21results about How to "Increase constraints" patented technology

The invention discloses a model fine tuning method and system based on task priority and elastic weight solidification, belongs to the technical field of machine learning, and aims to solve the technical problem of how to provide a stable, efficient and low-power silk screen manufacturing model which can adapt to various complex production scenes. Comprising the following steps: collecting sample data, and dividing the sample data into a plurality of subsets; service weights and priority coefficients of all new tasks in the silk screen manufacturing process are analyzed based on the sorted subsets; constructing a knowledge graph based on the entities and the relationship between the entities; calculating the importance degree of the model parameters to the old task based on the Fisher information matrix, and adjusting the priority level of the model parameters based on the knowledge graph; and on the basis of the Fisher information matrix, updating a loss function of the silk screen manufacturing model under the new task through EWC, and correcting the current silk screen manufacturing model on the basis of entities and relationships in the knowledge graph, so that multi-task model fine tuning is realized.

The utility model belongs to the field of semiconductor power device manufacturing, and particularly relates to a GaN HEMT device with a back barrier structure with gradually changed components, which comprises a SiC substrate layer, a GaN buffer layer connected above the SiC substrate layer, a first AlGaN back barrier layer connected above the GaN buffer layer, and a GaN channel layer connected above the first AlGaN back barrier layer. According to the GaN-based HEMT device, the back barrier layer with gradually-changed components is introduced, so that the electron constraint in a channel is effectively enhanced, the short-channel effect of the device is inhibited, the direct-current characteristic of the device is improved, and the power characteristic of the GaN-based HEMT device is further improved. For the interface scattering problem caused by the back barrier structure, the gradient component back barrier structure can reduce interface scattering and parasitic parameters compared with a fixed component back barrier structure, the influence on a GaN main channel is reduced, and the small signal and large signal characteristics of the device are improved.

The utility model discloses an I-shaped steel reinforcing assembly which comprises transverse I-shaped steel fixedly installed on the left side and the right side of I-shaped steel, a round sleeve is installed on the front side of the I-shaped steel, a square sleeve is installed on the surface of the round sleeve, and a plurality of bolts are arranged on the surface of the I-shaped steel in a penetrating mode. The round sleeve and the square sleeve are fixedly connected with the I-shaped steel through bolts, a connecting frame is fixedly connected to the surface of the square sleeve, an upper top plate is arranged at the top of the connecting frame and fixedly connected with the I-shaped steel, and a lower bottom plate is arranged at the bottom of the connecting frame and fixedly connected with the I-shaped steel. The round sleeve, the square sleeve and the connecting frame are used in cooperation, the periphery of a hole formed in the I-shaped steel is reinforced, then the supporting plate, the supporting frame and the triangular supporting plate are used in cooperation, the triangular supporting plate can support the lower bottom plate, and finally the round sleeve is stably installed on the periphery of the hole formed in the I-shaped steel.

The utility model discloses an electric pump which comprises a pump cover, a stator assembly and a pump shell, the electric pump is provided with a pump inner cavity, the stator assembly is located in the pump inner cavity and comprises a stator winding and an isolation sleeve, the isolation sleeve is located in the pump inner cavity, the pump inner cavity comprises a first cavity and a second cavity, the isolation sleeve isolates the first cavity and the second cavity, and the first cavity and the second cavity are not communicated. The stator winding is located in the second cavity and comprises a stator iron core, the electric pump comprises a restraining part, the restraining part is fixedly connected with the stator iron core, the restraining part is located at the end of the stator iron core in the axial direction of the electric pump, the stator iron core is arranged on the periphery of the isolation sleeve in a sleeving mode in the radial direction of the electric pump, and the restraining part is in interference fit with the isolation sleeve. By means of the mode, constraint on the isolation sleeve can be increased, and therefore vibration and noise of the electric pump can be reduced.

The application provides a PAN precursor contribution quantification method based on real atmospheric observation constraints, relates to the technical field of atmospheric pollution tracing, and comprises the following steps: obtaining real atmospheric observation data of a target region, including PAN, nitrogen oxides, ozone, volatile organic compounds, carbonyl compounds, photolysis frequency and meteorological parameters; constructing an observation constraint vector, establishing an observation constraint box model containing PAN conversion, VOCs oxidation, free radical circulation mechanism and the like and driving operation, recording the relationship between reaction flux and species to form a dynamic reaction network; determining a generation flux based on PAN mass conservation, reversely tracing peroxyacetyl radicals, intermediate products to observation precursors from the PAN generation reaction, and distributing the contribution according to the flux proportion; merging the contribution of the same precursor and its oxidation products, and quantifying the contribution amount and contribution rate of each precursor to PAN generation. The application overcomes the defects of high uncertainty and artificial assumption, and accurately quantifies the actual contribution of real atmospheric PAN precursors.

The invention relates to the technical field of metal plastic forming, in particular to a stub-bar-free cross wedge rolling composite near-net forming method for a high-performance shaft, which comprises the following steps: heating a metal bar to a temperature interval where the metal bar is subjected to plastic deformation, and performing taper angle pre-forming treatment on the end part of the metal bar to obtain a pre-processed blank; the pre-machined blank is placed in a rolling area of a cross wedge rolling machine tool, radial rolling is conducted on the pre-machined blank through wedge-shaped molds of an upper roller and a lower roller so that the pre-machined blank can generate axial plastic extension to obtain a formed shaft part, and cooling and appearance finishing are conducted on the formed shaft part to obtain a high-performance shaft part; whether the forming stability of the high-performance shaft meets the requirement or not is determined based on the relative roundness error of the end of the high-performance shaft; determining whether to increase the relative difference rate of the upper and lower rolls; whether the relative temperature drop rate of the end in the rolling forming process is increased or not is determined; and whether the jacking force on the end part of the pre-machined blank is reduced or not is determined. The forming stability of the high-performance shaft part is improved.

This utility model relates to the field of prefabricated steel structure buildings, specifically to a floor connection component. The first strip-shaped hole provides vertical adjustment margin, allowing for fine-tuning of the position of the first connection unit relative to the column during installation, effectively compensating for manufacturing and installation errors and improving assembly fault tolerance. The first connection unit of the square tube structure has good compressive, bending, and torsional resistance, and as the main load-bearing component, it can effectively transmit the axial force and bending moment of the column. The second connection unit of the U-shaped structure provides stable lateral support for the beam, enhancing the constraint on the beam end. The circular hole ensures uniform stress distribution at the connection with the beam, thereby improving the overall stiffness of the floor connection component. This solves the problems of high installation alignment accuracy requirements, poor fault tolerance, and insufficient stiffness of connection nodes in existing floor connection components, which affect the overall structural performance.

The invention discloses a volcanic rock inorganic mineral substance full-pouring bus support hoisting structure, and relates to the technical field of bus support hoisting, the volcanic rock inorganic mineral substance full-pouring bus support hoisting structure comprises a wall body and a pouring bus body, the pouring bus body is arranged on one side of the wall body, and mounting plates are fixedly connected to the two sides of the pouring bus body at equal intervals; the triangular supports are arranged on the side, close to the mounting plate, of the pouring bus body at equal intervals, it can be guaranteed that the pouring bus body does not deviate in the mounting process, negligence caused by long-time high-strength operation of workers is avoided, and if the stable state of the pouring bus body cannot be maintained, the pouring bus body can be prevented from being damaged. The problem that the position of the pouring bus body cannot be ensured to be absolutely accurate when the pouring bus body is fixed is solved, so that the expected mounting effect is perfectly achieved, the tight connection and good contact between the pouring bus bodies are ensured, the contact resistance is reduced, and the reliability of electrical connection can be ensured by the accurate mounting of the pouring bus body. And the risk of electrical discharge, short circuit or system fault caused by installation offset is reduced.

This utility model discloses a device for enhancing the uniformity and deposition rate of magnetron sputtered thin films on curved parts. The device includes a fixed rod, a disk permanent magnet, mounting posts, and a housing. The disk permanent magnet is mounted on the fixed rod, and adjacent disk permanent magnets are spaced apart by mounting posts. The disk permanent magnet is completely encased in the housing. The magnetic poles of adjacent disk permanent magnets are the same on their relative circular surfaces. During the coating process, an additional magnetic field can be provided on the substrate surface, enhancing the constraint of discharge particles on the substrate surface, thereby solving the problem of low plasma density and uneven distribution near the concave curved surface.

The utility model provides a battery pack and an energy storage system, and relates to the technical field of energy storage. The battery pack comprises a battery box body, a plurality of battery modules, a first support, a second support and a third support, a battery bin is formed in the battery box body, the battery modules are located in the battery bin, and each battery module comprises a plurality of single batteries arranged in the first direction; wherein in the first direction, a support is arranged between every two adjacent battery modules and at least one of one end of each battery module and the other end of each battery module, a plurality of energy absorption grooves are formed in each support, and crushing guide grooves are formed in the groove walls of at least part of the energy absorption grooves, so that the buffering and vibration reduction performance of the battery pack is improved.

The invention discloses a cardiovascular acute event early warning method based on reinforcement learning, and aims to solve the problems that an early warning scheme is difficult to consider individualized trigger opportunity and evaluation frequency and is difficult to balance false alarm and missed alarm. The method comprises the following steps of: outputting risks, trends and uncertainty of current and future time windows by using multiple heads of an uncertainty perception deep network, constructing a state in combination with historical alarm records, individualized risk thresholds and evaluation time intervals, and inputting the state into a security constraint reinforcement learning network; in the same forward reasoning, whether early warning, threshold value adjustment amount and next evaluation time interval are determined in a combined manner, and a constraint weight updating strategy is adaptively adjusted based on early warning income, false alarm and missed alarm indexes and prediction uncertainty after a real label is obtained. The technical effects of reducing the false alarm rate, optimizing the early warning advance and the evaluation frequency and improving the early warning accuracy and reliability of the cardiovascular acute event on the premise that the missed alarm risk is controllable are achieved.

The invention discloses a battery aging trajectory adaptive extrapolation method fusing physical time sequence prior, which comprises the following steps: collecting full life cycle constant current charging data of a battery, extracting an incremental capacity curve, and constructing characteristic peak evolution to form multi-dimensional physical time sequence prior; a state space backbone network is introduced, and trend perception is achieved through a priori initialization implicit state; constructing an attention anomaly detection module to discriminate real relaxation and noise; monotonic constraint is released through a micro-relaxation layer for a relaxation event, and data is corrected through Kalman filtering for noise; and constructing a dynamic trade-off loss function to adaptively adjust the physical constraint strength, and iteratively training the model based on the corrected data to realize stable extrapolation prediction. The method effectively solves the problems that in the prior art, real physical relaxation and measurement noise cannot be distinguished, and prediction deviation is caused by physical constraint rigidity.

The utility model discloses a main and auxiliary steel beam connecting node structure of a bamboo-hat-shaped roof steel structure. The main and auxiliary steel beam connecting node structure comprises a main steel beam, a fixing plate, an auxiliary steel beam, a fastening bolt and a triangular connecting plate, one end of the fixing plate is fixedly connected with a web of the main steel beam, the plate face direction of the fixing plate is parallel to the plate face direction of a web of the auxiliary steel beam, and the web of the auxiliary steel beam is attached to the fixing plate and locked and fixed through a plurality of fastening bolts, so that the main steel beam is fixedly connected with the auxiliary steel beam. The two triangular connecting plates are fixedly connected between the upper flange plate and the lower flange plate of the main steel beam and between the upper flange plate and the lower flange plate of the auxiliary steel beam respectively, and the two triangular connecting plates are arranged on the two sides of the auxiliary steel beam respectively. The utility model relates to the technical field of steel structure construction, and can solve the problem of low structural strength of a connection node of main and auxiliary steel beams in the prior art.

The application discloses a vehicle-infrastructure collaborative positioning method based on multi-view degradation perception, relates to the technical field of intelligent transportation and automatic driving, and comprises the following steps: acquiring vehicle end point cloud data, and constructing a vehicle end point-surface residual error factor; acquiring measurement data of a vehicle-mounted inertial measurement unit, and constructing an IMU pre-integration factor through mechanical arrangement; acquiring roadside point cloud data, constructing a roadside residual error based on the roadside point cloud data and the vehicle end point cloud data, and constructing a roadside DICP factor; constructing a factor graph model based on the vehicle end point-surface residual error factor, the IMU pre-integration factor and the roadside DICP factor, updating the pose information of a vehicle, and obtaining a positioning result. The application can effectively overcome the defects of drift and unstable registration caused by the insufficient observability of a laser radar in a degradation scene, solve the problems of high delay and insufficient constraints of multi-view vehicle-road collaborative positioning, and realize long-time, stable and high-precision real-time positioning in a complex traffic environment.

The invention discloses a pressure pipeline inner surface crack risk assessment method based on extended finite elements and stress triaxality, and aims to solve the problems that an existing method is insufficient in precision and ignores a coupling effect, and comparison files are limited to extreme working conditions. The method comprises the following steps: establishing and verifying an XFEM basic model, constructing a pipeline XFEM model containing inner surface cracks, revealing a crack size-angle coupling rule and identifying a critical angle of 8-12 degrees, predicting crack propagation based on stress triaxial degree, and carrying out risk grading and detection priority ranking in combination with the coupling rule. According to the method, the size-angle coupling effect is innovatively quantified, the stress three-axis degree is upgraded into an independent extension prediction tool, a complete engineering closed loop is formed, the evaluation error is smaller than or equal to 10%, the method is suitable for conventional operation and maintenance of 09MnNiDR low-temperature steel pipelines and the like, an RBI plan can be directly optimized, and the practicability and the universality are remarkably superior to those of a simulation technology optimization guiding scheme of a comparison file.

The utility model discloses a gun drill guiding auxiliary device of promoting pressure injection rod deep hole machining precision belongs to gun drill processing technical field, including auxiliary frame, the auxiliary frame upper surface rotationally connected with the fixed frame, and the auxiliary frame upper surface slidingly connected with the lower guide block, and the lower guide block upper end slidingly connected with the upper guide block, and the upper guide block outside is pasted with the fixed frame inner wall simultaneously, the upper guide block inside is all set up with the lower guide block inside connection groove, and the connection groove surface fixedly connected with the reinforcing rib. The gun drill guiding auxiliary device of promoting pressure injection rod deep hole machining precision, through setting up the auxiliary frame and the fixed frame, effectively avoid the gun drill and appear deviation when initial drilling, from the machining starting stage just has guaranteed the position precision of deep hole machining, restricts the displacement of gun drill in each direction simultaneously, controls position deviation in the extremely small range, effectively reduces the accumulation of position deviation, guarantees the straightness and cylindricity of deep hole machining.

The utility model discloses a circular-tube-shaped vibration electrode for electrostatic detection, and belongs to the field of electrostatic measurement. The vibration electrode with the circular tube-shaped structure comprises a circular tube-shaped fixed electrode, a circular ring-shaped piezoelectric ceramic piece and a circular tube-shaped vibration electrode, the circular-ring-shaped piezoelectric ceramic piece is located between the circular-tube-shaped fixed electrode and the circular-tube-shaped vibration electrode, the circular-ring-shaped piezoelectric ceramic piece completely covers the end face of the vibration electrode, and the circular-tube-shaped fixed electrode, the circular-ring-shaped piezoelectric ceramic piece and the circular-tube-shaped vibration electrode are coaxially arranged and fixedly connected into a whole in sequence. The whole static voltage sensor is located in the static voltage sensor shell; the circular-tube-shaped fixed electrode is fixed on the circuit board, a cylindrical electrostatic charge induction electrode is arranged in front of the circular-tube-shaped vibration electrode, and the central axes of the circular-tube-shaped fixed electrode and the cylindrical electrostatic charge induction electrode coincide; one end of the cylindrical electrostatic charge induction electrode faces the circular tube-shaped vibration electrode, and the other end faces the static power supply; the circular tube-shaped fixed electrode is electrically connected with the circuit board; and the circular tube-shaped vibration electrode is connected with zero volt voltage.

The utility model provides an energy storage battery bunching device, which relates to the technical field of battery bunching and comprises a bunching device body, and an inner cavity is arranged in the middle of the bunching device body. The top of the bunching device body is in threaded connection with the bottom of the threaded rod and is rotationally connected with an upper clamping plate, a lower clamping plate is arranged below the upper clamping plate, and corresponding bunching holes are formed in the opposite sides of the lower clamping plate and the upper clamping plate. Compared with the prior art, the distance between the lower clamping plate and the upper clamping plate can be adjusted, energy storage battery wire harnesses with different diameters / volumes can be conveniently and rapidly restrained, and according to the use requirements of different wire harness positions, a plurality of wire harnesses can be rapidly used in a parallel or overlapped mode.

The present invention provides a light source unit, a system and a surface defect detection method based on photometric stereo, comprising a lamp group composed of a plurality of lamp beads, a PCB board, a fixing part and an adjustment component, wherein the lamp group and the fixing part are respectively mounted on two opposite sides of the PCB board, and the adjustment component is connected to the fixing part to drive the angle adjustment of the lamp group. By adopting a structure of two adjusting screws, the angle between the LED lamp group and the center of the light source can be adjusted, thereby solving the adjustment pain point of a fixed photometric stereo light source, realizing the adjustment of the shooting angle of the lamp group, and realizing the application of illumination at more angles.

The invention discloses an image restoration method based on dynamic matching of multiple receptive fields and damaged features. The invention comprises the following steps: using a damaged image as an input, constructing a unified scale multi-receptive field feature extraction module, and extracting a unified scale feature group with different receptive fields; using a damaged image mask as a constraint, establishing a linear mapping relationship through a damaged feature mapping module, and extracting a one-dimensional weight feature corresponding to the current damaged feature; using the weight feature to weight the feature group, and obtaining a weighted unified scale multi-receptive field feature group; using a feature screening module, establishing a dynamic matching relationship between damaged features and features, and obtaining a re-screening feature with a high matching degree; using the re-screening feature as an input, decoding is performed through a decoder group based on fast Fourier convolution, and obtaining a restored image; through the setting, the constraints of the restoration network are strengthened, the artifacts appearing in the repaired image are effectively removed, and the semantic consistency of the restored image is enhanced.