38results about How to "Reduce problem size" patented technology

The invention discloses a power prediction method considering photovoltaic power clustering, which comprises the following steps of: firstly, acquiring historical data of a photovoltaic power station, decomposing a photovoltaic power time sequence by using an STL time sequence decomposition method, performing DBSCAN clustering on the decomposed photovoltaic power data, identifying a low-power cluster and a high-power cluster, and predicting the power of the photovoltaic power station according to the low-power cluster and the high-power cluster; establishing a photovoltaic power prediction model based on a time convolutional network and Times Net for low power and high power respectively; secondly, a badger algorithm is improved, Chebyshev chaotic mapping is adopted to improve initialization of a badger population, and a Gaussian mutation strategy and a lens imaging reverse learning strategy are introduced to obtain MHBA; and optimizing hyper-parameters of the TCN photovoltaic low-power prediction model and the TimesNet photovoltaic high-power prediction model by using an improved badger algorithm, and predicting photovoltaic power by using the optimized models to obtain a final prediction value. Compared with the prior art, the precision and effectiveness of the photovoltaic power prediction model can be improved.

According to the energy storage coupling heat pump steam generation method and system, the mode that the heat pump device and the energy storage device cooperatively operate is adopted, and efficient and economical steam supply is achieved. The method comprises the steps that high-temperature water is prepared and stored in the off-peak electricity period through the heat pump device, and the high-temperature water is further heated through the energy storage device to generate superheated steam for a user to use. The energy storage device comprises a plurality of energy storage modules, and a heat exchange pipeline of the energy storage device can be switched between a parallel connection mode and a series connection mode through switching, so that deep heat release and flexible regulation and control are achieved. According to the method, the energy storage capacity and heat release power of the energy storage module are dynamically adjusted, enthalpy increment distribution of the heat pump and the energy storage device is optimized in combination with load prediction, and diversified steam requirements are met. And meanwhile, preheating or dynamic supplement is conducted on the energy storage module through the heat pump device, waste heat recovery is achieved, and the energy efficiency is improved. According to the method, the operation cost is remarkably reduced, the capacity of the system for adapting to load fluctuation is enhanced, and the method is suitable for industrial steam use scenes and has the advantages of being efficient, stable and flexible.

The invention discloses a wastewater ozone catalytic oxidation treatment process and device, and relates to the technical field of wastewater treatment.According to the wastewater ozone catalytic oxidation treatment process and device, oxygen-enriched gas exhausted by a tail gas destructor is purified and then reused for an ozone generator, the scale of an oxygen preparation system such as an air compressor and an oxygen generator is reduced, and the production efficiency is improved. Therefore, the equipment investment and the operation cost are reduced, and the pollution of harmful gases such as volatile organic compounds (VOCs) in the tail gas to the environment is effectively avoided; meanwhile, carbon dioxide is removed in the tail gas treatment process, and carbon emission is further reduced; besides, through the chemical reaction of treating carbon dioxide by sodium peroxide, not only is harmful gas successfully removed, but also oxygen is additionally generated and is used for supplementing the oxygen concentration which is reduced due to ozone reaction consumption in the mixed gas; while the tail gas recycling efficiency is improved, the dependence on external oxygen is remarkably reduced, and the energy consumption of oxygen production equipment and ozone generation is reduced.

The reciprocating piston type heat engine comprises a cylinder body, a piston, a connecting rod, a crankshaft, a lever and a compensation shaft, the piston is arranged in an inner cavity of the cylinder body, one end of the connecting rod is hinged to the piston, the crankshaft is hinged to the connecting rod, the other end of the connecting rod is hinged to the lever, and the other end of the lever penetrates through a sliding way transversely arranged on the compensation shaft. The lever can slide back and forth in the slideway of the compensation shaft. The piston thrust can be converted into output torque as much as possible, the output torque is effectively increased, and the combustion efficiency of the reciprocating piston type heat engine is improved.

An intelligent compensation method for coupling errors of a dynamic gravimeter comprises the following steps: selecting typical environment interference influencing the dynamic gravimeter in a complex measurement environment, and analyzing an influence mechanism of each typical environment interference on the measurement accuracy of the dynamic gravimeter; according to the influence mechanism under each typical environment interference, analyzing a coupling error influence mechanism between the typical environment interferences; based on the coupling error influence mechanism, constructing a coupling error comprehensive compensation model; adding the coupling effect loss into a data-driven loss function, and defining a model coefficient constrained by physical information; and performing model training by using the training data, and calculating the total loss of the verification set on the obtained model to obtain a final coupling error comprehensive compensation model. According to the invention, the coupling error of the dynamic gravimeter in a complex environment is compensated based on coupling error mechanism analysis and a physical constraint long-short-term memory model, and the measurement precision and environmental adaptability of the dynamic gravimeter under complex multi-factor interference conditions are improved.

An electrode assembly folding apparatus according to one embodiment of the present disclosure includes: a supply unit for supplying the electrode assembly that comprises two sheet-shape separators, a second electrode continuously located between inside surfaces of the separators facing each other, and a first electrode alternately located up and down on outside surfaces of the two separators, wherein a first unit body and a second unit body are alternately connected to each other, and wherein the first electrode of the first unit body is located at an upper side and the first electrode of the second unit body is located at a lower side, a holding unit that holds the first unit body supplied from the supply unit and transfers it through a swing motion, thereby folding the electrode assembly in a zigzag shape, and a stack unit that stacks the first unit body transferred by the holding unit.

The utility model provides a kind of isopropanol wastewater treatment system.The system includes distillation module, denitrification module and AO biochemical module;The inlet of denitrification module is also connected with nitrate nitrogen wastewater source, and the outlet of denitrification module is connected with the inlet of AO biochemical module.The system can recover most isopropanol in isopropanol wastewater, realize resource utilization, and convert nitrate nitrogen in nitrate nitrogen wastewater into nitrogen by denitrification, remove carbon source at the same time, greatly reduce oxygen consumption and sludge production during subsequent AO biochemical treatment, reduce wastewater treatment cost;At the same time, the treatment effect of the treatment system is better, can greatly reduce COD value and TN content in wastewater.

The invention belongs to the technical field of equivalent modeling, and discloses a hybrid wind power plant equivalent method and device based on oscillation energy decomposition, and the method comprises the steps: obtaining the operation measurement data of a hybrid wind power plant, carrying out the calculation or frequency scanning of each fan grid-connected branch, and obtaining a broadband equivalent impedance characteristic; constructing a multi-machine oscillation energy equivalent unit comprising a plurality of grid-connected branches according to the broadband equivalent impedance characteristic and the actual grid-connected topology of the wind power plant; establishing an energy function and a dynamic equation according to the multi-machine oscillation energy equivalent unit; performing modal decomposition and branch decomposition on the system oscillation energy, and calculating the oscillation energy contribution degree and the dissipated energy participation degree of each fan grid-connected branch in a dominant oscillation mode to obtain an energy characteristic quantity; performing fan grouping to obtain a fan grouping result of the hybrid wind power plant; and performing differential equivalence and parameter aggregation according to a fan grouping result to obtain a small disturbance equivalence model. The calculation model scale and the calculation complexity of the hybrid wind power plant can be reduced.

An electrode assembly folding apparatus for folding an electrode assembly in a zigzag shape according to one embodiment of the present disclosure includes: a supply unit for supplying the electrode assembly that comprises two sheet-shaped separators, a second electrode continuously located between inside surfaces of the separators facing each other, and a first electrode alternately located up and down on outside surfaces of the two separators, wherein a first unit body and a second unit body are alternately connected to each other, wherein the first electrode of the first unit body is located on an upper side and the first electrode of the second unit body is located on a lower side, a holding unit that holds and transfers the first unit body supplied from the supply unit, and a stack unit that stacks the first unit body transferred by the holding unit.

This invention provides a non-uniform discrete equivalent circuit modeling method and system for electromigration reliability. It employs an equivalent circuit-based approach to analyze the electromigration stress evolution of general complex interconnect structures, fully considering thermal field effects in the model. This achieves equivalent circuit modeling for stress analysis under multi-physics coupling. Furthermore, by using an adaptive mesh method with non-uniform spatial discretization and a steady-state / quasi-steady-state filtering mechanism, it solves the problem of wasted computational resources in traditional uniform discretization methods when dealing with large-span interconnect structures, and avoids unnecessary transient simulations of immortal interconnect trees, reducing the computational load of full-chip-level analysis. This enables efficient and high-precision transient analysis of the electromigration reliability of large-scale interconnect networks under multi-physics coupling effects.

The application discloses a whole-body-local multi-scale biomechanical model coupling construction method for human body impact simulation, comprising the following steps: constructing a whole-body dynamics model, obtaining the kinematic response and dynamic response of the human body under a preset impact working condition; determining the connection constraint position corresponding to a target local area in the whole-body dynamics model, and extracting the constraint force and / or constraint moment at the connection constraint position; performing equivalent conversion on the constraint force and / or constraint moment to obtain equivalent load for representing the whole-body dynamics effect acting on the target local area; constructing a local simulation model of the target local area, and applying the equivalent load to the local simulation model to realize simulation analysis of the human body impact response. Through the conversion of the whole-body dynamics constraint force to the local simulation model, the application reduces the calculation complexity of the human body impact simulation while retaining the whole-body inertia effect, and improves the simulation efficiency and calculation stability.

The invention discloses a copper alloy material failure evaluation and corrosion rate prediction method under different seawater flow rates, and belongs to the field of marine environment copper alloy corrosion. The method comprises the following steps: firstly, analyzing an XRD (X-Ray Diffraction) spectrum of the copper alloy corroded by flowing seawater, and determining a micro-strain level and dislocation density; atomic-scale axial tensile strain is applied to the crystal face (111) according to the density functional theory, and microstructure models under different seawater flow velocities are established. And analyzing the dissolution potential, surface energy, work function and dissolution activation energy barrier of copper atoms on the alloy surface, and exploring the relationship between the dissolution rate and the activation energy barrier. And establishing a prediction model of the seawater flow velocity and the copper alloy corrosion rate in the seawater dynamic environment. According to the method, the corrosion dynamic characteristics of the copper alloy at different seawater flow velocities are obtained, and the analysis of the atomic-scale axial tension induced copper alloy corrosion failure dynamic process is realized, so that a theoretical basis is provided for the anti-corrosion design of a copper alloy component.

The application discloses a centrifugal pump off-design condition diagnosis method and system based on a vibration signal, and the method comprises the following steps: collecting vibration signals of a centrifugal pump under multiple flow conditions, and constructing a vibration sample sequence with a working condition label; reconstructing the vibration signal by using a variational mode decomposition (VMD); performing a continuous wavelet transform (CWT) on the reconstructed signal, and mapping a one-dimensional vibration signal into a two-dimensional time-frequency graph; constructing a Light-CBAM-CNN diagnosis model, and training the same; inputting the vibration signal collected in real time into the trained diagnosis model after pretreatment, VMD and CWT, outputting a working condition category and a diagnosis confidence, and outputting an "uncertain" label and triggering a retest mechanism or an alarm when the confidence is lower than a threshold value. The application can realize online diagnosis of high-energy consumption off-design conditions of the centrifugal pump only by relying on the shell vibration signal, and has the advantages of small modification workload, low deployment cost, high identification reliability and strong energy-saving scheduling guiding significance.

This application provides a mobile robot task scheduling method and system based on a heterogeneous computing architecture, relating to the field of intelligent scheduling technology. The method includes: constructing a workspace geometric model based on the geometric structure of a planting row; obtaining the motion state parameters of the mobile robot using a first computing unit; determining the look-ahead compensation amount by combining the statistical value of the inference time of the target recognition task by a second computing unit; determining a spatiotemporal pre-aiming slice in the visual field region based on the look-ahead compensation amount and the workspace geometric model, and cropping image data located within the slice from the original image data; the second computing unit performing target recognition on the image data, generating a work instruction corresponding to the recognized target, and controlling the lateral actuator to perform the work according to the work instruction. This can alleviate the spatiotemporal asynchrony problem between perception results and execution position in a moving work scenario, reduce missed and misaligned tasks, and reduce the communication and computational load of the heterogeneous computing system.

The invention belongs to the field of image analysis, and particularly relates to an unmanned aerial vehicle power inspection foreign matter detection method and device based on YOLO and an unmanned aerial vehicle. The method comprises the following steps: S1, constructing a foreign matter detection network based on YOLO, wherein the network comprises an input layer, a backbone network, a feature fusion network and a prediction head; s2, a large number of real electric power inspection images collected by the unmanned aerial vehicle are obtained, and a target selection box and category marking information are manually added to foreign matter existing in the images to serve as sample images; and S3, dividing the data set containing a large number of sample images into a training set and a test set, and training and testing the foreign matter detection network. S4, retaining the model parameters of the network meeting the requirements after the test; and the system is used for carrying out real-time processing on real-time images collected in the line patrol process of the unmanned aerial vehicle and identifying foreign matters contained in the real-time images. The problem that an existing target detection algorithm is difficult to be suitable for a foreign matter detection task in an electric power inspection image is solved.

This invention proposes a hardware implementation method, apparatus, electronic device, and computer-readable storage medium for the Sigmoid function, relating to the field of data processing technology. The method constructs a function mapping table for the Sigmoid function, containing discrete non-negative floating-point values ​​within the unsaturated interval, based on the symmetry and saturation of the Sigmoid function, thereby substantially reducing the scope of the lookup table. On this basis, the absolute value of the floating-point values ​​belonging to the unsaturated interval is taken to obtain the value to be processed. This value is then converted to the target fixed-point format, and integer operations are used for table lookup, interpolation, and symmetry restoration to obtain the output value of the Sigmoid function based on the target floating-point value, improving computational efficiency. While ensuring the function output capability across the entire range, the size and storage overhead of the function mapping table are effectively reduced, while linear interpolation under the fixed-point format improves output accuracy.

The invention relates to the field of low-orbit satellite navigation enhancement and conduction fusion positioning application, and provides a low-orbit satellite positioning time service signal receiving and processing method based on double sliding windows. According to the method, segmentation and burst modulation characteristics of signals are fully analyzed, and sequential processing of the segmented signals is realized by using a data caching method; real-time detection and identification of burst signals are realized by using double sliding FFT windows; partially matched filtering and frequency domain searching are used to realize accurate synchronization and measurement of pilot frequency ranging signals. Rapid demodulation of a message data frame is realized by using closed-loop tracking and dynamic assistance; and after data such as Doppler observed quantity, pseudo-range observed quantity and telegraph text data frames are obtained, receiving processing of the low-orbit satellite positioning time service signals is finally completed. According to the method, the difficulty of burst and real-time receiving and processing of the low-orbit satellite positioning time service signal is solved, meanwhile, the scale and operand of a correlator for signal processing are reduced, the occupied resources are few, and the operation efficiency is high.

This application provides a deep learning-based method and system for evaporator defect detection, relating to the field of anomaly detection. The method includes: acquiring multi-view image data of the evaporator; extracting multi-scale features, including the number of fins, their arrangement direction, and the spatial location of notches, through the backbone network layer of a deep learning target detection model; nonlinearly enhancing the multi-scale features through the model's feature enhancement network layer to obtain enhanced multi-scale features; and then extracting reconstructed features through a hybrid attention mechanism layer; based on the reconstructed features, obtaining the detection results of evaporator fins and notches through the model's detection layer, and comparing them with preset structural parameters to obtain the defect detection results. This application, used in evaporator defect detection, solves the technical problem of existing technologies being unable to stably identify evaporator defects under changing environmental and evaporator structural parameters.

The embodiment of the invention provides an ultra-low noise tile-type phased array plane based on a silicon-based chip, and the phased array plane comprises an antenna part PCB, and the antenna part PCB comprises an array antenna, an LNA chip, a board-level filter, a BFIC chip based on a silicon-based technology, and a power division network which are connected in sequence; wherein the array antenna is provided with a patch antenna, and a feed point of the patch antenna is directly connected with the LNA chip through a plated-through hole; the LNA chip is connected with the board-level filter through a plated-through hole, the board-level filter is connected with the BFIC chip through a plated-through hole, and the BFIC chip is connected with the power division network through a plated-through hole; the LNA chip is integrated with a dynamic tuning matching circuit, and dynamic impedance matching can be carried out according to the working condition of the patch antenna. Therefore, the tile-type phased array plane provided by the embodiment of the invention realizes the consideration of high integration level, low noise, low loss, dynamic matching and interference resistance, and overcomes the defects of the traditional tile-type phased array plane in the aspects of integration level, noise, loss, matching flexibility, filtering and the like.

The invention discloses a false star filtering and star map recognition method based on multi-sky-area parallelism, which is applied to autonomous attitude determination of a star sensor and comprises the following steps: constructing a multi-sky-area parallelism index database of a whole celestial sphere, dividing a whole celestial sphere space represented by a unit spherical surface, and generating a plurality of candidate sky-area sets which cover the whole celestial sphere and are allowed to be overlapped with one another; the method for extracting star pattern features based on the discretized radial angular distance comprises the following steps: extracting position parameter information of observation star points based on star map data acquired by a star sensor, and constructing a radial feature vector for representing a local spatial distribution relationship of fixed stars for each observation star point in a star map; performing feature matching of the multi-day-area parallel star mode feature vectors, and calculating a matching evaluation index based on the feature matching; and a final star map recognition result is obtained based on iterative screening recognition and false star elimination. And outputting a final star map recognition result for subsequent attitude calculation. The invention further discloses a corresponding system, electronic equipment and a computer readable storage medium.

The invention discloses a water-rich rock mass fracture precursor identification method based on a lightweight neural network, and belongs to the technical field of deep rock mass engineering safety monitoring, and the method comprises the steps: collecting pulse waves and interference waves of deep rock mass fracture to form a data pair, building a waveform library based on the data pair, building a lightweight model for predicting rock fracture, and carrying out the prediction of rock fracture. The method comprises the following steps of: extracting and splicing data pairs to obtain a fused feature map, extracting local features, fusing to obtain a time sequence feature map, compressing by utilizing global pooling, converting into a nonlinear channel association vector, normalizing to obtain a weight vector, combining with the time sequence feature map, and compressing to obtain a channel statistical vector; and performing nonlinear feature transformation and numerical regularization by using a full connection layer, and obtaining ternary probability distribution through a Softmax function. According to the invention, key technical support is provided for real-time monitoring, and urgent demands of water-rich rock mass engineering on intelligent, real-time, high-reliability and safe monitoring are precisely met.

The utility model discloses a brake cylinder cover press-fitting device, and belongs to the technical field of brake cylinders. The brake cylinder cover press-fitting device comprises a support mechanism which comprises a support and a supporting column, and the supporting column is arranged on the support; the clamping mechanism comprises a first clamping piece, a second clamping piece and a clamping driving cylinder, at least one of the first clamping piece and the second clamping piece is rotatably arranged on the supporting column in a sleeving mode, the first clamping piece is provided with a first clamping part, the second clamping piece is provided with a second clamping part, the first clamping part and the second clamping part are oppositely arranged, and the clamping driving cylinder is arranged on the supporting column; a cylinder bottom and a cylinder rod of the clamping driving cylinder are rotationally connected with the first clamping piece and the second clamping piece respectively; and the press-fitting mechanism comprises a press-fitting driving cylinder and a jacking piece, the press-fitting driving cylinder is connected with the support, the jacking piece is connected with a cylinder rod of the press-fitting driving cylinder, and a positioning hole matched with the piston rod is formed in the jacking piece. The brake cylinder cover press-fitting device is easy and convenient to operate, safe, efficient and low in cost.

The invention relates to the technical field of deep learning, and discloses a feature analysis method for a deep learning model, a storage medium and electronic equipment. The method comprises the steps of obtaining a mapping relationship between a neuron activation value of a deep learning model and a feature region of a classification object; determining a suspected feature region of the classification object according to the mapping relation; the suspected feature region represents a feature region which has high contribution to activation of a plurality of important neurons of the deep learning model at the same time; performing model training on the deep learning model based on the classification object to obtain an output data variable quantity; and determining a key feature region from the suspected feature regions according to the output data variable quantity. The pruning operation can be more targeted, and the scale and the calculation complexity of the model are reduced on the premise that the model performance is guaranteed.

The invention discloses a nuclear reactor neutron transport calculation method and device, and the method comprises the steps: carrying out the space merging of a fine network multi-group neutron transport model in each time of source iteration in a process of calculating the fine network multi-group neutron transport model through a source iteration mode, and obtaining a coarse network multi-group neutron transport model; correcting the first coarse mesh multi-group neutron standard flux through the coarse mesh multi-group correction factor to obtain a second coarse mesh multi-group neutron standard flux, and solving the coarse mesh multi-group neutron transport model to obtain a third coarse mesh multi-group neutron standard flux; performing iterative solution on the coarse mesh multi-group neutron transport model to obtain a target coarse mesh multi-group neutron standard flux so as to determine a fine mesh multi-group correction factor; and correcting the initial fine-net multi-group neutron standard flux through the fine-net multi-group correction factor to obtain a target fine-net multi-group neutron standard flux, and calculating the fine-net multi-group neutron transport model through the target fine-net multi-group neutron standard flux to improve the efficiency of nuclear reactor neutron transport calculation.

The invention relates to a shale condensate gas reservoir circulation gas injection process optimization method and device and a storage medium. The method comprises the steps that geological characteristic parameters and fracturing construction microseismic monitoring data corresponding to a target shale condensate gas reservoir are obtained; establishing a geological numerical model and a crack model based on the data, establishing a component mathematical model, performing state calculation based on a state equation, performing condensation flash calculation, performing full implicit numerical solution, performing relative permeability and capillary pressure model, and optimizing a circular gas injection process. According to the method, microseism inversion, phase equilibrium calculation and numerical simulation are integrated, the minimum total volume of stratum gas condensate and the maximum cumulative gas production rate serve as optimization targets, the circulation gas injection rate is quantitatively determined, and the shale gas condensate reservoir circulation gas injection process is quantitatively optimized; the problems that in the prior art, due to the fact that optimal gas injection technological parameters need to be found through a large number of field tests, mining time is long, efficiency is low, and cost is high are solved.

A plunger type gate device comprises a plunger and a crown part, the plunger extends into the section of a water passing hole, the plunger comprises a gate leaf, the gate leaf is composed of a plunger supporting beam, an upstream face steel plate and a water stop ring, the water stop ring is formed by arranging H-shaped steel or channel steel towards the hole wall to form a circle of groove, U-shaped rubber is inversely buckled in the groove, and a hollow rubber bag is installed in the groove. A plunger supporting beam is arranged in the water stop ring, and an upstream face steel plate is welded to the front side of the water stop ring; the crown is arranged outside the hole, the upward end of a crown supporting beam of the crown is hinged to a base through a connecting rod, the base is embedded in the outer wall of the hole, the top point of a herringbone beam faces outwards and is fixed to the crown supporting beam, the top point of the herringbone beam is connected with the crown supporting beam through a pull rod, and the crown supporting beam is connected with a plunger through a pull rod. The device is simple in civil engineering work, the precision requirement for gate manufacturing and mounting can be lowered, the load and scale of supporting walking components and opening and closing facilities are reduced, and the construction, operation and maintenance cost of the gate is remarkably reduced.

The invention relates to the application field of low-earth-orbit satellite navigation enhancement and low-earth-orbit satellite precision single-point positioning, and provides a low-earth-orbit satellite composite code-shift keying signal receiving and processing method based on extension overlapping. According to the method, the engineering implementation requirement of the composite code-shift keying signal is met, the signal modulation characteristic of the composite code-shift keying signal is fully utilized, and rapid synchronization of the signal is realized based on a low-speed message signal of the composite code-shift keying signal; secondly, replacing traditional coherent matched filtering with FFT (Fast Fourier Transform) conversion which is easier to engineer by using a principle of replacing time domain convolution with frequency domain multiplication, so as to reduce the complexity of cyclic correlation; and finally, code phases of mapping data bits are expanded and overlapped, so that the parallelism of related processing is improved, the number of coherent matching times is reduced, and the calculation efficiency is improved.

The invention discloses a sketch recognition method. The method comprises the following steps: acquiring sketch original data, converting the sketch original data into sketch sequence data, removing repeated points and isolated nodes in the sketch sequence data, and acquiring sketch sequence optimization data; sampling the sketch sequence optimization data by adopting a vertex distance maximization sampling method, and normalizing the sampled sketch sequence optimization data to obtain sketch sequence standard data; according to the sketch sequence standard data, constructing sketch data with a time sequence structure; and constructing a sketch recognition model, and obtaining a sketch recognition result according to the sketch data with the time sequence structure and the sketch recognition model. According to the method, when the graph structure of the sketch is constructed, the time sequence information is introduced and is directly processed by GNNs, so that participation of a traditional neural network model is not needed, the scale and calculation overhead of the model are reduced, and the accuracy of subsequent sketch recognition is improved.

The invention relates to the field of unmanned aerial vehicle image processing, in particular to a corn phenological period monitoring method based on unmanned aerial vehicle images and deep learning, and adopts the technical scheme that a corn planting test point is tested by taking a year as a time unit, and N planting densities are set for M hybrid varieties of the test point, so that the diversity of a data set is improved; features in different scenes are considered; the training subset is used for training a MaizePhenoNet model, the verification subset is used for monitoring model training, and the phenomenon of overfitting of model training is prevented; compressing the hidden dimension of the classification head from a 1280 channel to a 512 channel so as to reduce parameters of a full connection layer; scaling expansion channels in the FUSDIB module and the UIB module and carrying out 8-times alignment, so that main parameter overhead caused by point convolution is reduced; under the condition that a backbone topological structure and a down-sampling strategy are not changed, the method effectively reduces the model parameter scale and storage requirements, and is suitable for an end-to-end corn phenological period identification task.