632results about How to "Short time" patented technology

The invention discloses a soybean nodulation number related major single nucleotide polymorphism (SNP) site and application thereof, the SNP site is positioned on a soybean chromosome 2, the physical distance of the SNP site in a soybean genome version number Glycine max Wm82.a2.v1 is Chr0243129492, the SNP site corresponds to a 20th basic group of a sequence shown as SEQ ID NO.1, when the site is T / T homozygosis, the corresponding genotype is A, when the site is T / T homozygosis, the corresponding genotype is B, and when the site is T / T homozygosis, the corresponding genotype is C; when the site is C / C homozygous, the corresponding genotype is B; the number of root nodules is as follows: the genotype A homozygous soybean is greater than or candidate to be greater than that of the genotype B homozygous soybean. By detecting the SNP, a tested soybean population can be accurately divided into a low root nodule number (C / C) and a high root nodule number (T / T). Therefore, the molecular marker can be used as a basis for preliminarily identifying the number of soybean nodules, has high efficiency and applicability such as simplicity in operation, low cost and short consumed time, and has important values for digging germplasm of high-nodulation soybeans, solving the problem of nitrogen fertilizer abuse and promoting sustainable development of agriculture.

The invention discloses a design method of a multi-constraint guidance law driven by a depth deterministic strategy gradient, and belongs to the technical field of missile guidance. The method comprises the following steps: S1, establishing a three-dimensional relative kinematics model of a missile interception target; s2, based on the S1, designing a fractional order sliding mode control law and analyzing the stability of the fractional order sliding mode control law; s3, designing and optimizing an A3DPG algorithm; and S4, inducing a strategy network to learn an action strategy through a reward mechanism in a dynamic environment, and realizing a multi-constraint guidance law with shortest time, small miss distance, field of view FOV constraint, acceleration change suppression, sliding mode surface convergence reward, end attack angle constraint and total reward. According to the method, the guidance law with the convergent line-of-sight angle is designed by using the fractional order sliding mode theory, and then the sliding mode surface parameters and the reaching law parameters are dynamically learned by using the reinforcement learning algorithm, so that the guidance law with dynamic adjustment is realized. And inducing a strategy network to learn an optimal guidance law under the evaluation of the indexes through factors such as a reward mechanism in a dynamic environment.

The invention relates to a database structure migration method and device, computer equipment, a computer readable storage medium and a computer program product. The method is applied to a target database and comprises the following steps: in response to a received structure migration instruction from a source database to the target database, determining a plurality of to-be-migrated types corresponding to the structure migration instruction; determining a migration sequence corresponding to the to-be-migrated types according to the dependency relationship among the to-be-migrated types; and based on the migration sequence, obtaining a target object corresponding to each to-be-migrated type, and executing a target structured query statement corresponding to the target object until execution of the target objects corresponding to the to-be-migrated types is completed, the target structured query statement is obtained by processing an initial structured query statement corresponding to the target object in the source database according to the parameters of the target database. The method can improve the migration efficiency and shorten the time.

The invention provides a fuzzy privacy set intersection method and system, a medium and electronic equipment, and the method comprises the steps: carrying out the encryption processing of an initial data set X through a homomorphic encryption algorithm, obtaining first data, and obtaining a processed data structure D through a defined space Hash algorithm; the receiver performs homomorphic addition or scalar multiplication operation processing on the first data by using own data Y, and performs confusion processing by using a random number R to obtain second data; after the sender carries out Augment algorithm processing on the random number, a Hash algorithm is carried out to obtain a corresponding Hash value, and the Hash value is recorded as third data to be sent to the sender; and the sender decrypts the second data, processes the second data through an Augment algorithm, executes a PSI protocol with the receiver to obtain fourth data, and compares the third data with the fourth data to obtain intersection data. According to the method, an addition homomorphic technology is adopted for a novel application scene of intersection of the fuzzy privacy set, and compared with a full homomorphic technology in the prior art, the time consumption is less, and the communication is small.

The invention provides a voice-based psychological state recognition method, device and equipment. The method comprises the following steps: acquiring voice data of a user; preprocessing the voice data to obtain target voice data; extracting voice features of the target voice data; inputting the voice features into a pre-trained voice psychological state model; based on the voice psychological state model, the psychological state of the user corresponding to the voice feature is determined, clinical evaluation of professional doctors is not depended on any more, consumed time is short, efficiency is high, and popularization and promotion are facilitated.

The embodiment of the invention discloses a building detection method, device and equipment based on a remote sensing image and a laser point cloud. A specific embodiment of the method comprises the following steps: carrying out single building segmentation processing on a building remote sensing image; for each building segmentation remote sensing image in the building segmentation remote sensing image group, the following processing steps are executed: an image edge node coordinate sequence and surface elevation information of the building segmentation remote sensing image are determined, and each edge node coordinate in the edge node coordinate sequence represents a key node of the building segmentation remote sensing image; extracting single building point cloud data corresponding to the building segmentation remote sensing image from the laser point cloud information; and generating building detection information according to the single building point cloud data and the corresponding standard building three-dimensional data. According to the embodiment, the building area edge detection precision and the building detection efficiency are improved.

The invention belongs to the technical field of fish genetic identification, discloses a multi-site large yellow croaker wild domestication population identification system based on an agarose gel electrophoresis technology, and particularly discloses a group of InDel molecular markers for large yellow croaker wild and domestication population identification, and the InDel molecular markers comprise InDel11-5053756, InDel20-14004876, InDel3-28345321 and InDel21-17327392. The InDel molecular marker detection primer is used for amplification to obtain a PCR product, agarose gel can be used for detecting strip polymorphism, extra restriction enzyme digestion or complicated polyacrylamide gel electrophoresis operation is not needed, and identification of large yellow croaker wild and domesticated populations can be rapidly and accurately achieved.

The invention relates to a part turning method based on a software CAM milling module and application, and belongs to the technical field of turning programming. The part turning method comprises the steps that S1, a software CAD function is started, and a curve model of a part is established; s2, starting a software CAM function, selecting a plane milling mode, setting in the plane milling mode, and generating a tool path code; s3, post-processing the tool path code to generate a numerical control program; and S4, the numerical control program is modified, a part is clamped on a machine tool, turning is started according to the modified numerical control program, and the part is an unconventional curved surface part. According to the turning programming method, whether a modeling module or a machining module is adopted, parameter setting is simple, errors are not prone to occurring, programming work of the unconventional curved surface can be efficiently completed, and the accuracy of a program is guaranteed.

The utility model relates to a two-stage road pavement structure for an unstable roadbed, which is particularly suitable for an unstable area easy to settle and comprises a first-stage pavement structure and a second-stage pavement structure positioned behind the first-stage pavement structure for construction, wherein the first-stage pavement structure is provided with a pre-sedimentation pavement base layer and a prefabricated high-strength concrete interlocking block surface layer, the prefabricated high-strength concrete interlocking block surface layer is arranged on the pre-sedimentation pavement base layer, the second-stage pavement structure comprises a leveling layer and an asphalt surface layer, the leveling layer is located on the prefabricated high-strength concrete interlocking block surface layer, and the asphalt surface layer is located on the leveling layer. And the asphalt surface layer is positioned on the leveling layer. The utility model has the advantages of low manufacturing cost, high bearing capacity, good durability, convenience in construction, easiness in maintenance and replacement and strong adaptability to settlement deformation, and can fully ensure the use of the whole life cycle of the road; subgrade settlement is accelerated; construction is completed in a short time, and the influence on current traffic is small; dismounting and reconstruction are not needed, and resources are saved.

The invention discloses a preparation method and application of a Gd-doped rutile type ruthenium dioxide nanosheet, and the Gd-doped rutile type ruthenium dioxide nanosheet is successfully prepared by taking RuCl3 and GdN3O9 as raw materials and adopting a salt template method. When being used as an oxygen evolution catalyst for producing hydrogen by electrolyzing water, the oxygen evolution catalyst shows excellent (OER) oxygen evolution performance, realizes the overpotential of 230mV under the current density of 10mA cm <-2 >, and has excellent conductivity, wider potential window and high redox reversibility, thereby having a wide application prospect in the field of electrocatalysis. In addition, the preparation method disclosed by the invention has the characteristics of simple process steps, safety, easiness in control, easiness in obtaining raw materials, few chemical reagents used in the whole preparation process, small environmental pollution, mild experimental conditions, relatively short time consumption, low cost and the like, so that the preparation method is suitable for industrial popularization and application.

The invention belongs to the technical field of wood modification, and particularly relates to a carbon quantum dot optimized furfuryl alcohol modified material and a preparation method thereof. The preparation method comprises the following steps that wood is treated with carbon quantum dot / furfuryl alcohol modified liquid through a vacuum pressurization impregnation method, then the wood is wrapped with a high-temperature-resistant waterproof material, and aging is conducted at the room temperature. And finally, the wood is heated and cured, and therefore the carbon quantum dot optimized furfuryl alcohol modified material is obtained. According to the carbon quantum dot / furfuryl alcohol modified liquid, the carbon quantum dots serve as a catalyst to catalyze curing of furfuryl alcohol in wood and also serve as a wood modifier to endow the wood with excellent corrosion resistance, and multiple effects are achieved through one agent. According to the carbon quantum dot optimized furfuryl alcohol modified material, the problem that the toughness of the furfuryl alcohol modified material is reduced is solved, overall balance and optimization of the dimensional stability, the strength, the toughness and the biological degradation resistance of wood are achieved, and the problem that the material color of the modified material is blackened is solved.

The present invention discloses a method and device for feature extraction and matching of satellite video images, which can solve the problem of real-time or near-real-time application of satellite video. The satellite video image to be matched is determined, corner points are extracted from the satellite video image to be matched; the corner points are filtered; based on the filtered corner points, the satellite video image to be matched is characterized, and a feature vector representing the satellite video image to be matched is obtained; and the feature vectors of the satellite video image to be matched are matched.

The embodiment of the invention discloses a movement control method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining grid maps of a plurality of levels, determining an initial starting point grid and an initial end point grid, then determining a target grid in a grid map of a target level, and then determining a target grid in the grid map of the initial level, the moving sub-path is determined in the mapping range of the target grid, and the space searched for determining the moving sub-path is small, so that the time complexity can be effectively reduced, the time consumption for determining the moving sub-path is reduced, and the smoothness of the target object navigation process is improved; the at least two moving sub-paths are determined in different time fragments, namely, the determination actions of the at least two moving sub-paths are split, so that the blockage phenomenon of the main thread is avoided, the smoothness of the navigation process of the target object is further improved, and the navigation efficiency of the target object is improved. The method can be widely applied to scenes such as cloud technology, artificial intelligence, games and intelligent transportation.

The invention discloses a method for preparing a C / C-SiC composite material through multi-stage recovery of carbon / carbon composite material processing waste, and belongs to the technical field of C / C composite material preparation. The method comprises the following steps: carrying out multi-stage screening treatment on the carbon / carbon composite material processing waste, and screening and separating short carbon fibers and pyrolytic carbon powder; mixing the screened short carbon fibers and pyrolytic carbon powder with phenolic resin, industrial silicon powder and SiC powder, and preparing a fiber reinforced resin block in a mold pressing manner; and carrying out one-step continuous carbonization and ceramic heat treatment on the fiber reinforced resin block to finally obtain the C / C-SiC composite material. According to the method, high-added-value recycling of the C / C processing waste is achieved, the process is simple, the cost is low, operation is convenient and fast, the problems of resource waste and environmental pollution caused by existing waste treatment are solved, and the prepared C / C-SiC composite material is excellent in performance and can be widely applied to the fields of aerospace, high-end manufacturing and the like.

The invention discloses a method for detecting homocysteine by a nano-pore sensor based on silver-sulfur polymerization, which comprises the following steps: pretreating a solution containing the homocysteine by 4-dimethylamino cinnamyl aldehyde, and then putting a sensor with a cylindrical nano-pore structure into the solution containing the homocysteine to carry out polymerization reaction; establishing a linear equation of the ratio of the conductivity before and after the polymerization reaction and the concentration of the homocysteine, and calculating the concentration of the homocysteine in the liquid to be detected through the linear equation. The liquid to be detected contains homocysteine and cysteine, and is pretreated by 4-dimethyl amino cinnamyl aldehyde before detection; and an Ag (I) probe is modified in a cylindrical nanopore channel of the sensor. According to the method, a special pretreatment method is utilized to cooperate with a constructed functional probe and a special nanopore structure, and current signal change is amplified based on silver-sulfur polymerization reaction, so that rapid and accurate detection of Hcy is realized.

The invention provides a binding protein specifically binding to mycoplasma pneumoniae P1 protein as well as a preparation method and application thereof, and relates to the technical field of biology. The binding protein contains a heavy chain variable region and a light chain variable region, wherein a complementary determining region of the heavy chain variable region comprises amino acid sequences consistent with VH-CDR1, VH-CDR2 and VH-CDR3 of the heavy chain variable region shown as SEQ ID NO.1; the complementarity determining region of the light chain variable region comprises amino acid sequences consistent with VL-CDR1, VL-CDR2 and VL-CDR3 of the light chain variable region as shown in SEQ ID NO.2 or 3. The binding protein has good specific binding capacity with the mycoplasma pneumoniae P1 protein, can be combined with mycoplasma pneumoniae whole bacteria, and can be used for detection of mycoplasma pneumoniae or mycoplasma pneumoniae P1 protein and detection and diagnosis of mycoplasma pneumoniae or mycoplasma pneumoniae infection.

The utility model discloses a mobile pasteurization tank, belongs to the technical field of food processing equipment, and provides a mobile pasteurization tank capable of effectively reducing energy consumption, the mobile pasteurization tank comprises an inner container, a temperature sensor is arranged in the inner container, and two bridging rings close to the upper end and the lower end respectively are fixedly connected to the outer side face of the inner container; the inner container is sleeved with an electric heating ring located between the two bridging rings, a heat insulation plate located outside the electric heating ring is further fixedly connected between the two bridging rings, the outer side faces of the two bridging rings are jointly and fixedly connected with a shell, the upper end of the shell is fixedly connected with an upper cover, and the upper end of the upper cover is provided with a feeding port communicated with the interior of the inner container. And a sealing cover is arranged at the feeding hole of the upper cover. The electric heating ring sleeved outside the inner container is provided with the heat shield structure with the reflecting layer, so that the heat production utilization rate of the electric heating ring is effectively improved, the heat loss is reduced, the heating time is shortened, the energy consumption during the working of the sterilization tank is reduced, and the sterilization tank has a better energy-saving characteristic.

The utility model provides a reversing conveying system which comprises a first mover module, a second mover module and a third mover module, the first mover module comprises a first mover body, a first permanent magnet array and a first sensing element, and the first permanent magnet array and the first sensing element are both arranged on the first mover body; the first connection module comprises a first driving part, a second driving part, a first connection stator and a second connection stator, the second driving part is fixedly connected with the driving end of the first driving part, the first connection stator is fixedly connected with the driving end of the first driving part, and the first driving part drives the first connection stator and the second driving part to move linearly; the second driving piece is connected with the second connection stator, and the second driving piece drives the second connection stator to rotate; the second connection stator comprises a second stator body, a second coil and a second detection element. According to the technical scheme, the problem that in the related technology, when workpieces are transferred between different conveying lines, the conveying efficiency is low is effectively solved.

The invention discloses a partition calculation method of an undirected graph reachable matrix, and relates to the technical field of data mining. According to the method, the undirected graph is divided into a plurality of sub-graphs through a partitioning algorithm, and the reachable matrix of each sub-graph is independently calculated, so that the complexity and resource requirements of global calculation are reduced, and the calculation efficiency is improved. In addition, a multi-core and distributed computing environment is utilized, so that network topology analysis scenes, such as power grid operation analysis, dynamic traffic scheduling and the like, with high requirements on real-time performance of a large-scale network can be efficiently processed. For network dynamic changes, a boundary equivalent data updating mechanism is introduced, rapid adjustment of a local matrix is achieved, global recalculation is not needed, the response speed of the system in a frequent topology change scene is remarkably improved, and the method is suitable for the business fields of power grid operation analysis, dynamic traffic scheduling and the like. And efficient operation and real-time response capability of the system are effectively guaranteed.

The utility model discloses an adjustable suspension device of a racing car. The adjustable suspension device comprises a steering mechanism, a damping mechanism and a suspension mechanism, a steering wheel of the steering mechanism is fixedly connected with a steering shaft, the steering shaft is rotationally connected with the frame through a bearing, and a steering angle signal sensor is arranged on the steering shaft, recognizes the steering angle of the steering wheel and converts the steering angle into an electric signal; one end of the damping mechanism is fixedly connected with the frame, and the other end is fixedly connected with a hydraulic telescopic rod in the suspension mechanism; the suspension mechanism comprises a tripod, a supporting plate is welded to the tripod, and the supporting plate is fixed to the other end of the hydraulic telescopic rod and bears force transmitted by the suspension; the hydraulic telescopic rod stretches out and draws back through cooperation of a hydraulic pump and an oil valve controller, and the oil valve controller receives an electric signal of the corner signal sensor and further controls oil supply of the hydraulic pump so as to achieve ascending and descending of the inner side car body and the outer side car body of the racing car; the heights of the suspensions on the inner side and the outer side of the racing car are correspondingly increased and decreased, the time consumed by the racing car on a curve and the time consumed by speed increasing after the racing car goes out of the curve are shortened, and then the racing result is improved.

The application discloses an electromagnetic multilayer fast multipole hierarchical GPU parallel acceleration implementation method, relates to the field of computational electromagnetics, and comprises the following steps: acquiring parameters of incident electromagnetic waves and a grid file of a target to be measured; adopting a multilayer fast multipole algorithm to construct an octree corresponding to the grid file and to construct a process tree; performing hierarchical division on the octree based on a ternary parallel strategy to obtain a plurality of parallel layers corresponding to the octree; adopting a mapping allocation method to allocate the operation load of each parallel layer to all operation processes of the process tree; according to the operation load corresponding to the operation process and the number of GPUs in the operation process, distributing the operation load corresponding to the operation process to each GPU in the operation process; and splicing the operation results of all GPUs and solving the radar scattering cross section of the target to be measured according to the splicing result. The application can effectively reduce the communication and synchronization overhead of the multilayer fast multipole algorithm, improve the multi-card load balancing and the overall computing efficiency.

The invention relates to the technical field of feed production, in particular to an intelligent quality detection machine for feed production, which comprises a mounting frame and a housing, a housing is fixedly connected to the mounting frame; a rectangular groove I is formed in the mounting frame; a second rectangular groove is formed in the housing, and the first rectangular groove coincides with the second rectangular groove. The device further comprises a power assembly, a U-shaped frame, a roller, a sealing plate and the like. A power assembly is connected in the housing; the power assembly is connected with a U-shaped frame; a roller is rotationally connected to the U-shaped frame; a sealing plate is rotationally connected to the roller through a rotating shaft. According to the intelligent quality detection machine for feed production obtained through the design, the feed rolls inside through rotation and swing of the roller, so that the hardness, brittleness and water content of the feed are detected, time consumption is low, efficiency is high, production line parameters are convenient to adjust in time, a large amount of unqualified feed is prevented from being produced on a production line, and resource waste is reduced.

The application relates to a processor, component, device and method for filter processing of an IIR filter. The processor comprises a first configuration register, a second configuration register, a general register and a matrix multiplication accumulation unit. The general register sequentially reads and stores each row of coefficients of a coefficient matrix of each order of the IIR filter. The matrix multiplication accumulation unit obtains the same input element corresponding to the stored current row of coefficients when the first configuration register is configured in a broadcast mode, copies the stored current row of coefficients when the first configuration register is configured in a copy mode, multiplies the current row of coefficients with the corresponding input element in parallel to obtain corresponding products, and sequentially accumulates the product results of each row of coefficients to obtain a final output value, so that at least four output variables of sequential sampling moments are obtained each time. In this way, the time consumption of IIR filter operation can be significantly shortened, and sufficient acceleration effect is provided.

Embodiments of the present application relate to the technical field of human-computer interaction, and disclose a method and device for determining an interactive gesture and an electronic device. The method comprises the following steps: first, target component detection is performed; then, gesture detection and recognition are performed in a local area (around a boundary box) near the target component. In addition, for N gesture detection boxes (multiple users) in a current video frame, M target gesture detection boxes are selected therefrom to perform hand component positioning and gesture recognition processing, which can reduce detection time consumption and improve real-time performance. On this basis, the interactive gesture is determined according to the gesture detection result of the current video frame and the gesture detection results of a plurality of historical video frames, that is, gesture recognition is selectively performed on the user in each frame in a plurality of continuous video frames to determine the interactive gesture, which can reduce the computing power, reduce the time consumption, improve the real-time performance, and improve the accuracy and stability of the interactive gesture. That is, in the multi-user human-computer interaction scene, the interactive gesture can be accurately and timely determined.

This invention discloses a method for multi-source data fusion and power prediction of wind turbine units, belonging to the field of power prediction technology. It addresses the technical problems of poor multi-source data fusion and power prediction analysis in existing solutions. By filtering and retaining key features through mutual information, redundant information can be effectively reduced. Dynamic weighting via an attention mechanism makes the fused features more adaptable to real-time operating conditions. Differential weight adjustments for wind speed, terrain, and equipment status ensure high relevance of the fused features even in complex scenarios. Spatial feature extraction captures local spatial correlations, temporal features capture temporal dependencies, and random forests handle nonlinear mappings, solving the problem of insufficient generalization ability of single models. The attention mechanism automatically assigns weights to different time steps and features, avoiding complex mathematical processes such as matrix operations and noise covariance estimation.

The invention discloses an automatic regulation and control side coaming device of a multi-point mold, and relates to the technical field of molds, the automatic regulation and control side coaming device of the multi-point mold comprises a lower mold bottom plate used for bearing a multi-point mold body, the upper end of the lower mold bottom plate is fixedly connected with two fixed outer coamings which are oppositely arranged, and two fixed inner coamings are fixedly arranged between the two fixed outer coamings; the upper end of the lower die bottom plate is further connected with four movable surrounding plate assemblies, and each movable surrounding plate assembly comprises a movable surrounding plate, a base and an electric cylinder push rod. The fixed outer coaming, the fixed inner coaming and the movable coaming assembly abut against the outer side of the multi-point mold body, so that the bending resistance of the multi-point mold body can be enhanced, the usable range of the mold is widened, the forming precision of the mold is improved, and the service life of the mold is prolonged.

The present application belongs to the field of advanced coating and preparation, and particularly relates to a high-transparent and scratch-resistant multifunctional antistatic coating and a preparation method thereof. The present application provides a high-transparent and scratch-resistant antistatic coating, raw materials of the antistatic coating comprising fluorion-containing silsesquioxane monomer, functional monomer and catalyst; wherein the functional monomer is a crosslinking agent or an anti-crystallization agent. The coating obtained by the present application has a light transmittance of >98% (550 nm); a surface pencil hardness of greater than 9H, no scratches after more than 1000 times of steel wool rubbing; a surface resistivity of about 10 8 ~10 9 Ω / sq, and a fast electrostatic discharge rate.

This invention relates to a constrained automated optimization method for semiconductor measurement, comprising: step 3) optimizing auxiliary parameters using a training set; step 4) selecting one or more second candidate parameter combinations to form a candidate parameter combination pool, and saving the regression coefficient set corresponding to each second candidate parameter combination; step 5) using the auxiliary parameters in the optimal second candidate parameter combination as independent variables, using the corresponding regression coefficient set as the initial coefficients of the equation, and the key parameters as dependent variables to establish a second multiple linear regression equation; performing the first-stage optimization and the second-stage optimization to select the optimal predicted values ​​and optimal regression coefficients of the key parameters; and step 6) using a test set to perform the first-stage optimization and the second-stage optimization to verify the optimal candidate parameter combination and its corresponding optimal regression coefficients. The two-stage optimization strategy fundamentally solves industry pain points such as difficulty in convergence, difficulty in reproducibility, strong subjectivity, and long processing time.

The utility model discloses a high -order curtain wall discoloration glass intelligence optical performance detection device, include: bottom plate, the top fixed connection of bottom plate has the limit slot piece, the top fixed connection of limit slot piece has the detection body through bolt, wheel replacement subassembly, the bottom rotation connection of wheel replacement subassembly with the top of bottom plate, wheel replacement subassembly is used for the user in the device is carried out rotation replacement, loading debugging component, the surface fixed connection of loading debugging component one side with wheel replacement subassembly, the utility model relates to glass detection technical field. This high -order curtain wall discoloration glass intelligence optical performance detection device, utilize the setting of loading debugging component, and the staff can be in the detection process, in the outside, the large high -order curtain wall discoloration glass is inserted into the plug -in block in swing frame, avoids the installation of glass in the detection equipment and spends a lot of time, before detection, only need to insert the plug -in block of external frame bottom into the installation slot piece.

The application provides a Gaussian ball clustering method for three-dimensional reconstruction of a highway traffic accident site, comprising the following steps: S1, collecting multi-view images of the accident site, solving camera parameters and initializing a three-dimensional Gaussian ball scene; S2, constructing a rendering training loop based on the initialized three-dimensional Gaussian ball scene, and adaptively adjusting the Gaussian ball density through cloning or splitting operations; S3, extracting the spatial and color features of the three-dimensional Gaussian ball, and constructing a multi-modal physical feature set; S4, calculating an adaptive discrete step length based on the multi-modal physical feature set, and mapping the continuous features into discrete coordinates; S5, generating a multi-modal hash index based on the discrete coordinates, and realizing continuous rearrangement of the same type of Gaussian ball in the display memory through GPU parallel sorting; S6, performing physical conservation formula parameter aggregation on the same type of Gaussian ball continuously stored in the display memory, releasing redundant display memory and simplifying the three-dimensional Gaussian ball set; and S7, completing all training iterations, and exporting the final simplified three-dimensional Gaussian ball set into a general PLY format file.