Real Load Research Articles

Load forecasting method based on CNN and extended LSTM

With the continuous development of power systems and the growing demand, accurate power load prediction has become essential to ensure stable system operation and optimize energy allocation. The collection and analysis of historical load data have become increasingly important, and the factors affecting load changes have grown more complex. To address this challenge, this paper proposes a short-term load prediction method based on convolutional neural networks (CNN) and multilayer extended long short-term memory (LSTM) neural networks. The method first extracts local features from historical data using 1D convolution operations in the CNN. Then, a multilayer extended LSTM network captures long-term dependencies in the data from multiple dimensions for more accurate load prediction. The prediction experiments use real load and related factor data from a city in eastern Mongolia. Results demonstrate that the proposed method outperforms other deep learning-based load prediction methods in direct single-step, four-step, and 32-step load predictions in terms of MAPE, RMSE, and MAE indexes. Additionally, the proposed model can be extended to more prediction application scenarios, including intelligent transportation and financial analysis.

Do we need new standards in dental implant testing?

Abstract The ISO 14801 standard test is repeatedly criticized for not reflecting the actual conditions and loads realistically enough to which dental implants are exposed (e. g. during chewing). The aim of this paper is to provide a brief review of the test parameters of the ISO standard and to compare them critically with the parameters that prevail in vivo. The research revealed significant simplifications of the test conditions, wich can partly be justified with better reproducibility and comparability. Nevertheless, the realization remains that the test deviates considerably from real physiological loading conditions and should be adapted accordingly or supplemented by further tests.

A Computationally Efficient Rule-Based Scheduling Algorithm for Battery Energy Storage Systems

This paper presents a rule-based control strategy for the Battery Management System (BMS) of a prosumer connected to a low-voltage distribution network. The main objective of this work is to propose a computationally efficient algorithm capable of managing energy flows between the distribution network and a prosumer equipped with a photovoltaic (PV) energy production system. The goal of the BMS is to maximize the prosumer’s economic revenue by optimizing the use, storage, sale, and purchase of PV energy based on electricity market information and daily production/consumption curves. To achieve this goal, the method proposed in this paper consists of developing a rule-based algorithm that manages the prosumer’s Battery Energy Storage System (BESS). The rule-based approach in this type of problem allows for the reduction of computational costs, which is of fundamental importance in contexts where many users will be coordinated simultaneously. This means that the BMS presented in this work could play a vital role in emerging Renewable Energy Communities (RECs). From a general point of view, the method requires an algorithm to process the load and generation profiles of the prosumer for the following three days, together with the hourly price curve. The output is a battery scheduling plan for the timeframe, which is updated every hour. In this paper, the algorithm is validated in terms of economic performance achieved and computational times on two experimental datasets with different scenarios characterized by real productions and loads of prosumers for over a year. The annual economic results are presented in this work, and the proposed rule-based approach is compared with a linear programming optimization algorithm. The comparison highlights similar performance in terms of economic revenue, but the rule-based approach guarantees 30 times lower processing time.

Reliability Index for Steel Structure’s Technical State and Residual Life Assessment

The article is dedicated to problem of technical state assessment of real steel truss under operation for residual life prediction after technical investigation. The reliability index is calculated based on design, real and prospective loads. Method of statistical data analysis adopted for stress-strain determination also for all kinds of applied loads. The residual life with the reliability index determined as a transition from satisfactory to emergency technical state. The method provides the possibility for estimation of future safety steel structures operation with technical investigation results only that exclude any subjective opinion.

Variable stroke fatigue test for aircraft landing gears

The fixed stroke fatigue test cannot simulate the real loading conditions of the landing gear, and the variable stroke fatigue test has increasingly become mainstream. A modular test device for fatigue tests for aircraft landing gears was designed, and the automatic control of the buffer stroke was realized with ectopic displacement control technology effectively. The horizontal loading cylinder could follow the wheel axis actively with the displacement active control technology. The follow-up loading along the vertical direction was realized by constructing a polar coordinate system and making the loading line automatic alignment. A split-type dummy wheel was designed to improve the loading accuracy, and a booster cylinder was designed to realize the control loading of high pressure combined with the hydraulic control system. The test results show that the technical scheme meets the design requirements, and the fatigue test was completed successfully. The device for landing gear variable stroke fatigue testing is appropriate for other landing gear static/fatigue tests.

Transient biomass-SOFC-energy storage hybrid system for microgrids peak shaving based on optimized regulation strategy

To address the issues of energy supply instability and peak-shaving in remote microgrids, this paper proposes a biomass-SOFC (Solid Oxide Fuel Cell) -energy storage hybrid system to meet the power demands of the microgrids. Additionally, it integrates the long short-term memory (LSTM) prediction algorithm for peak shaving in the microgrids. Firstly, transient models of biomass gasifier and SOFC and the vanadium redox flow battery (VRFB) model are established, and the experimental results are used to verify the models. Subsequently, the dynamic responses and sensitivity analysis of the biomass gasification process and the internal temperature field of the SOFC are investigated. Next, an analysis is conducted of the differences in VRFB parameters obtained based on the forecasted load and the real load. Finally, the ultimate peak regulation strategy is determined, and the VRFB parameters are reasonably designed. The research results indicate that, following optimization, the designed capacity and maximum power of the VRFB are 6.7 MWh and 1.5 MW, respectively. This result represents a reduction of 23 % and 11.76 % compared to the values before optimization, leading to a total investment decrease of 26.21 % for the energy storage system. In addition, the annual profit from the peak shaving operation of the system is 189,879 dollars.

Optimal Integration for DGs with EVs planning in Distribution Networks with Hybrid MC-GA Technique using Realistic Load Models

With the growing integration of Distributed Generations (DGs) and Electric Vehicles (EVs) into distribution grids, the need for efficient and intelligent management strategies has become paramount. The need for power consumption is increasing every day in accordance with the standard demand pattern. As a result, either creating a lot of electricity or minimizing losses is crucial. In order to minimize real and reactive power losses from the system's point of view, this paper presents the optimal performance index-based size and location determination of DGs, EVs and their coordination in distribution networks using realistic load models (RLMs) such as RLM-1, RLM-2, RLM-3, RLM-4, and RLM-5, respectively. Improvement of the system power factor (SPF) is defined in this analysis as the power system's performance both with and without different kinds of DGs and DGs with EVs for various RLMs. PHEVs type of EVs used in this paper. The hybrid Monte Carlo-Genetic Algorithm (MC-GA) is the basis of the simulation technique that is being suggested. Networks with IEEE-16 and 37 buses in their distribution systems have been used to test the suggested methods. Comprehensive simulations on a test distribution network are used to show that the suggested method is effective. The system's performance can be optimized in terms of grid stability, efficiency, and reliability, according to the results. By stressing the significance of employing a hybrid MC-GA technique to meet realistic load circumstances, this research advances optimal DG and EV solutions.

Multi-Objective Optimization Algorithm Based Bidirectional Long Short Term Memory Network Model for Optimum Sizing of Distributed Generators and Shunt Capacitors for Distribution Systems

In this paper, a multi-objective grey wolf optimization (GWO) algorithm based Bidirectional Long Short Term Memory (BiLSTM) network machine learning (ML) model is proposed for finding the optimum sizing of distributed generators (DGs) and shunt capacitors (SHCs) to enhance the performance of distribution systems at any desired load factor. The stochastic traits of evolutionary computing methods necessitate running the algorithm repeatedly to confirm the global optimum. In order to save utility engineers time and effort, this study introduces a BiLSTM network-based machine learning model to directly estimate the optimal values of DGs and SHCs, rather than relying on load flow estimates. At first, a multi-objective grey wolf optimizer determines the most suitable locations and capacities of DGs and SHCs at the unity load factor and the same locations are used to obtain optimum sizing of DGs and SHCs at other load factors also. The base case data sets consisting of substation apparent power, real power load, reactive power load, real power loss, reactive power loss and minimum node voltage at various load factors in per unit values are taken as input training data for the machine learning model. The optimal sizes of the DGs and SHCs for the corresponding load factors obtained using GWO algorithm are taken as target data sets in per unit values for the machine learning model. An adaptive moment estimation (adam) optimization approach is employed to train the BiLSTM ML model for identifying the ideal values of distributed generations and shunt capacitors at different load factors. The efficacy of the proposed ML-based sizing algorithm is demonstrated via simulation studies.

Impact of triaxial stress state at various pavement depths on the dynamic modulus of asphalt mixture

The mechanical characteristics of asphalt mixture are closely related with its stress state, temperature, and loading rate, therefore, the loading condition of the asphalt mixture's dynamic modulus test should be consistent with the actual temperature, loading frequency, and stress state that are applied to the mixture in actual pavement, thus can the test produce an objective result that can accurately reflect the actual stress-strain relationship for the asphalt mixture. However, due to limitations in the loading capacity of current dynamic modulus test equipment, the present modulus tester cannot yet set such a loading condition whose stress state is consistent with that of asphalt mixture in actual pavement. As a result, the modulus testing results of the current test may not accurately reflect the actual stress-strain characteristics of an asphalt mixture under real traffic loads. Therefore, this paper aims to figure out the impact of stress state on asphalt mixture’s dynamic modulus, and firstly conducts a numerical analysis to ascertain asphalt pavement’s triaxial stress state at different depth, then, utilizing the nonlinear elasticity theory of Soil Plasticity, proposes a theoretical model that can reflect triaxial stress state’s effect on asphalt mixture’s dynamic modulus, and then, arranges a series of triaxial dynamic modulus tests for asphalt mixture in different stress state to verify the model’s effectiveness, and meantime analyzes stress state’s influencing rule to asphalt mixture’s dynamic modulus. Their results indicate, the asphalt mixture of the actual pavement is in an obvious triaxial stress state, and that the higher the triaxial stress state, the larger the dynamic modulus, particularly for the mixture near the surface, whose high stress state will lead to their dynamic modulus to be significantly larger than that of the underlying course, while the model proposed in this paper can to a large extent reflect this impact.

Optimised Deep Learning for Time-Critical Load Forecasting Using LSTM and Modified Particle Swarm Optimisation

Short-term electric load forecasting is critical for power system planning and operations due to demand fluctuations driven by variable energy resources. While deep learning-based forecasting models have shown strong performance, time-sensitive applications require improvements in both accuracy and convergence speed. To address this, we propose a hybrid model that combines long short-term memory (LSTM) with a modified particle swarm optimisation (mPSO) algorithm. Although LSTM is effective for nonlinear time-series predictions, its computational complexity increases with parameter variations. To overcome this, mPSO is used for parameter tuning, ensuring accurate forecasting while avoiding local optima. Additionally, XGBoost and decision tree filtering algorithms are incorporated to reduce dimensionality and prevent overfitting. Unlike existing models that focus mainly on accuracy, our framework optimises accuracy, stability, and convergence rate simultaneously. The model was tested on real hourly load data from New South Wales and Victoria, significantly outperforming benchmark models such as ENN, LSTM, GA-LSTM, and PSO-LSTM. For NSW, the proposed model reduced MSE by 91.91%, RMSE by 94.89%, and MAPE by 74.29%. In VIC, MSE decreased by 91.33%, RMSE by 95.73%, and MAPE by 72.06%, showcasing superior performance across all metrics.

Load profiles of residential off-grid solar systems on the Navajo Nation

Standalone off-grid electrical systems, no matter where they are deployed or for what user class, are designed based upon the load they are expected to serve. State-of-the-art computerized off-grid system design tools require the user to specify the expected load profile, that is, how the power consumption changes throughout the day. Often, this is at an hourly resolution, and some characterization of the distribution of power around the average values may be required. Specifying realistic and reasonable load profiles is a barrier to the appropriate design of standalone systems. This research extends previous studies on daily energy consumption of residential solar-powered off-grid systems on the Navajo Nation to provide hourly load profiles, statistical characteristics, and probabilistic models. The data analyzed come from 90 homes over a two-year period. K-means clustering is used to identify prototypical normalized load profiles when the data are grouped by year, season, weekday, and weekend. Eight parametric probability density functions are fit to the grouped data at an hourly resolution. Their fit to the data is evaluated using the Cramér-von Mises (CvM) statistic. The results show that the load profiles tend to be night-peaking and that Log Normal and Gumbel distributions can reasonably model variation in the data. The load profiles and probabilistic models can be used in off-grid design software and to synthesize load profiles for design and future research.

Mechanical performance analysis method for ribbed H-section aluminum alloy members with initial curvature and torsion angle

In this study, an experimental investigation was conducted on the axial compression performance of ribbed H-section aluminum alloy members with initial curvature and torsion angle under varying boundary conditions, including one end hinged with the other rigidly connected, and both ends rigidly connected. Ultimate bearing capacity and failure modes were identified under real loads and subsequently compared with previous findings from our research group on members with hinged ends. To account for initial imperfections introduced during processing and transportation, 3D scanning technology was utilized to capture the precise geometrical dimensions, constructing an accurate numerical simulation model. The experimental results were corroborated with numerical simulations, leading to the proposal of an analytical method for members with initial curvature and torsion angle. Furthermore, extensive parametric analysis elucidated the impact of initial curvature, torsion angle, and slenderness ratio on the ultimate bearing capacity, culminating in the formulation of the stability factor and calculated length factor based on numerical outcomes. The study discovered significant variances in bearing capacity under different boundary conditions, with one-end hinged and one-section rigidly connected, and two-end rigidly connected conditions exhibiting 1.4 and 2.1 times the capacity of the hinged-at-both-ends scenario. Under different boundary conditions, the axial compression members were subjected to flexural-torsional buckling failure. Moreover, when the ultimate bearing capacity was reached, the lower flange of the member and the web near the lower flange appeared obvious buckling phenomenon. The numerical analysis aligned well with experimental data, validating the simulation method's reliability and revealing the stress distribution and evolution during member failure. These findings offer vital theoretical insights and technical support for engineering design and practical applications.

Techno-economics of solar re-powering and retro-fitting an existing district heating network

Most of the district heating systems today use higher operating temperatures than those in new-built systems, possibly limiting compatibility with solar energy. This study evaluates the cost-effectiveness in terms of unit heat cost of integrating solar heating into an existing district heating system compared to not using solar energy, under changing economic boundary conditions such as collector and fuel cost, in addition to discount rate. This is investigated for both a scenario where the solar heating and a boiler replacement is done concurrently, as well as a scenario where solar heating is added to an existing system without replacing the boiler. A theoretical district heating supply of 3 MW is modelled and simulated based on a real system and load profile. The heat supply is varied to include storage with or without solar heating. Results for a 3 % discount rate indicate that; Replacing a 3 MW boiler with a slightly smaller boiler of 2.5 MW and adding a storage is cost effective and yields a unit heat cost of 58.0 EUR/MWh (16.1 EUR/TJ) which is a reduction of about 6 %. Installing solar heating together with the boiler replacement yields a unit heat cost as low as 55.7 EUR/MWh (15.4 EUR/GJ) which is a reduction of about 8 %. When replacing the boiler, all system configurations have similar unit heat costs compared to a boiler-only system, so factors such as emission reductions due to solar heating are relevant when considering alternatives. Furthermore, adding solar flat plate collectors corresponding to a 13 % solar fraction without replacing the boiler can reduce the unit heat cost as low as 34.8 EUR/MWh (9.7 EUR/TJ), which is 32 % lower than without solar. Evacuated tube collectors can increase this solar fraction to 17 % with similar system size, although at a higher cost. At a discount rate of 5 % solar heating is cost-competitive when fuel cost is above 26 EUR/MWh (7.2 EUR/TJ) and at 7 % competitive when fuel cost is above 32 EUR/MWh (8.9 EUR/TJ). Increasing solar heating system size reduces the backup-boiler fuel use during summer maintenance and makes fuel type less relevant for the overall unit heat cost.

Robust PD/A biofilms sustain advanced real municipal wastewater treatment: Reinforcement of interspecific communication and electron transfer

In this study, a robust partial denitrification/anammox (PD/A) biofilm system was efficiently established within merely 83 days using solely conventional activated sludge as the inoculum, employing a multivariate regulation strategy involving the addition of biofillers, optimization of municipal wastewater feeding ratios, prolongation of hydraulic retention time, and introduction of hydrazine. The performance variability of PD/A system within a long-term, intricate mainstream substrate environment was further investigated by incrementally elevating municipal wastewater proportion (50 % → 75 % → 100 %). The results demonstrated that PD/A system ultimately achieved an impressive total nitrogen removal efficiency of 93.6 % and an effluent total nitrogen concentration of 8.6 mg/L, accompanied by anammox contributing over 85 % to nitrogen removal. The biofilm’s layered spatial structure nurtured Candidatus Brocadia, enabling it to reach an abundance of 2.5 % under high real municipal wastewater loading. Metabolic function analyses revealed that biofilm enhanced the abundance of genes associated with interspecies electron transfer and quorum sensing communication in PD/A consortia after prolonged exposure to municipal wastewater. These findings underscore the significances of interspecific interactions, bacterial communications, and synergistic metabolic traits within PD/A biofilm systems in the stabilization of real municipal wastewater treatment.

Hydraulic parameter estimation for district heating based on laboratory experiments

In this paper we consider calibration of hydraulic models for district heating networks based on operational data. We extend previous theoretical work on the topic to handle real-world complications, namely unknown valve characteristics and hysteresis. We generate two datasets in the Smart Water Infrastructure Laboratory in Aalborg, Denmark, on which we evaluate the proposed procedure. In the first data set the system is controlled in such a way to excite all operational modes in terms of combinations of valve set-points. Here the best performing model predicted volume flow rates within roughly 5 and 10 % deviation from the mean volume flow rate for the consumer with the highest and lowest mean volume flow rates respectively. This performance was met in the majority of the operational region. In the second data set, the system was controlled in order to mimic real load curves. The model trained on this data set performed similarly well when evaluated on data in the operational range represented in the training data. However, the model performance deteriorated when evaluated on data which was not represented in the training data.

The relevance of shock absorber friction on the total friction of a MacPherson suspension system during lateral manoeuvres

The friction potential of the suspension is typically considered to be invariant as the vertical dynamics of a vehicle are analysed. However, the current research on MacPherson suspensions indicates a strong dependence of the suspension friction on the present driving situation, which is primarily characterised by the wheel deflection relative to the chassis, and the tyre forces. Although it is well known that the shock absorber is generally subjected to side loads that can vary and initiate a certain amount of friction, the precise contribution of the shock absorber friction to the total suspension friction has yet to be investigated in detail. It is therefore essential to conduct a quantitative study on the importance of the shock absorber in the context of specific driving situations regarding the development of new, friction optimised shock absorber generations. The present study investigates the relationship between total suspension friction and the friction of the shock absorber, with a particular focus on the case of wheel lateral forces that aim to represent a simplified cornering manoeuvre of the vehicle. This is realised by using a multi-body simulation approach and a strain gauge application in order to estimate the side force at the shock absorber’s top mount. Subsequently, the shock absorber friction is determined using a test setup that considers the realistic load and mounting situation from the perspective of the MacPherson suspension subsystem. The combination of simulation and shock absorber friction measurement is sufficient to fully explain the characteristic friction trend on suspension (subsystem) level. The contribution of the shock absorber friction to the total suspension friction is determined to be a maximum of 80%, emphasising the importance of considering the side force at the shock absorber while aiming for suspension friction minimisation and ride comfort performance enhancement. On the other hand, the shock absorber’s contribution is also observed to be limited depending on the actual wheel force, indicating that the friction of the other components remains a crucial factor in the MacPherson suspension concept.

Intelligent aerodynamic identification method based on wavelet transform and deep learning

Abstract Wind tunnel force testing is particularly crucial for the aerodynamic characteristics of hypersonic vehicles, and the intelligent and high-precision aerodynamic identification technology serves as the key supporting technology. Due to the short effective test time of the pulse combustion wind tunnel, the inertial force vibration generated by the force measurement system is difficult to be attenuated completely at the start of the wind tunnel. At the same time, there are many interference signals in the output signal during the test, which makes it difficult to achieve high-precision aerodynamic load identification. Therefore, in order to eliminate the influence of interference signals on aerodynamic identification accuracy and obtain accurate aerodynamic loads, this paper introduces an intelligent aerodynamic identification method based on wavelet transform and deep learning. Firstly, meyer wavelet is selected as the wavelet basis, and signal processing is performed on the output signal through wavelet transform to eliminate interference signals and improve signal quality. Then, through RNN-GRU network model to identify the load. The reliability of the method is verified by the rod balance test platform. The results show that the method can effectively reduce the influence of interference signal, and the relative error between the identified load and the real load is about 2%. At the same time, the identification results of the method are compared with those of the mean value method, which further verifies the accuracy of the proposed method.

Improved stomatognathic model for highly realistic finite element analysis of temporomandibular joint biomechanics

BackgroundMechanical response analysis of the temporomandibular joint (TMJ) is crucial for understanding the occurrence and development of diseases. However, the realistic modeling of the TMJ remains challenging because of its complex composition and multivariate associations. ObjectiveThis study aims to develop a highly realistic stomatognathic model that accurately represents the geometric accuracy, structural integrity, and material properties. And further optimizes the interference and establishes the application range of the simplifications and the assumptions. MethodsGeometric reconstruction of the bone was based on high-resolution image data, with the accuracy of the occlusal surface ensured by plaster cast model registration. Soft tissues such as cartilage, the disc, the periodontal ligament (PDL), and disc attachments often need to be approximated or assumed. Therefore, the finite element methods (FEM) was used to optimize these assumptions, including 1) the biomechanical effects of the thickness and modulus of the PDL, 2) the approximation of the geometry and material behavior of the disc, and 3) the simplification of the loading and boundary conditions. Results1) The deformation of the PDL causes tooth movement, which spreads to the distal condyle and further effects the TMJ load situation, 2) Disc reconstructed by MRI and hyperelastic material behavior are necessary for accurate TMJ loading analyses, 3) The loss of relative sliding movement between teeth interferes with realistic TMJ loading. ConclusionThe improved stomatognathic model delivers highly realistic and validated simulation, offering theoretical guidance for virtual treatments and TMJ multivariate overload studies.

Computational Models of Dynamic Load Sources for Modeling of Construction Structures Operation Used in Monitoring of Technical Condition of Buildings and Structures

This research aims to develop principles for assessing the impact of mega-cyclic vibrodynamic loads on the reliability of construction structures. The relevance of the reasons for the development and improvement of algorithms for numerical modeling of multicycle dynamic loads on building structures is due to the steadily increasing intensity of such loads on buildings and structures in megacities, as well as the acute practical problem of significant differences in the dynamic characteristics of buildings and structures obtained as a result of mathematical modeling and determined by experimental methods. The article presents research materials on computational equivalent models of dynamic load sources for numerical modeling of the behavior of construction structures under their influence, using the method of vibroacoustic analogies. The article examines models of sources of dynamic impact on construction sites. Algorithms and final formulas for computational modeling of the simplest sources of dynamic load are developed using the method of vibroacoustic analogies. The dynamic properties of the simplest dynamic load sources were analyzed. A significant difference between the computational models of real and ideal dynamic load sources. The article presents research and development results intended for calculating the distribution of dynamic loads on elements of construction structures in industrial and civil engineering projects located in areas with high levels of transport vibrodynamic impacts. An important property of the proposed computational equivalent models of sources of dynamic impact on building structures is the possibility of computational verification of critical elements, points, or nodes of load-bearing structures of buildings and structures under dynamic overloads. The position of these critical elements, points, and nodes of load-bearing structures under dynamic loads can differ significantly from their position determined using static and quasistatic computational modeling methods.

Simulating noise and vibrations on passenger ships using Statistical Energy Analysis

Various sources aboard a passenger ship contribute significantly to noise and vibration concerns. Notably, fluctuating pressure loads on the hull can result in elevated noise levels within rooms near the propulsion. Expensive solutions of sound insulation such as floating floors are employed to reduce these levels. Nowadays, designing these treatments is mainly based on empirical studies. However, the limited availability of the ship, the laborious nature of experiments, and the possibility of solutions being oversized have only increased the method's constraints. Utilizing calculation methods can save time, reduce additional weight on the ship, and provide more tailored design solutions. A Statistical Energy Analysis (SEA) has been used as deterministic approaches are often not suitable for the frequency range of interest for full-scale structures. The validation process for the wave-based SEA method involves several steps. Initially, simple models are constructed to establish correlations between local airborne and structure-borne energy transmission using simple acoustic loads, as well as to assess the method's scope of application. Subsequently, this approach is applied to larger models, such as the aft full part of the ship, enabling the prediction of noise levels and energy transmission paths under real loads.