2518results about How to "Reduce dependence" patented technology

The invention discloses a multi-modal large model automatic evaluation method and system based on Agentic Workflow, relates to the technical field of artificial intelligence model evaluation, and solves the technical problems that an existing large model evaluation method lacks image-text evaluation, is relatively high in subjectivity and is easy to cause illusion. The method comprises the following steps that S1, an input task is analyzed, and a structured task template is generated; s2, generating a multi-round multi-modal task flow chart based on the task template generated in the step S1; s3, based on the flow chart in the step S2, distributing the multi-modal large model component to each task node, executing a task and obtaining an intermediate result; s4, performing multi-dimensional scoring on the intermediate and final results in the step S3 by a plurality of scoring Agents, and fusing to generate an evaluation result; and S5, generating a feedback report based on the process data in the steps S1 to S4, identifying a model short plate, and providing an optimization suggestion. The method has the advantages of comprehensive evaluation of multiple modes, real-time feedback optimization and the like.

The invention discloses a photovoltaic-energy storage-charging multi-stage scheduling and market bidding optimization method and a photovoltaic-energy storage-charging multi-stage scheduling and market bidding optimization device. The method comprises the following steps: constructing a data-driven random environment model reflecting photovoltaic output, electricity price fluctuation and charging load uncertainty by adopting a mode of combining time sequence clustering and a non-homogeneous Markov chain based on historical operation data; modeling a scheduling and bidding problem of the optical storage and charging integrated station into a multi-stage Markov decision process model which comprises day-ahead decision and joint optimization of multiple intra-day rolling adjustment; a deep reinforcement learning algorithm is utilized to train the network, and a strategy regulation and control network which can adapt to various uncertain scenes and meet equipment physical constraints is obtained; and deploying the trained strategy regulation and control network in an energy management system to realize global coordinated scheduling and bidding of the optical storage and charging integrated station. According to the method, the economic benefit is remarkably improved, the robustness is greatly enhanced, the decision is globally coordinated and optimized, the real-time decision capability is strong, and the expandability and portability are good.

The invention relates to the technical field of deep foundation pit monitoring, and discloses a deep foundation pit micro-deformation InSAR remote sensing method and system and a storage medium. The method comprises the following steps: synchronously acquiring data through a ground radar and an InSAR satellite, and obtaining a point target set and an original scattering characteristic sequence; obtaining a net deformation phase sequence without influence through registration residual correction, phase deduction and weighted fusion processing, and carrying out phase consistency quality control and unwrapping processing to obtain an accumulated deformation quantity sequence; constructing a deformation mechanism correlation characteristic matrix based on correlation analysis of the cumulative deformation quantity sequence and radar reflection intensity amplitude attenuation characteristics; and fusing the deformation mechanism correlation feature matrix and the real-time differential interference image group, extracting deformation rate and phase gradient change, and further generating a deep foundation pit local micro-deformation risk early warning index set. According to the invention, the tiny deformation of the surrounding area of the foundation pit can be effectively detected, potential risks are warned in advance, and the safety and monitoring efficiency in the construction process are improved.

The invention relates to the technical field of industrial vision multispectral imaging detection, in particular to a defect detection method and system based on multispectral fusion imaging, and the method comprises the steps: synchronously collecting a multi-channel spectral image of the surface of an object to be detected, and generating an enhanced feature map fusing multispectral information; identifying and preliminarily marking a suspected defect region by applying a region segmentation algorithm based on anomaly detection; extracting a multi-dimensional spectral response curve of each region in each original spectral channel, and constructing a spectral feature vector; matching the vector with a standard template of a pre-established defect spectral feature database, and classifying and confirming defect types; and carrying out contour refined analysis, calculating the geometric dimension and position of the defect, and generating a structured detection report. According to the method, the sensitivity of defect detection and the classification accuracy are improved through accurate matching of multispectral fusion and spectral features, and automatic and high-precision defect identification is realized.

The invention relates to the technical field of numerical control machining and automatic control, in particular to a numerical control machining self-adaptive control system and method based on multi-source data, and the method comprises the steps that FPGA hardware synchronously collects main shaft current, servo current of each feed shaft, cutting vibration and grating ruler pulse, and time bases are aligned and packaged into a machining multi-dimensional state matrix; the multi-physical field error is decoupled through a three-layer cascade machine tool dynamics model, a tool nose contour deviation vector is solved, and servo lag and cutting force deformation factors are fused to construct a dynamic constraint boundary. When the deviation amplitude exceeds the limit, generating a speed feed-forward gain compensation instruction based on an inverse control model to perform instant suppression; when it is monitored that oscillation is eliminated and returns to a steady state, full-time-domain data stream interception logic is triggered, and a process optimization instruction is generated through backtracking. According to the method, the problems of nonlinear time-varying error decoupling and real-time compensation lag under multi-physics field coupling are solved, and the dynamic precision and process stability of numerical control machining under complex working conditions are remarkably improved.

The invention discloses a self-calibration and drift compensation method for an automatic underground water monitoring station, and the method comprises the following steps: 1, transmitting a reference measurement starting instruction to an underground bubble method device by an edge calculation gateway, forming a reference data pair by a bubble method water level reference value and a reference acquisition moment, and writing the reference data pair into a reference data cache region; 2, the edge computing gateway creates a state storage structure and a covariance storage structure; and step 3, the edge computing gateway reads a drift amount estimation value from the state storage structure as a drift compensation amount, and when a drift estimation standard deviation exceeds a set threshold, bubble method hardware reference acquisition is triggered. According to the invention, the problems of high manual calibration cost and lack of independent reference support in pure software estimation in the prior art are solved, so that a monitoring station can stably operate for a long time under an unattended condition and continuously output high-precision underground water level monitoring data.

The invention discloses a heat supply network pipeline hydraulic and thermal working condition abnormal fluctuation intelligent identification and fault early warning method, and relates to the technical field of computer processing, and the method comprises the following steps: S01, collecting hydraulic and thermal parameters of a heat supply network pipeline in real time; s02, acquiring abnormal data of the time sequence sample set based on the thermodynamic mechanism model, detecting and repairing the abnormal data; s03, extracting characteristic parameters capable of reflecting abnormal fluctuation of the hydraulic and thermal working conditions of the heat supply network pipeline from the repaired time sequence sample set; s04, establishing an abnormal fluctuation identification model by utilizing an intelligent algorithm, and judging whether the hydraulic and thermal working conditions of the heat supply network pipeline have abnormal fluctuation or not; and S05, when the abnormal fluctuation is identified, giving a grading early warning mechanism and risk prevention and control suggestions of the damage degree grade of the associated pipeline. According to the invention, by fusing multi-source data and an intelligent algorithm, full-chain intelligent management of the hydraulic and thermal working conditions of the heat supply network pipeline from abnormal accurate sensing, fault intelligent diagnosis to risk active early warning is realized.

The invention relates to the technical field of data processing, and discloses a roller press parameter adjusting method and system, rolling equipment and a storage medium, and the method comprises the following steps: obtaining target parameters of a roller press reaching a preset operation state, and carrying out parameter configuration on roll gaps at all levels to form a first parameter sequence; on the basis of the first parameter sequence and the target parameter, determining speed ratio configuration among all sections of rollers of the roller press to form a second parameter sequence; the roller press is controlled to execute feeding operation according to the first parameter sequence and the second parameter sequence, and diaphragm detection data formed through processing of the roller press are collected; and correcting the parameter sequence according to the difference between the diaphragm detection data and the target parameter to obtain a target diaphragm parameter. According to the invention, automation and standardization of debugging of the roller press are realized, the precision of parameter configuration and the production consistency are remarkably improved, the dependence on manpower is reduced, the targets of membrane thickness and surface density are stably realized, and the machine debugging efficiency and the product yield are improved.

The invention provides a multi-reflection superposition enhanced microwave displacement measurement method and system, and the method comprises the steps: S1, constructing a microwave multi-reflection propagation path, enabling a microwave radar to transmit electromagnetic waves to the microwave multi-reflection propagation path as a transmission signal, and receiving a corresponding echo signal; s2, extracting length change information of a microwave multi-reflection propagation path according to the echo signal; s3, establishing a linear superposition relation model between the total length of the microwave propagation path and respective displacements of W reflection surfaces in the microwave multi-reflection propagation path; w2; and S4, based on the linear superposition relation model, displacement information of the corresponding reflection surface is obtained and output. According to the invention, a larger phase variation can be obtained under the condition of small displacement or weak vibration, so that the measurement sensitivity is improved and the signal-to-noise ratio is improved.

The invention discloses a multi-modal contrast learning fault diagnosis method for a small sample scene, and the method comprises the following steps: carrying out the enhancement of a one-dimensional signal and two-dimensional image fused fault data set through physical simulation for the small sample scene with scarce industrial fault data, and constructing a positive and negative sample pair through a plurality of data enhancement strategies; based on heterogeneous multi-modal fault data, designing a double-flow encoder architecture of a time sequence branch and an image branch, extracting depth features and mapping the depth features to a unified feature space through a projection head; performing supervised contrast learning pre-training based on intra-modal and inter-modal dual contrast loss; supervision fine tuning is carried out based on multiple loss functions such as physical guidance, so that accurate diagnosis of equipment faults is realized in a small sample scene. According to the fault diagnosis method under the unbalanced sample and limited labeling conditions, the problem that a traditional data driving model depends on large-scale labeling samples is effectively relieved through supervised comparative learning and cross-modal information alignment.

The invention discloses a multi-source heterogeneous data dynamic causal blood relationship tracing method and a related device, and belongs to the technical field of data blood relationship detection.The method comprises the steps that static blood relationship metadata and dynamic index time sequence data of aviation test data are collected, a dynamic causal blood relationship map is constructed, each edge of the map is associated with a causal strength weight, and the dynamic causal blood relationship is obtained; and reverse traceability is carried out through the dynamic causal blood relationship map, and root cause nodes causing abnormity are determined. Static consanguinity and dynamic indexes are collected, a dynamic causal consanguinity map with nodes as triads and edges carrying causal strength weights is constructed, and dynamic attributes of the map break through static relevance limitation, so that traceability can reveal state change and causal dynamics; the influence degree between nodes is quantized by introducing a causal strength weight, and the root cause is determined by combining a probabilistic reasoning model, so that the spanning from fuzzy association to quantitative positioning is realized, and the problems that the root cause positioning is fuzzy and is lack of quantitative analysis are solved.

The invention discloses a voiceprint recognition system attack resisting method and device, equipment and a medium, by constructing a shadow voice sample set and a substitution model, dependence on target speaker data and query capability is reduced, and privacy risk and query overhead are reduced; meanwhile, two types of substitution models are trained based on an information theory method, so that the mobility and stability of the models are improved, and the generalization ability is enhanced; besides, the attacker model adopts a specific architecture and combines an alternate training strategy, so that attack loss and prediction probability difference are optimized, imperceptibility and attack effectiveness are balanced, and tone quality reduction or attack failure is avoided; and finally, through systematic training and optimization, a more uniform evaluation caliber is expected to be provided, and more reliable guidance is provided for engineering landing and risk evaluation.

The application relates to the field of information technology, in particular to an intelligent processing method and system for new media data. The method comprises the following steps: reading historical new media material data; analyzing the historical new media material data, calculating an availability estimation value of the historical new media material, and sorting and screening the historical new media resource according to the availability estimation value; storing the analyzed historical new media material data in a classified manner, and establishing a multidimensional index of the historical new media material data; receiving a new media content generation demand, converting the new media content generation demand into a new media material matching condition; screening and sorting new media materials based on the new media material matching condition, and selecting the first N new media material data as basic new media material data; and fusing a theme content based on the basic new media material, generating multi-form new media content, and realizing intelligent processing of high-efficiency, high-quality and self-optimizable new media content.

The invention provides a tumor medical image classification method of zero sample evolution NAS based on double-index collaborative evaluation, which comprises the following steps: collecting tumor medical image data, and constructing a data set containing a tumor focus image and a normal tissue image; constructing an extended search space based on a cell structure, wherein the search space comprises a basic operation set, the cell structure and a network overall structure; steady-state evolution algorithm parameters are set, and steady-state evolution initialization is carried out to obtain an initial candidate architecture; taking a double-index comprehensive score calculated by the information rate and the FIM stability as a comprehensive index for evaluating the current candidate architecture, updating the candidate architecture through a steady-state evolutionary algorithm based on a dynamic optimization mechanism until the maximum evolutionary algebra is reached, and obtaining an optimal architecture; and training the optimal architecture by using the data set, and carrying out tumor medical image classification by using the trained optimal architecture. According to the method, the dependence on the annotated data is reduced, the consumption of computing resources is reduced, and meanwhile, the diagnosis efficiency and accuracy are improved.

The invention discloses a bio-based recoverable high-thermal-conductivity dynamic ordered network composite material and a preparation method thereof, and is characterized in that the material takes a bio-based multifunctional epoxy system and a carboxylic acid compound containing disulfide bonds as reaction monomers, and the reaction monomers are reacted under the conditions of high temperature and metal salt catalyzed epoxy-carboxylic acid esterification reaction to obtain the bio-based recoverable high-thermal-conductivity dynamic ordered network composite material. A dynamic ordered network structure simultaneously containing disulfide bonds and beta-hydroxyl ester bonds is constructed, and the dynamic ordered network can be used as a springboard for phonon transmission, serves as an efficient'thermal bridge ', and plays a role similar to an ordered'crystal domain', so that higher heat-conducting property is shown when heat-conducting filler is not introduced, high flexibility and intrinsic heat conduction are realized, and the service life of the material is prolonged. And the material is endowed with reprocessing and recycling characteristics. Due to the synergistic existence of carboxyl, hydroxyl generated by epoxy ring opening and metal coordination in the system, and in combination with electrostatic interaction, Van der Waals force and hydrophobic interaction, the material has the adhesion characteristic suitable for a thermal interface at the same time.

The invention belongs to the technical field of truss girder hoisting, and particularly relates to intelligent hoisting equipment for a suspension bridge truss girder in an extreme wind environment and a hoisting construction method.The intelligent hoisting equipment comprises a floating crane, and a hoisting tool component is fixedly installed at the lower end of a steel wire rope of the floating crane and comprises a posture adjusting device and an offset adjusting device. According to the intelligent hoisting equipment for the suspension bridge truss girder in the extreme wind environment and the hoisting construction method, by arranging the sling component and integrating the posture adjusting device and the offset adjusting device, multi-degree-of-freedom active adjustment can be conducted on the hoisted truss girder. The offset adjusting device generates reverse inertia force by driving the mass block to move, so that horizontal swinging of the truss girder caused by wind load is effectively counteracted; the posture adjusting device accurately corrects pitching and rolling postures of the truss girder by adjusting the length of the steel wire rope set, and therefore the overall wind resistance stability and posture keeping capacity of the hoisting system are remarkably enhanced in the extreme wind environment.

The invention is suitable for the technical field of vehicle dynamics modeling and control, and provides a preview active suspension stiffness control method based on road surface grading and MAP mapping, and the method comprises the steps: collecting a road image through a vehicle-mounted visual sensor, dividing regions, and recognizing the road surface unevenness grade of each region through a convolutional neural network; obtaining the vehicle speed; and according to the current pavement grade and the vehicle speed, a vehicle speed-pavement grade-rigidity mapping MAP table generated in advance is inquired, and the target suspension rigidity is rapidly obtained and adjusted. According to the invention, the dependence on a high-precision sensor and a high-computing-power processor is reduced, and low-cost and fast-response rigidity optimization control of the preview active suspension is realized.

The invention relates to the technical field of underground water simulation, in particular to an underground water numerical simulation method coupled with vegetation ecology. The method realizes bidirectional coupling between groundwater and vegetation ecology through iteration, and specifically comprises the following steps: inverting vegetation coverage based on groundwater burial depth; inversion of actual evapotranspiration is carried out based on underground water burial depth and vegetation coverage; and finally, updating the net supply amount of the groundwater model based on evapotranspiration until convergence. According to the method, a statistical function relation is established to replace a complex physical model, and drainage and overflow cell water levels are corrected in iteration to ensure calculation stability.

The invention relates to the technical field of motors, in particular to a magnetic self-centering motor assembling method which is applied to a motor and comprises the following steps: forming a rotor assembly; a rotor assembly is rotatably and preliminarily installed on a base through a rotating shaft assembly, an end cover is preliminarily installed on the base, a bearing chamber of the end cover supports an outer ring of a bearing on the rotating shaft assembly through a radial elastic supporting unit, and the radial elastic supporting unit allows the outer ring of the bearing to slightly and radially float; a driving current is applied to the stator assembly, so that the rotor assembly generates a rotation trend or rotates at a low speed, and the rotor assembly is driven to overcome the micro radial rigidity of the radial elastic supporting unit by utilizing the unbalanced magnetic pulling force of interaction between the rotor assembly and the stator assembly; and the rotation axis of the stator assembly is automatically adjusted to a position which tends to be coaxial with the central axis of the stator assembly. The method can actively compensate radial dimension chain errors and assembly deviations accumulated in the previous process.

The invention relates to the technical field of automatic welding, in particular to a self-adaptive intelligent welding system based on a dynamic coefficient model. The system comprises an information acquisition module used for acquiring point cloud data of a workpiece to be welded and extracting geometric and physical characteristics; the welding seam recognition module is used for judging the optimal welding seam type; the process parameter generation module is used for activating the single-channel fillet weld, multi-layer and multi-channel or groove welding sub-module through a process calculation agent and generating initial process parameters by utilizing a built-in dynamic coefficient model; the welding execution and feedback module is used for issuing a process instruction and performing real-time fine adjustment on voltage, current or speed based on a sensing signal linkage process coefficient; and the incremental learning module is used for continuously updating the model coefficient by analyzing the bias of the positive sample data packet. The full-closed-loop intelligent welding method solves the problems that traditional welding depends on a static expert database and cannot cope with assembly errors, complex heat accumulation and the like, and full-closed-loop intelligent welding from geometric perception to parameter generation and real-time correction is achieved.

The invention discloses a power grid monitoring alarm event handling method and system based on a large model, and relates to the technical field of power systems, and the method comprises the steps: obtaining a historical sample of a power grid monitoring alarm event, and carrying out the preprocessing; constructing a balance training sample set based on a first large language model and resampling combined sample enhancement method; based on the pre-training language model, utilizing the balance training sample set to carry out fine tuning training, and constructing a power grid monitoring alarm event diagnosis model; on the basis of a second large language model, a power grid monitoring alarm event disposal model is constructed, and the disposal model is used for retrieving reference information from a historical disposal knowledge base in combination with an event type diagnosis result output by the diagnosis model and generating an auxiliary disposal suggestion for an input alarm event text; and integrating the diagnosis model and the disposal model to form a power grid monitoring alarm event disposal framework. According to the invention, intelligent diagnosis and auxiliary disposal of the power grid monitoring alarm are realized, and the diagnosis accuracy and the disposal efficiency are improved.

This application discloses a method and apparatus for constructing a multi-scale adaptive gating MambaPlus network, belonging to the field of artificial intelligence and machine learning technology. The method includes: initializing the network configuration and constructing the basic structure; preprocessing the input data to generate a standard dataset; mapping the input data to the hidden space via an input mapping layer, and extracting backbone features from the Mamba backbone; constructing at least two parallel scale branches in the hidden space to obtain a multi-scale feature set; inputting the backbone features and multi-scale features into an adaptive gating module, dynamically allocating weights and adaptively fusing them through a hierarchical gating mechanism to generate fused features; further enhancing the features through cross-scale attention and feedforward enhancement, and then superimposing the residuals to generate the final discriminative features; finally, completing category prediction and model training evaluation. This application, while retaining the advantages of Mamba's long-range dependency modeling, addresses the problems of insufficient utilization of multi-scale information, poor adaptive feature fusion, and low robustness in complex scenarios.

The invention discloses an intelligent view collection method and system based on an unmanned aerial vehicle cluster, and relates to the technical field of unmanned aerial vehicle task planning, and the method comprises the steps: carrying out the three-dimensional space grid division of a target region, determining the flight attribute and priority weight of each grid unit, and obtaining a target grid unit set; candidate viewpoints are generated for the target grid units, and a coverage relation matrix is established; performing viewpoint selection under energy constraint by adopting a set coverage algorithm to obtain a viewpoint task set; distributing viewpoint tasks for each unmanned aerial vehicle through a distributed negotiation mechanism according to the unmanned aerial vehicle cluster state, and forming a task execution sequence; monitoring the execution state of the unmanned aerial vehicle cluster and dynamically adjusting task allocation; and constructing a spatio-temporal index structure of the view data, calculating the comprehensive coverage quality of each target grid unit, identifying a missing region and generating a supplementary shooting task. According to the method, the task allocation of coverage optimization and load balancing is realized under the energy constraint, and the efficiency and reliability of view acquisition are improved.

The invention discloses a real-time prediction and control method and system for the deformation state of a superplastic forming part based on digital twinning, and belongs to the technical field of intelligent manufacturing. The method aims at solving the technical problems that in the superplastic forming process, the deformation state of a part is invisible, control depends on experience, and quality is unstable due to open-loop control. The core lies in that a multi-task learning artificial intelligence prediction model taking air inflow time sequence data as input and taking pressure in a mold cavity and a part full-field deformation state as output is constructed by fusing parameterized finite element simulation and physical experiment data in an offline stage; in the online stage, the model is utilized to dynamically predict deformation states such as a strain field and a thickness field in the part according to air inflow data collected in real time, and process transparency is achieved; and a model prediction control algorithm is further combined, a prediction state is compared with an ideal path, the air inlet pressure is reversely optimized and adjusted in real time, and a closed-loop intelligent control system is formed. According to the method, perspective and active accurate control of the internal state in the black box forming process are achieved, and the quality consistency and the yield of parts can be improved.

The invention discloses an autonomous evolution type 5G protocol fuzzy test system based on a large language model, and the system comprises a protocol extraction unit which employs the large language model to analyze a 5G protocol specification document, and extracts structured protocol knowledge elements containing a message format, state machine logic and parameter constraints; the vulnerability analysis unit generates a potential vulnerability description list through big language model reasoning based on the knowledge elements; the test strategy generation unit drives a large language model to generate a targeted test case based on protocol knowledge, vulnerability description and historical test knowledge in the dynamic test knowledge base; the test execution unit executes the use case and captures a result; and feeding back an analysis result of the internalization unit, extracting a causal rule, generating new test knowledge and feeding back the new test knowledge to the knowledge base. According to the system disclosed by the invention, test targets and resources can be autonomously scheduled according to continuous updating of the knowledge base, so that the test process has an iterative optimization capability, and full-process automation and intelligentization of 5G protocol fuzzy testing are realized.

The invention discloses a satellite autonomous task planning method and system based on a ground digital mirror image, and belongs to the technical field of spacecraft measurement and control. Through the states of a satellite-ground link synchronous satellite and a ground digital mirror image, a task request is received in the ground digital mirror image, a planning algorithm is operated, a candidate scheme is generated for high-fidelity simulation deduction and verification, a feasible scheme is evaluated and selected and converted into an instruction sequence to be uploaded to a satellite, the satellite executes an instruction and feeds back the state, and a closed loop is formed. According to the method, a complex planning and verification process is placed in a ground digital mirror image, the efficiency and reliability of task planning are remarkably improved by using the strong computing power of the ground, meanwhile, the autonomous operation capability of a satellite is improved, and the problems that traditional satellite task planning is slow in response and poor in reliability, and the autonomous planning capability of a pure satellite is weak can be solved.

The invention discloses an unmanned aerial vehicle signal detection and identification method based on spectrum feature enhancement. The method comprises the following steps: firstly, carrying out time-frequency transformation on a received signal to obtain an original time-frequency graph; performing learning enhancement on the degenerated time-frequency graph by adopting a coding-decoding type deep neural network, and realizing noise suppression and structure recovery by combining pixel reconstruction, structural similarity and a texture perception loss function; and finally, performing target area detection and positioning on the enhanced time-frequency graph, directly outputting a structured result containing a time-frequency range, a category and confidence, and completing conversion from a frequency spectrum to a linkable engineering target. According to the unmanned aerial vehicle signal detection and recognition method based on spectrum feature enhancement, a spectrum feature enhancement mechanism is introduced before traditional spectrum analysis and feature recognition processing, and region-level detection and judgment are executed under the enhanced spectrum constraint condition; reliable discovery, positioning and identification of an unmanned aerial vehicle control link and an image transmission link in a complex electromagnetic environment are realized.

The invention belongs to the technical field of bar production, and particularly relates to a production method of a steel bar for a heavy-load automobile half axle, and the steel bar comprises the following chemical components in percentage by mass: 0.39-0.42% of C, 0.25-0.35% of Si, 1.45-1.70% of Mn, less than or equal to 0.013% of P, less than or equal to 0.006% of S, less than or equal to 0.25% of Cr, less than or equal to 0.25% of Ni, 0.15-0.20% of Mo, 0.06-0.10% of V, less than or equal to 0.15% of Cu and the balance of Fe and inevitable impurities. The method has the advantages that high-performance, high-purity and low-cost production of the steel bar for the heavy-duty automobile half shaft is realized through green short-process production of'ecological electric furnace + LF + RH + continuous casting 'in combination with low-carbon component design and a controlled rolling and controlled cooling process. And macrostructures and non-metallic inclusions of the finished steel meet the requirements of high-end products.

The invention relates to the technical field of image data processing, in particular to a hub surface defect detection method and system based on image processing, and the method comprises the steps: shooting the surface of a hub, obtaining an enhanced grayscale image, and carrying out the superpixel segmentation of the enhanced grayscale image, so as to obtain a connected domain; constructing a connected domain set of a target connected domain of an enhanced grayscale image shot over against the surface of the hub; respectively inputting the connected domains in the connected domain set into the trained SVM model, and outputting the defect probability that each connected domain belongs to each defect type; and performing weighted fusion on the defect probability of each connected domain belonging to the target defect type to obtain a comprehensive probability of the target connected domain belonging to the target defect type, and when the comprehensive probability is greater than a defect threshold, judging that the target connected domain has the defect of the target defect type. According to the method, the adaptive measurement distance is constructed, so that the superpixel segmentation process is more fit with the hub surface physical structure and the defect real boundary, and the defect detection accuracy is improved.

The invention provides a clothing pattern intelligent generation method and system based on multi-modal visual identification, and relates to the technical field of visual identification. The method comprises the following steps: firstly, acquiring clothing image data and text description data, performing unified preprocessing, obtaining a style category label, an appearance attribute label and a local contour key point set through multi-modal visual identification, generating structure semantic elements and geometric anchor points by utilizing a semantic-to-structure mapping knowledge base, and constructing a plate type structure diagram; further analyzing the structural constraint and executing constraint solution to generate a candidate pattern; and in combination with manufacturability verification and constraint correction, closed-loop optimization is formed, a target plate type file meeting the structure and manufacturing requirements is output, and effective connection from semantic recognition to plate type structure generation is realized.