Forward Calculation Research Articles

Dynamic Modeling and Model-Based Control with Neural Network-Based Compensation of a Five Degrees-of-Freedom Parallel Mechanism

In this paper, a spatial parallel mechanism with five degrees of freedom is studied in order to provide a promising dynamic model for the control design. According to the inverse kinematics of the mechanism, the dynamic model is derived by using the Lagrangian method, and the co-simulation using MSC ADAMS and MATLAB/Simulink is adopted to verify the established dynamic model. Then the pre-trained deep neural network (DNN) is introduced to predict the real-time state of the end-effector of the mechanism. Compared to the traditional Newton’s method, the DNN method reduces the cost of the forward kinematics calculation while ensuring prediction accuracy, which enables the dynamic compensation based on feedback signals. Furthermore, the computed torque control with DNN-based feedback compensation is implemented for the trajectory tracking of the mechanism. The simulations show that, in the most complicated case that involves friction and external disturbance, the proposed controller has better tracking performance. The results indicate the necessity of dynamic modeling in the design of control with high precision.

High-precision 3D reconstruction of multiple magnetic targets based on center weighting method

A high-precision three-dimensional reconstruction method of multiple magnetic targets is studied in this paper. The inversion space is divided into cubic grid cells, and various complex shapes of magnetic targets are simulated by the combination of grid cells. The forward calculation of the magnetic gradient tensor in the spatial domain is performed. Then, the correlation imaging method is used to calculate cross correlation between the observed magnetic gradient tensor and the theoretical magnetic gradient at each grid cell tensor due to a magnetic dipole. The spatial distribution and center position of magnetic targets are preliminarily estimated by the resultant correlation coefficients, and use them to construct the center weighting function as a constraint to the inversion equations. The solution is completed by using the conjugate gradient method to estimate the magnetic susceptibility of each grid cell, and the 3D outline of the magnetic target is depicted by using the visualization technique. The experimental results show that the proposed method can achieve the 3D reconstruction of multiple simple magnetic targets with high precision.

Wavefield distortion imaging of Earth's deep mantle

The seismic wavefield, as recorded at the surface, carries information about the seismic source and Earth's structure along the seismic path, essential for the understanding of the interior of our planet. For 40 years seismic tomography studies have resolved the 3D seismic velocity structure in growing detail using seismic traveltimes and waveforms. These studies have been driving our understanding of the dynamics and evolution of the planet, but are limited in their spatial resolution to imaging scales of a few 100 s to 1000 km due to the constraints of the tomographic inversion. Detailed studies of seismic waveforms can resolve finer scale structure but are often reliant on serendipitous source-receiver combinations and provide very uneven coverage of the planet. Therefore, we often lack an understanding of the fine scale structure of the Earth that is important to understand structures and processes such as mantle plumes or details of slab recycling. Here we show evidence that we can exploit slowness vector deviations of the seismic wavefield to extend our knowledge of Earth structure to smaller scales using large datasets. Analysing seismic array data, we show strong and measurable focussing and defocussing effects of the teleseismic P and Pdiff wavefield sampling the deep Earth. We compare the P-wave results to additional S and Sdiff data and find good agreement between both wavetypes. We can link the wavefield deviations to strong velocity variations assuming sharp boundaries are sampled along the path in the deep mantle. The dataset samples the Pacific and Gulf of Mexico well and shows strong horizontal incidence (backazimuth) deviations in the Pacific (up to 14° westwards) and beneath the Gulf of Mexico (up to 5 to 8° east- and west-ward). The backazimuth deviations are also reflected in slowness deviations in the range of ± 0.8 s/° relating to velocity variations in the range of ± 9 km/s. Using 3D raytracing we are able to forward model the detected backazimuth variations of the P and Pdiff dataset. The high frequencies of the P-waves, density of the ray-paths, and low computational cost of our forward calculation allow us to construct a higher resolution and more detailed model of velocity anomalies under Hawaii than was possible with previous methods. The best-fitting velocity model for the Pacific contains two low-velocity regions located at N25°/W155° and N25°/W165° beneath the tip of the Hawaii Emperor chain. The Pacific anomalies have diameters (D) of 6° and 2° with velocity reductions (dVP) of 8% and 4% with heights (H) above the CMB of 70 km and at least 200 km, respectively. We also detect a fast region of 3% velocity increase in the North Pacific rising at least 300 km above the CMB with a diameter of 12° at N60°/W175°. Beneath the Gulf of Mexico we find ambiguous results with either a slow region (N25°/W85°, H = 200 km, dVP=-3%, D = 2°) or a fast region (N15°/W75°, H = 200 km, dVP = 3%, D = 4°) able to explain the data. We thus show that the directivity information of the seismic wavefield - largely underexploited - can be used to resolve the fine scale velocity structure of the Earth's interior with great accuracy and can deliver additional insight into deep Earth dynamics.

Shear wave velocity structure at the Fukushima forearc region based on H/V analysis of ambient noise recordings by ocean bottom seismometers

SUMMARYThis study presents the shear wave velocity (VS) structures of sedimentary sequences and a section of the upper crustal layer in the Fukushima forearc region of the Japan Trench subduction zone, which were obtained by analysing the horizontal-to-vertical (H/V) spectral ratios of ambient vibration records. The H/V curves were derived using 31 d of continuous seismic data from 3 broad-band and 16 short-period ocean bottom seismometer (OBS) stations. Using the broad-band data, H/V ratios from 0.01 to 10 Hz were derived, but the ratios below 0.1 Hz frequencies were unusually large and temporally unstable. Characterization of seismic noise energy from ∼1 yr of seismic data of three broad-band OBSs revealed variable and elevated energy conditions below 0.1 Hz due to typical long-period oceanic noise; we link these observations with the unstable H/V ratios below this frequency. Therefore, H/V analysis was performed in the frequency range of 0.1–10 Hz for both broad-band and short-period OBSs to obtain subsurface VS profiles. For the forward calculation of the H/V ratios in the inversion process, we used the recently developed ‘hvgeneralized’ method, which is based on the diffuse field assumption, and accounts for the water layer on top of stratified media. Moreover, available prior geological and geophysical information was utilized during the inversion of the H/V curves. We found that subsurface VS ranged from approximately 30 m s−1 at the seabed to approximately 4900 m s−1 at 7000 m below the sea floor (mbsf). Starting with the best model candidate at each OBS location, the effect of the water layer on the H/V curve in the deep ocean was investigated by comparing synthetic H/V curves with and without the water layer. The synthetic H/V analysis revealed that the water layer had a significant effect on H/V amplitudes at higher frequencies (>1 Hz), whereas comparatively little effect was observed at lower frequencies (<1 Hz). This study provides an empirical basis for H/V analysis using OBS data to determine VS down to several kilometres of sedimentary sequences to the upper crust with high-resolution.

New Insights into the Anomaly Genesis of the Frequency Selection Method: Supported by Numerical Modeling and Case Studies

The frequency selection method (FSM) is the further development of the audio frequency telluric electricity method (TEFM). However, debate continues regarding the mechanisms leading to anomaly genesis. Therefore, the present study intends to explore this using 2D forward modeling of magnetotelluric (MT) sounding and practical applications of FSM on three Chinese case studies in karst and granitic settings. In the first stage, the profile curves and pseudo-section of apparent resistivity ( $${\rho }_{s}$$ ) and horizontal electric field component $${(E}_{y})$$ in MT mode are obtained by forward calculation. As a result, the static shift in $${\rho }_{s}$$ is observed over the near-surface inhomogeneities, as documented in the literature. Additionally, the profile curves of $${E}_{y}$$ showed an obvious static shift in the rectangular coordinate system (i.e., the curve rises with the increase in frequency) which is a well-known phenomenon. The pseudo-sections of $${E}_{y}$$ also showed static shift characteristics at the horizontal position above the anomaly, referred to as the “noodles phenomenon.” The FSM results were obtained from case studies related to groundwater and low-resistivity clay-filled karst body identification. The $$\Delta V$$ section curves and pseudo-section showed a significant low potential, and a “noodles phenomenon,” respectively, above the low-resistivity anomalous body. These abnormal characteristics of $$\Delta V$$ are the basis for delineating the horizontal position of the groundwater aquifer by applying FSM. It is concluded that the anomaly of FSM is the reflection of the static shift in MT, and hence, the FSM can be categorized as a “static shift method.” Therefore, this inspired us to investigate whether the static shift feature of the surface $${E}_{y}$$ component could be utilized to explore near-surface geological bodies such as clay-filled or water-filled cavities.

Learning to multi-vehicle cooperative bin packing problem via sequence-to-sequence policy network with deep reinforcement learning model

In the logistics bin packing scenario with only rear bin doors, the packing sequence of items determines the utilization of vehicle packing space, but there is relatively little research on optimizing the packing sequence of items. Therefore, this article focuses on the bin packing sequence problem in the multi-vehicle cooperative bin packing problem(MVCBPP) and proposes a deep reinforcement learning model based on the sequence-to-sequence policy network with deep reinforcement learning model(S2SDRL). Firstly, the sequence-to-sequence neural networks model is constructed, which determines the packing probability of all items. The items will be packed by combining the bidirectional LSTM model and the attention module to construct the encoder and decoder. Secondly, the bin packing strategy of the items is obtained by the constructed reinforcement learning packing framework. Finally, the Seq2Seq policy network is updated and optimized by the policy gradient method with a baseline to obtain the current optimal packing strategy. In several bin packing scenarios, S2SDRL improves the average vehicle space utilization by more than 4.0% compared with the traditional packing algorithm, and the forward computation time of the model is much smaller than that of the traditional heuristic algorithm, so the model also has more realistic application value. Ablation experiments also confirm the effectiveness of the modules in the S2SDRL model for optimization of the packing order. The sensitivity analysis shows the model's some stability when the input data changes.

Improving Accuracy of Horizontal Flow Field Using Acoustic Tomography With Real-Time Station Position Correction

High-precision and spatiotemporal flow field measurement in small-scale waters has a decisive impact on navigation and water resource management. Multitransmission rays and position drift, as two main difficulties in improving the accuracy of small-scale flow field, need to be solved urgently. In this article, four coastal acoustic tomography (CAT) systems are arranged in Qingyang Lake, Hunan, China, on March 1, 2022, for a short-term observation. As the GPS error and shipping caused station drift contaminate the reconstructed flow field, a real-time station position correction method, based on travel time and temperature, depth (TD) sensors, is applied in our inversion. The inversion range average current (RAC) is in agreement with forward calculation, which proves the reliability of the reconstructed horizontal flow field. Besides, comparisons of net flow and acoustic Doppler current profiler (ADCP) velocity are performed to validate the improvement of the real-time station position correction algorithm. The errors of the corrected net flow are less than 3.818 m 2/s, with a minimum of 0.0042 <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\text{m}^{2}$ </tex-math></inline-formula> /s. The root-mean-square error (RMSE) between the CAT inversion results and the ADCP velocity is 0.123 m/s after correction, relatively reduced 5.95% than before correction. The smaller the RMSE, the higher the inversion accuracy, so the CAT method with real-time station position correction is feasible for high-precision horizontal flow field reconstruction.

High-Dimensional Uncertainty Quantification in Electrical Impedance Tomography Forward Problem Based on Deep Neural Network

In electrical impedance tomography (EIT), the uncertainty of conductivity distribution may cause the uncertainty in the forward calculation and further affect the inverse problem. In this paper, an improved univariate dimension reduction method based on deep neural network (DNN-UDR) is proposed for the high-dimensional uncertainty quantification in EIT forward problem. Firstly, DNN is studied to build a substitute model for EIT forward problem in order to solve the high-dimensional problem. Three normalized circular finite element models are established with random uniform conductivity distribution. Then UDR is used to analyze and quantify the uncertainty in the simulation with the form of probability. Compared with Monte Carlo simulation (MCS), the probability distribution of voltage is fitted, and the quantification indicators such as mean, variance, variation coefficient and covariance, are also consistent. On the other hand, with the increase of parameter dimensions, DNN-UDR accelerates the computations obviously. This indicates that DNN-UDR is effective and has high structural stability, accurate prediction results and high computational efficiency.

Fabrication, Testing, and 3D Comprehensive Analysis of Swept-Tip Tiltrotor Blades

This paper covers the design, fabrication, testing, and modeling of a family of Froude-scale tiltrotor blades. They are designed with the objective of gaining a fundamental understanding of the impact of a swept tip on tiltrotor whirl flutter. The goal of this paper is to describe the development of the blades needed for this purpose. The rotor is three bladed with a diameter of 4.75 ft. The blades have a VR-7 profile, chord of 3.15 inches, and linear twist of –37° per span. The swept-tip blades have a sweep of 20° starting at 80% R. The blade properties are loosely based on the XV-15 design. A CATIA and Cubit-based high-fidelity three-dimensional (3D) finite element model is developed. It accurately represents the fabricated blade and is analyzed with X3D. Experiments in a vacuum chamber were carried out to demonstrate the structural integrity of the blades. Measured frequencies and strains were validated with X3D predictions proving the fidelity of the 3D model. Thus, even though the wind tunnel facilities were closed due to COVID-19, hover and forward flight calculations for the blade stress could be performed using the high-fidelity 3D structural model. The results prove the blades have sufficient structural integrity and stress margins to allow for wind tunnel testing.

Research on Small-Loop Transient Electromagnetic Method Forward and Nonlinear Optimization Inversion Method

The small-loop transient electromagnetic (TEM) method has great application prospects in shallow surface geophysical exploration. However, the theoretical research on the forward and inversion of this method is obviously lagging behind, which seriously affects the application effect in practical engineering. For the oscillation problem of the small-loop TEM response calculated by using the forward theory of the conventional TEM methods, the dipole integral superposition method was proposed to solve the problem in this article. The transmitter source is divided into an infinite number of electric dipoles, and the calculation expressions of the electromagnetic field components satisfied by the electric dipoles are deduced, and the forward calculation problem of the layered model of the small-loop TEM method is solved. For the small-loop TEM method inversion problem, the simulated annealing (SA) method was presented for inversion calculation in this article. First, the influence of the selection of the initial temperature on the inversion result of the SA algorithm was analyzed. Then, we demonstrate the algorithm’s flexibility by choosing different initial models and search ranges and concluded that the different initial models and search ranges do not affect the inversion results of the SA algorithm. Next, the typical layered models were established, and the SA algorithm was used to carry out calculations. The research results show that the SA algorithm is an excellent global optimization algorithm, which can easily jump out of the local minima and converge to the global minima. Finally, we processed the measured data through inversion calculations and compared it with the results of the other three methods to verify the effectiveness and accuracy of the SA algorithm for small-loop TEM data processing. The research content solves the theoretical problem of the small-loop TEM method forward and inversion, promotes the development of the small-loop TEM method technology, and objectively and effectively solves practical engineering application problems.

Reducing Biases in Thermal Infrared Surface Radiance Calculations Over Global Oceans

Thermal infrared (IR) environmental satellite data assimilation and remote sensing of the surface and lower troposphere depend on accurate specification of the spectral surface emissivity within clear-sky forward calculations. Over ocean surfaces, accurate modeling of surface-leaving radiances over the sensor scanning swaths is complicated by a quasi-specular bidirectional reflectance distribution function (BRDF). Recent findings at the Joint Center for Satellite Data Assimilation (JCSDA) have also revealed significant zonally varying systematic biases (≈ |0.5| K) on a global scale over cold ocean waters, these the result of temperature dependence in the thermal IR optical constants. This paper proposes practical solutions to these problems by modeling thermal IR “effective emissivity” in a manner that accounts for both surface emission and quasi-specular reflectance, along with temperature dependence, while meeting the latency and computational constraints of operational global data assimilation and retrieval systems. We overview the theoretical basis of the model and validate it against ship-based Marine Atmospheric Emitted Radiance Interferometer (MAERI) spectra obtained from cold and warm water ocean campaigns.

Fast Recognition on Shallow Groundwater and Anomaly Analysis Using Frequency Selection Sounding Method

The validity of the frequency selection method (FSM) in shallow (&lt;150 m) groundwater exploration was illustrated by practical applications, and the relationship between potential electrode spacing MN and groundwater depth in FSM sounding method was analyzed and preliminary theoretical research was carried out by a simple geologic-geophysical model of sphere. Firstly, under the combined action of horizontal alternating electric field and alternating magnetic field, a simplified geophysical model of low resistivity conductive sphere in homogeneous half space was established, and the forward calculation was performed on the FSM sounding curve. Then, the water yield of 131 wells in the application of FSM in the Rural Drinking Water Safety Project of 12th Five-Year Plan in Guangxi Province was counted. In addition, detailed tabular statistical analysis was carried out on the drilling results of 98 drilling wells, and the relationship between potential electrode spacing MN at abnormal sounding curve and actual drilling water depth was compared and studied. Theoretical analysis and practical application show that FSM has obvious effectiveness in shallow groundwater exploration, and it is an effective method to determine shallow groundwater well locations in the future. The cause of FSM anomaly is the comprehensive effect of the natural 3D alternating electromagnetic signal underground. At the same time, the practical statistics show that there is 1:1 approximation between the size of potential electrode spacing MN at the anomaly curve of the frequency selection method and the actual drilling water depth, which verifies the correctness of the theoretical simulation results. FSM could be widely used in the shallow groundwater exploration in the future, and it is an effective, non-destructive, fast, and low-cost geophysical method.

Application of a global–local full‐waveform inversion of Rayleigh wave to estimate the near‐surface shear wave velocity model

AbstractWe present a global–local full‐waveform inversion (FWI) of surface waves to estimate a high‐resolution shear wave velocity model of the near‐surface at the site of Grenoble (France). The seismic data we use have been acquired in the framework of the InterPACIFIC project. A first attempt is made by employing the multichannel analysis of surface waves (MASW) to invert the fundamental modes of Rayleigh waves. This inversion is solved through a global optimization driven by the neighbourhood algorithm. However, the limiting 1D assumption, together with the severely ill‐posedness of the MASW inversion motivate us to exploit all the information content of the recorded Rayleigh waves (i.e., travel‐times and amplitudes) by implementing a global–local FWI approach. We first perform a global FWI in which a genetic algorithm (GA) is used to minimize the L2 norm difference between observed and simulated seismic waveforms up to 30 Hz. This inversion employs a two‐grid scheme in which the subsurface is described by a fine grid input to the forward modelling, whereas the unknowns are defined on a coarse grid. A bilinear interpolation brings the coarse‐grid model into the fine grid. To accelerate the convergence, two strategies are considered; offset marching and frequency marching. Then, the long‐wavelength velocity profile provided by the GA inversion is used as the starting point for a local FWI for further refinement of the predicted model. In this case, we employ a frequency‐marching strategy up to a maximum frequency of 60 Hz. Both the global and local approaches use a finite difference code for the forward computation. Despite the limited a priori information infused into the inversion framework, the presented global–local FWI scheme provides reliable results as demonstrated by the good match between recorded and predicted seismograms and by the agreement of the estimated velocity with both nearby borehole data and stratigraphic information on the study site. Our global–local strategy also achieves better data fitting and improved matching with the available borehole data with respect to the velocity profile estimated when the local FWI is started from the MASW predictions.

Constraining 3-D variations in mantle attenuation using normal modes: forward modelling and sensitivity tests

SUMMARY Robust estimates of the Earth’s temperature, composition and partial melt are difficult to obtain from seismic tomography models based only on wave velocities. Seismic attenuation is able to add crucial additional information and allows us to constrain 3-D variations in temperature as well as answer fundamental questions regarding the presence of water and partial melt within the mantle. A major problem with measuring attenuation is that scattering and focusing effects need to be included in order to distinguish between intrinsic attenuation (transformation of energy to heat) and scattering (redistribution of energy). Here, we will use whole Earth oscillations or normal modes, exploiting the fact that small scale scattering becomes less important at longer periods. In addition, focussing is implicitly included by simultaneously measuring the elastic and anelastic splitting function of a given normal mode. Normal mode data has rarely been used before to constrain 3-D variations in attenuation, hence we first need to assess the feasibility of using whole Earth oscillations to image 3-D variations in attenuation in Earth’s upper and lower mantle. Here, we use simple models in forward calculations to evaluate the sensitivity of normal modes to 3-D attenuation variations, and find observable differences in normal mode spectra when 3-D variations in mantle attenuation are included. We also test if 3-D attenuation variations can be recovered using splitting function measurements for synthetic normal mode spectra and investigate potential problems with unaccounted earthquake magnitude variations, noise and wide-band cross-coupling between modes. We find that we are able to recover input synthetic anelastic splitting functions, even when they are an order of magnitude smaller than the elastic splitting functions. Biases in the earthquake seismic moment solutions may lead to ‘ghost’ anelastic splitting functions being recovered, especially for high degree structure. However, this ‘ghost’ signal is only relevant if the size of the input anelastic splitting function is 50 per cent smaller than the scaled amplitude signal of expected upper mantle attenuation. Random noise and normal mode coupling also lead to spurious effects in the anelastic splitting functions. However, they manly affect high degree anelastic structure in a non-branch consistent manner. Thus, we are able to robustly recover low anelastic degree structure (smax ≤ 4) in all tested cases.

A machine learning approach to joint gravity and cosmic-ray muon inversion at Mt Usu, Japan

SUMMARY The ability to accurately and reliably obtain images of shallow subsurface anomalies within the Earth is important for hazard monitoring and a fundamental understanding of many geologic structures, such as volcanic edifices. In recent years, machine learning (ML) has gained increasing attention as a novel approach for addressing complex problems in the geosciences. Here we present an ML-based inversion method to integrate cosmic-ray muon and gravity data sets for shallow subsurface density imaging at a volcano. Starting with an ensemble of random density anomalies, we use physics-based forward calculations to find the corresponding set of expected gravity and muon attenuation observations. Given a large enough ensemble of synthetic density patterns and observations, the ML algorithm is trained to recognize the expected spatial relations within the synthetic input–output pairs, learning the inherent physical relationships between them. Once trained, the ML algorithm can then interpolate the best-fitting anomalous pattern given data that were not used in training, such as those obtained from field measurements. We test the validity of our ML algorithm using field data from the Showa-Shinzan lava dome (Mt Usu, Japan) and show that our model produces results consistent with those obtained using a more traditional Bayesian joint inversion. Our results are similar to the previously published inversion, and suggest that the Showa-Shinzan lava dome consists of a relatively high-density (2200–2400 km m–3) cylindrical anomaly, about 300 m in diameter. Adding noise to synthetic training and testing data sets shows that, as expected, the ML algorithm is most robust in areas of high sensitivity, as determined by the forward kernels. Overall, we discover that ML offers a viable alternate method to a Bayesian joint inversion when used with gravity and muon data sets for subsurface density imaging.

Back Analysis of Surrounding Rock Parameters of Large-Span Arch Cover Station Based on GP-DE Algorithm

Due to the characteristics of soil–rock composites and large-span arches, the surrounding rock parameters of stations are difficult to obtain accurately under soft upper and hard lower geological conditions when the arch cover method is used to carry out the construction of a large-span underground excavation station. To optimize the design of stations and guide the next step of construction, an intelligent inverse analysis method, the Gaussian process differential evolution co-optimization algorithm (GP-DE algorithm), is proposed for the arch cover method for station construction. Taking the Shikui Road station of the Dalian Metro Line Five as the engineering background, the finite element model of FLAC3D is established. By combining the measured data of the sensor and the monitoring data obtained using the orthogonal scheme, this algorithm is used for the joint back analysis of displacement stress and the accuracy of the inversion parameters is verified by forwarding the calculation for FLAC3D. By using the obtained surrounding rock parameters, the demolition length of the center diaphragm to the Shikui Road station is optimized. Under different numbers of training samples, the inversion effect of the GP-DE algorithm and the other three common back-analysis algorithms is compared and analyzed. Finally, based on the iteration rate and convergence effect, the value range of the differential evolution algorithm parameters F and CR is given. The results show that the forward calculation results of the parameters obtained from the back analysis are in good agreement with the actual values, and the accuracy of the back-analysis results is high.

Reconstruction algorithms for grain mapping by laboratory X-ray diffraction contrast tomography.

X-ray-based non-destructive 3D grain mapping techniques are well established at synchrotron facilities. To facilitate everyday access to grain mapping instruments, laboratory diffraction contrast tomography (LabDCT), using a laboratory-based conical polychromatic X-ray beam, has been developed and commercialized. Yet the currently available LabDCT grain reconstruction methods are either ill-suited for handling a large number of grains or require a commercial licence bound to a specific instrument. To promote the availability of LabDCT, grain reconstruction methods have been developed with multiple reconstruction algorithms based on both forward and back calculations. The different algorithms are presented in detail and their efficient implementation using parallel computing is described. The performance of different reconstruction methods is assessed on synthetic data. The code to implement all the described algorithms has been made publicly accessible with the intention of fostering the development of grain mapping techniques on widely available laboratory instruments.

A deep learning-based network for the simulation of airborne electromagnetic responses

SUMMARYAirborne electromagnetic (AEM) method detects the subsurface electrical resistivity structure by inverting the measured electromagnetic field. AEM data inversion is extremely time-consuming when huge volumes of observational data are involved. Forward modelling is an essential part and represents a large proportion of computational cost in the inversion process. In this study, we develop an AEM simulator using deep learning as a computationally efficient alternative to accelerate 1-D forward modelling. Inspired by Google's neural machine translation, our AEM simulator adopts the long short-term memory (LSTM) modules with an encoder–decoder structure, combining the advantages in time-series regression and feature extraction. The well-trained LSTM network describes directly the mapping relationship between resistivity models with transceiver altitudes and time-domain AEM signals. The prediction results of the test set show that 95 per cent of the relative errors at most sampling points fall in the range of ±5 per cent, with average values within the range of ±0.5 per cent, indicating an overall prediction accuracy. We investigate the effects of the distributions of both resistivity and transceiver altitude in the training set on the prediction accuracy. The LSTM-based AEM simulator can effectively handle the resistivity characteristics involved in the training set and yields great sensitivity to the variations of transceiver altitudes. We also examine the adaptability of our AEM simulator for discontinuous resistivity variations. Synthetic tests indicate that the application effect of the AEM simulator relies on the completeness of the training samples and suggest that enriching the sample diversity is necessary to ensure the prediction accuracy, in cases of observation environments dominated by extreme transceiver altitudes or under-represented geological features. Furthermore, we discuss the influence of network configuration on its accuracy and computational efficiency. Our simulator can deliver ∼13 600 1-D forward modelling calculations within 1 s, which significantly improves the simulation efficiency of AEM data.

Robust deep learning-based forward dose calculations for VMAT on the 1.5T MR-linac

In this work we present a framework for robust deep learning-based VMAT forward dose calculations for the 1.5T MR-linac. A convolutional neural network was trained on the dose of individual multi-leaf-collimator VMAT segments and was used to predict the dose per segment for a set of MR-linac-deliverable VMAT test plans. The training set consisted of prostate, rectal, lung and esophageal tumour data. All patients were previously treated in our clinic with VMAT on a conventional linac. The clinical data were converted to an MR-linac environment prior to training. During training time, gantry and collimator angles were randomized for each training sample, while the multi-leaf-collimator shapes were rigidly shifted to ensure robust learning. A Monte Carlo dose engine was used for the generation of the ground truth data at 1% statistical uncertainty per control point. For a set of 17 MR-linac-deliverable VMAT test plans, generated on a research treatment planning system, our method predicted highly accurate dose distributions, reporting 99.7% ± 0.5% for the full plan prediction at the 3%/3 mm gamma criterion. Additional evaluation on previously unseen IMRT patients passed all clinical requirements resulting in 99.0% ± 0.6% for the 3%/3 mm analysis. The overall performance of our method makes it a promising plan validation solution for IMRT and VMAT workflows, robust to tumour anatomies and tissue density variations.

A Comprehensive Study of Local, Global, and Combined Optimization Methods on Synthetic Seismic Refraction and Direct Current Resistivity Data

Most geophysical inversions face the problem of non-uniqueness, which poses a challenge in the mapping and delineation of the subsurface anomalies. To tackle this challenge, a combined local and global optimization approach is considered for jointly inverting two-dimensional direct current resistivity (DCR) and seismic refraction (SR) data that aim to estimate the corresponding physical model parameters. In this combined approach, the output of the local optimization method is used to determine the search space and tuning parameters for the global optimization algorithm. The multi-objective genetic algorithm (non-dominated sorting genetic algorithm) was utilized to jointly optimize the objective functions of two different methods. Because the genetic algorithm is a population-based optimization method, it requires numerous forward calculations. To deal with the expected high computational cost associated with this approach, parallel computing was utilized for the forward function evaluations to reduce the run time of the entire process. The proposed approach was tested using synthetic two-dimensional resistivity and velocity models that had three different types of anomalies (dyke, positive, and combined positive and negative). The results showed an improvement in the anomaly delineation in the output of the combined local and global optimization method compared with the local optimization method. Additionally, similar synthetic models were tested using only the single objective global optimization algorithm (conventional global optimization), which showed promising anomaly delineation.