Application of orthogonal sampling to incomplete data interferometric tomography: numerical simulation and experimental analysis

Wire sawing is an important process in the cutting of NdFeB magnets and the sawed surface topography and surface roughness (SR) are important indicators for assessing surface quality. This paper analyzed the effects of process parameters on the sawed NdFeB surface topography and SR based on orthogonal experiments and then presented an SR prediction model called ISSA-BP, which was based on a BP neural network using an improved sparrow search algorithm (ISSA). For the problem of insufficient optimization capability of the traditional sparrow search algorithm (SSA), Cubic chaotic mapping, Latin hypercube sampling, the sine–cosine algorithm, Levy flight, and Cauchy mutation were introduced to improve the traditional sparrow search algorithm (SSA) to obtain ISSA, improving algorithm convergence speed and global optimization. The ISSA was then used to optimize the initial weights and thresholds of the BP neural network for predicting Ra. Research shows that the sawed surface topography reflects a combination of brittle and ductile material removal. As the workpiece feed speed and size decrease and the wire speed increases, there is a reduction in SR. Compared with the SSA-BP and traditional BP models, the ISSA-BP prediction model has reduced various error indicators such as mean absolute error (MAE) and mean square error (MSE). The mean absolute error (MAE) of the prediction model optimized by the ISSA is 0.064475, the mean square error (MSE) is 0.0072297, the root mean square error (RMSE) is 0.085028, and the mean absolute percentage error (MAPE) is 3.7171%. The research results provide an experimental basis and technical support for predicting the SR and optimizing the process parameters in diamond wire-sawing NdFeB.

An orthogonal projection sampling method is proposed in this paper and is applied in the reconstruction of incomplete data field by using the prior knowledge algorithm based on the modified ART, by which the satisfying reconstruction results can be obtained even with less sampling directions and limited sampling angular range. The Mean Square Error (MSE) and peak value between the origin field, the reconstructed field using orthogonal projection sampling method and traditional projection sampling method are analyzed with the numerical simulation of computer. The results indicate that the orthogonal projection sampling can reduce the MSE greatly in the reconstruction of incomplete data. Thus our researches provide a new means for the reconstruction of incomplete data field.

A new algebraic reconstruction technique (ART), nonlinear auto-adjusting iterative reconstruction technique (NAIRT), is proposed and applied to reconstruct a section of an actual thermal air flow field. With numerical simulation, NAIRT was tested to reconstruct a complicated field to demonstrate its superior reconstructive capability. In contrast, three typical ARTs, the basic ART, simultaneous ART (SART), and a modified SART (MSART), were simulated to demonstrate the reconstructive capability improvement attained through the use of the proposed NAIRT. The calculated results were discussed with mean square error (MSE) and peak error (PE). A thermal air flow field was produced with an alcohol burner and was detected by a laser beam. With laser beam projections, a cross-section of the field was reconstructed by NAIRT. As a result, the reconstructive capability was improved much by NAIRT. The MSE decreased by 95.5%, and PE by 97.2% from that of the basic ART. Only NAIRT converged without filters while its reconstructive accuracy improved. By increasing the projections from 42 to 84, the accuracy of NAIRT without filters was improved significantly. NAIRT could effectively reconstruct the section of the thermal field. The proposed NAIRT needed no filter for its convergence and it had the highest reconstructive accuracy and simplest iterative expression of those analyzed.

<p>This paper introduces a novel hybrid approach that integrates Latin hypercube sampling (LHS) and Bayesian optimization for optimizing artificial neural networks (ANNs) in fault detection, classification, and location for transmission lines. The proposed method advances the accuracy and efficiency of fault diagnosis in power systems, representing a significant step forward compared to conventional approaches. The test system is a 400 kV, 50 Hz, 300 km transmission system, and the simulations were carried out in MATLAB/Simulink environment. Using the strategic insight of LHS, optimal initial points were determined, which subsequently formed the basis for the Bayesian optimization to further refine the learning rate and training epochs. Within the fault detection domain, the model showcased remarkable precision when deployed on an evaluation dataset of 168 cases, accurately detecting every instance of normal and faulty scenarios. This culminated in an astounding 100% accuracy in fault detection. In terms of fault classification, the ANN model, trained on a dataset of 495 instances, revealed perfect regression coefficients across the training, testing, and validation subsets. When tested against unseen data, it demonstrated its proficiency by correctly classifying 154 out of 154 cases, showcasing a 100% F1 score. Also, the accuracy figures in terms of fault location fluctuated between 99.826% and 99.999%, with a mean absolute percentage error (MAPE) of 0.053%. The model's mean square error (MSE) stood at 0.0083, while the mean absolute error (MAE) was calculated at 0.0717. A deep dive into diverse fault types reaffirmed the model's precision, underscoring its consistent performance across various fault scenarios. The trained models were deployed on three different transmission lines and the models exhibited remarkable precision in all the cases tested. In summary, the innovative hybrid optimized ANN model, weaving together the strengths of LHS and Bayesian optimization, signifies an advancement in the field of power system fault analysis, ushering in heightened efficiency and reliability.</p>

This paper pioneers a novel approach in electromagnetic (EM) system analysis by synergistically combining Bayesian Neural Networks (BNNs) informed by Latin Hypercube Sampling (LHS) with advanced thermal-mechanical surrogate modeling within COMSOL simulations for high-frequency low-pass filter modeling. Our methodology transcends traditional EM characterization by integrating physical dimension variability, thermal effects, mechanical deformation, and real-world operational conditions, thereby achieving a significant leap in predictive modeling fidelity. Through rigorous evaluation using Mean Squared Error (MSE), Maximum Learning Error (MLE), and Maximum Test Error (MTE) metrics, as well as comprehensive validation on unseen data, the model's robustness and generalization capability is demonstrated. This research challenges conventional methods, offering a nuanced understanding of multiphysical phenomena to enhance reliability and resilience in electronic component design and optimization. The integration of thermal variables alongside dimensional parameters marks a novel paradigm in filter performance analysis, significantly improving simulation accuracy. Our findings not only contribute to the body of knowledge in EM diagnostics and complex-environment analysis but also pave the way for future investigations into the fusion of machine learning with computational physics, promising transformative impacts across various applications, from telecommunications to medical devices.

This paper aims to develop an artificial neural network (ANN) to predict the energy consumption and cost of cross laminated timber (CLT) office buildings in severe cold regions during the early stage of architectural design. Eleven variables were selected as input variables including building form and construction variables, and the values of input variables were determined by local building standards and surveys. ANNs were trained by the simulation data and Latin hypercube sampling (LHS) method was used to select training datasets for the ANN training. The best ANN was obtained by analyzing the output variables and the number of hidden layer neurons. The results showed that the ANN with multiple outputs presented better prediction performance than the ANN with single output. Moreover, the number of hidden layer neurons in ANN should be greater than five and preferably 10, and the best mean square error (MSE) value was 1.957 × 103. In addition, it was found that the time of predicting building energy consumption and cost by ANN was 80% shorter than that of traditional building energy consumption simulation and cost calculation method.

The number of aerodynamic analysis in the design optimization of propeller has been reduced significantly by using the surrogate model, such as kriging. In this paper, a more efficient aerodynamic design optimization method of propeller is proposed by using the HK (hierarchical kriging) model. The high-fidelity model is defined as a RANS (Reynolds- Averaged Navier-Stokes) simulation for propeller and the low-fidelity model is defined as the results by blade-element/vortex theory. Initial samples are selected for two levels of fidelity via LHS (Latin Hypercube Sampling). The aerodynamic performance of varying fidelity is used to build variable-fidelity surrogate models for functions of objective (e.g. thrust) and constraint (e.g. shaft power). Infill sampling criteria, including MSP (minimizing of surrogate prediction), EI (expected improvement), PI (probability of improvement), LCB (lower-confidence bounding) and MSE (mean-squared error) are used to obtain new samples, and the surrogate models are repetitively updated until a global optimum is found. High-altitude propellers are optimized by the kriging model and hierarchical kriging model, respectively. Compared to kriging model, the number of RANS solving by the hierarchical kriging model is reduced 37.5%, and the optimization time is reduced 24.3%. The results have shown that the proposed design method for propeller can significantly improve the optimization efficiency.

This work presents a complete hybrid testing methodology for flow assessment that combines computational fluid dynamics (CFD) simulations, stochastic variational Gaussian processes (SVGP), and multi-fidelity Gaussian processes (MFGP) to achieve accurate assessments with fewer measurements. Focused on minimizing energy consumption and testing times, the methodology utilizes SVGP to identify informative measurement locations based on CFD data, enabling a data-driven design of experiments (DoE) approach. This optimizes sensor placement and minimizes redundant measurements. Acquired experimental data are then fused with CFD data using a MFGP model, providing a comprehensive flow field reconstruction with associated uncertainty estimates within a Bayesian framework comparable to traditional methodologies. Two test cases demonstrate the methodology's effectiveness: the H25 axial compressor, used as a development test case due to availability of preexisting experimental data, and a state-of-the-art ECL5 ultra-high bypass ratio fan, for which the methodology was applied in a blind validation scenario. Results show a significant reduction in required measurements (around 70% for the H25 and 50% for the ECL5) without sacrificing accuracy, leading to a substantial decrease in experimental testing time. The SVGP DoE framework outperforms random and Latin hypercube sampling, effectively capturing relevant flow features with fewer measurements. The methodology also adapts to varying flow conditions, accurately predicting flow fields and quantifying uncertainties across different Reynolds numbers. This work introduces a shift in experimental campaigns, strategically optimizing measurement strategies during the planning phase, emphasizing machine learning trends in shaping new research paths toward more efficient, cost-effective, and sustainable experimental practices for fluid dynamics.

The paper presents the research of the numerical simulation techniques and the preliminary experimental validation on the start characteristics for a typical two-dimensional (2D) hypersonic inlet. In order to obtain the start and unstart hysteresis loop, numerical simulations methods using the incoming flow field, the zero-velocity flow field, and the latest convergent flow field as the initialized flow field separately are adopted to calculate the inlet start characteristics until the flow fields converge. The calculation software is AHL3D, a parallel computational fluid software self-developed by CARDC. The numerical methods also include the method from unstart flow field to start one with Mach-numbers gradually increasing till the inlet is start and the method from start flow field to unstart one with Mach-numbers decreasing till the inlet is unstart. According to the comparison with different initialized flow field and with different step-size of Mach-number, the inlet start characteristics are obtained and the numerical techniques are analyzed. Based on the comparison and the analysis, it is suggested that the numerical techniques of inlet start characteristics should use the zero-initialized flow field with one step to the final Mach-numbers to predict the minimal self-start Mach-number of the inlet and use the incoming flow-initialized flow field to predict the minimum start Mach-number. The results show that the numerical techniques are high-efficient and easily operational. To verify the effect of the numerical techniques, wind tunnel tests are arranged to research the start characteristics of a typical 2D hypersonic inlet. The preliminary experimental result shows that the numerical techniques of predicting inlet start characteristics are consistent with the experiments very well, which means that the application of the numerical techniques can be further carried out for such hypersonic inlets as mentioned in the current paper.

Establishing a concise and accurate analytical model is the key to developing a feasible progressive collapse design for engineering practice. However, existing models either focused on an individual force mechanism or required complicated computer programming. Among existing machine learning (ML) techniques, multi-gene genetic programming (MGGP) can be trained to obtain explicit formulas for engineering problems. In this study, a comprehensive database was established by data collection, Latin hypercube sampling and structural design, and was used to train the mathematical model for quantifying progressive collapse resistance of reinforced concrete (RC) beam-column substructures under middle column removal scenarios. Further, an energy-based error index was proposed to validate the accuracy of the MGGP model among others. The research outcomes can provide references for the development of simplified analytical models for calculating the progressive collapse progress of RC frame structures, and promote the development of the practical design method.

This study presents the hydrodynamic analysis and optimization for a single-channel submersible pump impeller. The numerical analysis involved the use of the shear stress transport turbulence model to solve three-dimensional Reynolds-averaged Navier-Stokes equations. The optimization process was based on a radial basis neural network model and was conducted by considering two design variables to control the area from the inlet to the outlet of the impeller. The hydrodynamic performance of the pump impeller was optimized by selecting the efficiency as the objective function. The objective function was numerically assessed via Latin hypercube sampling at design points selected in the design space, and the optimization yielded a maximum increase in efficiency of approximately 1% in terms of the design flow rate, as compared to the initial design. The performance curve for the efficiency was also improved in the high flow rate region. Further details of the internal flow fields between the reference and optimum designs are also analyzed and discussed in this work.

A novel multi-objective optimization design approach is presented integrating the Improved Whale Optimization Algorithm (IWOA) with Back Propagation Neural Network (BPNN) for the centrifugal pump with vaned diffuser. The primary objectives of this optimization are to enhance pump efficiency and reduce pressure pulsation. Under head constraints, seven critical geometric design variables of the vaned diffuser centrifugal pump were identified as optimization parameters. Sample data were generated using Latin Hypercube Sampling (LHS) to construct a high-accuracy surrogate model, which was subsequently globally optimized using IWOA. The optimization results were validated through numerical simulations, revealing that the optimized model achieves a 1.56% increase in efficiency and a 28.6% reduction in pressure pulsation intensity. These enhancements significantly improve the operational stability and reliability of the pump. Furthermore, the research delves into entropy production analysis and the comparison of unsteady flow characteristics, elucidating the mechanisms behind the reductions in internal energy losses and the improvements in flow field properties resulting from the optimized design. This work provides a novel methodology for designing high-efficiency, low-vibration centrifugal pumps, offering substantial engineering application value and valuable theoretical insights.

In fluid mechanics, modal decomposition, deeply intertwined with the concept of symmetry, is an essential data analysis method. It facilitates the segmentation of parameters such as flow, velocity, and pressure fields into distinct modes, each exhibiting symmetrical or asymmetrical characteristics in terms of amplitudes, frequencies, and phases. This technique, emphasizing the role of symmetry, is pivotal in both theoretical research and practical engineering applications. This paper delves into two dominant modal decomposition methods, infused with symmetry considerations: Proper Orthogonal Decomposition (POD) and Dynamic Mode Decomposition (DMD). POD excels in dissecting flow fields with clear periodic structures, often showcasing symmetrical patterns. It utilizes basis functions and time coefficients to delineate spatial modes and their evolution, highlighting symmetrical or asymmetrical transitions. In contrast, DMD effectively analyzes more complex, often asymmetrical structures like turbulent flows. By performing iterative analyses on the flow field, DMD discerns symmetrical or asymmetrical statistical structures, assembling modal functions and coefficients for decomposition. This method is adapted to extracting symmetrical patterns in vibration frequencies, growth rates, and intermodal coupling. The integration of modal decomposition with symmetry concepts in fluid mechanics enables the effective extraction of fluid flow features, such as symmetrically or asymmetrically arranged vortex configurations and trace evolutions. It enhances the post-processing analysis of numerical simulations and machine learning approaches in flow field simulations. In engineering, understanding the symmetrical aspects of complex flow dynamics is crucial. The dynamics assist in flow control, noise suppression, and optimization measures, thus improving the symmetry in system efficiency and energy consumption. Overall, modal decomposition methods, especially POD and DMD, provide significant insights into the symmetrical and asymmetrical analysis of fluid flow. These techniques underpin the study of fluid mechanics, offering crucial tools for fluid flow control, optimization, and the investigation of nonlinear phenomena and propagation modes in fluid dynamics, all through the lens of symmetry.

A new deflection tomography system was put forward and named Auto-adapted Iterative Deflection Tomography (AIDT). It was tested to approach accurate reconstruction of a high frequency field containing shock waves with sufficient projections. A high frequency field was constructed with Gauss and rectangle window functions. It was simulated to project in all directions with angle resolving rate per degree. Two cross projections was selected to simulate reconstructing the model field by AIDT. Then, the number of projections was increased step by step in order to approach the accurate values of the model. All reconstructed results were studied. As a result, all high frequency sections are distorted much. With a few projections, such as two projections, the reconstructed results lost the basic characteristics of the model. With projections increasing, the distortions get lighter and lighter. Near the critical projections, such as twenty-four projections, the reconstructed field becomes similar to the model except some high frequency sections. When eighty-four projections are employed, the reconstructed result is in accord with the model. In this case, the Mean Square Error(MSE) is 0.00003609. But up to one hundred and eighty projections, the reconstructed result almost stopped improving. In this case, the MSE is 0.00003231. As we know, these are most accurate reconstructions to now. AIDT can accurately reconstruct high frequency fields with shock waves.

Underwater Optical Wireless Communication (UWOC) becomes an emerging underwater communication technology, with high-data rates over relatively medium transmission ranges. When optical wireless signal transmitted in ocean water channel, it will suffer from drastic scattering and absorption due to water molecules, dissolved particles, air bubbles, and turbulence. Absorption and scattering of the transmitted wireless optical signal in underwater channel led to attenuation in optical signal power. Optical signal attenuation over underwater channel is an aggregate of` different parameters effects that changed frequently, then practical measuring of this attenuation is complicated, difficult, expensive, and time-consuming process. In this work, improved neural network optimized with future search algorithm (FANN) was proposed, as an efficacious solution to obtain an accurate, relabel values of attenuation coefficient in different water types and conditions. The proposed FANN model provides a good much results to the practical measured values. The performance of the proposed FANN model was evaluated using mean square error (MSE), root mean square error (RMSE) and mean absolute error (MAE) error indices. The errors in attenuation coefficient values obtained by the proposed FANN model had been calculated and its values are very acceptable which are lie lower than 10-4. The performance of the proposed FANN model shows excellent results which indicate the superior performance of the proposed FANN model.