73results about How to "Valid conversion" patented technology

The invention relates to the technical field of image processing, and discloses a detection method, system and equipment for intelligent welding of pipe pile splicing and a medium. The method comprises the steps that noise reduction is conducted on a target weld seam image through a filtering algorithm, registration and splicing are conducted on the image through feature matching, and a weld seam unfolding image is obtained; segmenting the weld joint expanded image by using a deep learning network, and extracting a weld joint center line and surface features to obtain weld joint feature data; defect detection and recognition are conducted on the weld joint feature data, hierarchical calculation is conducted through a rule set, and weld joint quality data are obtained; parameter mapping is conducted according to the weld joint quality data, a welding parameter instruction is generated, welding process variables are adjusted, and target welding parameters are obtained; and the welding seam quality data and the target welding parameters are stored, and a quality report is generated through data analysis. The detection accuracy of intelligent welding for pipe pile splicing is improved.

The invention provides a semiconductor equipment data acquisition method and a semiconductor equipment data acquisition system supporting standardized communication transformation, which are used for realizing data acquisition and process control on different types of semiconductor equipment, formulating corresponding communication rules for equipment conforming to a communication standard, and converting an equipment communication protocol by a cluster system so as to match a standard interface. For equipment which does not completely meet the standard but can realize communication through programming, the trunking system directly acquires required data and converts the required data to meet the standard interface specification. For equipment without communication capability, data is acquired through an equipment interface by adopting an image recognition technology, data input and function operation are realized through simulating keyboard and mouse operation, and finally, protocol conversion from data to a standard interface is completed by a cluster equipment control system. And the equipment data is converted into a message data packet conforming to a standard semiconductor equipment communication interface, so that docking with a semiconductor equipment data analysis and process control system is realized, and the production efficiency and quality of the semiconductor equipment are improved.

The invention provides a traditional Chinese medicine formula granule quality tracing system based on data fusion, and relates to the technical field of data processing. Multiplying the actual value of each key feature by the corresponding weight and the dynamic adjustment factor to obtain a weighted key feature value; summing all the weighted key characteristic values to obtain a comprehensive quality index; comparing the comprehensive quality index with a corresponding quality standard to obtain a comparison result; according to a comparison result, performing quality judgment on the new data of the traditional Chinese medicine formula granules, and if the comprehensive quality index is lower than a quality standard, judging that the traditional Chinese medicine formula granules in the corresponding batch have quality problems; and analyzing the weighted key feature values to obtain key features having negative effects on the comprehensive quality index. According to the invention, the efficiency and accuracy of quality problem processing are improved.

The invention discloses a traffic flow prediction method based on feature embedding and a space-time multi-head self-attention mechanism, and relates to the technical field of intelligent traffic and flow prediction, and the method comprises the steps: obtaining historical traffic flow sequences of N sensors at the first T moments in a traffic network; taking historical traffic flow sequences of N sensors in a traffic network at the first T moments as input, and generating features including time and space through a feature embedding layer; the generated features including time and space are input into a stacked space-time coding layer, calculation is carried out through a decoupled time multi-head self-attention mechanism and a space multi-head self-attention mechanism, new time features and space features are obtained, the time features and the space features are fused, and space-time features are obtained; inputting the spatial-temporal features into an MLP module, and further extracting high-level features through multi-layer MLP; and outputting the high-level features through a regression layer to obtain a prediction result. According to the method, the problem that the traditional prediction model is difficult to process the time-space complexity at the same time is effectively solved by utilizing a decoupling type time and space multi-head self-attention mechanism, and the accuracy of traffic flow prediction is improved.

The invention relates to the technical field of microelectronics, and discloses a processing method of a passive chip for magnetic field energy collection, which comprises the following steps: S1, providing a chip with a flexible substrate, and depositing a first dielectric layer on the surface of the substrate to isolate the substrate from a coil; s2, forming a three-dimensional spiral coil structure on the dielectric layer by adopting a three-dimensional micro-nano processing technology, wherein the coils are stacked layer by layer in an asymmetric spiral form along a direction perpendicular to the substrate; and S3, in the forming process of the coil, a multi-wire-harness cross arrangement design is used, so that the inductance increment of a unit area is maximized. Through the design of the three-dimensional spiral coils and the asymmetric coils, the limitation of a traditional planar coil structure on magnetic field coupling and induction efficiency is broken through, the three-dimensional spiral coils are arranged in a multi-layer mode in the vertical direction, magnetic flux coupling between the coils is tighter, and the inductance value and the induced electromotive force are remarkably improved.

The invention discloses a patch mode revocation-based vulnerability benchmark test set automatic construction method and system, and the method comprises the steps: 1, obtaining a target program, and collecting historical vulnerabilities in other versions of programs of a project where the target program is located; 2, searching a historical vulnerability patch related to the historical vulnerability in a patch database; 3, locating a bug patch in a target program according to the historical bug patch and the historical bug; and 4, transplanting the historical vulnerabilities to the positions of the vulnerability patches in the target program according to a vulnerability patch revocation rule to obtain a real vulnerability test set. According to the method, automatic construction of the real vulnerability test set is realized through vulnerability patch positioning and vulnerability patch revocation rules, and a more real and comprehensive test benchmark is provided for evaluation of a fuzzy test tool.

The invention discloses a multi-combined supply system and method integrating Carnot cell energy storage and liquid hydrogen cold energy. The multi-combined supply system comprises a vapor compression refrigeration cycle charging system, a Rankine cycle discharging system, a low-temperature energy storage system, a high-temperature energy storage system, a liquid hydrogen cold energy utilization system, a fuel cell waste heat utilization system and a liquid hydrogen pipeline. The charging system comprises a first compressor, a first condenser, a second compressor, a second condenser, a first throttle valve and a first evaporator which are sequentially communicated and circulated; the high-temperature heat storage tank and the low-temperature heat storage tank of the high-temperature energy storage system exchange heat with the discharge system through the second evaporator and the third evaporator; a high-temperature cold storage tank and a low-temperature cold storage tank of the low-temperature energy storage system exchange heat with the charging system and the discharging system through a first evaporator and a third condenser respectively; the liquid hydrogen cold energy utilization system achieves the liquid hydrogen gasification cold energy power generation process. The efficient comprehensive energy system is constructed by using liquid hydrogen cold energy and fuel cell waste heat.

The invention relates to an intelligent card chip energy consumption prediction and optimization method and a related device. The method comprises the following steps: acquiring operation data based on a target working mode acquired by an intelligent card chip in a target time window, and performing time slice division processing and clustering processing on the operation data to obtain a multi-dimensional feature vector in the target time window; obtaining a trained energy consumption prediction model, inputting the multi-dimensional feature vector into the energy consumption prediction model for processing, and predicting to obtain a dynamic energy consumption distribution diagram corresponding to the energy consumption parameters of the smart card chip; and determining each energy consumption unit of the smart card chip, and optimizing the energy consumption parameters based on the association relationship between each energy consumption unit and the dynamic energy consumption distribution diagram to obtain optimized energy consumption parameters. By adopting the method, energy consumption prediction and optimization can be efficiently and effectively carried out on the intelligent card chip.

The invention discloses a software test case generation method based on a bidirectional long and short memory algorithm, and belongs to the technical field of software testing. The method comprises the steps of obtaining a to-be-processed file; preprocessing the to-be-processed file to obtain preprocessed text data; inputting the preprocessed text data into the trained Bi-LSTM network for feature extraction to obtain text features; and generating a test case according to the text features. According to the method, the software test case is automatically generated by innovatively applying the bidirectional long and short memory network, and deep understanding of software requirement documents, codes and logs is realized through a deep learning algorithm, so that the correlation and accuracy of the test case are improved. According to the method, the test process can be optimized, and it is ensured that the test cases can cover more test scenes including conventional operation and potential abnormal conditions.

The invention belongs to the technical field of generators, and particularly relates to a hybrid nano generator and a power management circuit thereof, comprising a shell, and an electromagnetic power generation unit, a friction power generation unit and a piezoelectric power generation unit which are located in the shell; and a corresponding power management circuit. The interior of the shell is connected with the top end and the bottom end in a penetrating mode through a glass fiber connecting shaft. A friction power generation unit, an electromagnetic power generation unit and a piezoelectric power generation unit are sequentially mounted at the bottom end, opposite to the shell, of the glass fiber connecting shaft; the piezoelectric power generation unit comprises at least four cantilevers and a disc, one end of each cantilever is fixed on the disc, and a mass block is placed on the upper surface of the other end; the center of the disc is connected with a glass fiber connecting shaft, and the inner sides of the upper surfaces of the four cantilevers are covered with PZT materials. According to the hybrid nano generator, vibration energy in the environment can be converted into electric energy, and energy is supplied to wireless sensor network nodes.

The invention discloses energy-saving tubular pile concrete admixture proportioning equipment and a method thereof, and relates to the technical field of concrete mixing equipment, the energy-saving tubular pile concrete admixture proportioning equipment comprises a mixing drum, a plurality of supports and a feeding hopper, the supports are fixedly connected to the outer surface of the mixing drum, and the feeding hopper is fixedly connected to the outer surface of the mixing drum. Transmission can be shared by the four second gears through the first gear, then balance and sharing of the transmission are achieved, the labor-saving effect is achieved, the purpose of saving energy for the motor is achieved, the rotating speed of the second gears can be higher than that of the first gear through different rotation ratios, rotation of the motor can be effectively converted, and the motor is more stable in rotation. The energy utilization rate of the motor is increased, the cement mortar can be fully mixed within a short time when being stirred by the stirring shaft, the cement mortar can form up-and-down backflow in the stirring barrel through the spiral stirring effect of the stirring shaft, and the mixing efficiency of the cement mortar is effectively improved.

The invention discloses a signal modulation and demodulation method based on an electromagnetic metasurface, and relates to the technical field of wireless communication, and the method comprises the following steps: S1, electromagnetic metasurface design: a programmable electromagnetic metasurface is designed, the electromagnetic metasurface is composed of a plurality of unit structures, and each unit structure is independently controlled; according to the signal modulation and demodulation method based on the electromagnetic metasurface, through a series of complete operation processes of encoding, modulation, transmission, demodulation, recovery and the like on digital information, the requirements for the transmission rate, stability and accuracy of the digital information in different application scenes are met, the signal modulation and demodulation strategy is improved, and the signal modulation and demodulation efficiency is improved. The continuity and accuracy of digital information transmission are ensured, and the situations of jamming, disconnection or data errors are avoided.

The invention discloses a 3D target detection method, a computer program product and a terminal, belongs to the technical field of image processing, and jointly calibrates a laser radar and an image acquisition device. Mapping the global feature of the point cloud to an image feature map by using a transformation kernel, constructing a sparse matrix, and flattening the image feature map into a matrix; performing batch processing and normalization processing on the sparse matrix and the matrix, and then performing connection processing to obtain fusion features of the image and the point cloud; performing region proposal by using the fusion features to generate candidate target regions; and carrying out classification and bounding box regression on the candidate regions, determining the category and the position of the target, and obtaining a 3D target detection result. The mapping relation between the point cloud and the monocular image is constructed by using the sparse point cloud, and it is ensured that only image pixel point coordinates and point cloud coordinate pairs sharing the same point can be paired, so that accurate alignment of features is realized, the accuracy of target detection is improved, and the precision of 3D target detection is ensured.

The present invention relates to a coating composition for forming a UV light conversion coating for a photovoltaic cell, a UV light conversion coating prepared therefrom, a method for preparing the UV light conversion coating, and a photovoltaic cell and a photovoltaic module comprising the coating. The composition comprises by mass: (A) 100 parts of a first epoxy resin that is liquid at room temperature, which is an epoxy resin containing at least two glycidyl ether groups per molecule on average; (B) 33-66 parts of a second epoxy resin that is liquid at room temperature, which is an alicyclic epoxy resin containing at least two epoxy cycloalkyl groups per molecule on average; (C) 33-66 parts of a vinyl ether monomer that is liquid at room temperature; (D) 6-16 parts of a cationic initiator; (E) 3-15 parts of a UV light conversion agent; and (F) 0-200 parts of an inert diluent. The light conversion coating prepared by the coating composition can reduce the UV band light transmittance and increase the visible light wavelength transmittance when used in a photovoltaic cell or a photovoltaic module, improve the light conversion efficiency and have excellent adhesion and aging resistance.

The utility model discloses an underwater bearing device based on a power generation device, relates to the technical field of power generation devices, and solves the problem that the power generation device in the prior art is lack of a protective structure when used in water. The device comprises a spherical shell used for wrapping a power generation device, a connecting structure is arranged in the circumferential direction of the spherical shell, and the spherical shell is connected with a plurality of blades through the connecting structure; and the spherical shell is provided with a flexible connection wire outlet connector which meets the current output requirement. By arranging the spherical shell, an existing power generation device can be wrapped, so that the internal power generation device is protected, the device can play a very good protection role, meanwhile, water can be isolated, the problems of corrosion, liquid inflow and the like can be avoided, meanwhile, water flow power can be fully utilized to be transmitted to the internal power generation device, and the device is simple in structure, reliable, durable and suitable for popularization and application. The economical efficiency is high.

The invention discloses a text embedding optimization-based sensitive content filtering and defending method for a text generation graph model, and relates to the technical field of large model security. The method comprises the steps of preparing a positive and negative sample data set and a common data set; parameters of a text encoder are adjusted, and a teacher-student model is used for training; stabilizing a text encoder by using a common data set; and designing a comprehensive loss function, performing training optimization in combination with the plurality of data sets to generate an adjusted student model, and generating a final security image through text embedding. According to the method, when text input containing sensitive vocabularies is processed through fine tuning of a text encoder, text embedding of the sensitive vocabularies is converted into text embedding of corresponding benign vocabularies, and the benign text embedding is used for guiding generation of a security image; on the premise of not changing the model generation capability, the generation of images containing sensitive pornography or other images protected by copyright by the text graph model is avoided, and the effective filtering, conversion and defense of sensitive contents are realized.

The present invention relates to an interior trim assembly wherein the interior trim assembly comprises a trim part (1) having an upper surface and a lower surface, the trim part (1) comprising graphene, the lower surface of the trim part (1) being integrated with a semiconductor cooling plate (2) for absorbing heat from the lower surface of the trim part (1), the semiconductor cooling plate (2) being thermally insulated from the upper surface of the trim part (1) when the temperature of the upper surface of the trim part (1) rises. The temperature difference between the upper surface and the lower surface of the decorating part (1) can make graphene generate electric energy, and after the electric energy between the upper surface and the lower surface of the decorating part (1) is supplied to the semiconductor cooling plate (2), the semiconductor cooling plate (2) can make the temperature of the decorating part (1) reduced.

The invention provides a modularized multi-level direct-current transformer topology suitable for a multi-voltage-level direct-current power grid and a control method. The topology comprises a bridge arm inductor, an upper bridge arm, a lower bridge arm and a filter inductor. One end of the bridge arm inductor is connected with the high-voltage side anode, and the other end is connected with one end of the upper bridge arm; the other end of the upper bridge arm is connected with one end of the lower bridge arm and one end of the filter inductor; the other end of the lower bridge arm is respectively connected with the high-voltage side cathode and the low-voltage side cathode; the other end of the filter inductor is connected with the low-voltage side anode; each of the upper bridge arm and the lower bridge arm comprises a sub-module string, each sub-module string comprises a plurality of sub-modules which are sequentially connected in series, and the voltage of each of the upper bridge arm and the lower bridge arm is adjusted by controlling the input number of the sub-modules, so that direct-current voltage conversion among different voltage levels is realized. Through the application of the upper and lower chain type sub-modules, the use of complex components is reduced, the cost and the control difficulty are reduced, and meanwhile, the safety, reliability and efficiency of the system are improved.

The invention relates to a method for generating an infrared sea ice area by simulating a visible light sea ice area, and relates to the technical field of image processing. The method comprises the following steps: determining a skyline in a visible light sea ice image; and dividing the image into a sky image and a sea ice image according to the determined sky line, performing gray scale gradient processing on the sky image, performing simulation processing on the sea ice image, and generating a long-wave infrared imaging effect. According to the method for generating the infrared sea ice area by simulating the visible light sea ice area, an effective infrared sea ice area image can be obtained, and subsequent image fusion and target detection and recognition are facilitated.

The utility model provides a loudspeaker structure based on vibration sound transmission, which relates to the technical field of electroacoustic conversion devices, and comprises an upper cover assembly, a voice coil assembly, a magnetic circuit assembly and a lower cover assembly, and the surface of the upper cover assembly is provided with a heat dissipation hole I, a positioning hole and a fixing hole. According to the utility model, the magnetic circuit assembly and the voice coil assembly which are modularly designed are matched, an innovative structure that the voice coil is fixed and the magnetic circuit is movable is adopted, an electric signal is directly converted into mechanical vibration to be conducted to a human body, effective conversion from a sound signal to somatosensory vibration is realized, and the limitation that a traditional loudspeaker can only transmit sound waves through air is overcome; the heat dissipation holes and the ventilation grooves arranged in the upper and lower cover assemblies form convection heat dissipation channels, the heat conduction performance of the metal heat dissipation fins is matched, the heat dissipation problem during working of the vibration horn is effectively solved, long-time stable operation of equipment is guaranteed, meanwhile, the elastic supporting structure of the elastic pieces is utilized, vibration loss is reduced, and vibration conduction efficiency is improved.

The utility model provides an energy storage system, and relates to the technical field of power grid management, and the energy storage system comprises an energy storage device, a switching device, and an AC function unit. Wherein the energy storage device is connected with an alternating current port of the energy storage system and is used for alternating current and direct current conversion and electric energy storage; the alternating current function unit is used for providing a supporting operation function for the energy storage system; the AC function unit is connected with the AC port through the switching device, and the switching device is used for controlling the commutation or disconnection of a power supply path of the AC function unit at a target phase. Therefore, the alternating current function unit is used for providing a supporting operation function for the energy storage system, the stability and the reliability of the energy storage system in the operation process are guaranteed, meanwhile, the application scene of the energy storage system is enriched, and the system is particularly suitable for the scene that a single-phase important load is connected to micro-grid energy storage three-phase alternating current.

The invention provides a visual vibration measurement method with high generalization ability, which comprises the following steps: S1, selecting labels with high generalization ability: collecting historical labels and making the historical labels into a data set; a classic deep learning algorithm is selected for training to decide an optimal label and rule; s2, high-efficiency target detection: designing a high-efficiency detector; a simple target tracker is provided; and converting vibration displacement. According to the method, the practicability, the reliability and the accuracy of vision measurement vibration are remarkably improved, and a solid technical foundation is provided for application in related fields.

The invention belongs to the technical field of thermal protection of high-speed aircrafts, and particularly relates to a thermal protection and heat-work conversion system of a high-speed aircraft. Comprising a thermal protection layer, a turbine generator and a spray pipe, the thermal protection layer comprises a cooling working medium storage layer and a catalytic decomposition reaction heat absorption layer which are arranged inside and outside, the cooling working medium storage layer is connected with the inlet end of the catalytic decomposition heat absorption layer, and the outlet end of the catalytic decomposition heat absorption layer is connected with the turbine generator and the spray pipe; a cooling working medium exchanges heat with the aircraft, is subjected to an endothermic decomposition reaction, then enters the catalytic decomposition heat absorption layer and is subjected to an endothermic chemical reaction in the catalytic decomposition heat absorption layer, and generated high-temperature and high-pressure gas enters the turbine generator and the spray pipe for power generation and acting respectively, so that conversion of heat and work is realized. The weight of the thermal protection layer is gradually reduced, meanwhile, gas generated through decomposition is used for other purposes, and multi-stage utilization of pneumatic waste heat and a cooling working medium is achieved.

The invention relates to the technical field of mowing and road cleaning all-in-one machines, and particularly discloses an intelligent mowing and road cleaning all-in-one machine which comprises a moving assembly, the bottom of the moving assembly is fixedly connected with a mowing and road cleaning assembly, and the top of the moving assembly is fixedly connected with a lifting power assembly. The moving assembly comprises a compression pushing assembly, the bottom of the compression pushing assembly is fixedly connected with a power assembly, the mowing road cleaning assembly comprises a fan assembly, a rotating arc-shaped assembly is installed at the bottom of the fan assembly, and a linkage assembly is installed at the bottom of the cover plate assembly. In the process that the linkage assembly drives the cutting assembly to rotate, the positioning round rod and the rotating bearing can rotate relative to the second positioning rod, it is guaranteed that the annular cover plate ascends and descends to drive the cutting assembly to rotate so that the cutting assembly can form a cutting blade and a barrel, and effective conversion between structures can be guaranteed conveniently.

The invention discloses a hydroelectric overhaul tool sorting method based on a SqueezeNet convolutional neural network model. The method is suitable for various tools such as wrenches, screwdrivers and sleeves. The method comprises the following steps: firstly, acquiring a maintenance tool image in a backlight environment, and removing background interference through an image preprocessing technology to generate a clear target binary image; in order to solve the problem of insufficient samples, data enhancement means such as expansion, corrosion and rotation are adopted to expand a training set. And then, deep learning training is carried out by using a SqueezeNet model, parameters are adjusted in combination with a cross entropy loss function and a gradient descent optimization algorithm, and finally, a high-precision recognition model is obtained. In addition, hand-eye calibration of the visual system and the mechanical arm is completed through a nine-point calibration method, and accurate conversion from image coordinates to mechanical arm action coordinates is ensured. In actual operation, stacked tools are separated through the flexible vibration disc, then the visual system collects images of the single tools, the types and the positions of the single tools are recognized through the model, and the mechanical arm is guided to conduct accurate grabbing.

The invention discloses a virtual machine instruction conversion method and device, electronic equipment and a storage medium. The method comprises the following steps: splitting a first executable file of the 16-bit virtual machine to obtain a plurality of components, and determining a first method component from the plurality of components; splitting the first method component to obtain a plurality of executable methods; performing instruction screening processing in the instruction sequences of all the executable methods to obtain a target integer instruction comprising an integer variable; reconstructing a target integer instruction, and modifying the number of transmitted parameters to obtain a reconstructed target integer instruction; and on the basis of all instructions including the reconstructed target integer instruction, instruction rearrangement processing, instruction updating processing and method component generation processing are performed in sequence, and a second executable file used in the 32-bit virtual machine is generated on the basis of a regenerated second method component. Incompatible instructions in the first executable file of the 16-bit virtual machine can be conveniently and effectively converted and reconstructed, and the second executable file suitable for the 32-bit virtual machine is obtained.

An embodiment of the present application provides a method and device for determining a radar deflection angle, wherein the determination method includes: measuring a first deflection angle between a straight lane and a longitudinal axis of a first coordinate system; a plane formed by the longitudinal axis and the transverse axis of the first coordinate system is parallel to a plane formed by the longitudinal axis and the transverse axis of an electronic map coordinate system, and the origin of the first coordinate system coincides with the origin of the radar coordinate system; acquiring radar data, the radar data including radar coordinates of a target detected by the radar; selecting radar data of a motion trajectory of the target within the range of at least one straight lane from the radar data; estimating a trajectory line according to the selected radar data, and determining a second deflection angle between the trajectory line and the longitudinal axis of the radar coordinate system; and determining the radar deflection angle according to the first deflection angle and the second deflection angle.

The utility model discloses a multifunctional drilling machine maintenance work station which comprises a concave movable support block, a lifting hydraulic supporting device, a pair of lifting limiting expansion pieces and a concave supporting platform, and relates to the technical field of drilling machine maintenance, a lifting threaded pipe is matched with a lifting threaded rod, so that the lifting process is more stable, and shaking and inclination are avoided; the convex telescopic shaft rod is telescopically adjusted in the concave telescopic shaft pipe, so that supporting can be performed according to different road surface conditions, and the overall stability is enhanced; and by opening different hydraulic valves, the convex telescopic shaft rods on the inner sides of different concave telescopic shaft pipes can be drained and pushed, and the flexible adjustment of the supporting effect is achieved.

The invention provides a group chat robot system based on a multi-layer memory module and a data processing method. The method comprises the implementation steps of initializing parameters; the group chat message extraction module converts original chat information sent by each user; the memory module obtains the first memory information and counts the number of the user memory information of different memory layer numbers; the forgetting and conversion module obtains second memory information; the evaluation module obtains third memory information; the reply module generates reply information. A field adding module in the memory module enables memory to be layered, and management and protection of information are facilitated. The statistical module realizes statistics and forgetting of memories of different layers, and the conversion module detects statistical results, deletes information with lower scores under the condition of more information, and converts'memory layers' in each piece of information with higher scores, so that overload of memory information and excessive accumulation of invalid information can be prevented. And the treatment efficiency is effectively improved.

A wave energy conversion device, comprising a housing. A water inlet cavity (25) and a water outlet cavity (26) are provided in the housing, and a one-way flow guide grille is mounted on an opening of each of the two cavities; a water turbine is mounted between the two cavities, and the upper part of the water turbine can be connected to a main shaft of a permanent magnet generator (2); the one-way flow guide grilles of the water inlet cavity (25) and the water outlet cavity (26) are arranged in a transverse V-shaped manner along an up-down direction, and the housing is provided with a water inlet and outlet (23) at a V-shaped opening; the water inlet cavity (25) and the water outlet cavity (26) are combined in an asymmetric manner, and the water inlet cavity (25) and the water outlet cavity (26) are respectively provided with a portion connected to a top portion (19) of the housing, and an air discharge opening is disposed on said portion.