30 results about "Quasi-Newton method" patented technology

The present invention relates to the field of microwave filters, and discloses a two-layer optimization algorithm applied to the coupling matrix transformation of microwave filters, which abstracts the coupling matrix topological transformation problem into an optimization problem and adopts the idea of ​​hierarchical optimization to solve it; in the first layer optimization, an adaptive differential evolution algorithm is used to solve the optimization problem, and a solution with better performance is searched in a larger optimization space. The adaptive differential evolution algorithm is an optimization algorithm based on swarm intelligence, which can approach the optimal solution through continuous iteration of the swarm. By using this algorithm, a relatively good solution can be found. In the second layer optimization, based on the solution obtained by the first layer optimization, an adaptive quasi-Newton method is used to solve the optimization problem, and a solution with better performance is quickly searched in the local optimization space. The quasi-Newton method is iterated based on gradient information, and only needs to calculate the approximate value of the inverse matrix of the Hessian matrix, so the calculation efficiency is high.

The invention relates to an ammeter error solving method and device and a storage medium, and the method comprises the following steps: obtaining a measurement switch in a metering box and data of all sub-meters corresponding to the measurement switch, and carrying out the preprocessing of the data of all sub-meters; constructing a split-phase energy conservation equation set and a total energy conservation equation set based on the preprocessed data of all the sub-tables; establishing a simultaneous equation set of a split-phase energy conservation equation set and a total energy conservation equation set, and solving through a BFGS quasi-Newton method to obtain an initial fitting residual sequence; screening out the metering period data based on the data in the initial fitting residual error sequence to obtain an optimal data set, solving the optimal data set through the BFGS quasi-Newton method again, carrying out repeated iteration, and stopping iteration until an iteration stopping condition is met to obtain a final optimal data set; and calculating the error of each sub-table of the measurement switch based on the final optimal data set.

The application belongs to the technical field of concrete strength detection, and relates to a micro core sample method strength conversion method and system for UHPC. The method comprises the following steps: preparing a plurality of different micro core sample samples and pouring a plurality of standard test piece samples matched with the micro core sample samples; performing a compression test on all the micro core sample samples and the standard test piece samples to calculate a measured conversion coefficient; constructing a mathematical model of a predicted conversion coefficient, determining a parameter to be calibrated and a value range of the parameter, constructing an objective function, performing parameter inversion by using a differential evolution algorithm and a limited memory quasi-Newton method, determining a calibrated value of the parameter, and updating the mathematical model of the predicted conversion coefficient; taking a micro core sample from a concrete entity structure on site, performing a compression test on the micro core sample, calculating the predicted conversion coefficient, and then calculating a standard test piece compression strength. The application can accurately establish a conversion relationship between the micro core sample compression strength and the standard test piece compression strength, and improve the strength detection precision of the UHPC entity structure.

The invention belongs to the technical field of preparation of carbon-based adsorbents, and relates to an adsorption performance prediction method and optimization method of a carbon-based CO2 adsorbent. According to the adsorption performance prediction method, the continuous distribution characteristics of the pore structure of the carbon-based material are represented by an N2 adsorption isotherm under 77K, and the construction of a global mapping framework of preparation conditions-the continuous distribution characteristics of the pore structure-CO2 adsorption separation performance is realized by combining a convolutional neural network and a multi-layer perceptron deep learning algorithm. And meanwhile, multiple rounds of preparation parameter optimization are carried out by adopting a limited memory quasi-Newton method algorithm, and dynamic closed-loop optimization of model performance and parameter regulation and control is realized in combination with an active learning strategy. Through multiple rounds of active learning iteration processes, the prepared biochar sample has a narrow micropore-mesopore coordinated hierarchical pore structure, and the CO2 and N2 separation capacity is remarkably improved. The invention provides a new method support for directional design and efficient application of the carbon-based CO2 adsorbent.

The invention discloses an optimization iteration solving method for an aero-engine model based on an LM algorithm and a quasi-Newton method, and the method employs a variable cycle engine model as an example, builds a complete engine model, and compiles a complete engine model function library. A function library comprises a complete engine model, a thermodynamic function, a model solving algorithm and a GUI interface, a volume model is adopted for an engine, a balance equation and a co-working equation of the engine are established, a novel algorithm enabling an initial guess value to be more accurate is adopted when the model is solved, the convergence speed can be increased by 13.7%-29.1%, and the convergence speed can be increased by 13.7%-29.1%. And meanwhile, the LM algorithm or the quasi-Newton method is automatically matched in the subsequent solution, so that the model solution is faster and more accurate, the steady-state error of the solved model does not exceed 0.8%, and the simulation precision is higher.

The application discloses a kind of based on storage optimization's hybrid beam forming method, system and storage medium, its method includes steps: according to the number of radio frequency link, antenna quantity and channel state calculation transmitting end hybrid precoding matrix constraint condition;According to constraint condition, the objective function of transmitting end hybrid precoding matrix optimization is established;According to the relationship of time delay and / or spectral efficiency and / or memory occupancy and quasi-newton method storage capacity calculation optimal storage capacity;According to the optimal storage capacity corresponding finite memory quasi-newton method, the optimal solution of transmitting end hybrid precoding matrix optimization objective function is calculated;According to the optimal solution of transmitting end hybrid precoding matrix optimization objective function, transmitting end hybrid precoding matrix is obtained and is used for hybrid beam forming.The application solves the problem that the system performance of existing hybrid beam forming method is poor and the calculation time delay is high.

The invention relates to the technical field of petroleum and natural gas engineering drilling, and discloses a track profile dynamic optimization method and system for drilling, and the system comprises a track model library, an initial value generation unit, an optimization solving unit, a constraint evaluation unit and a result output unit. The orbit model library constructs three-section and five-section orbit section models based on a constant tool face angle, and the optimization solving unit fuses a quasi-Newton method and an artificial bee colony algorithm to realize orbit parameter optimization under multiple variables and multiple constraints. According to the method, an adaptive track profile type is selected through real-time drilling data, and optimized parameters are output after initial value solving, algorithm optimization and constraint evaluation. According to the method, the feedback of the drilling process can be dynamically responded, the target window drilling rate is remarkably improved, the track complexity and the dragging pressure risk of a long horizontal section are reduced, the drilling requirements of directional wells and horizontal wells under complex stratums are met, and technical support is provided for efficient and safe development of drilling operation.

The invention provides a while-drilling electromagnetic wave resistivity logging inversion method based on a quasi-Newton method, and relates to the technical field of oil-gas exploration and development, and the method comprises the steps: collecting the actual measurement response data of an isotropic or anisotropic target stratum, constructing a three-layer TI medium stratum model, and determining the to-be-inverted parameters; performing global optimization by adopting a genetic algorithm, and outputting an optimal result as an initial value; constructing a quasi-Newton inversion model containing a regularization term, dynamically updating a Jacobian matrix by using a Broyden updating rule, introducing a diagonal adaptive damping matrix to replace a fixed damping factor, and adaptively adjusting the damping strength according to a parameter relative increment ratio; and inputting the initial value into the improved model for iterative operation until a termination condition is met. The method gets rid of manual experience dependence, avoids local extremum convergence, improves the inversion precision, reduces the number of iterations, improves the calculation efficiency, accurately inverts key parameters, meets the requirements of real-time geosteering engineering, and improves the oil-gas exploration and development efficiency.

The invention discloses an abnormal power fluctuation detection method and system fused with time sequence modeling. The method comprises the following steps: acquiring daily electric quantity year-on-year data of each industry and an industry power consumption data set; the method comprises the following steps: constructing a time sequence model based on year-on-year data of daily electric quantity of each industry, calculating posterior distribution of a predicted value of the time sequence model through an L-BFGS quasi-Newton method, and calculating a data basic confidence interval of a next time point; inputting each characteristic value of the time sequence model one day before the date to be predicted, the mean absolute error of the predicted value of the short-term historical window time sequence model, the residual standard deviation and the industry power consumption data set into a full-connection network to obtain a heteroscedasticity confidence interval; and calculating a final threshold interval based on the two confidence intervals to carry out industry electricity consumption abnormity judgment. According to the method, the problem that the phenomena of false alarm and missing report occur frequently due to the fact that the power utilization characteristic difference between different industries is not considered and the change of the electric quantity rule along with time is difficult to adapt in the prior art is solved.

The present invention discloses a method for automatic multi-parameter debugging of optical modules based on deep learning, which belongs to the field of optical communication technology. The method comprises the following steps: setting an adjustment parameter space of an optical module; constructing an optical module parameter debugging data set; constructing a deep neural network proxy model; training the deep neural network proxy model, and optimizing the optimal observation value set selected after training by using a Bayesian optimization acquisition model and a quasi-Newton method to obtain predicted performance parameters; judging whether the predicted performance parameters meet the sample addition condition, and if so, adding adversarial training samples to the optical module parameter debugging data set and retraining, otherwise saving the current optimal observation value set; repeating the training and saving the optimal observation value set until the optimal observation value set is separated by a certain time. t The optimal observation value sets saved in each round meet the convergence constraint, and the parameter debugging results are written into the Flash and stored. The present invention solves the problems of low efficiency and insufficient accuracy in optical module debugging.

This invention discloses a DC motor speed control identification method and system based on an improved quasi-Newton method, belonging to the technical field of DC motor speed control system modeling. First, a time-varying nonlinear DC motor speed control system is constructed, and an identification model of this system is obtained. Then, an identification method flow based on the quasi-Newton method under step-size line search is constructed. The optimal step size is obtained through step-size line search for parameter estimation, and a correction strategy is designed to ensure the positive definiteness of the approximation of the Hessian matrix inverse. This invention avoids the oscillation or slow convergence problems caused by a fixed step size by dynamically adjusting the step-size factor, and simultaneously introduces a correction strategy to forcibly guarantee the positive definiteness of the approximation of the Hessian matrix inverse, thereby obtaining more accurate motor model parameters.

The invention discloses an AC / DC power system load flow calculation method and system based on a quasi-Newton method, and the method comprises the steps: building load flow calculation models of a conventional DC converter station and a flexible DC converter station based on the input and output characteristics of an AC side and a DC side; a rectangular coordinate equation which only considers node voltage in traditional load flow calculation is expanded to an augmented rectangular coordinate system which considers node voltage and current, and a unified load flow calculation model of AC-DC load flow calculation is established; determining the direction and step length of the first iteration through a Newton method, and obtaining an iteration initial value of the quasi-Newton method; the quasi-Newton iteration matrix is continuously updated in the iteration process, so that repeated derivation calculation of the Jacobian matrix and inversion are not needed; and when the power flow solution is iterated to meet the requirement, outputting a power flow calculation solution. Through the calculation method provided by the invention, the unified iteration of the alternating current and direct current power flow can be realized by utilizing the alternating current and direct current unified power flow model, meanwhile, the time required by each iteration is saved by adopting the quasi-Newton method, and the method has the characteristic of super-linear convergence.

The application discloses an electric-gas comprehensive energy system cascade fault prediction method, and relates to the field of comprehensive energy system simulation control. The method comprises the following steps: firstly, an electric-gas comprehensive energy system cascade fault model is established, which comprises power system primary frequency modulation dynamics, natural gas fault propagation dynamics, coupling element dynamics and a fault model. Then, based on Newton interpolation and backward difference principles, an electric-gas cascade fault analysis model based on a numerical differential discrete formula is derived and established. Secondly, aiming at the high-dimension, nonlinearity and rigidity characteristics of the model, a variable order and variable step length calculation strategy is proposed to optimize the solving process, and a matrix updating and decomposition strategy based on the quasi-Newton method is adopted to reduce the operation cost. Finally, a key event positioning technology based on a backward difference vector is used to accurately lock the cascade generator tripping fault time. The application further discloses an electric-gas comprehensive energy system cascade fault prediction system, and the application can efficiently and accurately predict the cascade fault evolution process.

The invention provides a six-degree-of-freedom parallel platform based on iterative initial value estimation and a forward motion solution method thereof. The six-degree-of-freedom parallel platform takes 6-UPU configuration as a research object; comprising a lower platform, an upper platform, six Hooke hinges fixed on the lower platform, six Hooke hinges fixed on the upper platform, and six electric push rods connected with the Hooke hinges of the lower platform and the Hooke hinges of the upper platform. The six-degree-of-freedom parallel platform motion positive solution method based on iteration initial value estimation comprises forward kinematics modeling of a six-degree-of-freedom parallel platform, an iteration initial value estimation method based on space geometry, an iteration initial value estimation method based on particle swarm optimization, an iteration initial value comparison method and a quasi-Newton method based on iteration initial value estimation.

The application discloses a material process parameter self-adaptive optimization method and system based on dynamic Bayesian optimization, and belongs to the field of artificial intelligence assisted material research and development. In view of the problems of high cost and easy falling into local optimization under complex phase change in traditional material research and development, the application constructs an initial set by acquiring material historical process and performance data; a nonlinear proxy model is constructed by using Gaussian process regression and a composite kernel function fusing smooth and rough priori; a function acquisition strategy is dynamically switched according to a prediction variance to balance global exploration and local utilization; within the extracted physical equipment hard constraint boundary, a multi-start quasi-Newton method is used to perform global optimization, and the next round of recommended experimental parameters are output and closed-loop iteration is supported. The application realizes accurate optimization of multi-dimensional parameters with few experimental times, and is mainly used for complex chemical and material process parameter optimization such as polymerization and alloy smelting.

Systems and methods reconstruction for a medical imaging system using a quasi-newton method. An unrolled iterative reconstruction process is used to reconstruct an image from the scan data. The unrolled iterative reconstruction process includes a plurality of cascades that include at least a data-consistency step and a regularization step. The data-consistency step is modified based at least in part on information of already calculated gradients of one or more previous cascades of the plurality of cascades using a quasi-newton computation.

The invention provides a power system small interference stability minimum damping ratio margin monitoring method, which comprises the following steps: firstly, considering load growth mode constraints and power system operation requirements, and establishing a small interference stability minimum damping ratio margin optimization model; secondly, constructing an augmented Lagrangian function by using a Lagrangian multiplier method and an exterior penalty function method, and deriving a first-order optimality condition; and finally, a nested double-layer iteration solving method is designed, the outer layer iteration is solved by adopting a quasi-Newton method, and the inner layer iteration is solved by adopting an extended three-stage strategy. According to the method, the most dangerous load increasing mode and the corresponding small-interference stability minimum damping ratio margin can be found while the system operation requirement is met, and the method is easy to implement and has high application value.

The application discloses a kind of mechanical arm device iterative learning control system and method based on quasi-newton method, target trajectory determination module in the system is used to determine the expected target trajectory of the mechanical arm device, nonlinear equation determination module is used to determine the nonlinear equation related to mechanical arm nonlinear system, pose acquisition module is used to obtain the position of the mechanical arm device under current iteration, control module is used to update the control input of the mechanical arm device when next iteration according to the expected target trajectory of the mechanical arm system, control input under current iteration, position output under current iteration and gain matrix, the mechanical arm device is controlled.The application can make that mechanical arm is in finite time period at any time to the trackable target trajectory Realize high-precision tracking.

The application discloses a solid internal temperature field and thickness efficient joint reconstruction method based on a quasi-Newton method, belongs to the field of ultrasonic nondestructive detection, and comprises the following steps: constructing a multi-parameter inversion problem with boundary heat flow and thickness as decision variables, defining an objective function containing a regularization term, and converting the inversion problem into a nonlinear optimization problem; for the nonlinear optimization problem, the quasi-Newton method L-BFGS is used to invert the heat flow, and the steepest descent method DG is used to invert the thickness, so that the internal temperature of a high-temperature structure and the size and thickness thereof are reconstructed. The application can realize synchronous high-precision and high-efficiency reconstruction of the internal temperature field and thickness size of a solid structure.

The invention discloses a solid internal temperature field and thickness efficient joint reconstruction method based on a quasi-Newton method, and belongs to the field of ultrasonic non-destructive detection, and the method comprises the steps: constructing a multi-parameter inversion problem with boundary heat flow and thickness as decision variables, defining a target function containing a regularization item, and converting the inversion problem into a nonlinear optimization problem; and aiming at the nonlinear optimization problem, combining a quasi-Newton method L-BFGS inversion heat flow with a steepest descent method DG inversion thickness, and reconstructing the internal temperature and the size thickness of the high-temperature structure. According to the invention, synchronous high-precision and high-efficiency reconstruction of the internal temperature field and the thickness size of the solid structure can be realized.

The present invention relates to a high-precision ray tracing method and device for an optical system based on a NURBS surface. The method and device include: generating an optical-mechanical surface expression based on NURBS by forward and reverse construction methods; formulating a ray packet allocation and parallel computing strategy according to the scale of the ray; constructing a primary BVH bounding box using the information of the optical-mechanical surface node; refining the surface element to generate a secondary BVH bounding box containing sub-surface elements; and using a quasi-Newton method to iteratively solve a high-precision value of the intersection of the ray and the optical-mechanical surface. The method and device are based on the NURBS mathematical representation method, combined with a large-scale ray subpackaging strategy, a parallel computing strategy, a specific discrete sampling scheme, etc., and by designing a hierarchical BVH acceleration data structure, high-precision tracing accuracy is achieved by using the quasi-Newton method while obtaining excellent iterative initial values, while effectively improving the interaction efficiency.

The invention belongs to the technical field of environmental monitoring and hydrological simulation, and particularly relates to a hydrological parameter and particle tracking fused pollution diffusion traceability system, which is characterized in that real-time hydrological observation data and high-resolution hydrodynamic model output are subjected to space-time alignment by constructing a dynamic coupling mechanism; a three-dimensional Lagrange particle tracking model is driven; a quasi-Newton method and finite element method collaborative optimization objective function inversion framework is adopted, and the space coordinates, the release time and the discharge flux of a pollution source are synchronously determined in a multi-source and strong disturbance scene; and non-tampering evidence storage is carried out on the key traceability evidence chain based on the alliance chain architecture. The structural defects of hydrological parameter staticization, inversion algorithm simplification, data credibility deficiency, response mechanism lagging, cross-domain cooperation difficulty and the like in a traditional pollution diffusion traceability system are overcome, and full-chain automation and precision support from pollution identification, path tracking, source inversion to responsibility determination are achieved.

The invention belongs to the technical field of generation power optimization of a wind power plant, and provides a wind power plant power optimization method and system based on BFGS and knowledge generalization mechanism data driving. According to the method, an optimization framework of a quasi-Newton method is adopted, and according to wind power plant power generation data information measured in real time, the gradient estimated by synchronous perturbation stochastic approximation and Hessian matrix inversion estimated by BFGS are utilized to complete updating of all fan control actions; a knowledge generalization mechanism is designed, and real-time generalization and full utilization are carried out on learned action knowledge; the method does not depend on a wind power plant generation power model, and solves the problems that an existing optimization scheme is slow in convergence speed and difficult to effectively improve the limitation that the modeling wind power plant generation power is difficult to analyze.

The present application relates to optical remote sensing satellite micro-vibration isolator dynamics parameter system level optimization method, belongs to spacecraft vibration control technical field;Establish the finite element model of the whole satellite;Modal analysis is carried out to the isolation system, and the function relation of the natural frequency of the isolation system and the axial stiffness k w , lateral stiffness k v Is fitted;The constraint condition of natural frequency f 横向 And f 纵向 Is set;The function relation of the angular displacement of high-resolution camera and the axial stiffness k w , lateral stiffness k v Of the isolator is fitted;The function relation of the angular displacement of the main components and the axial stiffness k w , lateral stiffness k v Of the isolator is fitted;The objective function F is constructed;With the constraint condition of lateral natural frequency f 横向 And longitudinal natural frequency f 纵向 As constraint, the objective function F is solved by nonlinear optimization using quasi-newton method, and the lateral natural frequency f 横向 , longitudinal natural frequency f 纵向 Of the optimized isolation system;The present application controls micro-vibration from the system level, and gives the specific parameter requirement of isolator design and development.

This invention provides a robot trajectory planning method based on second-derivative regularization, comprising: parameterizing the joint space path with path parameters, establishing a joint velocity and acceleration constraint model, and obtaining the original path velocity square distribution and feasible region based on the TOPPRA algorithm; designing a second-derivative regularized objective function in the path velocity square space, including a data fidelity term and a curvature penalty term; constructing a bounded constraint optimization model based on the upper and lower bounds of the feasible region and the values ​​of the first and last ends, and analytically deriving the gradient of the objective function; solving the optimization model using a bounded constraint quasi-Newton method, restoring the obtained optimal path velocity square to a smooth path velocity, recovering the kinematic time series of each joint, and completing the trajectory smoothing planning. This invention can effectively suppress sharp abrupt changes in joint acceleration, achieving robot trajectory planning that balances time efficiency and motion stability.

The application discloses a power flow calculation method and system of AC / DC power system based on quasi-Newton method, establishes a power flow calculation model of conventional DC converter station and flexible DC converter station based on input and output characteristics of AC side and DC side; extends the rectangular coordinate equation of only considering node voltage in traditional power flow calculation to the augmented rectangular coordinate system considering node voltage and current, and establishes a unified power flow calculation model of AC / DC power flow calculation; establishes the direction and step length of the first iteration through Newton method, and obtains the initial value of quasi-Newton iteration; continuously updates the quasi-Newton iteration matrix in the iteration process, so that the calculation of Jacobian matrix and inversion is not required repeatedly; when the power flow solution iteration meets the requirements, the solution of power flow calculation is output. Through the calculation method, the unified iteration of AC / DC power flow can be realized by using the unified power flow model of AC / DC, meanwhile, the quasi-Newton method is adopted to save the time required by each iteration, and has the characteristics of super-linear convergence.

The application provides a method for electromagnetic wave resistivity logging while drilling inversion based on quasi-Newton method, and relates to the technical field of oil and gas exploration and development, which comprises the following steps: collecting measured response data of isotropic or anisotropic target formation, constructing a three-layer TI medium formation model, and determining the parameters to be inverted; performing global optimization by using genetic algorithm, and outputting the optimal result as an initial value; constructing a quasi-Newton inversion model containing a regularization term, dynamically updating the Jacobian matrix by using Broyden updating rule, introducing a diagonal adaptive damping matrix to replace the fixed damping factor, and adaptively adjusting the damping strength according to the relative increment of the parameters; inputting the initial value into the improved model for iterative operation until the termination condition is met. The application is free from dependence on artificial experience, avoids local extremum convergence, improves inversion accuracy, reduces the number of iterations, improves calculation efficiency, accurately inverts key parameters, meets the requirements of real-time geosteering engineering, and improves the efficiency of oil and gas exploration and development.

The invention discloses a malicious software security detection method based on a graph neural network, and the method comprises the following steps: S1, collecting static behavior data and dynamic behavior data of to-be-detected software, and constructing a software graph structure; s2, carrying out standardization and feature coding on the software graph structure, and extracting node, edge and global structure features; s3, initializing the graph neural network model, setting structural hyper-parameters and initializing weight parameters; s4, executing forward propagation to obtain embedded features and preliminary classification results of the software samples; s5, performing quasi-Newton method joint optimization on weight parameters and structural hyper-parameters of the graph neural network model; and S6, processing new sample data, inputting the optimized model, and outputting a malicious software detection result. According to the malicious software detection method fusing the graph neural network and the quasi-Newton method, efficient recognition and accurate classification of unknown and variant samples are achieved.

The invention discloses a heterogeneous velocity modeling method based on seismic data Hessian optimization, and relates to the technical field of geophysical exploration velocity modeling, and the method comprises the following steps: S1, obtaining seismic observation data and seismic simulation data; s2, giving seismic wavelets of a required main frequency, and carrying out convolution on the seismic wavelets, the seismic observation data and the seismic simulation data to obtain seismic frequency division data; s3, establishing a target function by using the seismic wave field residual error of the seismic simulation data and the observation data after frequency division; s4, using the seismic wave field residue and based on a quasi-Newton method to optimize a high-order derivative item in the Hessian matrix, and using the optimized high-order derivative item to obtain an iterative update direction of the velocity model; and S5, obtaining an iteration step length of the speed model by using the target function based on a Barzilling-Borwein method and a non-monotonic criterion, and updating the speed model by using the iteration step length. According to the method, the main characteristics of the Hessian matrix are utilized, so that the speed inversion precision can be effectively improved.

The invention discloses a high-precision component-level model modeling method considering a volume effect, and relates to a modeling method of components such as a gas compressor, a combustion chamber, a turbine blade gas film cooler, a heat exchanger and a duct ejector. An initial guess value search algorithm (LSFLA) combining Levy flight and a shuffled frog leaping algorithm is adopted to solve an initial guess value of an engine common working equation, the initial guess value of the common working equation is solved, and a quasi-Newton method is used to replace a Newton-Raphson method to accelerate a model iteration process. The modeling method explains the influence of the volume effect on the engine transition state characteristic according to the geometric structure of the engine, and dynamic simulation and steady state simulation are carried out, which explains that the volume effect is not negligible.