155 results about "Projection model" patented technology

The invention discloses an on-orbit spacecraft attitude estimation method and system based on ISAR image feature selection, and belongs to the field of aerospace control systems. The method comprises the following steps: acquiring an ISAR image of a spacecraft, and acquiring a complex linear structure set and three-dimensional feature points of an on-orbit spacecraft; then, according to the obtained three-dimensional-two-dimensional projection model, the CRLB of each reference structure in the complex linear structure set is deduced to carry out attitude estimation error analysis; calculating the trace of the CRLB covariance matrix of each reference structure to select an optimal feature structure, correcting the scattering point trace of the optimal feature structure by using polynomial fitting, and switching the reference structures as a new optimal feature structure according to the scattering point loss rate and a preset sequence; and optimizing the spacecraft attitude angle solving function by using a particle swarm and LM hybrid algorithm to obtain attitude angle parameters. The target with high-precision target attitude real-time estimation can be completed aiming at the problems that high-order frequency change in a dynamic environment is difficult to capture and resolution and noise suppression are contradictory due to a fixed window time-frequency analysis method.

The invention relates to a large language model dynamic routing method and device based on context learning model characterization and a readable storage medium. Query is embedded and mapped to a language model input space by using a projection model, semantic alignment is realized, a representative evaluation set covering multi-dimensional capability is automatically screened from a benchmark question bank, and the representative evaluation set is used for evaluating the multi-dimensional capability. Performance characteristics of the model on an evaluation set are efficiently obtained at a time, and high-quality context model capability representation is formed; then real-time query embedding and context model capability representation are combined, a lightweight routing language model is used for supervised learning, so that the fine-grained model distinguishing capability is achieved, an increment embedding updating mechanism is designed, and when a new model is accessed or an old model is upgraded, cold start can be rapidly completed only through a very small number of fixed questions, so that the efficiency is improved. The calculation and maintenance cost is greatly reduced, and the accuracy, real-time performance and flexible expansibility of model routing are effectively improved.

The invention provides a low-earth-orbit satellite global beam position distribution and scheduling system based on deep reinforcement learning, and the system comprises a low-earth-orbit satellite scene modeling beam position distribution module which is used for constructing a low-earth-orbit satellite communication scene and completing the global beam position division. A satellite constellation configuration and orbit parameters are set, global beam positions are divided based on a spherical network, a beam ground projection model and a user system model are established, and a scene basis and a beam position division basis are provided for beam resource scheduling; the deep reinforcement learning beam resource scheduling module is used for realizing intelligent allocation scheduling of beam resources, establishing a decision system based on a Markov decision process by constructing a beam resource allocation model, extracting features through a state space model, performing time sequence modeling, and optimizing a scheduling strategy based on deep reinforcement learning. And dynamically matching the user demand and the beam resource. Therefore, dynamic optimization of low-orbit satellite beam resources is realized, the resource utilization rate is improved, and the communication requirements of users are efficiently met.

The invention discloses a construction method based on a Chinese poem-semantic knowledge graph, and particularly relates to the technical field of graph reasoning. The method comprises the following steps: dividing a poem text into a plurality of feature sub-channels according to style dimensions such as rhythm, rhyme, image, allusion and the like, extracting a heterogeneous style feature vector group, and generating an affiliation probability matrix by using a style fingerprint projection model; generating a style mixing label under an affiliation fuzzy condition, and realizing style affiliation reasoning based on atlas embedding and semantic path traversal; if the style boundary sample exists, triggering an atlas structure self-updating mechanism to realize extension and boundary learning of style nodes; the method effectively improves the accuracy and interpretability of Chinese classical poem style affiliation and the adaptive ability of the map structure, and is suitable for scenes such as poem style analysis, author affiliation judgment and style evolution modeling.

The invention discloses an unmanned aerial vehicle dynamic projection map construction method and system based on spatio-temporal data fusion, and belongs to the technical field of unmanned aerial vehicle video monitoring and geographic scene fusion, and the method comprises the steps: obtaining multi-modal flight data in real time based on an unmanned aerial vehicle, and the multi-modal flight data comprise video image data and spatio-temporal reference data; and performing data preprocessing on the multi-modal flight data to obtain route data. And constructing an unmanned aerial vehicle dynamic projection model, and performing space-time synchronous projection rendering on the route data to obtain an unmanned aerial vehicle dynamic projection map. And constructing a dynamic calibration model, and correcting the dynamic projection map of the unmanned aerial vehicle. According to the method, an integrated dynamic geographic information sensing system is constructed through deep cooperation of a spatio-temporal data fusion framework and an intelligent calculation engine, accurate mapping of the geographic space driven by spatio-temporal reference fusion is realized, millimeter-level space registration capability is constructed through multi-modal data intelligent solution and terrain adaptive rendering, and the real-time dynamic geographic information sensing system is constructed. And a technical breakthrough is formed in the dimensions of accurate positioning, data fusion, real-time response and the like.

A space-time learning and lightweight direct-current optimal power flow projection scheduling method driven by a reactance perception graph kernel comprises the steps that the reactance perception graph kernel based on a susceptance matrix is constructed, a power grid node reactance perception spatial dependency relationship is extracted, and multi-period node feature tensors are organized; a space-time learning device comprising a graph convolution encoder and a time sequence encoder is adopted to carry out space-time joint modeling on the node features, and the cross-time evolution law of the node marginal electricity price and the voltage phase angle is learned; mechanism consistent physical regular terms of load flow safety and climbing consistency are introduced in training, and physical constraint influences are gradually enhanced by means of weight scheduling; and converting the phase angle trajectory into an equivalent unit output initial value through physical mapping, constructing a lightweight quadratic programming projection model with the minimum deviation initial value, and generating a feasible day-ahead scheduling plan under the constraint of meeting power balance and the like. According to the method, the whole link from prediction to scheduling is opened, the speed and the reliability are both considered, and an engineering available path is provided for day-ahead multi-period clearing and large-scale scene analysis.

A method for previewing the terrain in front of an emergency rescue vehicle in a geometric feature degradation scene relates to the technical field of road surface recognition, realizes accurate alignment of LiDAR point cloud and IMU data through timestamp synchronization and linear interpolation, and combines a spherical projection model and a degradation perception complementary feature selection algorithm to realize the previewing of the terrain in front of the emergency rescue vehicle. Converting the three-dimensional point cloud into a robust intensity image and extracting gradient significant features; based on a double-observation secondary filter frame, a motion state is predicted by utilizing IMU forward propagation, a point cloud geometric residual error and an intensity image luminosity residual error are synchronously fused, timestamp deviation is compensated through back propagation, and a multi-source observation model under a global coordinate system is constructed. Finally, the error state is iteratively optimized to realize collaborative output of the high-precision odometer and the three-dimensional terrain map, and the problems of data asynchronism, feature degradation and dynamic interference in a complex scene are effectively solved.

Systems and methods for intuitive search and recommendation including a content comprehension engine executing on a computer processor and configured to: receive a recommendation request identifying a source content item; generate a first embedding for the source content item in a first embedding space from content metadata and contextual data; apply a trained neural projection model to map the first embedding to a second embedding space, thereby producing a projected embedding; compute, for content item models stored in a repository, a similarity score between the projected embedding and the content item model, each content item model including word-vector collaborative-filtering representations of an available content item; select, based on the similarity scores, a subset of the content item models; and output a result set including the available content items corresponding to the subset and ordered by the similarity scores.

A method and device for vehicle camera autocalibration during vehicle travel for driver assistance systems and automated driving. A vehicle camera images a region of vehicle surroundings through a window. The method includes: providing a vehicle camera projection model including multiple extrinsic parameters, at least one intrinsic vehicle camera parameter and at least one window parameter; capturing a sequence of camera images during vehicle cornering; determining a curve type during the cornering; estimating the parameters using pixels in the sequence of images of stationary objects in the vehicle surroundings, the current vehicle movement, and the determined curve type, by minimizing an error function indicating the deviation between pixels which correspond to stationary objects in the vehicle surroundings and which are established from the sequence of images and pixels of the stationary objects which are projected by the projection model; and outputting at least one of the estimated parameters.

The invention relates to the technical field of industrial Internet of Things and edge computing, in particular to an intelligent risk identification method and system for an energy increasing station. The method comprises the following steps: deploying a sensor network at an energy adding station and synchronously acquiring signals, extracting multi-scale time-frequency and physical response characteristics, mapping heterogeneous characteristics to energy flow fingerprints through a hybrid projection model, and performing self-calibration enhancement by using a differential baseline; energy flow fingerprints are decoded and mapped into semantic features, candidate abnormal modes are recognized through short-term trend analysis and multi-source cross correlation, and local abnormal scores are calculated; constructing a physical semantic constrained local causal network based on candidate anomalies, deducing an anomaly propagation path, quantifying node risk contribution, grading and positioning a key risk source; and packaging the risk information into a token matching strategy library, executing a transactional action chain according to a hierarchical intervention principle, monitoring an effect in real time and supporting rollback upgrading. According to the invention, real-time identification, causal deduction and intelligent dynamic response of the energy station risk are realized.

The invention relates to a cultural and creative product design display method and system based on augmented reality, and the method comprises the steps: obtaining design drawings of cultural and creative products, determining the projection relation between the design drawings according to the projection relation of the design drawings, and generating a three-dimensional projection model of the cultural and creative products for displaying the cultural and creative products in a display region; in the process of user-oriented display, the visual angle of the user is used as a rendering target of the three-dimensional projection model, so that the three-dimensional projection model is rendered only within the visual angle of the user. The immersive cultural and creative product display experience is provided for the user, the three-dimensional projection model of the cultural and creative product can be viewed from any angle, and the cognition and interest of the user on the product are greatly improved.

The invention relates to the technical field of mechanical manufacturing and machining, in particular to a dressing wheel and honing wheel design method for improving contact line stability in the gear honing machining process. Based on workpiece gear parameters, a left and right tooth surface mathematical model is established, a honing process space coordinate conversion relation is established, a conjugate meshing relational expression is derived by combining the relative slip speed of meshing contact points and a normal vector, and a tooth surface contact line model is established; and the calculation modeling of the two-dimensional projection of the left and right tooth surface contact lines, the projection intersection point position and the variation under the same rotation angle is realized. And a honing wheel dressing contact line model under dressing wheels with different tooth numbers and a tooth surface contact line intersection projection model in the dressing and honing process under the tooth number difference are established. A honing wheel and variable tooth number dressing wheel design method is provided, and geometric and technological parameters of the honing wheel and the variable tooth number dressing wheel are determined; and a trimming and honing two-degree-of-freedom conjugate meshing relation and a modified tooth surface numerical model are constructed and verified, discretization and point cloud processing are conducted on the model, high-precision tooth surface point cloud is obtained, and accurate machining of a trimming wheel and a honing wheel is guided.

The invention relates to the technical field of simulation optimization, and discloses a charging module heat dissipation optimization method based on temperature field simulation, and the method comprises the steps: firstly discretizing a heat dissipation space into voxel units, and collecting flow heat data; calculating the flow resistance sensitivity according to the relationship between the flow velocity vector and the main wind direction; calculating an effective convection potential energy density and a gradient vector thereof by combining a Reynolds number and a temperature difference; then, generating a chasing cold source driving vector based on the heat flux demand of the component; establishing a thermal wake flow cone projection model, and calculating a thermal wake flow rejection vector; generating a flow resistance avoidance vector in combination with the regional flow velocity direction; and finally, performing weighted synthesis on the three vectors to obtain optimized displacement, and updating coordinates of the component. According to the technical scheme, through multi-physics field coupling analysis, while a driving component finds a high cooling potential area, flow channel smoothness and thermal wake avoidance are effectively considered, and deterministic optimization of the charging module layout is realized.

The application discloses a 4D video generation method fusing monocular vision and terrain elevation data, comprising the following steps: acquiring video frame images and corresponding camera pose data collected by a monocular camera; acquiring terrain elevation data of a target area, and constructing a three-dimensional elevation network model; based on the camera pose data, establishing a ray projection model for pixels in the video frame images, calculating the intersection of the rays and the three-dimensional grid model, and obtaining the reference distance from the monocular camera to the ground surface corresponding to the pixels; generating an initial relative depth map of the video frame images by using a depth estimation model; selecting high-confidence ground pixels from the video frame images as reference points, taking the reference distance corresponding to the reference points as the true value, and performing absolute scale correction on the initial relative depth map to obtain an absolute depth map; performing back projection calculation on the absolute depth map, the camera pose data and pixel color information to generate a three-dimensional point cloud, mapping the three-dimensional point cloud to a real geographic coordinate system, and generating 4D video data in time sequence.

This invention discloses a video tag dynamic overlay anti-shake method and system based on AR, comprising: acquiring real-time video stream and device pose parameters at corresponding timestamps; constructing a dynamic perspective projection model based on the device pose parameters and calculating the initial pixel coordinates of the target AR tag in the current video frame; planning a local feature search window based on the initial pixel coordinates; calculating the average pixel displacement vector of image feature points in the local feature search window of the current and adjacent historical video frames; performing reverse displacement compensation on the average pixel displacement vector to obtain the target rendering coordinates, and rendering the target AR tag to the current video frame based on the target rendering coordinates. Compared with the prior art, this invention can overcome the shortcomings of large drift errors caused by simply relying on mechanical pose, high computational power consumption caused by global matching, and high-frequency jitter of tags caused by slight device vibrations, thereby improving tag mapping accuracy, reducing system computational power consumption, and improving the visual stability of video overlay.

The application discloses a kind of 3D target detection methods of point cloud image fusion based on basic model, applied to the field of automatic driving technology, comprising: inputting multi-view image into basic model, obtain image depth map;Image depth map and radar point cloud are input into semantic projection model, obtain radar depth map;Image depth map and radar depth map are input into filtering module, obtain enhanced radar point cloud;Respectively obtain the multi-scale image BEV feature and multi-scale radar BEV feature corresponding to multi-view image and enhanced radar point cloud, and input into multi-scale feature fusion module, obtain multi-scale fusion feature;Fusion feature is input into detection head, obtain detection result.The application realizes the densification of sparse radar point cloud by basic model, and realizes the complete fusion of multi-scale and multi-modal BEV feature by multi-scale feature fusion module, improves the accuracy and robustness of target detection.

This invention provides an automatic calibration method and apparatus for a ship panoramic surround view system based on UAV collaboration and 3D surface mapping. A virtual 3D surface projection model is established in the ship's coordinate system, and the surface normal vectors of sampling points on this model are mapped outwards to generate the UAV's flight trajectory. The UAV, equipped with an RTK positioning module, acts as a dynamic calibration target with precise three-dimensional coordinates as it flies along this trajectory, its spatial distribution naturally matching the geometry of the surface projection model. During flight, images from the UAV are simultaneously acquired by the ship's onboard camera, and pixel coordinates are extracted using a target detection algorithm. Simultaneously, the time synchronization data between the UAV and the ship's onboard RTK module is used to establish the correspondence between pixel coordinates and three-dimensional coordinates in the ship's coordinate system, minimizing reprojection errors and jointly calculating the extrinsic parameters of each camera. This eliminates the dependence on fixed planar references on shore, enabling automated calibration with precise adaptation to the 3D surface projection model in any water area.

The invention relates to an ocean dynamic wave direction sensing method and system based on an optical flow method, and the method comprises the following steps: S1, segmenting an input image, and obtaining a water surface image; s2, establishing a perspective projection model of a water surface plane and an image plane by adopting a BEV visual angle conversion algorithm, and converting the segmented water surface image into an aerial view image of an overlooking visual angle; s3, using a CLAHE algorithm to reinforce wave texture features in the aerial view image; and S4, setting multi-layer pyramid scaling by adopting a dense optical flow algorithm, calculating an optical flow field of a continuous multi-frame image sequence in the enhanced aerial view image, and counting a main optical flow direction to determine a wave propagation direction. According to the method, wave feature distortion caused by camera inclination is eliminated through a BEV visual angle conversion algorithm, the recognition precision is improved, and the self-adaptive image enhancement module can complete wave direction recognition in strong and dark light environments through a CLAHE algorithm; the method can be realized only by a common industrial camera without special water body marks or auxiliary equipment.

The application discloses a parking assistance method and related equipment, the method comprises the following steps: when it is detected that a vehicle reaches a preset range of a parking space, starting a parking assistance function, and projecting a vehicle model of the vehicle according to the parking assistance function to obtain a vehicle projection model; the vehicle projection model is optimized and adjusted to obtain a target vehicle projection model, and the target vehicle projection model is combined with the parking space to obtain an occupancy situation; if it is confirmed according to the occupancy situation that the target vehicle projection model and the boundary of the parking space do not have an overlapping part, and the distance between the target vehicle projection model and the left and right adjacent vehicles of the parking space is greater than a predetermined safety value, it is determined that the parking space meets the safety parking requirement, and the driver is prompted to park. The application projects the vehicle projection corresponding to the vehicle in front of the vehicle, combines the virtual and the real with the parking space, and then enables the driver to project to accurately determine whether the parking space can be parked and facilitate opening the door, so as to improve the safety of parking.

The invention belongs to the field of aero-engine and gas turbine machining, and discloses a blade margin plate and adapter numerical control machining tool path generation method and a related device, and the method comprises the steps: building a projection part model of a blade; establishing a reference plane at a preset distance from the blade inner edge plate, generating an intersection curve, offsetting to obtain an offset curve, and constructing a driving curved surface through regular extension; the radius of the milling cutter is not larger than that of the switching fillet; and on the basis of the projection model, variable contour milling is conducted through the driving curved surface, and an integrated numerical control machining tool path of the margin plate and switching is generated. According to the method, the machining cutter connecting position is transferred to the blade body molded surface area from the sensitive transfer fillet, and the root molded surface thinning out-of-tolerance caused by polishing is prevented. And a standard cutter is adopted to process a qualified transfer fillet through cutter path control, so that the cutter customization cost is reduced. The finally generated tool path lines are uniform, transition is continuous, a smooth switching structure can be directly machined, dependence on polishing is remarkably reduced, and the machining quality and the percent of pass are improved.

The invention discloses a construction site walking working face slip risk quantification system and method based on an unmanned aerial vehicle. The system comprises a data acquisition module, a terrain safety risk quantification module, an earth surface safety risk quantification module, a slip risk quantification module and a walking working face visual management module. The method is based on a high-resolution orthographic projection model (DOM) image and an elevation digital terrain model (DEM) which are automatically acquired by an unmanned aerial vehicle, and is combined with a YOLOv8-seg deep learning model of multi-scale driving and multi-window slice optimization to realize refined identification and grading of the terrain and earth surface slip risk of a construction site. And generating a global slip risk quantification map through hexagonal block gridding and a preset two-dimensional fusion mapping matrix. According to the invention, the sliding hidden danger distribution can be accurately mastered, the high-risk area can be timely found, the targeted intervention can be adopted, the sliding accident rate of the walking working face can be obviously reduced, and the construction safety management level can be improved.

The embodiment of the invention discloses a monocular four-dimensional video generation method and device based on geometric constraints and electronic equipment. A specific embodiment of the method comprises the following steps: extracting video geometric dynamic information from a monocular video frame sequence to obtain video geometric dynamic information; performing alignment correction on the initial depth map sequence according to the static region three-dimensional point cloud to generate an aligned and corrected depth map sequence; inputting the aligned and corrected depth map sequence into a camera projection model to obtain a video frame three-dimensional point cloud sequence; generating a dynamic three-dimensional point cloud sequence and a static three-dimensional point cloud sequence; generating a dynamic initial Gaussian point cloud and a static background Gaussian point cloud; performing geometric constraint updating on the initial point cloud transformation model to obtain an updated point cloud transformation model; and generating a monocular four-dimensional video according to the static background Gaussian point cloud, the dynamic initial Gaussian point cloud and the updated point cloud transformation model. According to the embodiment, the quality and rendering efficiency of the monocular four-dimensional video can be improved.

This invention relates to the field of machine learning technology, and discloses a data processing method, apparatus, and medium based on model adapters in heterogeneous devices. The method includes: dividing a target model into several functional regions; sequentially selecting target functional regions from these regions using a least-last-updated algorithm; obtaining initial adapter parameters for each heterogeneous device corresponding to a target functional region based on a preset loss projection model, according to all target functional regions and the corresponding heterogeneous devices; obtaining target global adapter parameters for the target functional region based on all initial adapter parameters; and obtaining a global adapter parameter set based on all target global adapter parameters. By aggregating the optimal data processing strategies from all different heterogeneous devices, the global adapter parameter set can control the target model to use the optimal processing method for data from other devices, thereby reducing the computational burden on the model.

This invention discloses a dynamic monitoring and alarm system for electrical fire hazards based on big data analysis, belonging to the field of electrical safety monitoring and intelligent early warning technology. It includes modules for visual space modeling and interference prediction, temporal thermal anomaly extraction, structural contour recognition and consistency judgment, temperature disturbance assessment and credibility grading, composite credibility modeling and hotspot correction, and dynamic scoring and identification model update. The visual space modeling and interference prediction module acquires the field-of-view projection model of the thermal imaging monitoring area and establishes a high-risk pixel distribution model for reflection interference based on camera installation parameters and spatial geometric features. This invention constructs an intelligent identification closed-loop system from perception to decision-making, integrating spatial modeling, thermal behavior analysis, and dynamic credibility assessment to achieve accurate identification and continuous adjustment of false hotspots, effectively avoiding false alarms and missed alarms, and significantly improving the accuracy and system stability of electrical fire early warning.

This invention provides a final layer training method and apparatus for a learning model that utilizes text as a substitute for images, reduces annotation costs for data group information where class sets and pseudo-attribute sets are matched, does not require the collection of separate image datasets balanced between data groups, and mitigates bias in image classifiers. [Solution] The final layer learning method of a learning model, performed by the final layer learning device of a learning model, includes the step of inputting training text into the second learning model and training the final layer of the first learning model, based on a projection model that connects the first embedding space of the first learning model and the second embedding space of the second learning model.

The invention provides a surface mine slope deformation monitoring method based on sequential topographic parameters, and belongs to the technical field of slope deformation monitoring. According to the method, radar satellite images and unmanned aerial vehicle high-frequency DEM data of a mine side slope area are efficiently collected, and through data processing and feature calculation, one-dimensional deformation results in the LOS direction of rail ascending and rail descending radars and sequential topographic parameters including the gradient, the slope direction and the topographic curvature are obtained respectively; and dynamically updating time sequence terrain parameters based on a mining activity period, constructing a projection model of radar sight line direction deformation and surface adaptive deformation, solving a three-dimensional deformation result of the mine slope, and realizing dynamic monitoring. According to the method, the problem that the slope deformation monitoring precision is insufficient due to dynamic change of terrain parameters is solved, a slope deformation precise monitoring system from data acquisition, parameter updating to model solving is formed, the limitation of a traditional method in the aspects of precision and dynamic adaptability is broken through, and technical support is provided for safe operation of the surface mine slope.

This invention discloses a rapid calculation method for the rearward infrared radiation intensity of an S-curve nozzle, belonging to the field of infrared radiation characteristic simulation of aero-engine exhaust systems. The method comprises the following steps: determining the design parameters of the S-curve nozzle and exhaust mixer components; establishing projection models of the key shielding sections of the exhaust mixer and S-curve nozzle; calculating the projected area of ​​each component after shielding by the S-curve nozzle using an image method; calculating the flow field parameter distribution and component surface temperature of the S-curve nozzle; calculating the axial length and volume of the visible region of the exhaust gas in the S-curve nozzle; calculating the spectral transmittance and spectral radiance of the exhaust gas along the nozzle outlet centerline within the visible region of the exhaust gas, as well as the spectral radiance intensity of the exhaust gas; calculating the infrared spectral radiance intensity of the wall surface; and integrating the spectral radiance to obtain the total infrared radiation intensity. This method solves the problems of existing technologies where the simulation calculation of the infrared radiation characteristics of S-curve nozzles is complex and time-consuming, and rapid calculation methods are only applicable to axisymmetric nozzles and cannot simulate the shielding structural characteristics of S-curve nozzles.

Dental treatment monitoring systems and methods may include accessing an input image of teeth taken at a particular time during dental treatment, and determining virtual-camera parameters that represent an estimated position and orientation of a virtual camera for producing a generated image from a time-projected 3D model of the teeth. The virtual-camera parameters may be iteratively adjusted by: generating a first generated image by modifying the virtual-camera parameters based on a first jaw in the generated image; determining a pixel-associated cost based on a comparison of the first generated image to the input image; generating a second generated image by modifying the first virtual-camera parameters based a second jaw in the first generated image; and determining a pixel-associated cost based on a comparison of the second generated image and the input image. The generated image may be generated from the time-projected 3D model using the adjusted virtual-camera parameters.

A reactive power-aware graph kernel driven spatio-temporal learning and light DC optimal power flow projection scheduling method, comprising: constructing a reactive power-aware graph kernel based on the admittance matrix, extracting the reactive power-aware spatial dependence relationship of the grid nodes, and organizing the multi-period node feature tensor; using a spatio-temporal learner containing a graph convolutional encoder and a time series encoder to model the node features in space and time, and learn the cross-period evolution law of the node marginal price and the voltage phase angle; in the training process, the mechanism consistent physical regularization term of power flow safety and climbing consistency is introduced, and the influence of physical constraints is gradually enhanced through weight scheduling; the phase angle trajectory is converted into the equivalent initial value of unit output through physical mapping, and a light quadratic programming projection model with minimum deviation initial value is constructed to generate a feasible day-ahead scheduling plan under the constraints of power balance and other constraints. The method connects the full link from prediction to scheduling, balances speed and reliability, and provides an engineering usable path for day-ahead multi-period clearing and large-scale scenario analysis.