239results about How to "Avoid performance degradation" patented technology

The invention provides a composite connection method for thick titanium alloy plates, and belongs to the technical field of metal material welding and additive manufacturing. According to the technical scheme, the method comprises the steps that symmetrical X-shaped grooves are machined in thick plate titanium alloy workpieces to be connected; a laser-electric arc hybrid welding system is adopted for conducting double-face bottoming welding on the X-shaped groove, so that a double-face bottoming welding seam is formed; and a powder feeding type laser cladding system is adopted, and multi-layer and multi-channel additive filling is conducted in the X-shaped groove above the double-face bottoming welding seam. According to the method, efficient laser-electric arc composite bottoming and low-heat-input laser cladding additive filling are separated, high-quality penetration of the root is guaranteed firstly, then welding deformation and residual stress are remarkably reduced through low-heat-input filling, the microstructure of a filling area is optimized, and therefore a welded joint with excellent comprehensive performance is obtained.

The application provides a heterogeneous computing power automatic adaptation method and system for an intelligent chip, relates to the technical field of heterogeneous computing adaptation of the intelligent chip, and acquires computing graph structure and operator feature data and hardware state data of a to-be-deployed model; an execution knowledge base is constructed, a scheme list is generated in combination with real-time data and the execution knowledge base; execution code is distributed according to the decision weight of the scheme list; the current execution state is evaluated at a predefined execution decision point to determine whether to switch the adaptive path; the actual execution effect and environment data generated after each decision are summarized, execution track records are formed, and the execution knowledge base is updated to periodically optimize the basis for scheme generation. The application can realize automatic perception, decision, optimization and online learning of a computing task in a heterogeneous computing power environment for an intelligent chip, significantly reduces the adaptation cost and improves the resource utilization efficiency.

An integrated detection method based on a liquid-filled hollow-core optical fiber enhanced Raman signal comprises the following steps: positioning an optical coupling system, a liquid adapter and a forward Raman signal collection system along the same dovetail groove guide rail, and operating a sliding block fine adjustment mechanism to adjust the relative positions of the optical coupling system, the liquid adapter and the forward Raman signal collection system so as to enable an optical path to be axially aligned; the micro-flow injection system is operated, target liquid is conveyed to the liquid adapter through the micro-flow guide pipe, the liquid adapter keeps the collimation alignment relation between the hollow-core optical fiber and the laser, and then the target liquid is introduced into a fiber core of the hollow-core optical fiber; the laser light source is started to emit laser, and after the laser is collimated and focused through the optical coupling system, a Raman signal is excited and generated; operating a forward Raman signal collection system to capture a Raman signal output by the hollow-core optical fiber, filtering out an exciting light Rayleigh scattering signal in the Raman signal, and transmitting the purified Raman signal to a Raman spectrometer; and analyzing and processing the purified Raman signal, and outputting a detection result of the target liquid.

The utility model discloses an electromagnetic damping vibration isolation device belongs to vibration control technical field. The utility model solves the technical problem of the magnetic field intensity fixed of electromagnetic damping vibration isolation device and coil heating easily leads to performance attenuation. The utility model discloses an electromagnetic damping vibration isolation device, including coil subassembly and conductor block. Coil subassembly includes coil and coil clamp. Coil clamp is held in the outer periphery of coil. Coil clamp includes clamp body and water cooling pipeline. The heat dissipation block is equipped on the clamp body, and the heat dissipation block extends along the radial direction of coil, and multiple heat dissipation blocks are evenly arranged around coil. The water cooling pipeline is surrounded in the clamp body. The conductor block is located between at least two groups of coil subassembly, and the magnetic pole of every two groups of coil subassembly is opposite, thereby making the conductor block can be subjected to electromagnetic damping force when producing vibration. The utility model discloses a vibration isolation device can cover the vibration control demand of multiple frequency bands, avoids coil overheat aging, improves the life of vibration isolation device.

This application provides an electric motor, an electric propulsion device, and an aircraft. The electric motor includes: a stator having a cavity; a rotor rotatably connected to the stator via a shaft; and the rotor being insulated from the stator; a conductive component disposed within the cavity, one end of the conductive component being conductively connected to the shaft, and the other end extending to the outside of the electric motor. The electric motor of this application, by incorporating the conductive component, creates a complete low-impedance electrical path between the shaft and the conductive component, allowing current that might be input into the electric motor via the rotor hub to be output to the outside of the electric motor, thus improving the electrical protection performance of the electric motor.

The invention belongs to the field of wireless communication, and particularly relates to a wireless signal multi-dimensional array coding method and system based on dynamic rank optimization. The method comprises the following steps: detecting state information of a wireless channel, predicting state change of the channel by using a pre-trained deep learning model, and generating a dynamic optimization multiplier; adjusting a preset static optimization multiplier by using the dynamic optimization multiplier, constructing a target function for jointly optimizing the energy efficiency and the spectrum efficiency, and determining an optimal transmission rank by solving the target function; comparing the optimal transmission rank with the current transmission rank to determine a rank-up or rank-down operation; and reconstructing a transmission signal of the wireless channel according to a result of the rank increasing or decreasing operation. According to the method, the optimization target is dynamically adjusted by predicting the channel change, and the asymmetric ascending and descending rank algorithm is adopted, so that the adaptive transmission rank optimization in a complex time-varying channel environment is realized, and the communication performance is remarkably improved.

The utility model relates to an anti-tilt fuel cell system, which comprises a vehicle body and a fuel cell system arranged in the vehicle body, fuel cell system mounting racks are arranged on two sides of the fuel cell system, and the fuel cell system mounting racks are in running fit with the fuel cell system. Therefore, the fuel cell system is always adjusted to be in a vertical state under the action of gravity when the vehicle is in an inclined state. The system is simple in structure and reasonable in design, when a vehicle is located on a ramp with a gradient, a fuel cell system can still be in a vertical state, so that the drainage strategy of a fuel cell is consistent with that of the fuel cell on a horizontal road surface, the complexity of system control is not increased, and compared with a complex multi-drainage-valve scheme, the system is simple in structure and convenient to use. The device is easy to implement and maintain, and has high practicability and economical efficiency.

The application discloses a non-magnetic ring proportional pressure reducing electromagnetic valve for a hydraulic system, a first through hole, a second through hole and a first blind hole are formed in the valve body, a pressure relief hole and a pressure inlet hole are formed in the side wall of the valve body, the pressure relief hole and the pressure inlet hole are communicated with the first through hole, a working hole is radially formed in the side wall of the valve body, the inner end of the working hole is communicated with the first through hole, and the outer end extends to the outer surface of the valve body, a valve core is axially movably installed in the first through hole, and a fifth outer groove and a first annular protrusion are arranged on the outer periphery of the valve core, the first annular protrusion forms an annular gap with the inner wall of the first through hole, a coil assembly is arranged in the yoke, a stop iron is fixedly arranged in the yoke through a mounting plate, and a magnetic circuit optimization structure is arranged on the outer surface of the stop iron. The magnetic circuit optimization and the multiple anti-sticking and friction-reducing designs are integrated, the independent magnetic ring is cancelled, the structure is simplified, the electromagnetic force, the response speed and the service life of the electromagnetic valve are comprehensively improved, the processing difficulty is obviously reduced, and the application range of the electromagnetic valve is widened.

The embodiment of the present disclosure provides a kind of for detecting device and system, it is applied to etching equipment, for detecting etching end point, the etching equipment is housed in the structure to be handled, wherein detecting device includes: detection unit, detection unit includes laser, probe structure and optical structure, optical structure includes first optical structure and second optical structure, first optical structure is connected with laser, second optical structure is connected with probe structure;Optical structure is arranged at the position of both ends in the reaction cavity of etching equipment, and the laser emitted by laser can be reflected between optical structure.Wherein, detection unit is based on the attenuation condition of laser between the optical structure to the substance produced in the etching operation process of the structure to be handled is detected, to obtain detection result, and based on detection result determines whether etching operation reaches etching end point.

The application discloses a backdoor defense method, an image classification method, a device, equipment and a medium, and is applied to the technical field of computers, and comprises the following steps: inputting a constructed distribution external linear disturbance into a backdoor model, determining a target category of a backdoor attack according to a predicted category distribution output by the backdoor model; regarding the target category as a target of a benign backdoor, regarding a synthetic linear disturbance as a trigger, constructing the benign backdoor, and injecting the benign backdoor into the backdoor model for training to obtain a controlled offset model; the benign backdoor is a controllable trigger map injected by a defense party; restoring a sample label corresponding to the benign backdoor to a real label, training the controlled offset model, removing a backdoor effect of the controlled offset model by using a forgetting mechanism, and obtaining a purified model. The application does not need to explicitly construct or restore a backdoor trigger, is not sensitive to a trigger form, is therefore high in universality and high in calculation efficiency, removes the backdoor while maintaining model performance, and balances efficiency, robustness and performance.

The application discloses an antioxidant composite ceramic aerogel and a preparation method thereof, and belongs to the technical field of composite ceramic aerogels.The main body of the antioxidant composite ceramic aerogel is a silicon nitride aerogel, the nanostructure of the silicon nitride aerogel is one of one-dimensional silicon nitride nanowires and two-dimensional silicon nitride nanobands or a combination of the two nanostructures, and a mullite coating and a copper-nickel metal coating are sequentially coated on the nanostructure of the silicon nitride aerogel from inside to outside.The application can solve the technical problem that the prior art cannot prepare silicon nitride ceramic aerogels which can stably serve in a high-temperature oxygen environment at low cost and in batches.

The invention discloses a method for eliminating internal stress of a large-scale polycrystalline diamond block material. The method comprises the following steps that the block material is placed in a vacuum heat treatment furnace; carrying out vacuumizing and gas washing operation on the furnace chamber so as to remove residual active gas; introducing high-purity protective gas into the furnace and maintaining positive pressure; executing a stepped heating and heat preservation process: firstly, heating to a first heat preservation temperature T1 at a first heating rate and preserving heat, and then heating to a higher second heat preservation temperature T2 at a second heating rate and preserving heat; and after heat preservation is completed, the controlled cooling process is executed till the furnace temperature is reduced to the room temperature. A strategy of'stepped heating + long-time heat preservation 'is adopted, firstly, the stress of a part of a high-energy area is preliminarily relaxed at a low temperature T1, and cracking caused by stress concentration due to rapid heating is avoided; then, at a higher temperature T2, sufficient energy is provided for atomic diffusion and grain boundary slippage, and deep and uniform release of the internal stress is achieved. And the method is particularly suitable for large-scale blocks.

The utility model provides a portable modularized nitrogen making machine. The portable modularized nitrogen making machine comprises a gas inlet seat and a gas outlet seat, wherein the gas inlet seat and the gas outlet seat are respectively provided with a gas inlet channel and a gas outlet channel; the air inlet end and the air outlet end of the adsorption module are respectively communicated with the air outlet end of the air inlet channel and the air inlet end of the air outlet channel; the first connecting piece is detachably arranged at the air inlet end of the air inlet channel in a sealing manner; the second connecting piece is detachably arranged at the air outlet end of the air outlet channel in a sealed mode. Pipeline connection among a compressed air pipeline, an air inlet channel and an air inlet control valve is achieved through a first input channel, a first output channel and a first control cavity which are integrally integrated in a first connecting piece. And the second input channel, the second output channel and the second control cavity which are integrally integrated in the second connecting piece realize pipeline connection among the nitrogen pipeline, the gas outlet channel and the gas outlet control valve, so that the number of pipe fittings required by the nitrogen making machine is greatly reduced, rapid installation of the gas pipeline of the nitrogen making machine is realized, the production efficiency is improved, and the production cost is reduced. And connection points of gas pipelines can be reduced, and the gas leakage risk is reduced.

The invention discloses self-repairing aerogel as well as a preparation method and application thereof, and belongs to the technical field of new materials. The aerogel comprises a substrate aerogel layer, a self-repairing functional layer and a reinforced skeleton, wherein the substrate is SiO2 aerogel and is prepared by adopting a combustion synthesis method; the self-repairing functional layer is composed of a polyimide microcapsule-ultraviolet curing type siloxane repairing agent system embedded in the substrate and a nanofiber reinforced polyurea and temperature-sensitive repairing component composite coating coated on the surface; and the reinforced framework is a Ni-Ti memory alloy mesh support. Through a synergistic mechanism of'surface layer temperature-sensitive flow repair-deep microcapsule repair agent release-skeleton crack propagation inhibition ', crack repair is realized, stability is maintained under extreme temperature difference, and the radiation shielding efficiency is high. The problems that a traditional material is insufficient in repairing capacity under the extreme environment of the moon and depends on external maintenance are solved, the durability and safety of moon buildings are remarkably improved, and the important aerospace application value is achieved.

This invention relates to the field of cable technology, specifically to a high-performance polyvinyl chloride (PVC) cable material and its preparation method. The cable material is based on PVC resin, combined with plasticizers and multifunctional flame-retardant composite fillers, with the filler content being 85-95 parts by weight per 100 parts by weight of resin. The multifunctional flame-retardant composite filler employs a multi-layered interface design: using magnesium hydroxide / calcium carbonate as the core, it first undergoes silane treatment to introduce epoxy sites, then constructs a phytic acid-tin-calcium-zinc chelate layer, and immobilizes double-grafted modified carboxymethyl-β-cyclodextrin, ultimately forming a tannic acid-metal ion polyphenol network layer. The preparation method includes pre-plasticization, segmented mixing with filler addition, and extrusion granulation. This invention effectively improves the filler-matrix interface compatibility, enhances flame retardancy, mechanical properties, and thermal stability, and solves the problem of insufficient durability in high-filler systems.

The invention provides a multi-dimensional vibration platform with oppositely-arranged double-sandwich piezoelectric actuators and an excitation method of the multi-dimensional vibration platform, and belongs to the technical field of high-frequency vibration engineering and micro-nano machining.The vibration platform comprises a center platform, the two oppositely-arranged piezoelectric actuators, two end covers and two bolts, wherein the center platform is integrated with the tail end platform, the two flexible hinges and the two amplitude-change poles, and the center platform is slender and is in mirror symmetry about the longitudinal symmetry plane of the tail end platform. The piezoelectric actuator comprises a longitudinal vibration ceramic group, a bending vibration ceramic group around the Y axis and a bending vibration ceramic group around the Z axis, translation along the X / Y / Z axis and rotation around the Y / Z axis can be realized through in-phase or anti-phase excitation different combinations under the condition of not increasing extra driving channels, and five-degree-of-freedom high-frequency micro-vibration output can be achieved at most. The structure has high symmetry, low parasitic vibration, compact size and high integration, and is suitable for precise engineering scenes such as micro-nano processing, laser-assisted manufacturing, cell manipulation and active vibration suppression.

This invention provides a confined grafted network antifreeze hydrogel, its preparation method, and its applications. The preparation method of the hydrogel includes: firstly, acetalizing and crosslinking polyvinyl alcohol and glutaraldehyde under acidic conditions to obtain a polyvinyl alcohol-glutaraldehyde hydrogel with a three-dimensional network structure; then, immersing it in a solution containing acrylic monomers and the initiator cerium ammonium nitrate, and heating it to allow the acrylic monomers to graft and polymerize onto the backbone, forming a confined grafted network structure. The hydrogel has a freezing temperature as low as -31.58℃, an elongation at break greater than 300% at -30℃, a tensile strength of approximately 5.6 MPa, and an electrical conductivity of approximately 0.03 S / m, and exhibits a stable and sensitive resistance response to tensile strain. This invention does not require the addition of exogenous small molecule antifreeze agents, and can maintain excellent mechanical flexibility, high strength, and stable conductive sensing performance at low temperatures. It is suitable for flexible strain sensors and wearable electronic devices, and has broad application prospects, especially in cold or extreme low-temperature environments.

The application discloses a navel orange garden rainwater runoff resource water and soil conservation drought resistance and disaster reduction method, relates to the technical field of intelligent agricultural water resource management, and is used for solving the problems of water and fertilizer waste and water and soil loss. Through construction of a closed-loop dynamic regulation and control mechanism of sensing, decision-making, execution and verification, a thing networking monitoring network is deployed to collect multi-source environmental data, and then an edge computing gateway generates control instructions in real time according to a rule base, so that valves, water pumps and other facilities are linked to execute precise irrigation and runoff regulation. Meanwhile, key benefit indexes such as rainwater storage capacity, irrigation water saving rate and soil loss reduction amount are calculated based on execution feedback data, and the hash values of the indexes are stored in a trusted manner. Finally, when the benefits do not meet the standards or the data is abnormal, root cause analysis is performed according to the stored historical data, and the optimal repair work order is generated based on the cost priority rules, so that the system is self-optimized. The application significantly improves the precision, system robustness and long-term benefits of water resource management while reducing manual intervention.

The invention provides an enterprise space appeal matching method based on an AI model, and the method comprises the following steps: collecting the multi-source sensing data of an enterprise space, and building a structured time-space information matrix through an access control, energy consumption, position and personnel label sensing terminal; constructing a causal relationship graph by using a structural causal modeling algorithm, and identifying key causal factors directly related to the spatial appeal; causal invariant characterization is extracted based on a multi-environment invariant risk minimization method, and an intervention effect model is established in combination with controllable intervention variables such as partition adjustment, capacity limitation and use strategies; generating an anti-fact sample of the candidate space unit by using an anti-fact reasoning mechanism, and calculating a matching adaptation score; establishing a matching optimization model according to constraint conditions such as topology, capacity, safety and moving lines; according to the method, high-precision and interpretable matching distribution under the conditions of complex space resources and multi-dimensional appeals is realized, and the method has the effects of flexible deployment, high matching stability and sustainable optimization.

The application discloses a ship lock valve vortex suppression protection method, characterized in that, at the bottom edge of the valve in a half-open state, water flow is periodically swallowed outward through the set water inlet and outlet, thereby effectively suppressing the periodic generation and shedding of the separation vortex, and realizing protection of the valve facility. The application can suppress and control the generation of the separation vortex at the bottom of the valve from the source by actively intervening and regulating the water flow, thereby greatly improving the safety of the valve facility and realizing safety protection of the valve.

This utility model relates to the field of turboshaft hybrid engine technology, and particularly to an integrated heat dissipation structure for the hardware of a turboshaft hybrid engine control system. The structure includes a rubber pad, with a fixing component fixedly connected to the upper end of the rubber pad. Two mounting blocks are fixedly connected to the front and rear ends of the fixing component. Each of the four mounting blocks has through-holes at its upper end. A mounting component is movably engaged with the upper end of the fixing component. Bolts are threaded to the four corners of the upper end of the mounting component. An air-cooling structure is inserted and fixedly mounted in the middle of the upper end of the mounting component. This integrated heat dissipation structure for the turboshaft hybrid engine control system hardware utilizes a synergistic mode of water-cooled heat absorption and air-cooled heat dissipation, efficiently dissipating the large amount of heat generated by the hardware, ensuring that the hardware temperature remains stable within a safe range, and effectively preventing performance degradation or malfunctions caused by high temperatures.

The invention provides a medical image classification model training method and system, a medical image classification method, equipment and a medium, and the method comprises the steps: carrying out the time-sequence feature extraction of a time sequence of a resting-state functional medical image through an encoder, so as to obtain the multi-scale features of the resting-state functional medical image; performing feature extraction processing on the metabolic network of the functional state medical image to obtain features of the functional state medical image; processing the multi-scale features and the features of the functional state medical image by using an adapter to obtain feature pairs of the resting state functional medical image and the functional metabolism medical image; performing constraint processing on each feature pair by using comparative learning to obtain an aligned feature pair; and training the initial model by using the aligned features, and performing group-level modal distillation processing on the initial model to obtain a medical image classification model. According to the medical image classification model training method, the flexibility and robustness of cross-modal feature fusion can be improved.

This utility model discloses a tiered energy-dissipating anti-ship collision system based on metamaterial structures, including an ultra-high performance concrete panel, a tiered energy-dissipating anti-collision layer, a rubber fender, bridge piers, bridge abutments, and bridge pile foundations. The tiered energy-dissipating anti-collision layer includes variable stiffness units, lightweight energy-dissipating materials, vertical steel plates, an upper top plate, and a lower bottom plate. The variable stiffness unit is composed of multiple concave hexagonal units connected by transverse and longitudinal steel plates, with the hollow structure in the middle filled with lightweight energy-dissipating materials such as polyurethane foam, ceramsite, and rubber granules that can absorb and disperse impact energy. The variable stiffness units are welded to the steel plates to form a whole, and the protective device is connected to the protected bridge pier through the rubber fender. The overall stiffness of the protective device increases with structural deformation upon impact, thus this utility model has a variable stiffness tiered energy-dissipating anti-collision effect, effectively prolonging the collision time between the ship and the bridge and reducing the impact force of the ship, achieving two-way protection for both the ship and the bridge.

This invention relates to a temperature control system for FTU battery packs, belonging to the field of power distribution automation technology. It includes an FTU device, a standard 5-pin backup power interface, a DC24V power supply, a battery pack, a temperature sensor, a heating element, and a temperature controller. The battery pack is connected to the FTU via pins 4 and 5. The temperature sensor is connected to the temperature controller via pin 1. The positive and negative terminals of the heating element are connected to the temperature controller via pins 2 and 3. The temperature controller is connected to the DC24V power supply and compares the temperature signal collected by the temperature sensor with a temperature threshold to control the start and stop of the heating element, achieving closed-loop temperature control of the battery pack. This invention achieves a "zero-destructive" upgrade of existing equipment, eliminating the need for opening holes or external wiring in the FTU protective enclosure. It retains the original sealing and protection rating (IP rating) of the equipment at the factory, preventing changes in the operating environment such as water ingress, condensation, and dust that might be introduced during modification, thus ensuring the long-term reliability of outdoor electrical equipment.

The invention relates to the technical field of speech processing, and particularly provides a disguise speech recognition method based on a Transform model and mixed features, which is carried out from two aspects of feature extraction and model establishment. A resonance peak parameter is calculated by adopting a cepstrum method, a cepstrum coefficient (GFCC) is obtained through a Gammatone filter bank, then the resonance peak, the GFCC and a difference coefficient of the GFCC are combined into a mixed characteristic parameter, and complementary correlation between mixed characteristics is mined. From the perspective of model establishment, the mixed features are used as the input of the model, and the Transform network model is used as the acoustic model of the voiceprint recognition system, so that the feature distribution is better fitted, the classification effect is remarkably improved, and the performance of the camouflage voice voiceprint recognition system is effectively improved. The problem of performance degradation caused by feature redundancy and modal noise in a traditional method is solved.

The utility model relates to water body oxygenation technical field especially, more particularly to a fan circulating water oxygenation equipment. Its technical scheme includes: pedestal, suction chamber and suction pipe, be equipped with the nozzle on the pedestal, be equipped with the suction chamber on the nozzle, the other end of suction chamber is equipped with the mixing chamber, the other end of mixing chamber is equipped with the diffusion chamber, the mixing chamber, nozzle, suction chamber and diffusion chamber constitute the jet structure, jet structure is used for gas water mixture, increases water body oxygen content, the bottom end suction inlet of suction chamber is equipped with suction pipe, the bottom end of suction pipe is equipped with the impurity removal structure, and the impurity removal structure is used for removing the impurity that suction pipe orifice filters. The utility model discloses through the jet structure and carries out the water body oxygenation circulation of lasting, has improved water body dissolved oxygen content, the impurity that automatic cleaning filter plate surface adheres, ensures the filtering effect of filter plate sustained and stable, avoids the equipment performance decline caused by the impurity blockage or accumulation.

The invention provides a camera computing power dynamic allocation method and system based on heterogeneous computing power cutting. The method comprises the following steps: acquiring a camera task and acquiring task and environment parameters based on the camera task; determining a task real-time computing power demand and a computing power cutting coefficient based on the task and the environment parameters; determining the actual distribution computing power corresponding to each camera task based on the task real-time computing power demand and the computing power cutting coefficient; and determining a corresponding calculation unit set for each camera task based on the actual distribution calculation power, constructing an objective function based on the calculation unit sets, solving the objective function to obtain an optimal calculation unit corresponding to each camera task, and executing the camera task through the optimal calculation unit. The method can ensure that the computing power distribution of the equipment is always in an optimal state, and avoids performance reduction caused by scene fluctuation.

The invention discloses a multi-dialect large model adaptation method and system oriented to a Gluggin language, and the method comprises the steps: obtaining corpus data and model constraint information of each Gluggin dialect region, building a dialect mapping matrix, and positioning key semantic difference points; characterization offset features are extracted, contrast training sample pairs are constructed, and vector space analysis is implemented to determine a priority adjustment layer; identifying dialect difference features in the model constraint information to generate layered coding parameters, generating a layered representation structure through collaborative matching of a priority adjustment layer and the layered coding parameters, and determining a weight adjustment boundary; determining a parameter distribution density by combining a weight adjustment boundary and a priority coefficient, generating an optimal migration weight, extracting a negative migration node, converting the negative migration node into a dialect boundary mark, and constructing a gradient propagation path; and detecting the confusion expression set, generating representation decoupling configuration, executing progressive collaborative optimization based on the representation decoupling configuration and the gradient propagation path to output a multi-dialect adaptation scheme, and realizing accurate adaptation and collaborative optimization of the multi-dialect of the Gu-language family.

The application provides a camera computing power dynamic allocation method and system based on heterogeneous computing power cutting, the method comprises the following steps: obtaining a camera task and collecting task and environment parameters based on the camera task; determining task real-time computing power demand and computing power cutting coefficients based on the task and environment parameters; determining the actual allocation computing power corresponding to each camera task based on the task real-time computing power demand and the computing power cutting coefficients; determining the corresponding computing unit set for each camera task based on the actual allocation computing power, constructing a target function based on the computing unit set, solving the target function to obtain the best computing unit corresponding to each camera task, and executing the camera task through the best computing unit. The application can ensure that the device computing power allocation is always in the optimal state, and performance degradation caused by scene fluctuations is avoided.