Emulation Accuracy Research Articles

ЭМУЛЯЦИЯ ПОКАЗАНИЙ ДАТЧИКОВ КАЧЕСТВА ВОЗДУХА В ГОРОДСКОЙ СРЕДЕ УМНОГО ГОРОДА

Urban air pollution poses a serious threat to human health. To monitor it, both individual sensors and systems are used to assess the concentration of dust particles PM1, PM2.5, PM10 and organic compounds. However, the reliability of the sensor system cannot be 100 percent. From time to time, certain sensors in a distributed system fail. For this reason, it is very useful to emulate their readings based on the readings of the remaining sensors. The work describes a data set and proposes a machine learning model, which, based on the readings of functional sensors and weather conditions at the data collection sites, simulates the readings of a failed sensor. The accuracy of such emulation for certain types of pollution has been assessed (the coefficient of determination ranges from 0.43 to 0.61).

Deep learning insights into non-universality in the halo mass function

ABSTRACT The abundance of dark matter haloes is a key cosmological probe in forthcoming galaxy surveys. The theoretical understanding of the halo mass function (HMF) is limited by our incomplete knowledge of the origin of non-universality and its cosmological parameter dependence. We present a deep-learning model which compresses the linear matter power spectrum into three independent factors which are necessary and sufficient to describe the $z=0$ HMF from the state-of-the-art Aemulus emulator to sub-per cent accuracy in a wCDM$+N_\mathrm{eff}$ parameter space. Additional information about growth history does not improve the accuracy of HMF predictions if the matter power spectrum is already provided as input, because required aspects of the former can be inferred from the latter. The three factors carry information about the universal and non-universal aspects of the HMF, which we interrogate via the information-theoretic measure of mutual information. We find that non-universality is captured by recent growth history after matter-dark-energy equality and $N_{\rm eff}$ for $M\sim 10^{13} \, \mathrm{M_\odot }\, h^{-1}$ haloes, and by $\Omega _{\rm m}$ for $M\sim 10^{15} \, \mathrm{M_\odot }\, h^{-1}$. The compact representation learnt by our model can inform the design of emulator training sets to achieve high emulator accuracy with fewer simulations.

Protocol optimization for functional cardiac CT imaging using noise emulation in the raw data domain.

Four-dimensional (4D) wide coverage computed tomography (CT) is an effective imaging modality for measuring the mechanical function of the myocardium. However, repeated CT measurement across a number of heartbeats is still a concern. A projection-domain noise emulation method is presented to generate accurate low-dose (mA modulated) 4D cardiac CT scans from high-dose scans, enabling protocol optimization to deliver sufficient image quality for functional cardiac analysis while using a dose level that is as low as reasonably achievable (ALARA). Given a targeted low-dose mA modulation curve, the proposed noise emulation method injects both quantum and electronic noise of proper magnitude and correlation to the high-dose data in projection domain. A spatially varying (i.e., channel-dependent) detector gain term as well as its calibration method were proposed to further improve the noise emulation accuracy. To determine the ALARA dose threshold, a straightforward projection domain image quality (IQ) metric was proposed that is based on the number of projection rays that do not fall under the non-linear region of the detector response. Experiments were performed to validate the noise emulation method with both phantom and clinical data in terms of visual similarity, contrast-to-noise ratio (CNR), and noise-power spectrum (NPS). For both phantom and clinical data, the low-dose emulated images exhibited similar noise magnitude (CNR difference within 2%), artifacts, and texture to that of the real low-dose images. The proposed channel-dependent detector gain term resulted in additional increase in emulation accuracy. Using the proposed IQ metric, recommended kVp and mA settings were calculated for low dose 4D Cardiac CT acquisitions for patients of different sizes. A detailed method to estimate system-dependent parameters for a raw-data based low dose emulation framework was described. The method produced realistic noise levels, artifacts, and texture with phantom and clinical studies. The proposed low-dose emulation method can be used to prospectively select patient-specific minimal-dose protocols for functional cardiac CT.

A Novel Control Scheme for the Electric Motor Emulator to Improve the Voltage Emulation Accuracy

A Novel Control Scheme for the Electric Motor Emulator to Improve the Voltage Emulation Accuracy

Fast production of cosmological emulators in modified gravity: the matter power spectrum

We test the convergence of fast simulations based on the COmoving Lagrangian Acceleration (COLA) method for predictions of the matter power spectrum, specialising our analysis in the redshift range 1 ≤ z ≤ 1.65, relevant to high-redshift spectroscopic galaxy surveys. We then focus on the enhancement of the matter power spectrum in modified gravity (MG), the boost factor, using the Dvali-Gabadadze-Porrati (DGP) theory as a test case but developing a general approach that can be applied to other MG theories. After identifying the minimal simulation requirements for accurate DGP boost factors, we design and produce a COLA simulation suite that we use to train a neural network emulator for the DGP boost factor. Using MG-AREPO simulations as a reference, we estimate the emulator accuracy to be of ∼ 3% up to k = 5 h Mpc-1 at 0 ≤ z ≤ 2. We make the emulator publicly available at: https://github.com/BartolomeoF/nDGPemu.

Development of surrogate-optimization models for a novel transcritical power cycle integrated with a small modular reactor

In recent years, various types of surrogate optimization models have been proposed to reduce the computational time and to improve the emulation accuracy. In this study, by leveraging an ANN surrogate model developed earlier, a comprehensive and efficient optimization algorithm is conceived for the global optimal design of an integrated regenerative methanol transcritical cycle. It combines a unique converging/diverging classifier model into the surrogate model to form a surrogate-based model, which significantly improves the prediction accuracy of the objective function. Six binary classifiers are explored and the multi-layer feed-forward (MLF) neural network classifier is selected. In addition, within the five global optimizers being explored, the basin-hopping (BH) and dual-annealing (DA) are selected. The optimal surrogate-based model and global optimizers are then combined to form a unique surrogate-optimizer model. The surrogate-optimizer model is slightly outperformed by the physics-based model in terms of the optimization results, the time consumption of the surrogate-optimizer model during the optimization searching process is 99% less than that of the physics-based model. As the results, the surrogate-optimizer model is slightly outperformed by the physics-based model in terms of the optimization results, where the Levelized Cost of Energy (LCOE) of the Surrogate-DA and Surrogate-BH models are 77.912 and 78.876 $/MWh, respectively, compared to the 77.190 $/MWh of the Baseline model with fairly close penalties between them. In the meantime, the time consumption of the surrogate-optimizer model during the optimization searching process is 99% less than that of the physics-based model.

MF-Box: multifidelity and multiscale emulation for the matter power spectrum

ABSTRACT We introduce MF-Box, an extended version of MFEmulator, designed as a fast surrogate for power spectra, trained using N-body simulation suites from various box sizes and particle loads. To demonstrate MF-Box’s effectiveness, we design simulation suites that include low-fidelity (LF) suites (L1 and L2) at 256 and $100 \, \rm {Mpc\, ~}h^{-1}$, each with 1283 particles, and a high-fidelity (HF) suite with 5123 particles at $256 \, \rm {Mpc\, ~}h^{-1}$, representing a higher particle load compared to the LF suites. MF-Box acts as a probabilistic resolution correction function, learning most of the cosmological dependencies from L1 and L2 simulations and rectifying resolution differences with just three HF simulations using a Gaussian process. MF-Box successfully emulates power spectra from our HF testing set with a relative error of $\lt 3~{{\ \rm per\ cent}}$ up to $k \simeq 7 \, h\rm {Mpc}{^{-1}}$ at z ∈ [0, 3], while maintaining a cost similar to our previous multifidelity approach, which was accurate only up to z = 1. The addition of an extra LF node in a smaller box significantly improves emulation accuracy for MF-Box at $k \gt 2 \, h\rm {Mpc}{^{-1}}$, increasing it by a factor of 10. We conduct an error analysis of MF-Box based on computational budget, providing guidance for optimizing budget allocation per fidelity node. Our proposed MF-Box enables future surveys to efficiently combine simulation suites of varying quality, effectively expanding the range of emulation capabilities while ensuring cost efficiency.

Emulation of Integrated Starter/Generator Using Power Electronic Devices

Integrated starter/generator (ISG) is an important facility in transportation power systems. Using power electronics based motor emulator (ME) to test the motor controller is an emerging technique to optimize the design process of the motor drive system. In this paper, an ME is designed to test the control system of ISG in the full operation process. Facing the challenge of dual-mode emulation, a voltage-in-voltage-out interface is proposed to handle the causality problem caused by operation mode change during the starting/generating process. The LCL interface with predictive voltage closed-loop control is proposed to ensure emulation accuracy. A prototype of ME has been developed. The experiment has been done to validate the performance of the emulator, showing that the proposed method can be used to test the ISG controller in full operational mode with accuracy and fast dynamics.

Real-Time-Shift: Pseudo-Real-Time Event Scheduling for the Split-Protocol-Stack Radio-in-the-Loop Emulation

The incorporation of real radio hardware and physical emulated radio links into higher layer network and protocol simulation studies has been a largely untouched area of research so far. The Split-Protocol-Stack Radio-in-the-Loop emulation combines pure discrete-event protocol simulation with a hardware-based radio link emulation. Since the basic techniques involve contrary time concepts, event communication between the two domains requires a rethink of scheduling and synchronization. With the Real-Time-Shift conservative synchronization and time compensation scheme, the simulator is decoupled from real-time constraints and limitations by introducing predetermined pause times for event execution. In this work, we present the core synchronization and event scheduling approach allowing for scalable pseudo-real-time simulations with radio hardware in the loop. This enables discrete-event simulations for wireless host systems and networks with link-level emulation accuracy, accompanied by an overall high modeling flexibility.

PQ4wind - a novel test bench for component-level wind turbine converter testing

The Fraunhofer-Institute for Wind Energy Systems IWES is developing a new multi-megawatt testing facility for low voltage wind turbine converters. The new facility focuses on examining the frequency converters' electrical behaviour in a wide frequency range from fundamental frequency up to the 200th harmonic. This paper presents an overview of the project, the design principles of the facility and select preliminary simulations of its grid emulation capabilities. The simulations comprise of a demonstration of the grid emulator's voltage control and the emulation of both linear and non-linear frequency-dependent grid impedances. The simulations exhibit promising results of the new testing facility regarding accuracy of emulation and usable frequency range.

Robust deep neural network surrogate models with uncertainty quantification via adversarial training

AbstractSurrogate models have been used to emulate mathematical simulators of physical or biological processes for computational efficiency. High‐speed simulation is crucial for conducting uncertainty quantification (UQ) when the simulation must repeat over many randomly sampled input points (aka the Monte Carlo method). A simulator can be so computationally intensive that UQ is only feasible with a surrogate model. Recently, deep neural network (DNN) surrogate models have gained popularity for their state‐of‐the‐art emulation accuracy. However, it is well‐known that DNN is prone to severe errors when input data are perturbed in particular ways, the very phenomenon which has inspired great interest in adversarial training. In the case of surrogate models, the concern is less about a deliberate attack exploiting the vulnerability of a DNN but more of the high sensitivity of its accuracy to input directions, an issue largely ignored by researchers using emulation models. In this paper, we show the severity of this issue through empirical studies and hypothesis testing. Furthermore, we adopt methods in adversarial training to enhance the robustness of DNN surrogate models. Experiments demonstrate that our approaches significantly improve the robustness of the surrogate models without compromising emulation accuracy.

Improving the feel of 3D printed prototypes for new product development: A feasibility study of emulating mass properties by optimising infill structures and materials

Abstract Product prototypes and particularly those that are 3D printed will have mass properties that are significantly different from the product they represent. This affects both functional performance and stakeholder perception of the prototype. Within this work, computational emulation of mass properties for a primitive object (a cube) is considered, developing a baseline numerical method and parameter set with the aim of demonstrating the means of improving feel in 3D printed prototypes. The method is then applied and tuned for three case study products – a games controller, a hand drill and a laser pointer – demonstrating that product mass properties could be numerically emulated to within ~1% of the target values. This was achieved using typical material extrusion technology with no physical or process modification. It was observed that emulation accuracy is dependent on the relative offset of the centre of mass from the geometric centre. A sensitivity analysis is further undertaken to demonstrate that product-specific parameters can be beneficial. With tuning of these values, and with some neglect of practical limitations, emulation accuracy as high as ~99.8% can be achieved. This was shown to be a reduction in error of up to 99.6% relative to a conventional fabrication.

Towards 1% accurate galaxy cluster masses: including baryons in weak-lensing mass inference

Galaxy clusters are a promising probe of late-time structure growth, but constraints on cosmology from cluster abundances are currently limited by systematics in their inferred masses. One unmitigated systematic effect in weak-lensing mass inference is ignoring the presence of baryons and treating the entire cluster as a dark matter halo. In this work we present a new flexible model for cluster densities that captures both the baryonic and dark matter profiles, a new general technique for calculating the lensing signal of an arbitrary density profile, and a methodology for stacking those lensing signal to appropriately model stacked weak-lensing measurements of galaxy cluster catalogues. We test this model on 1400 simulated clusters. Similarly to previous studies, we find that a dark matter-only model overestimates the average mass by 7.5%, but including our baryonic term reduces that to 0.7%. Since two more variables are marginalized over when we include our baryonic term the posteriors on the cluster mass calibration are larger than the dark matter-only model. Additionally, to mitigate the computational complexity of our model, we construct an emulator (surrogate model) which accurately interpolates our model for parameter inference, while being much faster to use than the raw model. We also provide an open-source software framework for our model and emulator, called maszcal, which will serve as a platform for continued efforts to improve these mass-calibration techniques. In this work, we detail our model, the construction of the emulator, and the tests which we used to validate that our model does mitigate bias. Lastly, we describe tests of the emulator's accuracy.

Forge: the f(R)-gravity cosmic emulator project – I. Introduction and matter power spectrum emulator

ABSTRACT We present a large suite of cosmological simulations, the forge (F-of-R Gravity Emulator) simulation suite, which is designed to build accurate emulators for cosmological observables in galaxy clustering, weak gravitational lensing, and galaxy clusters for the f(R)-gravity model. A total of 200 simulations explore the cosmological parameter space around a standard Planck cosmology with a Latin hypercube, for 50 combinations of $\bar{f}_{R0}$, Ωm, σ8, and h with all other parameters fixed. For each parameter combination, or node, we ran four independent simulations, one pair using 10243 particles in $500\, h^{-1}\, \mathrm{Mpc}$ simulation boxes to cover small scales, and another pair using 5123 simulation particles in $1.5\, h^{-1}\, \mathrm{Gpc}$ boxes for larger scales. Each pair of initial conditions is selected such that sample variance on large scales is minimized on average. In this work we present an accurate emulator for the matter power spectrum in f(R) gravity trained on forge. We have verified, using the cross-validation technique, that the emulator accuracy is better than $2.5{{\, \rm per\, cent}}$ for the majority of nodes, particularly around the centre of the explored parameter space, up to scales of $k = 10\, h \, \mathrm{Mpc}^{-1}$. We have also checked the power spectrum emulator against simulations that are not part of our training set and found excellent agreement. Due to its high accuracy on small scales, the forge matter power spectrum emulator is well suited for weak-lensing analysis and can play a key tool in constraining f(R) gravity using current and future observational data.

On Efficient Non-Stationary Channel Emulation in Conductive Phase Matrix Setup for Massive MIMO Performance Testing

Massive multiple-input multiple-output (MIMO) is considered as a critical technique in 5 G New Radio (NR), due to its capability to significantly improve data rate and link reliability for the radio link. To make it a reality, it is essential to evaluate massive MIMO performance in realistic spatial fading channel conditions. Specifically, dynamic spatial channel emulation is crucial to evaluate the massive MIMO beam operations. Due to the high cost associated with both the traditional conduced testing method and the standardized multi-probe anechoic chamber (MPAC) based Over-the-Air (OTA) testing setup for massive MIMO performance testing, fading channel emulation for massive MIMO testing using a conductive phase matrix setup has been proposed in the literature. However, it is still a challenging problem to emulate dynamic spatial channels in a cost-effective and efficient manner for the conductive phase matrix setup, since traditional algorithms suffer from high computational complexity and system cost. This paper proposes a fast and accurate framework to emulate dynamic spatial channels in the phase matrix setup, where the basic idea is to mimic a virtual MPAC setup. By doing so, we can mimic MPAC setup with flexible probe configurations. Furthermore, a novel successive cancellation strategy and a closed-form algorithm are proposed to determine the probe locations and probe weights, respectively, which can significantly reduce the computational complexity associated with dynamic channel emulation. Simulation results show that the proposed algorithm is highly accurate and computationally efficient, compared to state-of-art solutions. Further, we apply our developed algorithm to emulate measured non-stationary channels and excellent emulation accuracy can be achieved. The proposed method can be employed for emulating dynamic spatial channels, which is of great importance for massive MIMO performance testing.

Inferring data model from service interactions for response generation in service virtualization

Context: Service virtualization has become a popular tool to provide testing environments for highly connected enterprise software systems. It enables the enterprise system under test to interact with and obtain responses from model-based service emulations instead of the actual services they use in production environments, providing accessibility and realness. Existing approaches consider only the control dependencies between messages (i.e., the service’s control model) and do not consider the relationships between data values of the messages (i.e., the service’s data model), limiting the accuracy of service emulation.

Improved Accuracy of Power Electronic-Based Motor Emulation With Compensation for Signal Transmission Delay

The power electronic based motor emulator (ME) has shown the prospect of replacing the mechanical back-to-back dynamometers with a power converter to validate the performance of the inverter under test (IUT). In real application of the ME to test the IUT, the controller of the IUT is independent from the ME, so that there’re time delays caused by the input voltage sampling of ME, ME calculation and the position signal transmission. Such a delay will cause cross coupling between dq axis which severely influence the emulation accuracy, especially under higher emulated motor speed. A compensation method is proposed to eliminate the cross coupling. A prototype is designed and simulation and experiment results are provided and compared to that of the normal motor emulator to verify the compensation effect. The experimental results have shown that the proposed compensation method can improve the emulation accuracy and a wider speed range of the motor can be emulated accurately.

A Parallel Topology Based Power-Stage PMSM Emulator With Accurate Ripple Current Reproduced

As a power hardware-in-the-loop (PHIL) technology, the power-stage motor emulator (ME) can be employed for the port characteristics emulation of motors. Meanwhile, the ripple current reproduction is one of the key technologies for achieving the high precision emulation of the port characteristics. Commonly, the high requirement for the hardware topology and the modulation strategy with rapid response and low harmonics are the main challenges. Aiming to obtain the precision ripple current, a ME scheme with dual-branch parallel topology is proposed in this paper, of which the equivalent switching frequency can be increased greatly by mean of interleaving switch. Furthermore, a new three-level space vector modulation strategy based on the idea of virtual transformation is presented. With the proposed strategy, the dual-branch parallel topology can be equivalently transformed into a specific three-phase three-level back-to-back topology. Finally, simulation and bench experiments have been conducted to validate the effectiveness of the proposed scheme. The results indicate that the proposed topology combined with the modulation strategy has a superior performance in improving the ripple current tracking accuracy of ME. Moreover, the proposed method can provide an excellent insight into improving the emulation accuracy of motor characteristics in practical applications.

Multifidelity emulation for the matter power spectrum using Gaussian processes

ABSTRACT We present methods for emulating the matter power spectrum by combining information from cosmological N-body simulations at different resolutions. An emulator allows estimation of simulation output by interpolating across the parameter space of a limited number of simulations. We present the first implementation in cosmology of multifidelity emulation, where many low-resolution simulations are combined with a few high-resolution simulations to achieve an increased emulation accuracy. The power spectrum’s dependence on cosmology is learned from the low-resolution simulations, which are in turn calibrated using high-resolution simulations. We show that our multifidelity emulator predicts high-fidelity (HF) counterparts to percent-level relative accuracy when using only three HF simulations and outperforms a single-fidelity emulator that uses 11 simulations, although we do not attempt to produce a converged emulator with high absolute accuracy. With a fixed number of HF training simulations, we show that our multifidelity emulator is ≃100 times better than a single-fidelity emulator at $k \le 2 \, h\textrm {Mpc}{^{-1}}$, and ≃20 times better at $3 \le k \lt 6.4 \, h\textrm {Mpc}{^{-1}}$. Multifidelity emulation is fast to train, using only a simple modification to standard Gaussian processes. Our proposed emulator shows a new way to predict non-linear scales by fusing simulations from different fidelities.

WiNE-Tap: Wireless LAN emulator with wireless network TAP devices

Conventional real-time wireless network emulators interact with physical network systems using wired LAN (Ethernet) virtualization technology, such as a TUN/TAP device, but they lack the capability of emulating the low-layer (data link sublayer and below) behavior of wireless network systems. For instance, a wireless device in ETSI ITS-G5 vehicular network systems adaptively changes its transmission power depending on the received signal strength, but existing wireless network emulators are incapable of simulating such behavior because they have no means of conveying transmission power information to the surrounding wireless devices in a virtual space. In addition, the implementations of existing network emulators vary even if they support the same network protocols and network configuration, and they rarely support multiple network simulators. The implementation of the network emulators limits the usability of network emulation because network simulators differ widely in the accuracy of their results. This paper proposes a wireless network emulation module called WiNE-Tap, which is implemented as a virtual wireless network device, to address these issues. WiNE-Tap runs in the Linux kernel space, allowing the underlying Linux system and a network simulator to exchange Linux kernel events and simulation events, such as frame transmission/reception and transmission power/RSSI changes, transparently to the user. This implementation allows the emulation of the adaptive behavior of wireless network low layers, as well as the use of any wireless network simulator. In this paper, we evacuate the performance of WiNE-Tap using Linux implementation with consideration of the simulation of radio propagation and network modes’ mobility. Some techniques for improving the timing accuracy of wireless network emulation are also described.