40 results about "Coefficient optimization" patented technology

The invention belongs to the technical field of blade ring structure optimization design of composite materials, and particularly relates to a structure optimization design method of a titanium-based composite material integral blade ring, which comprises the following steps of: analyzing and calculating an equivalent elastic constant of a titanium-based composite material on a micro-scale, and performing secondary development on Abaqus software by using a Python language on a macro-scale to obtain an equivalent elastic constant of the titanium-based composite material integral blade ring; and then compiling a static analysis plug-in of the integral blade ring structure, and finally realizing optimal design of the blade ring structure by utilizing an automatic plug-in. The method has the advantages of being simple, easy to operate, high in efficiency and the like, the static strength reserve coefficient-optimization parameter curve is formed by carrying out automatic finite element simulation of various integral blade ring structural forms, and the further obtained optimization conclusion provides effective guidance and basis for the structural forms of the integral blade ring.

The invention discloses a parameter self-adjusting robust active noise control method, and relates to the technical field of noise control. The method comprises the following steps: acquiring a sound pressure signal in real time, processing the sound pressure signal to form a discrete noise signal as a reference signal, and constructing a dynamically updated reference signal sequence; constructing a filter and generating a weight coefficient thereof, and filtering the reference signal to generate an output signal; processing an output signal of the filter through a secondary path to generate a de-noising signal, collecting a residual signal, feeding back the residual signal to the filter, providing error feedback for self-adaptive iteration of the filter, and driving a weight coefficient of the filter to be optimized and updated; determining an optimal parameter combination for the generalized robust adaptive loss function to realize adaptive optimal determination; and updating a filter weight coefficient vector. According to the method, through organic combination of a self-adaptive parameter adjustment mechanism and a robust control strategy, the fast convergence characteristic, the low steady-state error and the high impact resistance are achieved, and the noise suppression efficiency in the complex environment is remarkably improved.

The invention discloses a bulk cargo intelligent logistics scheduling method and system based on artificial intelligence. The method comprises the steps that firstly, bulk cargo transportation tasks are acquired, and to-be-scheduled task features are extracted through a task feature extraction network; obtaining transport capacity attribute data of to-be-selected transport capacity, and generating to-be-selected transport capacity features through the transport capacity feature extraction network; screening the optimal transport capacity of which the scheduling priority coefficient is higher than a reference value, and extracting the characteristics of the optimal transport capacity; optimizing the optimal transport capacity feature based on the adaptation coefficient of the task feature and the optimal transport capacity feature; and determining a final scheduling transport capacity and distributing tasks through matching of the optimization features and the to-be-selected transport capacity features. According to the design, multi-dimensional feature extraction of tasks and transport capacity is realized by utilizing an artificial intelligence network, the adaptation precision is improved in combination with a dynamic optimization mechanism, the problems of low matching efficiency and poor real-time performance of traditional scheduling are solved, and the intelligent scheduling level of bulk cargo logistics is improved.

The invention relates to the technical field of intelligent water affairs, and provides a water supply pipe network leakage monitoring method comprising the following steps: collecting pipe network multi-source data and pre-processing; constructing a pipe network directed graph topological structure, inputting data into a space-time graph convolutional neural network to extract space-time correlation characteristics, and predicting a dynamic reference pressure value and a flow value of each node in combination with environment temperature and historical water consumption; calculating a residual error between an actual monitoring value and a reference value, dynamically determining an adaptive threshold value based on historical residual error statistical characteristics, and judging that leakage occurs when the residual error continuously exceeds a limit; finally, the pipe network hydraulic micro model is called for reverse evolution, the virtual leakage amount of each pipe section is calculated through a leakage coefficient optimization algorithm, and the pipe section with the maximum value is determined as the leakage point. According to the method, deep learning and a hydraulic mechanism are fused, the anti-interference capacity is high, the false alarm rate can be effectively reduced, and early warning and accurate positioning of tiny leakage are achieved.

A combustion mechanism simplification and optimization method based on a two-stage deep neural network algorithm belongs to the field of combustion dynamics simulation and comprises three stages of component simplification based on a first-stage deep neural network, deep reaction simplification and coefficient optimization based on a second-stage deep neural network. Firstly, a deep neural network model with component retention states and initial conditions as input and combustion characteristics as output is constructed, and efficient screening and gradual simplification of mechanism components are achieved. Secondly, deeply simplifying a reaction path, setting a time scale threshold value, eliminating non-key rapid reactions, and constructing a reaction number controlled simplification mechanism; and finally, constructing a deep neural network model taking the key reaction kinetic parameters as input, and performing global optimization of mechanism parameters in combination with a genetic algorithm to improve the prediction precision of the mechanism on key combustion performance indexes. According to the method, a combustion mechanism with high simplification and high fidelity can be realized, and the method is suitable for rapid calculation in complex three-dimensional combustion simulation.

The invention relates to the technical field of grinding processes, and provides a planetary reducer gear shaft machining method capable of reducing machining errors, which comprises the following steps: setting an initial value of a PI coefficient, adjusting grinding parameters of grinding treatment by using a PID control algorithm, and collecting surface roughness before and after the grinding treatment and a vibration value of the grinding treatment; determining the machining effect stability degree of the grinding treatment and the predicted value of the surface roughness degree of the next grinding treatment; a PI coefficient reference value, vibration-grinding precision correlation and PI coefficient parameters of grinding treatment are determined; and according to the PI coefficient reference value of grinding treatment, the PI coefficient value, the vibration-grinding precision correlation and the PI coefficient parameter, a PI coefficient optimization value is determined, and planetary reducer gear shaft machining is achieved according to the PI coefficient optimization value. Cutting machining errors can be reduced, and the precision of the planetary reducer gear shaft is improved.

The invention discloses a meta-heuristic adaptive robust azimuth gyro drift calibration method, and provides a technical scheme for improving the azimuth gyro drift calibration precision on a software level, and the method comprises the following steps: establishing a Kalman filtering model of an augmented measurement equation based on an angular rate; in combination with a variational Bayes-based measurement noise variance matrix adaptive estimation method and an improved sine and cosine heuristic multi-correction coefficient optimization algorithm, drift calibration of multi-channel measurement error decoupling and multi-channel error correction is realized, so that the drift calibration of multi-channel measurement error decoupling and multi-channel error correction can be realized while hardware cost and system complexity are not increased. Therefore, the method is very suitable for large-scale application and popularization.

This invention discloses a ship cloud computing load balancing method based on a hybrid intelligent algorithm, relating to the field of ship intelligent computing and cloud computing integration technology. The method includes the following steps: task classification modeling, edge-cloud resource status monitoring, multi-objective optimization decision-making, and dynamic load adjustment. This invention reduces the load standard deviation of ship edge-cloud nodes through a multi-objective optimization algorithm and a dynamic adjustment mechanism. It optimizes task processing efficiency, reducing the average latency of real-time tasks and shortening the total processing time of non-real-time tasks. It reduces system energy consumption by optimizing the energy consumption coefficient, thereby reducing the total energy consumption of edge nodes and cloud servers. It adapts to ship scenarios, improving the robustness and adaptability of scheduling strategies to address the heterogeneity of ship tasks and network dynamics.

The present invention belongs to the field of laboratory environment control technology, and in particular relates to an artificial intelligence-based intelligent monitoring and regulation method for laboratory temperature and humidity. The method first establishes laboratory temperature and humidity PID control and designs temperature and humidity PID controllers. Next, an objective function is constructed, integrating indicators such as overshoot and rise time and performing normalization processing, and optimizing temperature and humidity regulation performance by adjusting weight coefficients. Then, an improved starfish optimization algorithm (SFOA) is used to optimize the proportional, integral, and differential coefficients of the PID controller through parameter setting and population initialization. Finally, a simulation is performed to obtain a response curve. Compared with traditional PID control technology, the present invention can more efficiently and accurately regulate temperature and humidity, effectively improving the stability of the experimental environment and the reliability of experimental results.

The invention relates to the technical field of eddy current sensors and voltage calibration. The eddy current sensor calibration method for the small-diameter metal ball comprises four parts of coordinate system establishment, calibration data acquisition, calibration sample set construction and calibration equation coefficient optimization. Through innovative coordinate system definition and position parameterization and in combination with an intelligent optimization algorithm, a voltage-position relation model capable of accurately representing the actual effective measurement axis of the small-diameter metal ball center deviation sensor is successfully established; and a nonlinear measurement error caused by sphere center eccentricity, high surface curvature and axis deviation of the sensor is effectively compensated.

According to the traffic equipment data processing method and device provided by the embodiment of the invention, a multi-dimensional data preprocessing mechanism is innovatively constructed, and standardized representation of characteristics such as equipment types and geographic positions is realized through abnormal value identification and one-hot coding. And designing a classification model based on a multi-label random forest, and establishing a feature division and decision tree construction strategy in combination with an information gain criterion and Gini coefficient optimization. And a voting integration mechanism is introduced, and accurate recommendation of illegal configuration is realized through fusion of multi-tree classification results. According to the method, the defects of the traditional technology in the aspects of data processing, classification modeling, result generation and the like are effectively overcome, and the accuracy and reliability of traffic equipment data processing are remarkably improved.

The application discloses a kind of based on depth reinforcement learning's double intelligent reflector dynamic reflection coefficient adjustment method.Current complexity and dynamic change of environment are important factors to influence intelligent reflector phase control, in dynamic and complex environment, long-term phase control is difficult thing.The method of the present application is based on depth reinforcement learning (DRL), constantly adjust through the feedback of environmental information, reach an optimal strategy, to carry out double intelligent reflector unit phase and amplitude adjustment.The method of the present application can provide a flexible, efficient, automatic method for double intelligent reflector wireless communication system to realize reflection coefficient optimization control, especially when facing changing environment and complex system dynamics.In this way, the energy efficiency of wireless network can be improved, the coverage range is expanded and the user experience is improved.In addition, the convergence rate is improved by distributing training for the method of the present application.Simulation results show that the method of the present application is effective and feasible, and the method of the present application has a faster convergence speed.

This invention discloses a dynamic automatic correction system for the nozzle area of ​​a steam turbine regulating stage, comprising: a detection module for real-time acquisition of operating parameters of the steam turbine regulating stage, including the pressure difference across the regulating valve, main steam flow rate, reheat steam temperature, unit power, and speed; and a control module, which is communicatively connected to the detection module and incorporates a load rate calculation model, a dynamic correction algorithm, and secondary correction logic. This invention ensures that the regulating valve opening remains within the high-efficiency range by dynamically correcting the nozzle area, combined with secondary correction logic. This verifies that throttling losses are reduced under low load conditions, directly reducing energy consumption and improving the unit's operational economy. Simultaneously, through high-frequency data acquisition, dynamic coefficient optimization, and flexible correction cycles, it achieves a "second-level" correction response. The specification clearly states that it can quickly meet the peak-shaving needs of the power grid and solve the problem of nozzle area decoupling from operating conditions under load fluctuations.

The application relates to the grinding process technical field and discloses a planetary reducer gear shaft machining method capable of reducing machining errors, which comprises the following steps: setting an initial value of a PI coefficient, adjusting grinding parameters of grinding treatment by using a PID control algorithm, collecting surface roughness before and after grinding treatment and vibration values of the grinding treatment; determining a processing effect stability degree of the grinding treatment and a predicted value of surface roughness of the next grinding treatment; determining a PI coefficient reference value of the grinding treatment, vibration-grinding precision correlation and a PI coefficient parameter; determining a PI coefficient optimization value according to the PI coefficient reference value of the grinding treatment, the value of the PI coefficient, the vibration-grinding precision correlation and the PI coefficient parameter; and realizing the planetary reducer gear shaft machining according to the PI coefficient optimization value. The application can reduce cutting machining errors and improve the accuracy of the planetary reducer gear shaft.

The invention discloses a thermal power generating unit primary frequency modulation control method based on economical efficiency and stability optimization, which belongs to the technical field of power system control, and is technically characterized by comprising the following steps: step 1, detecting power grid frequency fluctuation, and when the power grid frequency fluctuation exceeds a set normal range, executing step 2; controlling each device to perform primary frequency modulation according to the magnitude of power grid frequency fluctuation and the SOC state of the energy storage device; 2, carrying out dynamic droop coefficient optimization on the energy storage device, and carrying out system economy calculation and stability calculation to obtain an optimal value of a target function; and step 3, setting a dynamic weight adjustment mechanism which is used for dynamically adjusting the weights of economy and stability according to the magnitude of the power grid frequency fluctuation, setting constraint conditions and determining an optimal frequency modulation strategy by using a particle swarm optimization algorithm, and has the advantages of being capable of adjusting and calculating the weights and response strategies in real time according to the power grid frequency fluctuation and improving the frequency modulation accuracy. And the economical efficiency and the stability of the system are effectively improved.

The application discloses a three-dimensional anisotropic medium space-time domain seismic forward method, device, medium and product, relates to the technical field of forward simulation of geophysical exploration, and comprises the following steps: based on a preset improved time high-order difference operator and a preset improved space high-order difference operator, respectively discretizing time derivatives and space derivatives in a three-dimensional VTI medium acoustic wave equation; based on the three-dimensional VTI medium acoustic wave equation, simplifying time-discrete functions and space-discrete functions to obtain a simplified space-time domain dispersion relation; based on a target optimization algorithm, solving the simplified space-time domain dispersion relation to obtain optimized time item high-order difference coefficients and optimized space item high-order difference coefficients; and based on the optimized time item high-order difference coefficients, the optimized space item high-order difference coefficients, the time-discrete functions and the space-discrete functions, extrapolating and solving the three-dimensional VTI medium acoustic wave equation. The application can effectively improve the numerical precision in anisotropic wave field simulation.

The application provides a method for generating an in-situ stress field inversion of a grid, based on the superiority of GAN data enhancement, deep learning of the distribution of lateral stress coefficients, GAN analysis of lateral stress coefficients, fine tuning of ancient lateral stress coefficients by the introduced generative adversarial network, optimization of the present in-situ stress field, and meeting the needs of engineering analysis. The advantages of uniform design can greatly reduce the number of tests and keep the training samples uniformly dispersed. After each uniform design test, GAN training can be performed, the data of the test points are input into the GAN, and the optimized lateral stress coefficients can be obtained, the optimized ancient in-situ stress field can be obtained by calculation, and the present in-situ stress field can be obtained after nonlinear elastic-plastic excavation simulation.

The invention provides a machine pump bearing weak fault diagnosis method and system based on time-frequency self-adaptive segmentation. The method comprises the steps that collected vibration signals are converted and processed to obtain corresponding time-frequency distribution characteristics; quantitatively representing the adjacent correlation distribution relation of the time-frequency distribution characteristics, and realizing frequency domain interval division to obtain multiple parts of time-frequency distribution characteristics; determining fault characteristic frequency ratio indexes corresponding to time-frequency distribution characteristics of different parts by using a principal component analysis technology, and selecting sensitive components; and performing coefficient optimization processing by adopting a differential search algorithm, calculating to obtain a fault feature component, and extracting and diagnosing weak fault features of the pump bearing based on the fault feature component. By adopting the scheme, the problems of single theory, complex calculation and insufficient accuracy in the prior art can be solved, the weak fault diagnosis of equipment is realized by comprehensively utilizing the time-frequency adaptive segmentation technology and the frequency band coefficient optimization algorithm, the timeliness is good, the operation result is finer, and the recognition and diagnosis of the weak fault of the bearing are timely and reliably realized.

The invention discloses a dual-active bridge topology circuit and a constraint coefficient optimization modulation method thereof. The circuit comprises a direct current side circuit, an alternating current side circuit, a series inductor and a transformer, the direct current side circuit adopts a full-bridge structure and is electrically connected with the primary side of the transformer; the series inductor comprises a transformer leakage inductor and an external inductor and is electrically connected with the secondary side of the transformer; and the alternating current side circuit adopts a matrix half-bridge structure and is electrically connected with the secondary side of the transformer. According to the method, the correlation model of the constraint coefficient and the total loss of the system is systematically established, so that efficiency optimization in a full-working-condition and full-phase-angle range is realized.

The invention belongs to the technical field of electric vehicle chassis control, and particularly relates to a distributed electric drive vehicle multi-target torque distribution control strategy based on DBO. The strategy adopts a hierarchical control framework: in an upper motion control layer, total driving torque is calculated through a PID (Proportion Integration Differentiation) controller, and additional yawing moment required for maintaining the stability of the vehicle is calculated based on a sliding mode control algorithm; in the lower torque distribution layer, a multi-objective optimization function including the tire load rate, the torque deviation and the motor energy consumption is constructed, four-wheel torque distribution is decoupled into a two-side front and rear axle distribution coefficient optimization problem, and a dung beetle optimization algorithm is introduced to perform optimization solution under the condition that road adhesion and motor characteristic constraints are met; and in the execution layer, the calculated final torque instruction is sent to the hub motor. The method can solve the problem that a traditional distribution strategy is difficult to give consideration to energy consumption and stability, and remarkably reduces the motor energy consumption and improves the driving smoothness while guaranteeing the vehicle driving safety.

The invention provides a four-dimensional ultrasonic signal real-time enhancement method based on anisotropic multiphase decomposition and LTI constraint, and belongs to the field of medical image processing. The method comprises the following steps: constructing a space-time Laplacian pyramid and anisotropic multiphase decomposition, constructing a space-time multiphase filtering network based on dimension decoupling, performing LTI constraint solution based on transposer error minimization, performing cross-scale multiphase feature alignment based on Kronecker product, performing multiphase synthesis inverse rearrangement, and performing image reconstruction. The behavior of the system is described by using the pulse response in the signal processing theory, the shift characteristic introduced by multiphase decomposition is eliminated, the LTI strong constraint is introduced in the filter coefficient optimization stage, and the constraint forces convolution kernel parameters in the space-time multiphase filter network to form a specific complementary structure, so that the performance of the system is improved. According to the invention, different multi-phase components can counteract aliasing terms during synthesis, so that checkerboard artifacts are eliminated macroscopically, low-frequency structure direction information is ensured to accurately cover each high-frequency sub-pixel phase, and no phase deviation exists.

The invention provides a sub-codebook composite codebook generation method and system based on regional similarity clustering, and the method comprises the steps: obtaining a matrix training set based on a channel covariance matrix of a large-scale MIMO system; extracting an upper triangular element of each channel covariance matrix, separating a real part and an imaginary part, and constructing a high-dimensional covariance feature training set; pCA dimension reduction processing is carried out, and a K-Medoids clustering algorithm is combined with contour coefficient optimization and an optimal number of covariance mode clusters are identified; based on the sample density weight, distributing the total code word number to each cluster in the covariance mode cluster according to a proportion, and taking the total code word number as the exclusive code word number of each cluster; obtaining an exclusive sub-codebook corresponding to each cluster; and integrating the exclusive sub-codebooks of all clusters, and constructing a final composite codebook. By extracting the triangular elements on the covariance matrix and separating the real and imaginary parts, the channel covariance mode features are accurately captured, and the adaptive capacity of the codebook to complex propagation scenes with multipath reflection, shielding and the like is effectively improved.

The application relates to the technical field of eddy current sensors and voltage calibration. The eddy current sensor calibration method for small-diameter metal balls comprises four parts of coordinate system establishment, calibration data acquisition, calibration sample set construction and calibration equation coefficient optimization. Through innovative coordinate system definition and position parameterization, combined with an intelligent optimization algorithm, the application successfully establishes an accurate voltage-position relationship model capable of accurately representing the small-diameter metal ball center deviating from the actual effective measurement axis of the sensor, and effectively compensates for the nonlinear measurement error caused by the ball center eccentricity, surface high curvature and sensor axis deviation.

The invention belongs to the technical field of wind power system control, and particularly relates to a reactive current adaptive control method and system of a wind power system, and the method comprises the steps: collecting the multi-dimensional operation data of the wind power system in real time, constructing a multi-dimensional state evaluation model based on an entropy weight method, and dividing the three operation levels of a stable state, a transition state and a disturbance state; aiming at different state levels, constructing a reactive current control coefficient optimization model by adopting a deep Q network (DQN), and outputting an optimal control coefficient by taking minimization of voltage deviation, converter loss and harmonic content as targets; combining the unit capacity ratio and the state adaptation coefficient to construct a collaborative distribution model, and calculating a reactive current reference value of each unit; and dynamic adjustment is realized through closed-loop control. The method solves the problems of single state evaluation, fixed control coefficient, lack of unit collaboration and neglect of equipment loss in the prior art, improves the voltage stability, operation economy and electric energy quality of the wind power system, and is suitable for a high-proportion wind power grid-connected scene.

This invention provides a method and apparatus for optimizing the torque coefficient of a permanent magnet torque motor, relating to the field of permanent magnet motor design optimization technology. The method includes determining the range of values ​​for the parameters to be optimized in the permanent magnet torque motor; determining an optimization objective function based on the torque coefficient equation and torque ripple equation; minimizing the optimization objective function as the objective, using the range of values ​​as constraints, and obtaining the parameters to be optimized by training a surrogate model and performing multi-objective optimization. The apparatus executes the above method. The method and apparatus for optimizing the torque coefficient of a permanent magnet torque motor provided by this invention can accelerate the optimization convergence speed while taking into account torque ripple changes.

The invention relates to a dual-wavelength digital holographic imaging method based on unwrapping coefficient optimization, and the method comprises the steps: retrieving an unwrapping coefficient of a wrapped phase obtained through the measurement of a first wavelength, and obtaining the wrapped phase of a second wavelength through the phase guidance of the unwrapped first wavelength; and with the minimum value of the square of the difference between the induced phase of the second wavelength and the measured wrapped phase of the second wavelength as a target, obtaining an unwrapping coefficient of the first wavelength meeting the condition through a least square method, and correctly reconstructing an unwrapped phase with an expanded measurement range. According to the method, the searching process of the dual-wavelength phase unwrapping coefficient is optimized, the optimal unwrapping coefficient is quickly determined, and high-precision and large-range phase reconstruction is realized. Meanwhile, the signal-to-noise ratio and the calculation efficiency are remarkably improved, and the method is suitable for efficient measurement of the microstructure three-dimensional morphology.

The invention discloses a laser interferometer high-precision filter coefficient generation method based on an FPGA, and belongs to the technical field of digital signal processing. The method comprises the following steps: designing an ideal floating point FIR filter coefficient according to a measurement requirement and carrying out initial quantization; constructing a cost function containing passband ripples and stopband attenuation weights; under the constraint of keeping linear phase symmetry, performing bit-by-bit perturbation and iterative search on integer coefficients, and searching an optimal integer coefficient set which minimizes a cost function; and generating a fixed-point coefficient file adaptive to the FPGA hardware resources. The invention provides a coefficient optimization strategy in a discrete space aiming at the problem that the performance of a filter is reduced due to the limited word length effect of an FPGA (Field Programmable Gate Array) in a laser interference measurement system. On the premise that FPGA hardware resource consumption is not increased, quantization noise is remarkably reduced, the stopband suppression degree and the signal-to-noise ratio of the filter are improved, and therefore the final measurement precision of the laser interferometer is improved.

The application is suitable for the technical field of copper wire, and particularly relates to a die matching coefficient optimization method for multi-pass drawing of bare copper wire, which comprises the following steps: a coefficient offset and a coefficient reference difference are respectively determined through a die matching reference parameter and real-time production parameters and the die matching reference parameter and a preset die matching target parameter, so as to obtain a coefficient actual difference of each pass of the bare copper wire; the intermediate quantity of the die matching reference parameter is introduced, instead of directly obtaining the coefficient actual difference through the real-time production parameters and the preset die matching target parameter, so as to improve the accuracy of the coefficient actual difference; and finally, a die matching coefficient optimization scheme of the bare copper wire is obtained through the coefficient actual difference and a preset extension threshold value. Through the above method, the problem that the die matching coefficient optimization scheme obtained by directly using the calculation logic of the target die matching coefficient minus the actual die matching coefficient has deviation and cannot accurately achieve the die matching target can be solved.

The application provides an affine power flow calculation method for a domain contraction type electrical comprehensive energy system, establishes an affine model of an electrical network, a gas network and a coupling element considering the uncertainty of distributed power output fluctuation and load change; around the network basic operating point under the determined performance flow, firstly, the state quantity is completely predicted based on network sensitivity analysis, and a compression coefficient optimization model is introduced to compress the domain of the predicted state quantity, so as to reduce the conservativeness of the state quantity solution. In addition, an electrical affine power flow column iteration method is provided, the power matching at the coupling unit is used to realize the power flow interaction in the calculation, the calculation sequence of the network in the complex coupling system does not need to be combed, and efficient and simple electrical-gas multi-power flow solving is realized.

The invention discloses a sounding rocket empennage aerodynamic optimization method and system based on a machine learning proxy model, and provides a proxy model optimization framework combining an electric eel foraging optimization (EEFO) algorithm and a kernel extreme learning machine (KELM) to solve the problems of high cost and low efficiency caused by traditional aerodynamic and structure optimization separation. CFD simulation data is obtained through a Latin hypercube method to train a KELM model, model parameters are optimized through an EEFO algorithm, and a multi-target optimization model is constructed to balance a resistance coefficient, the empennage volume and a lift coefficient. Optimization results show that the method can significantly reduce the resistance, moderately control the lift force loss and improve the lift-drag ratio, and an efficient optimization scheme is provided for the design of the sounding rocket empennage.