arbitrary multivariate moments, which enables the model to achieve a much higher effective latent dimensionality for a fixed learned latent space dimension. We demonstrate moment pooling on the collider physics task of quark/gluon jet classification by extending energy flow networks (EFNs) to moment EFNs. We find that moment EFNs with latent dimensions as small as 1 perform similarly to ordinary EFNs with higher latent dimension. This small latent dimension allows for the internal representation to be directly visualized and interpreted, which in turn enables the learned internal jet representation to be extracted in closed form. Published by the American Physical Society 2024

Infrared-safe energy weighting does not guarantee small nonperturbative effects

Infrared and collinear (IRC) safety has long been used a proxy for robustness when developing new jet substructure observables. This guiding philosophy has been carried into the deep learning era, where IRC-safe neural networks have been used for many jet studies. For graph-based neural networks, the most straightforward way to achieve IRC safety is to weight particle inputs by their energies. However, energy-weighting by itself does not guarantee that perturbative calculations of machine-learned observables will enjoy small nonperturbative corrections. In this paper, we demonstrate the sensitivity of IRC-safe networks to nonperturbative effects, by training an energy flow network (EFN) to maximize its sensitivity to hadronization. We then show how to construct Lipschitz energy flow networks (L-EFNs), which are both IRC safe and relatively insensitive to nonperturbative corrections. We demonstrate the performance of L-EFNs on generated samples of quark and gluon jets, and showcase fascinating differences between the learned latent representations of EFNs and L-EFNs. Published by the American Physical Society 2024

Strategy-based optimization and life cycle environmental impact assessment of a stable photovoltaic power-to-hydrogen system

With the growing global emphasis on reducing greenhouse gas emissions and transitioning to a low-carbon economy, renewable hydrogen is expected to play a vital role in the chemical industry. However, there is a contradiction between the continuous and stable hydrogen demand of chemical process systems and the intermittent and fluctuating nature of renewable energy systems when coupling them together in the industrial chemical production. There are currently few studies addressing year-round stable hydrogen load systems while considering renewable energy intermittency and capacity optimization. Moreover, the comprehensive life cycle environmental impact study of such stable renewable-hydrogen production system is extremely limited. Therefore, this study focuses on a photovoltaic power-to-hydrogen system that can operate stably for 8,760 h per year, specifically targeting its application in the chemical industry. Two operating strategies of the hydrogen system were considered: fully off-grid and ultra-light grid-connected scenarios. The capacity configuration under each scenario was optimized using gird search algorithm, with the objective of minimizing the levelized cost of hydrogen. Based on the optimization results, the energy flow network, cost-effectiveness, and comprehensive life cycle environmental impacts of such system were analyzed. The results show that the ultra-light grid-connected hydrogen production system, due to the support of the power grid, has improved the reliability of the hydrogen production system and significantly reduced the capacity configuration of system components, thereby demonstrating better performance in terms of economics and overall lifecycle environmental impact compared to a completely off-grid hydrogen production system. Its global warming potential (GWP) and levelized cost of hydrogen under electricity sales scenario are 12% and 31% lower, respectively, compared to fully off-grid hydrogen production. Considering China’s future demand for large-scale renewable hydrogen, ultra-light grid-connected hydrogen production may become the primary transitional pathway in the near to medium term.

Evaluating the Effects and Mechanisms of the Eco–Substrate in Aquaculture Environment Restoration from an Ecosystem Perspective via the Ecopath Model

Aquaculture supplies high-quality and healthy proteins. With the increasing human demand for aquaculture production, intensive pond aquaculture developed rapidly and results in environmental deterioration. To solve this problem, the eco-substrate (ES), which is the biofilm carrier, has been utilized in aquaculture ponds. Studying the ecological mechanisms of ES from the perspective of the ecosystem may be conducive to the sustainable development of aquaculture. In this study, it was evaluated how ES makes a difference to the trophic structure, energy flow, and system characteristics of two different aquaculture pond ecosystems via the ecopath model. Three aquaculture ponds with ES were designed as the treatment ecosystem and three aquaculture ponds without ES were designed as the control ecosystem. There were 13 and 14 functional groups in the control and treatment ecosystems, respectively. The results showed that (1) the macrozooplankton and microzooplankton showed strong effects on the ecosystem in the keystoneness index; (2) energy transfer pathways in the treatment system with ES increased by 26.23% compared to the control system; (3) the ES improved the utilization rate of detritus, which was 14.91% higher than that of the control ecosystem; (4) the material and energy flow index and network information characteristics demonstrated the ES enhanced the complexity and stability of the treatment system. To improve the energy utilization efficiency, filter feeders can be introduced to ES ponds. Overall, the ES can alter the trophic structure, improve the energy utilization efficiency, and enhance the stability and maturity of aquaculture ecosystems, representing a sustainable practice. Considering the total area of aquaculture ponds on the earth reaching more than 5 million hectares, the application prospect of ES is broad.

On the design of 5GDHC substation control systems

ABSTRACT This paper presents challenges in the control of 5GDHC networks and proposes an approach for the robust control design of these networks. Temperatures are low and temperature differences are small in 5GDHC networks, in order to minimise heat losses in the distribution network and directly utilise low-temperature heat sources. Therefore, the robust control design of substations is crucial for stable and efficient operation of 5GDHC networks. We proposed the key steps to obtain a consistent control design that can be verified and tested prior to commissioning of the building substations and network. The design approach was elaborated in a case study in an existing neighbourhood in Haarlem, the Netherlands, constructed in the 1930s to 1970s. This neighbourhood will become independent of natural gas using PVT-rooftop panels, individual heat pumps (HP), a 2-pipe, bi-directional energy flow network operating at ultra-low temperatures and ATES system as a seasonal and daily storage facility.

Hypergraphs in LHC phenomenology — the next frontier of IRC-safe feature extraction

In this study, we critically evaluate the approximation capabilities of existing infra-red and collinear (IRC) safe feature extraction algorithms, namely Energy Flow Networks (EFNs) and Energy-weighted Message Passing Networks (EMPNs). Our analysis reveals that these algorithms fall short in extracting features from any N-point correlation that isn’t a power of two, based on the complete basis of IRC safe observables, specifically C-correlators. To address this limitation, we introduce the Hypergraph Energy-weighted Message Passing Networks (H-EMPNs), designed to capture any N-point correlation among particles efficiently. Using the case study of top vs. QCD jets, which holds significant information in its 3-point correlations, we demonstrate that H-EMPNs targeting up to N=3 correlations exhibit superior performance compared to EMPNs focusing on up to N=4 correlations within jet constituents.

The characteristics and influencing factors of the embodied energy flow network in fishery trade among countries along the 21st Century Maritime Silk Road

The characteristics and influencing factors of the embodied energy flow network in fishery trade among countries along the 21st Century Maritime Silk Road

Analysis of the characteristics and influencing factors of China's embodied energy flow network

Understanding the flow of embodied energy between provinces in China and the factors affecting the network has an important impact on reducing energy consumption in each province and promoting balanced regional development. This paper uses the multi-regional input–output model to construct the interprovincial embodied energy flow network in China and defines the energy flow between provinces under the trade of products and services. Then, it uses the ecological network model to carry out the ascendency and network environ analysis of the interprovincial embodied energy flow network in China and uses the Quadratic Assignment Procedure regression model to calculate the impact of various factors on the flow network. The main research conclusions are as follows: (1) The embodied energy flow in the middle reaches of the Yangtze River and the Yellow River is relatively high. The embodied energy flow value in Ningxia and Qinghai is low. Zhejiang, Guangdong, and Jiangsu often have the highest embodied energy outflow value, while Inner Mongolia, Shanxi, and other provinces have higher inflows. (2) The center of gravity of through flow in the flow system is biased to Jiangsu, Guangdong, and other places; Gansu, Ningxia, Qinghai, and Hainan are at the edge of the system. (3) The differences in the economic development level, population size, and energy structure have a positive impact on the embodied energy flow network between provinces in China; the geographic distance will have a negative impact on the flow network.

Hierarchical high-point Energy Flow Network for jet tagging

Jet substructure observable basis is a systematic and powerful tool for analyzing the internal energy distribution of constituent particles within a jet. In this work, we propose a novel method to insert neural networks into jet substructure basis as a simple yet efficient interpretable IRC-safe deep learning framework to discover discriminative jet observables. The Energy Flow Polynomial (EFP) could be computed with a certain summation order, resulting in a reorganized form which exhibits hierarchical IRC-safety. Thus inserting non-linear functions after the separate summation could significantly extend the scope of IRC-safe jet substructure observables, where neural networks can come into play as an important role. Based on the structure of the simplest class of EFPs which corresponds to path graphs, we propose the Hierarchical Energy Flow Networks and the Local Hierarchical Energy Flow Networks. These two architectures exhibit remarkable discrimination performance on the top tagging dataset and quark-gluon dataset compared to other benchmark algorithms even only utilizing the kinematic information of constituent particles.

Probing dark QCD sector through the Higgs portal with machine learning at the LHC

The QCD-like dark sector with GeV-scale dark hadrons has the potential to generate new signatures at the Large Hadron Collider (LHC). In this paper, we consider a singlet scalar mediator in the tens of GeV-scale that connects the dark sector and the Standard Model (SM) sector via the Higgs portal. We focus on the Higgs-strahlung process, qoverline{q} ′ → W* → WH, to produce a highly boosted Higgs boson. Our scenario predicts two different processes that can generate dark mesons: (1) the cascade decay from the Higgs boson to two light scalar mediators and then to four dark mesons; (2) the Higgs boson decaying to two dark quarks, which then undergo a QCD-like shower and hadronization to produce dark mesons. We apply machine learning techniques, such as Convolutional Neural Network (CNN) and Energy Flow Network (EFN), to the fat-jet structure to distinguish these signal processes from large SM backgrounds. We find that the branching ratio of the Higgs boson to two light scalar mediators can be constrained to be less than about 10% at 14 TeV LHC with mathcal{L} = 3000 fb−1.

Plant-wide byproduct gas distribution under uncertainty in iron and steel industry via quantile forecasting and robust optimization

In the modern iron and steel industry, the efficient distribution of byproduct gases faces significant challenges due to quantity- and quality-related uncertainties of gases. This study presents an optimal approach to gas distribution that addresses this issue by incorporating the energy flow network and the uncertain surplus gases from the manufacturing system. The uncertain optimization problem is formulated as a two-stage robust optimization (TSRO) model, including “here-and-now” decisions aimed at minimizing the start-stop cost of energy conversion units, as well as “wait-and-see” decisions aimed at minimizing the operating cost of gasholders and the penalties resulting from energy excess or shortage. To facilitate practical implementation, we propose a “first quantify, then optimize” approach: (1) quantifying the uncertainty of surplus gases via a conditional quantile regression (CDQ)-based T-step time series model, and (2) finding the optimal solution through a column-and-constraint generation algorithm. Furthermore, a case study is conducted on an industrial energy system to validate the proposed methodology. Computational results, using evaluation indicators, such as MAPE, RMSE, PICP, and PINAW, confirm the effectiveness of the data-driven time series model in accurately quantifying uncertainties in each period. Sensitivity analysis demonstrates that the proposed TSRO model achieves a favorable balance between robustness and flexibility by selecting the combination of “budget and quantile” and the parameters of storage and conversion units. Comparative results reveal: (1) the optimal objective of the TSRO closely aligns with that of stochastic programming (SP) and is 2.717 times higher than that of deterministic optimization (DO); and (2) the computation time of TSRO is 2.388 times longer than that of DO, yet significantly smaller than that of SP, being only 0.07 times longer. Consequently, TSRO can efficiently find a robust gas distribution solution with the desired level of conservativeness for integrated iron and steel plants.

Hierarchical Causal Graph-Based Fault Root Cause Diagnosis and Propagation Path Identification for Complex Industrial Process Monitoring

Modern complex industrial processes are increasingly integrated and automated. There are numerous control loops in the process system and the closed-loop feedback makes units strongly coupled. This makes the faults propagate and evolve with the material and energy flow networks. Thus, it poses a great challenge to the fault root cause diagnosis and propagation path identification for industrial processes. This paper proposes a fault root cause diagnosis and propagation path identification strategy for complex industrial process based on hierarchical causal graph (HCG). Firstly, the process is divided into several sub-blocks according to process knowledge. The monitoring model is built for fault detection. Secondly, the fault-related sub-blocks and variables are selected based on contribution plot, and the causal analysis based on transfer entropy is performed on them to construct HCG. In this way, the causal connections can be significantly reduced. Subsequently, a propagation path identification method based on causal feature long short-term memory (CF-LSTM) is designed. Finally, real data from hot strip mill process (HSMP) and their comparative experiments are adopted to validate the fault diagnosis performance of the proposed framework.

Hierarchical model to find the path reducing CO2 emissions of integrated iron and steel production

Improving CO2 emission status of the iron and steel industry is an important problem for China to be able to reach low-carbon development goals. A hierarchical model of CO2 emission was established to find the way to reduce emissions from the iron containing material flow network and the energy flow network in this paper. This model was developed not only based on the interconnection and interrelationship among the iron containing material flow, energy flow and carbon emission, but also added the impact of energies buffer and adjustment. Case study shown that, CO2 emission intensity was 2.23 t/t_s. Significant reasons affecting the CO2 emission intensity, which were the release of gases, the mismatch between coking and iron making capacity, and the less utilization of scrap in convertor, were found. Three optimization scenarios were proposed. Calculation results shown that, CO2 emissions intensity was reduced by 0.01 t/t_s, 0.02 t/t_s and 0.02 t/t_s, respectively. Comparing each scenario with the previous, blast furnace gas buffer capacities were 47.6 × 103tce, 8.4 × 103tce, 25.1 × 103tce, and coke oven gas buffer capacities were 56.3 × 103tce, 20.7 × 103tce, −38.5 × 103tce. At last, some strategy suggestions were put forward.

Energy-weighted message passing: an infra-red and collinear safe graph neural network algorithm

Hadronic signals of new-physics origin at the Large Hadron Collider can remain hidden within the copiously produced hadronic jets. Unveiling such signatures require highly performant deep-learning algorithms. We construct a class of Graph Neural Networks (GNN) in the message-passing formalism that makes the network output infra-red and collinear (IRC) safe, an important criterion satisfied within perturbative QCD calculations. Including IRC safety of the network output as a requirement in the construction of the GNN improves its explainability and robustness against theoretical uncertainties in the data. We generalise Energy Flow Networks (EFN), an IRC safe deep-learning algorithm on a point cloud, defining energy weighted local and global readouts on GNNs. Applying the simplest of such networks to identify top quarks, W bosons and quark/gluon jets, we find that it outperforms state-of-the-art EFNs. Additionally, we obtain a general class of graph construction algorithms that give structurally invariant graphs in the IRC limit, a necessary criterion for the IRC safety of the GNN output.

The dynamic evaluation and optimization model for steel enterprise’s energy flow network operations

In line with the global green steel development, a steel enterprise seeks to reduce its energy consumption by building a scientific, effective evaluation and optimization model for steel manufacturing’s energy utilizations. It highlights the enterprise and the steel sector’s green mainstream. This paper proposes a dynamic evaluation model for energy flow network operations and brings forward evaluation indicators based on analyses of each energy flow’s input and output, and utilization, conversion and loss during the steel manufacturing process. They provide support for making dynamic optimizations for the energy flow network to achieve an optimal operation efficiency. As a result, an optimization model for energy flow network operations has been built and plays a part in guiding the steel enterprise to contribute further to energy conservation and consumption reduction.

Method for evaluating the dynamic thermal performance of heating terminals based on an analysis of heat quantity and grade

Behavior energy conservation in buildings greatly impacts energy conservation and carbon reductions at the national level, and part-time space heating has become an important part of building operation. Under these conditions, heating terminals with different properties differ significantly in their heat consumption and heat grade demands. However, a comprehensive evaluation of the transference of quantities and grades of heat during the part-time heating of different terminals remains lacking. In this study, we developed a entransy-based analytical method and used it to compare the experimental and field-measured performances of three typical heating terminals (a radiator, radiant flooring, and a fan coil) from start-up to stabilization. The entire energy flow networks — from the heat source to the outdoor environment — were analyzed. Transient and cumulative entransy dissipation was analyzed to determine the energy utilization trends of the terminals. The dissipations between thermal nodes were also explored to intuitively reflect the key process(es) limiting heat transfer in a terminal. Further, we compared variations in heating performance under different operating durations to determine the applicability of each terminal type. Our results suggest that the radiant floor had the greatest entransy dissipation during start-up owing to its high thermal storage capacity and transfer resistance. However, because of the high thermal resistance of the terminal-to-chamber process, the entransy dissipation of the radiator was the greatest following stabilization (accounting for 57.0%), and the difference in the dissipation between radiators and the other terminal would increase with longer operating times.

A novel evaluation method for energy efficiency of process industry — A case study of typical iron and steel manufacturing process

Energy efficiency is an extremely important indicator for exploring energy conservation and consumption reduction. The traditional energy efficiency assessment methods for process industry lack an in-depth thinking on energy utilization of the whole system. Based on the proposed required energy, this paper established an energy efficiency assessment method for process industry. By establishing material and energy flow networks, energy efficiency of typical iron and steel manufacturing process (ISMP) is analyzed and its influencing factors are discussed. The results found that the required energy of coking, sintering, pelletizing, blast furnace iron-making, basic oxygen furnace steel-making and steel rolling are 2626.2 MJ/t-coke, 1122.9 MJ/t-sinter, 992.3 MJ/t-pellet, 9781.5 MJ/t-hot metal, 393.95 MJ/t-molten steel and 445.3 MJ/t-steel products, respectively. The energy efficiency of typical ISMP is 66.9%. The energy efficiency of the ISMP can be effectively improved by adjusting the steel ratio, recovering waste heat and residual energy, and developing interface technologies.

Material and Energy Flow Analysis (Mefa) of the Unconventional Method of Electricity Production Based on Underground Coal Gasification

Purpose In this paper, the application of Umberto NXT LCA software to devise a Material and Energy Flow Analyses (MEFA) for the technology of producing electricity from gas extracted in the process of shaftless underground coal gasification is presented. The Material Flow Analyses of underground coal gasification includes a range of technology, through obtaining process gas and its purification, to electricity production, and, additionally, the capture of carbon dioxide. Methods To evaluate electricity production based on Underground Coal Gasification, Material and Energy Flow Analyses (MEFA) was used. Modeling material and energy flow helps a high level of efficiency or technology of a given process to be reached, through the effective use of resources and energy, or waste management. The applied software for modeling material flow enables, not only, the simulation of industrial processes, but also the simulation of any process with a material or energy flow, e.g. in agriculture. Results MEFA enabled the visualization of material and energy flow between individual unit processes of the technology of electricity production from UCG gas. An analysis of material and energy flow networks presented in the form of Sankey diagrams enabled the identification of unit processes with the biggest consumption of raw materials and energy, and the greatest amount of emissions to the environment. Practical implications Thanks to applying material and energy flow networks with Umberto software, it is possible to visualize the flow of materials and energy in an analyzed system (process/technology). The visualization can be presented in the form of an inventory list of input and output data, or in the form of a Sankey diagram. In the article, a Sankey diagram has been utilized. MEFA is first stage of the plan to conduct analyses using Umberto software. The analyses performed so far will be used in the following stages of the research to assess the environmental impact using the LCA (Life Cycle Assessment) technique, to analyze costs using the LCC (Life Cycle Cost) technique, and to analyze eco-efficiency. It is important to highlight the fact that this is the first attempt of material and energy flow analysis of electricity production from UCG gas. Originality/value This is the first approach which contains a whole chain of electricity production from Underground Coal Gasification, including stages of gas cleaning, electricity production and the additional capture of carbon dioxide.

Material and Energy Flow Analysis (Mefa) of the Unconventional Method of Electricity Production Based on Underground Coal Gasification

Purpose In this paper, the application of Umberto NXT LCA software to devise a Material and Energy Flow Analyses (MEFA) for the technology of producing electricity from gas extracted in the process of shaftless underground coal gasification is presented. The Material Flow Analyses of underground coal gasification includes a range of technology, through obtaining process gas and its purification, to electricity production, and, additionally, the capture of carbon dioxide. Methods To evaluate electricity production based on Underground Coal Gasification, Material and Energy Flow Analyses (MEFA) was used. Modeling material and energy flow helps a high level of efficiency or technology of a given process to be reached, through the effective use of resources and energy, or waste management. The applied software for modeling material flow enables, not only, the simulation of industrial processes, but also the simulation of any process with a material or energy flow, e.g. in agriculture. Results MEFA enabled the visualization of material and energy flow between individual unit processes of the technology of electricity production from UCG gas. An analysis of material and energy flow networks presented in the form of Sankey diagrams enabled the identification of unit processes with the biggest consumption of raw materials and energy, and the greatest amount of emissions to the environment. Practical implications Thanks to applying material and energy flow networks with Umberto software, it is possible to visualize the flow of materials and energy in an analyzed system (process/technology). The visualization can be presented in the form of an inventory list of input and output data, or in the form of a Sankey diagram. In the article, a Sankey diagram has been utilized. MEFA is first stage of the plan to conduct analyses using Umberto software. The analyses performed so far will be used in the following stages of the research to assess the environmental impact using the LCA (Life Cycle Assessment) technique, to analyze costs using the LCC (Life Cycle Cost) technique, and to analyze eco-efficiency. It is important to highlight the fact that this is the first attempt of material and energy flow analysis of electricity production from UCG gas. Originality/value This is the first approach which contains a whole chain of electricity production from Underground Coal Gasification, including stages of gas cleaning, electricity production and the additional capture of carbon dioxide.

Equivariant energy flow networks for jet tagging

Jet tagging techniques that make use of deep learning show great potential for improving physics analyses at colliders. One such method is the Energy Flow Network (EFN) - a recently introduced neural network architecture that represents jets as permutation-invariant sets of particle momenta while maintaining infrared and collinear safety. We develop a variant of the Energy Flow Network architecture based on the Deep Sets formalism, incorporating permutation-equivariant layers. We derive conditions under which infrared and collinear safety can be maintained, and study the performance of these networks on the canonical example of W-boson tagging. We find that equivariant Energy Flow Networks have similar performance to Particle Flow Networks, which are superior to standard EFNs. However, equivariant Particle Flow Networks suffer from convergence and overfitting issues. Finally, we study how equivariant networks sculpt the jet mass and provide some initial results on decorrelation using planing.