31 results about "Linear variation" patented technology

The present invention relates to the technical field of transistor linear change detection, and discloses an AI detection method for transistor linear change based on a hybrid model. The method comprises: obtaining transistor linear sample image data and preprocessing it, constructing a hybrid model architecture comprising a parallel feature extraction component and a series classification decision component; using the preprocessed data to train an optimization model to determine a target hybrid detection model; inputting real-time image data into the model, extracting spatial distribution and shape contour features through the feature extraction component, determining the type of linear change after fusion through the classification decision component, and outputting a detection result. The present invention achieves multi-dimensional feature collaborative extraction and dynamic fusion through a hybrid model architecture, solving the problem of insufficient detection accuracy of traditional single models and improving the efficiency and accuracy of automated detection of transistor linear changes.

The invention discloses a frequency response test method and system for an optical fiber F-P cavity filter. The method comprises the following steps: S1, executing the following operations through a signal generator: S11, enabling an approximate linear change region of a transmission peak of the filter to be located near the working wavelength of a narrow linewidth laser; s12, a driving signal x (t) changing periodically is added to the filter, and the center of an approximate linear change area of a rising edge / falling edge of a transmission peak of the filter is made to be close to the working wavelength of the narrow-linewidth laser; s13, enabling the working wavelength of the narrow linewidth laser to be always located in an approximate linear change region of the rising edge / falling edge of the transmission peak; and S2, when the driving mechanism of the filter works at the driving frequency f of the driving signal x (t), detecting the narrow linewidth laser passing through the filter by adopting a photoelectric detector PD, and converting the narrow linewidth laser into a response electric signal y (t). According to the invention, the frequency response of the optical fiber F-P cavity filter can be effectively tested.

The application discloses a kind of visual inspection methods based on inductance surface defect, first color image is carried out color channel decomposition, using the gray scale of product image carries out template positioning, obtains coordinate;Second, according to the relative area of positioning coordinate, then carry out morphological operation;Finally, through gray scale stretching, the contrast of defect and background is enhanced, so that it finds out product defect and carries out elimination, obtains quality qualified product.The detection result obtained by the above-mentioned detection method of the application reduces the misjudgment rate, improves the yield, at the same time meets the control of product quality of customer, improves the detection efficiency of product.As the gray scale linear variation of part of defects in different spatial domain is more obvious, the application decomposes the image of different spatial domain through different color component information, so that the defect is more easily obtained and detected.

A kind of contact force measurement method of load launch process, comprising: step one: install load ring assembly on the mouth piece of launch tube through shear sensor;Step two: adjust the concentricity of load ring assembly and launch tube, adjust the levelness of shear sensor;Step three: set stress measuring element on shear sensor;Step four: hang standard weight on load ring assembly, fit linear variation curve between the signal of stress measuring element and the total mass of standard weight;Step five: place launch tube in water and launch load, obtain the contact force corresponding to load ring assembly by linear variation curve between the signal data of stress measuring element and the total mass of standard weight, accurately measure the contact force between the mouth end of launch tube and load during launch process, provide test data support for the establishment of contact force mathematical model of carrier platform and load separation process, improve the accuracy and reliability of load launch trajectory design.

The invention discloses a multi-modal feature-based urban competition effect redundant data elimination system and method, and relates to the field of data processing, and the method comprises the steps: analyzing a nonlinear dependency relationship of historical urban competition effect data through employing a marginal distribution technology, constructing a joint probability distribution function, and obtaining a linear change range of standard urban competition effect data; the method comprises the following steps: acquiring a heterogeneous data stream of a real-time urban competition site by using a multi-modal data collector, and extracting an audience behavior mode as real-time urban competition effect data based on the heterogeneous data stream; and dividing real-time city competition effect data by taking the linear change range as a reference point, generating dispersed redundant data and stacked redundant data, and performing elimination processing on the dispersed redundant data and the stacked redundant data by matching a redundant data elimination rule. According to the method, the data quality can be improved, invalid or interference information is filtered out, and clearer and higher-credibility data support is provided for urban competition effect evaluation.

The application discloses a high-performance airfoil suitable for active stability of a refueling cone, and meets the following conditions: the maximum relative thickness of the airfoil is 5.6%, the position of the maximum relative thickness is located at 10% of a chord length; the maximum relative camber of the airfoil is 4%, the position of the maximum relative camber is located at 31% of the chord length; the relative leading edge radius of the airfoil is 0.0081, and the relative trailing edge thickness of the airfoil is 1%. The airfoil designed in the application can realize that the lift force keeps linear variation with the angle of attack in the range of 16 degrees at the flight speed of 0.2-0.3 times the sound speed and the flight height of 1-4 kilometers, and can provide a large lift force for a cone cuplet, so as to provide effective support for suppressing the drift and swing of the cone during the oil adding / receiving process.

The embodiment of the invention provides an asset correlation determination method and a related device, and the method comprises the steps: determining a joint feature function between a variance factor and the logarithmic prices of two types of assets based on the variance factor and the logarithmic prices of the two types of assets through responding to the input of observation data; the variance factor obeys an overall equilibrium model of the interest rate; converting the joint feature function into a conditional density function about the observation data through Fourier inversion; updating a preset likelihood value based on the conditional density function to obtain an updated likelihood value; and determining the asset correlation based on the updated likelihood value. According to the embodiment of the invention, the dynamic relationship is recursive based on the data model, the method does not depend on any statistical approximation and linear hypothesis, is more in line with the actual market nonlinear change process, and improves the accuracy of asset correlation.

The invention discloses a DC power system fault detection method, system and device, and a storage medium, and relates to the technical field of fault detection. The method comprises the following steps: collecting node data in a direct-current power system, and generating a column vector according to a time sequence; performing standardization processing on the column vector to obtain a standard non-Hermitian matrix; calculating a standard matrix product and a corresponding characteristic value of the standard non-Hermitian matrix, and determining characteristic value distribution of the standard matrix product according to a single-ring theorem; the characteristic value distribution is limited by the inner ring radius and the outer ring radius; constructing a random variable based on the characteristic value distribution and the average spectral radius; when the random variable is smaller than the radius of the inner ring, judging that the system breaks down; and when the random variable is greater than or equal to the radius of the inner ring, determining that the system has no fault. According to the method, the defects of the traditional method in quickness, anti-interference performance and multi-end system adaptability can be overcome by quantifying the linear change trend of the fault correlation characteristics.

The invention relates to the technical field of oil exploration and development, and discloses a stratum collapse pressure earthquake prediction method and device based on rock mechanics nonlinear variation, and the method comprises the steps: carrying out the nonlinear fitting of compressive strength with rock cohesion and an internal friction angle through a triaxial mechanics experiment, and obtaining a corresponding fitting relation; then, through seismic inversion, a rock mechanical attribute body, an overlying stratum pressure body, a pore pressure body, a maximum horizontal principal stress body, a minimum horizontal principal stress body, and three-dimensional space rock cohesion distribution and internal friction angle distribution are obtained through calculation; and finally, according to the pore pressure body, the maximum horizontal principal stress body, the minimum horizontal principal stress body, the three-dimensional space rock cohesion distribution and the three-dimensional space rock internal friction angle distribution, predicting stratum three-dimensional collapse pressure body data in combination with a collapse pressure calculation model. By means of the method, the formation collapse pressure can be predicted more accurately, geological disasters caused by improper drilling operation can be reduced, and therefore the geological environment is protected, and potential influences on the earth surface and underground water resources are reduced.

This invention relates to the fields of wireless sensing and deep learning, specifically to a human pose estimation method based on a multimodal attention network. In video modality learning, a multi-resolution network (FCN) is designed to extract pose features. By applying a local self-attention network to the generated heatmap, the weights and attention levels of different regions are adaptively adjusted to capture more refined keypoint information. In CSI modality learning, a spatiotemporal attention network and a multimodal-guided linear spatial position variation layer are designed to facilitate self-learning of its spatiotemporal features. A teacher-student architecture is adopted in the overall multimodal learning network to allow the Wi-Fi signal-based pose estimation model to learn more from the deep learning capabilities of the video modality. By transferring the correct human pose estimation information from the video learning network to the Wi-Fi signal learning network, accurate human pose information can be estimated using only the CSI modality as input.

Systems and methods for measuring web tension distribution in roll-to-roll processes, for example, such as R2R processes employed in the fabrication of printed devices. Such systems and methods entail a web that travels between first and second rollers in a longitudinal direction of the web, inducing tension in the web in the longitudinal direction thereof such that tension is present in a flexible substrate of the web between the first and second rollers, and operating the system to determine an average tension and linear variation of tension present in the flexible substrate resulting from the tension induced in the web inducing a nonuniform tension distribution in the flexible substrate between the first and second rollers. The systems and methods utilize one or more devices that induce deflection in the web between the first and second rollers.

The invention provides a video coding and decoding method, device, equipment and system and a storage medium, and the method comprises the steps: determining a prediction mode of a current block when the current block is coded and decoded; if the prediction mode of the current block is a multi-model intra-frame block copy illumination compensation mode, determining a reference block of the current block, and determining N groups of linear model parameters, N being a positive integer greater than 1; the target group linear model parameters are selected from the N groups of linear model parameters, and the target group linear model parameters are used for performing linear change on the reference block to obtain the prediction block of the current block, so that the illumination compensation effect is improved, and the prediction accuracy and the coding and decoding performance are further improved.

This invention discloses an LSTM-based thermal error prediction method for lead screws based on an attention mechanism, relating to the field of lead screw technology. The method includes: collecting multi-source data such as temperature field, thermal error, and time-series operating conditions during lead screw operation; preprocessing the collected multi-source data and filtering key thermal features; constructing an improved LSTM basic prediction model embedded with a dual attention mechanism; fusing the improved LSTM model with a lightweight tree model at the decision layer; inputting the preprocessed and filtered key thermal features into the fusion model; and outputting the lead screw thermal error prediction result. This invention achieves high-precision time-series prediction of lead screw thermal errors, solving the problems of insufficient mining of the correlation between thermal features and thermal errors, incomplete capture of time-series features, resulting in prediction results being greatly affected by irrelevant factors, insufficient accuracy, and difficulty in adapting to the nonlinear variation law of lead screw thermal errors, thus failing to meet the operating requirements of high-precision equipment.

The invention relates to a time synchronization control method for geological exploration acquisition signals, and belongs to the technical field of second pulse signal time synchronization control between an exploration station and a satellite by using stably changing phase difference information between signals. According to the technical scheme, a stable and periodically-changed phase difference is formed between a frequency signal of a crystal oscillation source of an exploration station and an adjusting signal of the crystal oscillation source, integral operation is conducted on the phase difference, linear increasing voltage is formed, non-full-cycle time correction is conducted on pulse per second signals between the exploration station and a satellite by means of the stable and linear change characteristic of the voltage, and the time correction precision is improved. According to the method, the measurement resolution is high, the circuit structure is simple, the method is easy to implement, the counting error of + / -1 pulse existing in a common high-frequency counting method is greatly eliminated, the measurement accuracy is high, and the measurement accuracy is high. And high-precision time synchronization of the second pulse signal of the exploration station and the second pulse signal of the satellite is realized.

The invention discloses a photoelectric stabilized platform discrete time control method and equipment based on a disturbance observer. The method belongs to the technical field of photoelectric stabilized platform servo control. The method comprises the following steps: establishing an airborne photoelectric stabilized platform current closed loop, a motor and a shafting load model, and introducing a disturbance observer to convert the model into a standard series second-order linear control system; the error amount of the expected speed and the feedback speed and the integral of the error amount are input, and a state variable is determined; based on a generalized standard series discrete second-order linear control system, determining a boundary curve of a phase plane and a controlled quantity linear change region, determining a phase trajectory returning to a zero state for different initial phase points, and obtaining corresponding controlled quantities according to different regions in a motion process; inputting the obtained control quantity and the feedback speed into a disturbance observer, and calculating to obtain a current control quantity; and inputting the current control quantity into a current closed loop, calculating to obtain driving voltage, and inputting the driving voltage into a driving module to drive a motor and a shaft system to rotate.

The present application relates to a stranded air-blown micro cable, comprising an outer sheath (3), wherein the outer sheath (3) is internally provided with a plurality of optical units (2) stranded around a strength member (1) as the center; spirally distributed grooves (4) of designed dimensions are provided in a surface of the outer sheath (3); a plurality of grooves (4) are provided, and are evenly distributed on the circumference of the cross-section of the stranded air-blown micro cable; and one portion of the grooves (4) and another portion of the grooves (4) are symmetrically arranged using the diameter of the strength member (1) as the axis of symmetry. The stranded air-blown micro cable has a linearity factor β ≤0.02, and a dynamic friction coefficient less than 0.3. The overall unit density ρ of an optical cable is in the range of 0.0005-0.0015 g / mm3. By means of the above structure, the grooves (4) are symmetrically and uniformly arranged, such that an airflow generated by air blowing uniformly acts on the outside of the outer sheath (3) so as to increase an air blowing distance; moreover, since the linearity factor and the dynamic friction coefficient are less than design values, the combined effect of these two factors reduces linear variations during an air blowing process and improves an air blowing force. In addition, the dimensions of the grooves (4) and the overall unit density of the optical cable are specifically limited, thereby ultimately increasing an air blowing distance in routes having many bends.

The application relates to an electric meter box fault prediction method and device, computer equipment, a storage medium and a computer program product. The method comprises the following steps: acquiring a plurality of groups of historical acquisition data of electric meter box terminals; determining an initial value of a linear change rate threshold parameter and an initial value of an elastic network hybrid parameter based on the plurality of groups of historical acquisition data and a target function; determining an initial fault prediction function according to the initial value of the linear change rate threshold parameter, the initial value of the elastic network hybrid parameter and the target function; acquiring current acquisition data of the electric meter box terminals; determining an initial prediction result according to the current acquisition data and the initial fault prediction function; adjusting the initial value of the elastic network hybrid parameter and the linear change rate threshold parameter based on the initial prediction result to obtain an adjusted fault prediction function; and determining a fault prediction result of the electric meter box based on the adjusted fault prediction function. The method can improve the accuracy of the electric meter box fault prediction result.

This invention discloses a method for predicting and attributing the spatiotemporal variation of groundwater storage based on a Transformer-LSTM hybrid network. The method includes collecting and analyzing groundwater storage variation data and various potential factors influencing these variations in a target area; constructing a Transformer-LSTM hybrid deep learning model using high-dimensional driving factors as input and the spatiotemporal variation of groundwater storage as output; training and validating the Transformer-LSTM model; deeply interpreting the temporal evolution and spatial distribution characteristics of groundwater storage based on the prediction results of the Transformer-LSTM model; and quantifying the influence of potential driving factors on groundwater storage variation using the interpretable SHAP method to identify key driving elements. This method can effectively capture the strong spatiotemporal heterogeneity and complex nonlinear variation characteristics of groundwater storage at a large regional scale.

The invention discloses a method and a system for measuring a rolling angle of a magnetic suspension model based on a trapezoidal marker line. The system comprises a pose measurement sensor and a trapezoidal marker line. The trapezoidal marker line is a closed trapezoidal ring which is adhered to the mass center of the model in a surrounding manner, the width value, parallel to the axis of the model, of the trapezoidal marker line is monotonically and linearly changed, and the trapezoidal marker line is a trapezoid with unequal bottom edges in a geometric unfolding state. During measurement, the linear array CCD scans the surface of the suspension model and detects pixel serial numbers corresponding to two edges imaged by the trapezoidal marker line; according to the difference value of the pixel serial numbers corresponding to the two edges, the roll angle of the suspension model can be solved in combination with a preset geometrical relationship; the axial position of the suspension model can be calculated through the mean value of the two edge pixel serial numbers. On the premise that a hardware system is not changed, integrated measurement of the rolling posture and the axial position is achieved only by changing the marker line configuration and the resolving method, and the method has the advantages of being good in compatibility, low in cost and high in measurement real-time performance.

One aspect of the invention relates to a method for compensating the rate as a function of temperature of a watch (1) in which, in an internal volume V occupied by n moles of a gas with constant R, a waterproof case (2) contains a movement (3) with an oscillator (4), in which the pressure coefficient Cp and the humidity coefficient Ch of the movement (3) are determined, the optimal value Cto of the thermal coefficient Ct of the oscillator (4) is calculated so as to define its rate as a function of the relative linear variation of the temperature T in order to compensate for pressure and humidity deviations, and, for after-sales, the watch (1) is equipped with a compensation device (10) to change the pressure P and / or the constant R and / or the amount of gas and / or the temperature T in the case (2), or, in the factory, by modifying the thermal coefficient of the elastic return means of the oscillator (4) by modifying the oxide layer and / or applying or ablating a coating, and / or modifying the amount and / or the nature of the gas in the watch, and / or modifying the internal volume of the case (2).

Provided in the present application are a video encoding method and apparatus, a video decoding method and apparatus, and a device, a system and a storage medium. The methods both comprise: when the current block is encoded / decoded, determining a prediction mode of the current block; if the prediction mode of the current block is a multi-model intra-frame block copy illumination compensation mode, determining a reference block of the current block, and determining N groups of linear model parameters, N being a positive integer greater than 1; and selecting a target group of linear model parameters from among the N groups of linear model parameters, and performing a linear change on the reference block by using the target group of linear model parameters, so as to obtain a prediction block of the current block. Therefore, an illumination compensation effect is improved, thereby improving the prediction accuracy and an encoding / decoding performance.

Systems and methods for measuring web tension distribution in roll-to-roll processes, for example, such as R2R processes employed in the fabrication of printed devices. Such systems and methods entail a web that travels between first and second rollers in a longitudinal direction of the web, inducing tension in the web in the longitudinal direction thereof such that tension is present in a flexible substrate of the web between the first and second rollers, and operating the system to determine an average tension and linear variation of tension present in the flexible substrate resulting from the tension induced in the web inducing a nonuniform tension distribution in the flexible substrate between the first and second rollers. The systems and methods utilize one or more devices that induce deflection in the web between the first and second rollers.

This invention discloses a method for testing the oil content of an automotive air conditioning compressor, relating to the field of air conditioning compressor testing technology. The method includes acquiring the original oil content monitoring data sequence of the compressor collected by an intelligent sensor within a test cycle; correcting the standard Kalman filter prediction step by incorporating the nonlinear characteristics of oil content state transitions; and dynamically denoising the original data sequence using an improved Kalman filter algorithm to obtain a regular and reliable denoised oil content data sequence. The denoised data sequence is then time-series segmented to separate the oil content quantum sequences for the startup phase and steady-state operation phase within the test cycle. The average oil content value for each phase is calculated, thereby generating the final oil content test result for the compressor. This method adapts to the nonlinear variation characteristics of oil content, weakens interference components in the monitoring data, and achieves segmented quantification of oil content states according to operating conditions, aligning with the testing and evaluation needs of actual automotive air conditioning compressor operating conditions.

The invention discloses a ship small target detection method and device, equipment and a medium, and relates to the technical field of remote sensing target recognizing.A channel attention mechanism is embedded into a feature transformation process to capture a nonlinear dependency relationship among channels and generate attention weights of the channels, features are extracted based on the attention weights of the channels, and the target recognition accuracy is improved. According to the mechanism, the nonlinear expression capability is weakened through linear features, so that the extraction capability of morphological features of small targets is improved; the method comprises the following steps of: firstly, performing linear change on features of each channel in a feature map to generate a channel weight of each channel, extracting spatial features channel by channel and fusing the spatial features, and realizing direct weighted modulation of the feature channels through linear operation by the mechanism, thereby avoiding complex nonlinear transformation and improving the recognition precision of the position of a small target; therefore, accurate feature recognition is carried out on the form and the position of the small target, and finally, the detection and recognition precision of the small target can be improved.

The invention discloses a spread spectrum clock signal simulation method, device and equipment and a storage medium, and relates to the field of electronic computers.The spread spectrum clock signal simulation method comprises the steps that modulation parameters input in a command line mode are received on the basis of a parameter transfer function, and values can be assigned to variables in target program codes on the basis of the modulation parameters; and realizing the following logics: determining a linear variation and an incremental step length of a spread spectrum clock period according to the modulation parameter, and simulating to obtain a spread spectrum clock signal according to the linear variation and the incremental step length. Thus, simulation of the spread spectrum clock signal is realized in a software simulation mode, and hardware circuits such as a frequency divider and a voltage-controlled oscillator for simulating the spread spectrum clock signal do not need to be integrated in a front-end verification platform, so that the complexity of the front-end verification platform can be greatly reduced.

This invention discloses an epitaxial method for InGaAs layered linear gradient variation buffer layers based on molecular beam epitaxy, comprising the following steps: S1: Determine the total variation range of the specific In composition and the total thickness of the linear variation of In, and calculate the retreat composition of the retreat layer based on the variation rate of the In composition; S2: Design the InGaAs linear gradient buffer layer in layers and determine the thickness of the entire layer; S3: Determine the value of the In composition after actual epitaxy, and adjust the In composition and thickness in step S2 according to the value after actual epitaxy; S4: Determine the growth rate of each gradient layer of InGaAs; S5: Epitaxially construct the InGaAs layered linear gradient variation buffer layer based on the layer thickness and the growth rate of each gradient layer of InGaAs. This invention realizes the InGaAs layered linear gradient variation buffer layer. x Ga 1‑x As linear gradient buffer layer, the linear change of In composition in As reduces the influence of source furnace temperature changes on the uncertainty of In and Ga growth rate changes.

The invention discloses a tree body-double rope vibration frequency calculation method, and belongs to the technical field of branch mechanical property analysis. The method comprises the following steps: constructing an ideal variable-cross-section circular-truncated-cone-shaped branch model, regarding a branch as a cantilever beam, and defining the linear change of the radius of the cross section of the branch along the length direction; establishing a variable cross-section deflection curve differential equation based on the ideal model; carrying out integral solution on the variable cross-section deflection curve differential equation, and deducing a bending elastic modulus formula of the variable cross-section circular truncated cone-shaped branch in combination with boundary constraint conditions; establishing a tree body-double rope model; and deducing a tree body-double rope vibration frequency formula based on the tree body-double rope model and a bending elastic modulus formula. The method provides a theoretical basis for design of vibration type fruit picking equipment, vibration frequency parameters can be optimized, tree body damage is reduced, and picking efficiency is improved.

The application discloses a CORS ionosphere modeling method and equipment based on linear variation characteristics, first, a plurality of GNSS stations in linear distribution are selected, observation data is preprocessed and STEC is calculated, and then STEC is projected into VTEC through a single layer model; then, a VTEC dataset along a target direction at the same time is constructed, IPP distance normalization is performed on the ionosphere piercing point, a VTEC reference value is calculated by calling an international reference ionosphere model as a fitting weight, a dynamic order determination strategy is adopted to complete polynomial fitting through a weighted least square method, and a regional ionosphere model is obtained. The VTEC corresponding to a virtual reference station is obtained through interpolation, ionosphere delay is obtained, and the ionosphere delay is further applied in virtual observation value calculation, and linear regional high-precision network RTK calculation is realized. The application fully utilizes the physical characteristics that the ionosphere has variation regularity in the longitude / latitude direction, relaxes the distance requirement between stations, significantly improves the accuracy of the regional ionosphere model, and further improves the observation accuracy of the linear regional network RTK.

The invention provides a large model training method and system based on differential low-rank adaptation, and the method comprises the steps: connecting a low-rank adaptation module in a linear change layer of a pre-trained large model in parallel, freezing the original parameters of the pre-trained large model, setting the parameters in the low-rank adaptation module as trainable parameters, and carrying out the training of the pre-trained large model. Performing forward propagation on the input small-batch samples to obtain a loss function, calculating a gradient vector of each sample in the small-batch samples about the low-rank adaptation module based on the loss function, and further performing a differential privacy protection mechanism based on the gradient vector to update parameters in the low-rank adaptation module to obtain an optimized low-rank adaptation module; and performing iterative optimization based on the optimized low-rank adaptation module until a preset training iteration condition is met to obtain a target large model. According to the method, the memory ability of the model to high-sensitivity information is reduced, the method has good attack resistance and privacy protection strength, the conflict between privacy protection and model performance is solved, and the calculation efficiency and the parameter efficiency are optimized.

The invention discloses a method for researching a northern bay ocean heat wave interannual change rule. The method comprises the following steps: acquiring OISST data, ERA5 data, ORAS5 data and a Nio 3.4 index; spatial distribution characteristics and linear variation trend characteristics of the north bay ocean heat waves are explored; analyzing the relation between the northern bay ocean heat wave intensity and the ENSO process through lead-lag correlation; spatial evolution of northern bay heat wave intensity and ocean surface temperature abnormity in the ENSO event process is researched; analyzing the energy contribution of the ocean and the atmosphere to the ocean heat wave event in the ENSO event process; and determining the influence of latent heat flux, long-wave radiation and short-wave radiation modulated by the ENSO process on heat waves in the north bay. The method solves the problems of lack of analysis of long-term change rules of the northern bay ocean heat waves and incomplete understanding of a driving mechanism in existing research, analyzes spatial distribution characteristics and interannual change rules of the northern bay ocean heat waves, and reveals a forming power mechanism of the northern bay ocean heat waves.