2004results about How to "Adaptable" 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 utility model provides an underwater robot for bridge detection, which relates to the technical field of bridge detection, and comprises a robot main body, a floating component for providing floating force is arranged at the upper part of the robot main body; the propeller set comprises two side propellers and two tail propellers, the two side propellers are located on the front face of the robot body, and the two tail propellers are located at the same end of the buoyancy assembly and symmetrically arranged front and back; and the two symmetrical walking mechanisms are arranged on the robot main body through mounting structural members. The beneficial effects of the utility model are that through the reasonable layout of the propeller group and the cooperative work of the propellers at the two sides and the tail part, accurate water surface positioning and movement are realized, and the robot is ensured to smoothly approach and fit a pile foundation from the water surface; the posture of the whole machine does not need to be adjusted when the pile foundation is close to the pile foundation, and the pile foundation can be directly adsorbed and attached through the propellers on the two sides; a tensioning mechanism in the walking mechanism has self-adaptability, and an elastic support is matched with a connecting frame connected in a sliding mode.

The invention relates to the technical field of semantic segmentation, in particular to a night semantic segmentation method based on low illumination enhancement and edge optimization, and the method comprises the steps: inputting an image into a low-light enhancement repair network based on the Retinex theory, local contrast enhancement and adaptive feature fusion, obtaining a denoised and enhanced intermediate image, and carrying out the edge optimization of the intermediate image; inputting the intermediate image into a semantic segmentation network to obtain a category distribution diagram of each pixel; inputting a discriminator embedded with a channel attention module according to the category distribution diagram of each pixel, and optimizing a generator composed of a low light enhancement repair network and a semantic segmentation network through a multi-task joint optimization loss function; and inputting a night image to be detected and segmented into the optimized generator, and outputting a segmentation result. By adopting the method, the low-light enhancement repair network is combined with a local contrast enhancement and channel feature fusion mechanism, so that the overall brightness of the image is improved, the details and edge structures of the image are reserved, and the perception capability and robustness of the model are improved.

The invention discloses an electric energy metering error evaluation method and system for a direct current charging pile, equipment and a medium, and the method comprises the steps: constructing an electric energy disturbance response function based on transient voltage and transient current, and effectively quantifying the transient metering deviation caused by measurement lag, voltage drop, current peak and other factors in the actual operation of the direct current charging pile. Error reconstruction modeling has higher dynamic adaptive capacity and physical rationality, the limitation that in the prior art, the influence of the high-variable-load working condition on metering cannot be reflected is broken through, and the accuracy and timeliness of direct-current charging pile error monitoring are greatly improved. Besides, a standard error is introduced to carry out fitting calibration on a reconstruction error and carry out energy conservation verification, thereby realizing verifiability on a physical level and controllability in engineering implementation, enhancing credibility and authority of an evaluation result, constructing a closed-loop analysis mechanism from error evaluation to metering correction and then to conservation verification, and improving reliability and reliability of the evaluation result. And the closeness and the adaptive capability of the overall error evaluation are improved.

The invention discloses a reversible metal 4D printing method based on hydrogel reversible deformation and a metal part, and belongs to the field of metal additive manufacturing. The method comprises the following steps: firstly, constructing a three-dimensional model comprising a hydrogel driving layer and a metal blank layer; respectively preparing hydrogel light-cured ink and metal or precursor slurry thereof; printing the two layers of materials sequentially or alternately by adopting a photocuring printer to obtain a composite green body; the hydrogel layer is dehydrated and shrunk through drying, the component is driven to generate preset shape deformation, the hydrogel layer absorbs water and expands through soaking, and reversible recovery of the shape of the component is achieved; and degreasing, selective reduction and sintering heat treatment are conducted on the component, and the compact final metal component is obtained. According to the invention, the independent hydrogel layer is used as a deformation driver, the dependence on the intrinsic phase change characteristic of a metal material is avoided, the material universality is extremely high, the method can be widely applied to 4D printing of various materials including refractory metal and high-reflection metal, and reversible control of shape change is realized.

The invention belongs to the field of wastewater treatment and industrial automatic control, and particularly relates to an intelligent dosing control method based on water quality online monitoring. The method comprises the following steps: capturing a water body spectrum fingerprint spectrum by using a full-spectrum scanning module to obtain a characteristic signal sequence; a chemical informatics molecular algorithm is introduced for preprocessing and feature enhancement, and molecular fingerprint information is recognized; a convolutional neural network model is constructed to decouple chemical oxygen demand, ammonia nitrogen, total phosphorus and water toxicity indexes from the spectral signals, and multi-parameter parallel inversion is achieved; and based on the inversion index and the spectral dynamic trend, the dosage is calculated through an intelligent decision-making algorithm, and accurate dosing of the medicament is driven. According to the invention, through fusion of full spectrum scanning and deep learning, high-precision anti-interference water quality perception is realized, potential risks can be identified and preventive intervention can be carried out, and the intelligent level and economical efficiency of dosing control are improved.

The utility model relates to the technical field of agricultural machinery transmission system detection, and discloses an agricultural machinery transmission system intelligent running-in table which comprises an installation platform, a main power assembly is arranged on the rear side of the upper end of the installation platform, and a running-in tool is arranged in front of the output end of the main power assembly. A left loading assembly and a right loading assembly are arranged on the left side and the right side of the running-in tool respectively, a transmission axle to be tested is arranged on the running-in tool to be supported, the main power assembly is in transmission connection with the power input end of the transmission axle, and the two output ends of the transmission axle are in transmission connection with the left loading assembly and the right loading assembly respectively. According to the utility model, through the automatic design of the main power assembly, the left loading assembly and the right loading assembly, the full-process automatic operation from axle installation, power input to loading test is realized, and the main power assembly is driven by the first driving motor, the first gear and the first rack plate to automatically complete the butt joint with the input end of the axle. And the complexity and the time cost of manual operation are greatly reduced.

The utility model provides an air bag device for construction of a cross-fault tunnel foam concrete buffer layer. The air bag device comprises an inflatable flexible air bag, a first annular closed steel formwork and a second annular closed steel formwork. The first annular closed steel formwork and the second annular closed steel formwork are arranged at the two ends of the inflatable flexible air bag. An inflation connector is arranged at one end of the inflatable flexible air bag, and a first filling pipeline and a second filling pipeline are arranged on the first annular closed steel formwork and the second annular closed steel formwork. The construction efficiency is remarkably improved on the premise of not losing the shock absorption and isolation performance of the foam concrete, one-time forming is achieved, the construction quality is guaranteed, and the cost is saved; in addition, due to the reusable characteristic of the air bag device, the construction cost is further reduced, and the problems that in a traditional construction method, efficiency is low, the seismic mitigation and isolation performance of the expanded excavation section foam concrete buffer layer is difficult to guarantee, and cost is high are solved.

The invention discloses an intelligent trash-cleaning machine based on AI visual identification, and belongs to the field of visual identification, the intelligent trash-cleaning machine comprises an AI visual identification module, an intelligent control system and a rotary trash-cleaning execution mechanism, the AI visual identification module is in communication connection with the intelligent control system, and the intelligent control system is electrically connected with the rotary trash-cleaning execution mechanism; through cooperative work of the AI visual recognition module and the intelligent control system, precise recognition and efficient cleaning of dirt in water are achieved, and compared with a traditional dirt cleaning mode, the system has remarkable advantages. The method is excellent in recognition capability, sewage disposal efficiency, response speed, adaptability and the like, and can be widely applied to pretreatment scenes of pump station water inlets, urban river channels, reservoir spillways, sewage treatment plants and the like. Popularization and application of the trash remover can bring good economic, social and environmental benefits, powerful technical support is provided for water area treatment, and the trash remover has wide market prospects and development potential.

The invention provides a water plant raw water and clear water lifting pump station joint optimization energy-saving dispatching method and system for solving the problems that traditional pump station dispatching depends on artificial experience, global optimization is lacked, and energy waste is serious. Comprising the steps that a flow-lift and flow-efficiency relation model is established, and a variable frequency pump deduces a performance formula under any frequency based on the similarity law; in the first-stage optimization scheduling, the optimal flow of a water intake pumping station and the optimal water storage level of a clean water tank in each time period are determined by adopting a dynamic programming algorithm and taking the minimum total work of the system as a target, and the conditions of flow balance, water level constraint and the like are met; and in the second-stage optimization scheduling, the genetic algorithm is adopted to optimize the combination and operation frequency of the water pumps in each time period by taking the minimum total power of the pump set as a target, and a flow and lift deviation penalty term is introduced. According to the method, water taking, water storage and water supply whole-process collaborative optimization is achieved, the energy saving rate exceeds 10%, the operation cost is reduced, the method adapts to dynamic changes of the water source tide level and the water demand, and the dispatching intelligent level is improved.

The invention belongs to the technical field of geological disaster monitoring and early warning, and discloses a slope geological disaster multi-source remote sensing image intelligent early warning method and system, and the method comprises the steps: a multi-source remote sensing data adaptive collection step: constructing a multi-source remote sensing data cube; a multi-scale spatio-temporal feature coupling extraction step: generating a spatio-temporal coupling feature vector based on an improved Swin Transform architecture and an adaptive spatio-temporal attention mechanism; a disaster risk dynamic assessment and feedback step: calculating a disaster risk level based on a multi-modal deep learning network, and dynamically adjusting a feature fusion weight according to a risk assessment result to form closed-loop feedback; and an intelligent early warning decision and response step: constructing a multi-stage early warning threshold system to trigger early warning and generate an early warning report, the slope geological disaster early warning method realizes multi-scale spatial-temporal feature adaptive fusion and a closed-loop cooperation mechanism, and significantly improves the accuracy and intelligent level of slope geological disaster early warning.

The invention provides a user interface testing method and device, electronic equipment, a medium and a program product, and can be applied to the technical field of big data and the technical field of artificial intelligence. The method comprises the following steps: acquiring a target test script, and performing semantic conversion on an operation instruction in the target test script by utilizing a first model to obtain a target test step set containing business semantic description; the target test step set is executed, in the process of executing the target test step set, the first model is used for determining the positioning mode of the target interface element according to the business semantic description, corresponding interaction operation is executed, and actual interface output is obtained; in response to execution failure of the target test step set, generating a diagnosis result by using a first model at least based on the actual interface output and the business semantic description; and inputting the diagnosis result into a second model, generating a modification code snippet by the second model, and applying the modification code snippet to the target test script.

The invention relates to the technical field of wastewater treatment, in particular to a multi-stage treatment method and system for lead smelting wastewater. The method comprises the following steps: acquiring water quality data of the multi-source lead-acid storage battery smelting wastewater, and based on the water quality data, analyzing to obtain a battery wastewater multiphase feature set; on the basis, the information of the stabilized intermediate water body is obtained by dynamically adjusting the hydrogen ion concentration, breaking a complexing structure and preferentially separating fluoride; on the basis, the antimony valence state and the antimony concentration are monitored in real time, the oxidizing agent type and the reaction intensity are matched in a self-adaptive mode, oriented efficient conversion of antimony is achieved, and low-antimony pre-sedimentation water outlet information is obtained; on the basis, according to the dynamic proportion of the inlet water hydrogen ion concentration to the lead and antimony ions, a precipitation path control instruction is automatically switched, a purified water body with lead and antimony synchronously and deeply removed is obtained, and a lead smelting wastewater multi-stage treatment log is output. Harmful pollutants in the lead smelting wastewater process are efficiently removed, the method adapts to water quality fluctuation, consumption reduction and cost saving are achieved, compliance treatment of wastewater is achieved, and environment-friendly and economic values are achieved.

The invention discloses a two-stage multi-mode bearing fault diagnosis method based on a pre-training large model, and belongs to the technical field of bearing fault diagnosis. The method aims at solving the problems that a traditional method is poor in generalization and poor in robustness under multiple working conditions and small sample conditions. The method comprises the following steps: firstly, constructing a learnable multi-modal Tokens which comprises a multi-scale patch Token, a feature Token and a fault Token, and realizing efficient extraction and fusion of multi-modal features; a time-frequency semantic fusion module is introduced, and comprehensive time-frequency features are output through adaptive frequency coding, time coding and multi-modal fusion; and inputting the multi-modal feature sequence into a pre-training BERT model, and adopting a two-stage training strategy, in the first stage, performing self-supervised pre-training by taking mask signal reconstruction as a target, and in the second stage, performing parameter fine tuning by taking fault classification as a target. According to the method, the diagnosis accuracy and the cross-working-condition generalization ability under the small sample condition can be remarkably improved.

The invention provides a pipeline welding real-time feedback control method and system based on machine vision. According to the method, multi-modal data such as a three-dimensional point cloud, a temperature field and a welding seam image are synchronously collected, and key characteristics such as a pose deviation gradient, a cooling rate curve and a welding seam size sequence are extracted after time sequence alignment, filtering and correction fusion. And a CNN and LSTM mixed adaptive deep learning model is utilized to analyze the welding quality state in real time, and sub-optimal parameters are dynamically optimized based on a reinforcement learning strategy, or a three-level response mechanism including heat reduction input, expert knowledge base matching and precise repair is triggered to process defect early warning. The system has an incremental learning capability and can be self-optimized along with a control period. The full-flow intelligent monitoring, real-time feedback and self-adaptive accurate control of the welding process are achieved, and the welding quality and the automation level are remarkably improved.

The invention discloses an interactive load prediction method based on fusion of prior knowledge and data driving, and the method comprises the steps: reconstructing a load sequence into a trend load sequence and a disturbance load sequence through modal decomposition, and constructing a prior knowledge base of tag-feature weight according to SHAP interpretability analysis; and through a prior knowledge base and feature contribution analysis, channel-level weight learning is carried out by using an automatic correlation determination module, and key input features are obtained. And inputting the reconstructed load sequence and the key input features into an improved Transform model, mapping a priori knowledge base into a priori knowledge matrix, embedding the priori knowledge matrix into the model, and generating a load prediction result. And carrying out interpretability analysis on a prediction result, and feeding back and updating an analysis result to a priori knowledge base to realize dynamic collaborative updating of prediction and the knowledge base. According to the invention, organic combination of expert knowledge and a data-driven model is realized, and the precision and stability of load prediction are improved.

The invention discloses a test method for a high-load heating tube for a mold, and belongs to the technical field of single-head electric heating tube detection. The method specifically comprises the following steps: S1, historical data acquisition: acquiring cold-state electrical performance data of single-head electric heating tubes with different specifications and under different working conditions; s2, historical data preprocessing: performing preprocessing including data cleaning, data normalization and data standardization on historical data to obtain a standardized data set; the test process design covers a full-period closed loop of cold state detection, hot state leakage current test, reliability test, secondary hot state leakage current test and secondary cold state detection, the electrical safety (leakage current control) under a high-load working condition is verified through the hot state test, and the reliability of an actual use scene is simulated through a rapid on-off test. Meanwhile, the appearance, temperature, gas release and other states of the heating pipe are monitored in real time, abnormal testing is terminated in time, safety risks are avoided, and it is ensured that the detection result can comprehensively reflect the actual use performance of the heating pipe.

The application relates to a heterogeneous feature clustering method based on a privacy image, characterized in that a diffusion generation process is converted from a pixel domain to a feature domain through a featureized noise and matrix matching technology, a to-be-trained diffusion generator model with a conditional denoising UNet architecture as a core is constructed, and training is conducted in combination with a sample-level consistency constraint and a distribution-level alignment constraint, then a feature pair with mutually aligned semantics is constructed by using a pseudo image generated by the trained diffusion generator model; subsequently, a variational autoencoder model is trained, cross-domain feature mapping is realized through a double-layer alignment loss, and finally a unified domain feature dataset is formed and clustering is conducted; the method has the advantages that cross-domain feature unification is completed under the premise that no source domain data and no paired samples are available, original image leakage is effectively avoided, a high-quality unified feature set can be directly output, and a safe and efficient solution is provided for cross-domain visual analysis under a privacy limited scene.

The invention relates to the technical field of ferrous metallurgy casting, in particular to a drainage sand removing and recycling method and device. Comprising a mechanical arm, a lifting rotation controller, a controller arm, a sand suction cylinder, a sand conveying pipeline, a sand collecting tank and an injection pump. The lifting and rotating controller is fixedly mounted on the mechanical arm; the controller arm is connected with the lifting and rotating controller, and the lifting and rotating of the controller arm are controlled by the lifting and rotating controller; the sand suction cylinder is connected with the controller arm, the sand suction cylinder is of a U-shaped structure, and one end of the sand suction cylinder is a horn mouth used for sucking drainage sand; one end of the sand conveying pipeline is positioned above the tundish cover sprue gate, and the end part of the sand conveying pipeline is bent downwards to be connected with the other end of the sand suction cylinder; the other end of the sand conveying pipeline is connected with the sand collecting tank; the sand collecting tank is connected with an injection pump pipeline, and the injection pump provides suction power. The ladle filler sand, harmful gas in gaps of the ladle filler sand and a long nozzle and a small part of molten steel in contact with the ladle filler sand in a steel ladle can be discharged out of tundish molten steel and recycled.

The invention relates to an intelligent multi-source navigation method based on inertia and Beidou / visual information fusion, which comprises the following steps: firstly, establishing a multi-source information system of an inertial navigation system, and embedding an IMU (Inertial Measurement Unit) multi-parameter online calibration mechanism based on carrier dynamics in inertial navigation solution; establishing a tight coupling fusion framework taking an INS resolving result as a reference, and constructing tight combination measurement with the original observed quantity of the BDS and constructing re-projection error measurement with the visual feature points by utilizing INS high-frequency pose information; an intelligent adaptive filter based on inertial dynamics error model constraint is adopted, an INS error equation is used as a state prediction model, a Sage-Husa algorithm is introduced to estimate system noise statistical characteristics in real time, estimated error parameters are fed back to INS calculation through closed-loop correction, real-time correction of navigation results is achieved, and the system noise statistical characteristics are estimated in real time. Through deep tight coupling fusion and intelligent adaptive processing, navigation precision and system reliability in a complex environment are effectively improved.

This invention provides a coal tar pitch-based nickel catalyst, its preparation method, and its application, belonging to the field of solid waste resource recycling and synthetic carbon materials technology. This invention uses a mixed roasting method to prepare a coal tar pitch-based nickel catalyst with micropores and mesopores from the residual coal tar pitch after coal tar extraction. The synergistic effect of active nickel as an active site and the functional groups on the support surface significantly improves its catalytic liquefaction performance, exhibiting the advantages of "low cost, high temperature resistance, and strong adaptability." The preparation method provided by this invention not only offers an effective way for the high-value recycling of coal tar pitch but also provides a low-cost material for green conversion, environmental governance, catalysis, and other fields. It represents a green resource utilization strategy with significant economic and environmental benefits and broad commercial prospects.

The utility model relates to the technical field of rotary drilling head, and disclose adjustable drilling angle's rotary drilling head. This adjustable drilling angle's rotary drilling head, including drill bit body, first drill tooth and second drill tooth, the bottom fixedly connected with the first drill tooth of annular array distribution of drill bit body, the outside of drill bit body lower extreme is offered with the installation hole that is annular array distribution and with inside intercommunication, the inside of installation hole all is fixedly connected with the shaft that extends to the inside of drill bit body, this device have adjustable drilling angle, strong adaptability, adjustment convenient and so on advantage, solved the traditional rotary drilling head drill tooth drilling angle unadjustable, lead to face different hardness, bedding structure's rock and soil, cannot optimal rock breaking, in turn lead to drilling efficiency reduction and, and if need to change drilling angle, must dismantle drill bit and re -weld or install, time -consuming and labor -intensive and influence construction progress's problem.

The application relates to the field of assembly building technology and discloses an assembled sound insulation and shock absorption laminated floor slab, which comprises a pouring plate, the outer part of the pouring plate is provided with a cladding structure, the cladding structure comprises a main frame, a side stop, an upper stop rod and a connecting groove, the two upper stop rods are fixedly installed on the side stop, the upper stop rod can be detachably connected with the main frame and forms a box body structure to cladding the pouring plate, a connecting groove is arranged on one side of the main frame, and a protrusion is fixedly arranged on the other side of the main frame; and a cushion block assembly, the pouring plate is assembled with the main frame, the side stop and the upper stop rod at a construction site after being prefabricated in a factory, the modular design makes the components separable for transportation, the size and weight of a transportation unit are greatly reduced, the transportation cost and hoisting difficulty are reduced, the on-site assembly is flexible, the size deviation of the actual installation position can be fine-tuned, and the adaptability is high.

The application provides a hypersonic platform radar target detection method, device and electronic equipment, relates to the technical field of hypersonic radar target detection, and the method comprises the following steps: inputting a received first echo signal into a time domain echo signal model, converting the first echo signal from a first signal domain to a second signal domain by using the time domain echo signal model, and obtaining a second echo signal; filtering the second echo signal by using an optimal parameter filter, obtaining a filtered third echo signal, converting the third echo signal into a third signal domain, and obtaining a coherent accumulation result; determining whether a target is detected by using a constant false alarm rate detection based on the coherent accumulation result, and determining the distance of the target based on the position of a peak value in the accumulation result in the case that it is determined that the target is detected. The hypersonic platform radar target detection method, device and electronic equipment provided by the application can realize high-precision, high-efficiency and highly adaptable radar target detection.

A robustness measurement method for LeNet-5 networks based on adversarial boundary constraints is presented, relating to the field of deep learning model testing. The main steps include: for each training sample, dynamically generating adversarial examples based on the model's current state during training iterations; constructing a composite loss function based on standard cross-entropy loss and dynamic boundary constraint loss; performing end-to-end training on all parameters of the LeNet-5 network; and using the overall approximate robustness boundary as a measure of model robustness after training. This method effectively improves the resistance of the LeNet-5 model to fast gradient sign-based adversarial attacks without altering the basic structure of the LeNet-5 network by designing a new loss function.

The utility model belongs to the technical field of pot seedling transplanters, and particularly relates to a linkage type seedling taking device of a transplanter, which comprises an upper seat plate, a seedling taking claw support is slidably mounted on the bottom surface of the upper seat plate, a seedling taking claw component is slidably mounted on the seedling taking claw support, and the sliding directions of the seedling taking claw support and the seedling taking claw component are perpendicular to each other. The seedling picking device is characterized in that a support driving air cylinder is installed on the front side of the seedling picking claw support, a swing driving air cylinder is installed at the bottom of the seedling picking claw support, the structure of a driving device is simplified, the power transmission efficiency is improved, the daily maintenance cost is reduced, and the seedling picking device has high environment adaptability.

The application discloses a research report generation method based on natural language processing and relates to the technical field of intelligent document analysis and generation, which comprises the following steps: version storage of source documents and generation of timestamp snapshots; paragraph-level and sentence-level segmentation of the timestamp snapshots and extraction of evidence segments carrying original text positioning indexes; core claim extraction of the evidence segments to generate conclusion units containing claim texts and confidence; establishment of the association relationship between the conclusion units and the evidence segments; generation of chapter texts according to the associated conclusion units and calculation of evidence segment coverage; and structural adjustment and regeneration of the texts when the evidence segment coverage is lower than a preset threshold. The method realizes evidence traceability, multi-type semantic association and automatic optimization based on coverage feedback.

The utility model discloses a sectional type scoliosis correction device, including scoliosis correction arc plate, still include pelvic support, with the pelvic support both sides rotation connection's basin bone limiting arc plate, lower extreme with the pelvic support both sides rotation connection's basin bone limiting arc plate rear end fixed link and upper end with scoliosis correction arc plate lateral fixed link and lower extreme with basin bone limiting arc plate detachable connection's two second telescopic piece. The utility model has the advantages of: adopt sectional type structure, has overcome the existing scoliosis correction device wearing in the middle patient's spine and cannot be flexed and cannot be flexibly adjusted problem, makes scoliosis correction device local effective traction, has improved the correction effect and has improved the patient's comfort degree, the application structure simple, convenient operation, strong adaptability and so on, is applicable to different age stages and the scoliosis patient of the degree of illness.