Simulation Speed Research Articles

Optimization algorithm of 500kVGIS circuit breaker electric field simulation based on GPU acceleration

It proposes an optimization approach for electric field simulation in 500kV gas-insulated switchgear (GIS) circuit breakers that use GPU acceleration. Accurate modelling of electric field distribution within GIS circuit breakers is critical for assuring the dependability and safety of high voltage power systems. However, the computational complexity of simulating these systems presents substantial obstacles to established approaches. The goal of this work is to improve the speed, efficiency, and scalability of electric field simulations by leveraging GPU parallel processing capabilities. They methodically build and refine the simulation algorithm, utilizing techniques such as parallel algorithm design, GPU architecture optimization, and efficient memory management. Through rigorous performance studies, they demonstrate significant increases in computational efficiency relative to CPU-based techniques, with an average speed up of 15x. Furthermore, the scalability of the GPU-accelerated method is evaluated, demonstrating near-linear scalability up to 8 GPUs. Validation against experimental observations and analytical models indicates that the simulation findings are accurate and robust. This study helps the progress of electric field simulation tools for GIS circuit breakers by providing insights into electric field behaviour and aiding design optimization and performance evaluation in high voltage power systems.

Improved Real‐Time Post‐Processing for Quantum Random Number Generators

AbstractRandomness extraction is a key problem in cryptography and theoretical computer science. With the recent rapid development of quantum cryptography, quantum‐proof randomness extraction has also been widely studied, addressing the security issues in the presence of a quantum adversary. In contrast with conventional quantum‐proof randomness extractors characterizing the input raw data as min‐entropy sources, it is found that the input raw data generated by a large class of trusted‐device quantum random number generators can be characterized as the so‐called reverse block source. This fact enables us to design improved extractors. Two novel quantum‐proof randomness extractors for reverse block sources that realize real‐time block‐wise extraction are proposed specifically. In comparison with the general min‐entropy randomness extractors, the designs achieve a significantly higher extraction speed and a longer output data length with the same seed length. In addition, they enjoy the property of online algorithms, which process the raw data on the fly without waiting for the entire input raw data to be available. These features make the design an adequate choice for the real‐time post‐processing of practical quantum random number generators. Applying the extractors to the raw data generated by a widely used quantum random number generator, a simulated extraction speed as high as 300 Gbps is achieved.

An improved social force model for improving pedestrian avoidance by reducing search size

In order to solve the slow simulation speed of the social force model in complex indoor scenes, an improved social force model is proposed to reduce the iteration scale of obstacles. The proposed model takes into account the factor of a person’s field of view (FOV) in indoor scenes, and the acting force between pedestrians and obstacles in the social force model is improved to find the optimal number of obstacle iterations without affecting the pedestrian movement law in the traditional model, so as to accelerate the simulation speed. The self-built simulation platform and real evacuation experiments are used to compare the performance indicators of traditional and optimized models and to analyze the relationship between scene complexity and machine computing efficiency. The results show that the optimized model takes less time to predict the evacuation path than traditional models, and as the number of initial pedestrians and obstacles increases, the proportion of machine computation time between the two models will gradually increase. Second, the CPU computation time of the optimized model increases gently, indicating better stability. Therefore, reducing the obstacle iteration scale based on the pedestrian’s FOV factor provides an effective method to solve the slow simulation speed of traditional models.

Bounding Volume Hierarchy-Assisted Fast SAR Image Simulation Based on Spatial Segmentation

In order to improve the simulation efficiency under the premise of ensuring the fidelity of synthetic aperture radar (SAR) simulation images, we propose a BVH-assisted fast SAR image simulation method based on spatial segmentation. The beam scanning model is established based on RD imaging geometric relation, and the bounding volume hierarchy (BVH) algorithm is used to assist in obtaining the time-varying latticed radiation and shadow areas within the radar beam, combining them with the real-time position of the sensors to complete the simulation of the electromagnetic (EM) wave transmission. The ray tracing algorithm is used to calculate the multiple backscatter fields of EM waves, including various material properties of the target surface. The SAR spatial traversal is adopted to spatially segment the latticed radiation area, and the compute unified device architecture (CUDA) kernel function is designed using the echo matrix cell method to make each cell of the target echo matrix as a subfield of the backscattering field, and the position of the echo matrix cell is traversed to obtain the target backscattering field. The target simulated echo is processed by the range Doppler (RD) imaging algorithm to obtain the SAR-simulated image. The simulation results show that compared with a CPU single-thread simulation, the simulation speed of the proposed method is significantly improved, and the SAR simulation image has high structural similarity with the real image, which fully verifies the effectiveness of the proposed method.

Wind-Induced Cooling Effects on Photovoltaic Panel Performance

This investigation evaluated the performance of a photovoltaic (PV) panel system under varied cooling speeds of a calibrated wind generator. The objectives encompassed the calibration of wind speed, integration of the wind generator with the PV panel system, monitoring the performance of the PV panel with wind-induced cooling, and analyzing overall performance under different wind generator settings. Parameters assessed included open circuit voltage, short circuit current, PV panel surface temperature, and power output. The tests were carried out both with and without wind cooling and under artificial and natural lighting conditions. Under artificial lighting conditions, the solar PV panel demonstrated suboptimal short circuit current compared to natural lighting conditions, leading to an overall decrease in power output. Moreover, the findings revealed a significant relationship between simulated wind speed and the overall performance of the PV panel, with notable variations in surface temperature, voltage level, current level, and power output. The statistical analysis supported these findings, with all dependent variables exhibiting statistically significant differences. The practical implications of these observations are pertinent to the design of more efficient and sustainable PV and PV-wind cooling systems. By comprehending the influence of wind on PV panel performance, system designers and operators can make informed decisions to maximize energy production and enhance the overall efficiency of the PV system.

A study on the high power microwave effects of PIN diode limiter based on deep learning algorithm.

PIN diodes, due to their simple structure and variable resistance characteristics under high-frequency high-power excitation, are often used in radar front-end as limiters to filter high power microwaves (HPM) to prevent its power from entering the internal circuit and causing damage. This paper carries out theoretical derivation and research on the HPM effects of PIN diodes, and then uses an optimized neural network algorithm to replace traditional physical modeling to calculate and predict two types of HPM limiting indicators of PIN diode limiters. We proposes a neural network model for each of the following two prediction scenarios: in the scenario of time-junction temperature curves under different HPM irradiation, the weighted mean squared error (MSE) between the predicted values from the test dataset and the simulated values is below 0.004. While in predicting PIN limiter's power limitation threshold, insertion loss, and maximum isolation under different HPM irradiation, the MSE of the test set prediction values and simulation values are all less than 0.03. The method proposed in this research, which applies an optimized neural network algorithm to replace traditional physical modeling algorithms for studying the high-power microwave effects of PIN diode limiters, significantly improves the computational and simulation speed, reduces the calculation cost, and provides a new method for studying the high-power microwave effects of PIN diode limiters.

Turbulence Closure With Small, Local Neural Networks: Forced Two‐Dimensional and β‐Plane Flows

AbstractWe parameterize sub‐grid scale (SGS) fluxes in sinusoidally forced two‐dimensional turbulence on the β‐plane at high Reynolds numbers (Re ∼25,000) using simple 2‐layer convolutional neural networks (CNN) having only O(1000) parameters, two orders of magnitude smaller than recent studies employing deeper CNNs with 8–10 layers; we obtain stable, accurate, and long‐term online or a posteriori solutions at 16× downscaling factors. Our methodology significantly improves training efficiency and speed of online large eddy simulations runs, while offering insights into the physics of closure in such turbulent flows. Our approach benefits from extensive hyperparameter searching in learning rate and weight decay coefficient space, as well as the use of cyclical learning rate annealing, which leads to more robust and accurate online solutions compared to fixed learning rates. Our CNNs use either the coarse velocity or the vorticity and strain fields as inputs, and output the two components of the deviatoric stress tensor, Sd. We minimize a loss between the SGS vorticity flux divergence (computed from the high‐resolution solver) and that obtained from the CNN‐modeled Sd, without requiring energy or enstrophy preserving constraints. The success of shallow CNNs in accurately parameterizing this class of turbulent flows implies that the SGS stresses have a weak non‐local dependence on coarse fields; it also aligns with our physical conception that small‐scales are locally controlled by larger scales such as vortices and their strained filaments. Furthermore, 2‐layer CNN‐parameterizations are more likely to be interpretable.

A FEA-Based Fast AC Steady-State Algorithm for a Voltage-Driven PMSM by a Novel Voltage-Flux-Driven Model

Finite element analysis (FEA) has been used more frequently than ever before in electrical machine design covering different disciplines such as electromagnetics, rotor dynamics, thermodynamics etc. The accuracy and speed of FEA are in high demands, especially when the time-stepping FEA is employed in the electromagnetic design of the machine. In this paper, to accelerate numerical transient in FEA of a voltage-driven permanent magnet synchronous machine (PMSM), a fast AC steady-state algorithm has been proposed, in which the voltage supply is converted into the stator flux linkage and the machine can be regarded as driven by the flux linkage directly. The electromagnetic transient is eliminated by injecting additional voltage signals to compensate the oscillating stator flux linkage in transient and the time-stepping error caused by the backward Euler in FEA. And the additional voltage signals will finally disappear after the steady state is achieved within about a half electrical time period. In addition, the maximum torque per ampere (MTPA) control logic can also be realized by adjusting the initial phase angle of the voltage supply based on the vector diagram. The simulation speed of the proposed method in this paper was compared to other traditional models and the accuracy was verified by experimental results.

Exploring the influence of building form parameters on three-dimensional ventilation potential in urban centers: A case study of Nanjing, China

In urban areas, variations in the functional usage of buildings or their clusters directly influence the 2D/3D morphology of buildings which can lead to changes in the ventilation potential (VP) at different heights, but there is a lack of refined quantitative relationship analysis between 3D building form parameters (BFPs) and VP at different heights within a certain urban area. The objective of our study is to obtain simulated wind field data from different heights, considering distances from roads, and transverse and longitudinal sections. We employ partial correlation analysis, simple correlation analysis, and Geographic detectors to uncover the extent and mechanisms of the influence of BFPs on the ventilation effectiveness of neighborhoods in a major urban center in China, specifically Nanjing. The findings demonstrate that, at all heights, there is an inverse relationship between wind speed and building density (BD), with the location of higher wind speed generally aligning with the intersection points. The correlation coefficients between BD and simulated wind speed are 0.040 (1.5 ​m), −0.475 (10 ​m), and −0.314 (30 ​m), while the correlation coefficients between building height (BH) and simulated wind speed are 0.237. However, no significant correlation is observed between floor area ratio (FAR) and simulated wind speed. Geographic detector analysis reveals differences in the correlation between BFPs and simulated wind speed at different heights within various building functional areas. Our study provides some practical information for urban planners and designers to carry out wind sensitivity and adaptation planning and urban design.

Centrifugal compressor design and cycle analysis of large-scale high temperature heat pumps using hydrocarbons

Using natural refrigerants in heat pumps, including hydrocarbons, is becoming attractive because of the tightening regulations concerning synthetic refrigerants. In the literature there is lack of studies investigating detailed component design, including compressor design when using hydrocarbons. This study investigates numerically centrifugal compressor design with hydrocarbons for large-scale high-temperature heat pumps. The compressor analysis was combined with a thermodynamic analysis of a two-stage heat pump. The effect of temperature lift and flash intercooler temperature on compressor sizing, rotational speed, efficiency, axial force, and Mach number were investigated. Generally, the use of hydrocarbons with five carbon atoms resulted in higher coefficient of performance and required lower compressor rotational speeds and larger compressor dimensions when compared to the hydrocarbons with four carbon atoms. The simulated rotational speeds were ranging from 8 krpm to 30 krpm, impeller diameters from 0.15 m to over 0.7 m and compressor efficiencies from 77% to over 85%. The use of cyclic molecules, namely cyclopentane and cyclobutene resulted in the highest heat pump coefficient of performance. Also, the flash intercooler temperature had a notable effect on the compressor design and axial force which has to be taken into account in heat pump cycle design.

Data-Driven Dynamic Inversion Method for Complex Fractures in Unconventional Reservoirs

Abstract Hydraulic fracturing is a crucial technology for enhancing the recovery of oil and gas from unconventional reservoirs. Accurately describing fracture morphology is essential for accurately predicting production dynamics. This article proposes a new fracture inversion model based on dynamic data-driven methods, which is different from the conventional linear elastic fracture mechanics model. This method eliminates the need to consider complex mechanical mechanisms, resulting in faster simulation speeds. In the model, the fracture morphology is constrained by combining microseismic data and fracturing construction data, and the fracture tip propagation domain is introduced to characterize the multi-directionality of fracture propagation. The simulated fracture exhibits a multi-branch fracture network morphology, aligning more closely with geological understanding. In addition, the influence of microseismic signal intensity on the direction of fracture propagation is considered in this study. The general stochastic approximation (GSA) algorithm is employed to optimize the direction of fracture propagation. The proposed method is applied to both the single-stage fracturing model and the whole well fracturing model. The research findings indicate that in the single-stage fracturing model, the inverted fracture morphology aligns closely with the microseismic data, with a fitting rate of the fracturing construction curve exceeding 95%, and a microseismic data fitting rate exceeding 93%. In the whole well fracturing model, a total of 18 sections were inverted. The fitting rate between the overall fracture morphology and the microseismic data reached 90%. The simulation only took 5 minutes, demonstrating high computational efficiency and meeting the needs of large-scale engineering fracture simulation. This method can effectively support geological modeling and production dynamic prediction.

An ASCON AOP-SystemC Environment for Security Fault Analysis

Cryptographic devices’ complexity necessitates fast security simulation environments against fault attacks. SystemC, a promising candidate in Electronic System Levels (ESLs), can achieve higher simulation speeds while maintaining accuracy and reliability, and its modular and hierarchical design allows for efficient modeling of complex cryptographic algorithms and protocols. However, code modification is required for fault injection and detection. Aspect-Oriented Programming (AOP) can test cryptographic models’ robustness without modifications, potentially replacing real cryptanalysis schemes and reducing the time and effort required for fault injection and detection. Through the utilization of a fault injection/detection environment, this paper presents a novel approach to simulating the security fault attacks of ASCON cryptographic systems at the ESL. The purpose of this methodology is to evaluate the resistance of ASCON SystemC models against fault attacks. The proposed methodology leverages the advantages of AOP to enhance the fault injection and detection process. By applying AOP techniques, we inject faults into the SystemC models without making any changes to the main codebase. This approach not only improves the efficiency of testing cryptographic systems but also ensures that the main functionality remains intact during the fault injection process. The methodology was validated using three scenarios and SystemC ASCON as a case study. The first simulation involved evaluating fault detection capabilities, the second focused on the impact of AOP on executable file size and simulation time, and the third focused on the ESL impact on the ASCON design process. Simulation results show that this methodology can perfectly evaluate the robustness of the ASCON design against fault injection attacks with no significant impact on simulation time and file executable size. Additionally, the simulation results prove that the ASCON development life cycle at the ESL reduces the amount of time devoted to the design procedure by 83.34%, and the ASCON security attack simulations at the ESL decrease the simulation time by 40% compared to the register transfer level (RTL).

Field Programmable Gate Array-Based Acceleration Algorithm Design for Dynamic Star Map Parallel Computing

The dynamic star simulator is a commonly used ground-test calibration device for star sensors. For the problems of slow calculation speed, low integration, and high power consumption in the traditional star chart simulation method, this paper designs a FPGA-based star chart display algorithm for a dynamic star simulator. The design adopts the USB 2.0 protocol to obtain the attitude data, uses the SDRAM to cache the attitude data and video stream, extracts the effective navigation star points by searching the starry sky equidistant right ascension and declination partitions, and realizes the pipelined displaying of the star map by using the parallel computing capability of the FPGA. Test results show that under the conditions of chart field of view of Φ20° and simulated magnitude of 2.0∼6.0 Mv, the longest time for calculating a chart is 72 μs under the clock of 148.5 MHz, which effectively improves the chart display speed of the dynamic star simulator. The FPGA-based star map display algorithm gets rid of the dependence of the existing algorithm on the computer, reduces the volume and power consumption of the dynamic star simulator, and realizes the miniaturization and portable demand of the dynamic star simulator.

Designing ultra-broadband terahertz polarization converters based on the transformer model

Polarization, a fundamental property of terahertz (THz) waves, can be controlled using metasurfaces for signal transmission and sensitive measurements. However, conventional methods for designing polarization-conversion metasurfaces require numerous numerical simulation iterations and trial and error attempts that are computationally intensive and time-consuming, necessitating a more rapid and efficient approach. This study proposes a transformer-based method for the inverse design of terahertz polarization converters; the approach can accurately design the corresponding metasurface based on the target spectrum. The model also comprises a forward network that can precisely and intuitively predict the meta-atoms spectra with high simulation speed and reduced computational cost. Using this approach, two ultra-broadband polarization conversion devices were designed in approximately 0.006 s within a target frequency range of 0.5–4 THz, with polarization conversion ratios greater than 90% within 1.23–3.19 THz and 1.18–3.24 THz. Additionally, the average cross-polarization ratio was greater than 50%. This study provides insights into designing terahertz polarization converters, which are applicable in terahertz communications, polarization imaging, and biomedical procedures.

Edge illumination x-ray phase contrast simulations using the CAD-ASTRA toolbox.

Edge illumination x-ray phase contrast imaging (XPCI) provides increased contrast for low absorbing materials compared to attenuation images and sheds light on the material microstructure through dark field contrast. To apply XPCI in areas such as non-destructive testing and inline inspection, where scanned samples are increasingly compared to simulated reference images, accurate and efficient simulation software is required. However, currently available simulators rely on expensive Monte Carlo techniques or wave-optics frameworks, resulting in long simulation times. Furthermore, these simulators are often not optimized to work with computer-aided design (CAD) models, a common and memory-efficient method to represent manufactured objects, hindering their integration in an inspection pipeline. In this work, we address these shortcomings by introducing an edge illumination XPCI simulation framework built upon the recently developed CAD-ASTRA toolbox. CAD-ASTRA allows for the efficient simulation of x-ray projections from CAD models through GPU-accelerated ray tracing and supports ray refraction in a geometric optics framework. The edge illumination implementation is validated and its performance is benchmarked against GATE, a state-of-the-art Monte Carlo simulator, revealing a simulation speed increase of up to three orders of magnitude, while maintaining high accuracy in the resulting images.

Coupling deformation analysis of self-morphing bilayers with mismatch strain

Self-morphing bilayers driven by the inner mismatch strain have inspired numerous actuators and soft robots in the four-dimensional (4D) printing field. Analyzing their coupling deformation is crucial for the design of actuators, but current methods face inefficiencies or accuracy issues. This study provides convenient and efficient approaches, from both deformation theory and model aspects, to analyze the coupling deformations of self-morphing bilayers. First, we derived a simplified energy expression for self-morphing bilayers to reflect their deformation process. This concise form permits direct derivation of some deformation theories for self-morphing bilayers, thereby significantly simplifying the complexity of deformation analysis. Second, we established the deformation simulation model of bilayers by discretizing and minimizing the simplified energy expression. The checkerboard pattern method, a new discrete differential geometric (DDG) approach, was introduced and improved to discretize the derived energy expression. The Levenberg-Marquardt (LM) algorithm was used to establish the deformation simulation model for bilayers by minimizing the discrete energy function. Further, a series of numerical simulations of bending bilayers were conducted. Results validated that the proposed model achieved a faster simulation speed compared to the finite element method (FEM)-based model in our previous work, and was more accurate for finite strain situations. Finally, three bio-inspired bilayers and a bio-inspired gripper (successfully grasped a sphere) were designed using the resulting model, demonstrating its practical application capabilities. These proposed methods hold substantial potential to facilitate the advancement of the design field of 4D printing and actuators.

CALIBRATION OF MICROSCOPIC TRAFFIC SIMULATION OF URBAN ROAD NETWORK INCLUDING MINI-ROUNDABOUTS AND UNSIGNALIZED INTERSECTION USING OPEN-SOURCE SIMULATION TOOL

Microscopic traffic simulation models offer an effective way to analyze and assess different transportation systems thanks to their efficiency and reliability. As traffic management issues become more prevalent, notably in urban areas, simulation tools enable a significant opportunity to replicate real-world conditions before implementation. Therefore, the calibration of traffic simulation models plays a substantial role in obtaining accurate and confidential results. Nowadays, urban regions are facing the challenge of restricted space for developing traffic solutions. As a consequence of environmental restrictions, the use of mini-roundabouts rather than larger roundabouts is increasing. Based on the given literature review, it is seen that not much attention was given to the complex modeling and calibration of microsimulation models of mini-roundabouts and unsignalized intersections. The objective of this study is to offer the calibration of microscopic traffic simulation of urban road network, including closely located mini-roundabouts and unsignalized intersection. To this end, an open-source tool called SUMO (Simulation of Urban Mobility) was utilized as a simulation environment in this study. The necessary data for developing a microsimulation model in SUMO was gathered using a videography technique. The traffic count data and speed were considered performance measures between field observations and simulation outputs. The routeSampler tool of SUMO, which has recently emerged in the literature, was used to match traffic count data and the corresponding time interval for traffic volume data calibration. The calibration of car-following model parameters using a trial-and-error approach was employed based on mean absolute percent error (MAPE) between simulated speeds and field-measured speeds. According to the findings of the study, the simulation model fulfilled the calibration aims of the FHWA guideline and is suitable for further research.

Real-Time Digital Simulator Design for Differential Drive Mobile Robot using FPGA

This paper presents a real-time simulation of a differential drive mobile robot (DDMR). The permanent magnet DC motors that drive the robot’s left and right wheels were modeled and executed in real-time on a Field Programmable Gate Array (FPGA) based co-simulator platform, interfacing with the Webots robot simulator, which simulates the DDMR on the PC side. The electrical parameters, which are not available in robot simulators, were simulated and measured by the proposed co-simulator system in real-time under various environmental conditions and trajectories of the robot. Parameters such as current, voltage, and torque were measured instantaneously, enabling a more realistic simulation. Additionally, the cycle time of the robot simulator was determined to be 32 ms, and the developed FPGA-based simulation operated at approximately 2000 times the speed of the robot simulator. The results demonstrate the applicability of the developed platform in robotic applications.

The bar pattern speed of the Large Magellanic Cloud

Context. The internal kinematics of the Large Magellanic Cloud (LMC) have been studied in unprecedented depth thanks to the excellent quality of the Gaia mission data, thus revealing the disc’s non-axisymmetric structure. Aims. We seek to constrain the LMC bar pattern speed using the astrometric and spectroscopic data from the Gaia mission. Methods. We applied three methods to evaluate the bar pattern speed by measuring it via: the Tremaine-Weinberg (TW) method, the Dehnen method, and a bisymmetric velocity (BV) model. These methods provide additional information on the bar properties, such as the corotation radius as well as the bar length and strength. We tested the validity of the methods with numerical simulations. Results. A wide range of pattern speeds are inferred by the TW method, owing to a strong dependency on the orientation of the galaxy frame and the viewing angle of the bar perturbation. The simulated bar pattern speeds (corotation radii, respectively) are well recovered by the Dehnen method (BV model). Applied to the LMC data, the Dehnen method finds a pattern speed of Ωp = −1.0 ± 0.5 km s−1 kpc−1, thus corresponding to a bar that barely rotates and is only slightly counter-rotating with respect to the LMC disc. The BV method gives a LMC bar corotation radius of Rc = 4.20 ± 0.25 kpc, corresponding to a pattern speed of Ωp = 18.5−1.1+1.2 km s−1 kpc−1. Conclusions. It is not possible to determine which global value best represents an LMC bar pattern speed with the TW method, due to the strong variation with the orientation of the reference frame. The non-rotating bar from the Dehnen method would be at odds with the structure and kinematics of the LMC disc. The BV method result is consistent with previous estimates and gives a bar corotation-to-length ratio of 1.8 ± 0.1, suggesting that the LMC is hosting a slow bar.

Evaluation and correction of wind energy forecast in Yangjiang

Wind power generation is one of the most promising new energy sources. It is important to study effective short-term wind speed forecasting methods. In this paper, the CMA-GD model is applied to forecast the wind speed over South China in four typical months (January, April, August, and October) in 2022. Taking Longyue Wind Farm in Yangjiang as an example, the simulated wind speed and wind direction are evaluated based on the observation of this station at four heights (10 m, 30 m, 50 m, and 70 m). The results show that the CMA-GD model has a good effect on forecasting the next day’s 24 h wind speed of Longyue Wind Farm. Overall, the correlation coefficients (R) between forecast and observation are 0.77-0.81, and the root-mean-square errors (RMSE) are 1.80-2.08 m·s-1. With the increase in altitude, the simulation effect is a little better. For different months, the R is as follows: October>August>April>January, and the RMSE and mean absolute error (MAE) are as follows: October>January>August>April. For diurnal variation, the wind speed simulation effect in the night is better than that in the day. Due to the influence of the subtropical monsoon climate and local mountain microclimate in April and August, there are some deviations in the wind speed simulation during 4-8 p.m. The difference in wind direction between forecast and observation are as follows: August<January<October<April, the average wind direction differences are between 2-17°, with little difference in the vertical direction. After the forecast wind speed is corrected by the random forest (RF) algorithm, the R of wind speed at each height is 0.92, and RMSEs are 1.07-1.27 m·s-1. The revised wind speed forecast has the same diurnal variation characteristics as the observation, indicating that the random forest algorithm can effectively reduce the model forecast error.