14 results about "Signomial" patented technology

The invention provides a double-layer safe trainable quantum machine learning method based on polynomial dynamics Lie algebra, and the method comprises the following steps: 1, constructing core variation simulation meeting the constraint of the polynomial dynamics Lie algebra, and guaranteeing the trainability of a model; 2, performing truncated Chebyshev graph coding at an input end, and constructing a rugged loss function landscape by using graph state entanglement and a Chebyshev tower strategy to prevent a snapshot inversion attack; 3, executing dynamic local scrambling at an output end, applying time-varying random local unitary transformation before measurement, and confusing a linear relation between gradient and a snapshot to prevent recovery attack of the snapshot; and 4, measuring and calculating a loss function, and updating parameters. According to the method, an orthogonal decoupling strategy is adopted, a privacy protection mechanism is externally arranged on an input / output interface, and trainability is anchored to core configuration, so that the capability of resisting algebraic attacks is remarkably improved while model convergence is ensured.

The invention discloses a distributed photovoltaic aggregation frequency modulation standby optimal configuration method considering node frequency security constraints under transmission and distribution cooperation. The method comprises the following steps: describing uncertainty of photovoltaic aggregation frequency modulation capability by using a Gaussian mixture model, and establishing a power transmission network node frequency response model containing frequency space distribution characteristics; taking minimization of system power generation and frequency modulation reserve cost as a target, and combining opportunity constraint planning to construct an optimal configuration model; a response model is reduced based on a frequency divider theory, a Bernstein polynomial approximates frequency dynamics, and a differential algebraic equation is converted into an algebraic equation; then, binary expansion and Gaussian mixture model linear invariance are utilized to convert the model into a mixed integer linear programming model, and Bernstein polynomial convex hull property tightening constraint is combined; and finally, solving the model to obtain the optimal photovoltaic aggregation frequency modulation reserve capacity and control parameters. The frequency space distribution is accurately described, the problem of insufficient frequency safety risk identification is solved, and the frequency safety level of a power system is improved.

The application discloses a multi-region frequency safety unit scheduling method and device based on Bernstein polynomials, and the method is based on the optimal distribution of the dynamic primary frequency modulation response of a synchronous generator, a speed governor dead zone and a frequency modulation backup, and a multi-region frequency response model is constructed; the integral-differential algebraic equation of the multi-region frequency response model is approximated through Bernstein polynomials to obtain an algebraic expression of the multi-region frequency dynamics, and a multi-region frequency safety constraint is constructed according to the algebraic expression; the multi-region frequency safety constraint is embedded into a unit combination problem to construct a multi-region frequency safety constraint unit combination model; through a self-adaptive double-layer decomposition algorithm, a mixed integer programming problem of the multi-region frequency safety constraint unit combination model is decomposed into a main problem and a sub-problem; and through solving, a multi-region frequency safety unit scheduling scheme is obtained. The application accurately captures the frequency dynamic difference between regions and improves the frequency safety of a high-proportion new energy power system.

The invention provides a method and system for inverting a material shear constitutive relation based on solid round rod torsion, and relates to the technical field of constitutive relation inversion, and the method comprises the steps: obtaining torque-torsion angle curve data of a solid round rod test piece in a pure torsion experiment, and the length and radius of the solid round rod test piece; establishing a shear stress-shear strain and torque-torsion angle bipolynomial model, and forming a mathematical model based on a polynomial form; then deriving a conversion relation between polynomial coefficients based on a round rod torsion theory, and fitting experimental data through an optimization algorithm to solve a torque polynomial coefficient; and finally, obtaining a constitutive relation coefficient by utilizing mathematical conversion, and generating the shear constitutive relation of the material. The method does not need to preset a constitutive form, is suitable for various materials such as metals and macromolecules, and has the advantages of simple model, efficient inversion, low experiment cost and the like.

The invention relates to the technical field of scientific simulation, in particular to a finite element model parameter optimization method and system based on an intelligent algorithm, equipment and a medium, a traditional manual iterative optimization process is converted into an automatic process through parametric modeling and automatic generation of a Python macro command, the manual operation time is remarkably shortened, and the efficiency is improved. And meanwhile, seamless connection from parameterized modeling of the finite element model to optimization result output is realized, so that the dependence on a user algorithm and programming capability is reduced, and full-process automation is realized. Secondly, multi-objective optimization and nonlinear constraint are supported, common non-convex and high-dimensional optimization problems in engineering can be flexibly processed, and the limitation of a traditional gradient algorithm on the continuity of an objective function is broken through; the response surface agent model replaces high-time-consuming finite element calculation through chaos polynomial expansion, a black box problem is converted into an explicit mathematical relationship, the dependence of an optimization algorithm on actual simulation is reduced, and the iteration efficiency is improved.

The present application belongs to the technical field of multi-agent cooperative control and path planning, and particularly relates to a multi-agent collision-free prediction optimization method based on smooth potential field and parallel ADMM. The method is based on the parallel ADMM architecture, and decouples the optimization problem into two sub-problems of individual dynamics and cooperative collision avoidance. At the individual level, the dual decomposition technique is used to convert the constrained optimization into a parallel solution of dual quadratic programming; at the cooperative level, a second-order differentiable smooth potential energy function is used to replace the mixed integer programming to handle the collision avoidance constraint. The present application reduces the computational complexity from exponential to polynomial, and significantly improves the real-time planning capability and robustness of large-scale multi-agent clusters in complex environments.

The invention discloses a generator set coal consumption curve fitting method and system based on an APSO algorithm. The method comprises the steps of data acquisition and time sequence construction; preprocessing the data; calculating a target variable according to the acquisition quantity, and completing normalization and boundary constraint; a quadratic polynomial model and a piecewise linear model are constructed in parallel, weighted fusion is carried out through learnable weights, and mixed expression of a coal consumption curve is obtained; performing adaptive particle swarm optimization: performing global optimization on the quadratic polynomial and piecewise linear model parameters; the inertia weight and the learning factor are adaptively adjusted along with iteration, and disturbance is applied to partial particle velocity when a plurality of generations are not improved so as to avoid local optimum; the fitness function is the error measure of the predicted coal consumption rate and the measured value; outputting and issuing a result: outputting an optimal parameter and a residual error index; and performing first-order derivation on the fitted curve to obtain an economic load area, performing highlight display on an HMI interface, and writing the curve, the interval and the index into a real-time database to be called by a scheduling / energy efficiency system.

The invention discloses an interval power flow calculation method and system for an AC-DC hybrid system, and belongs to the technical field of power system analysis and calculation, and the method comprises the steps: obtaining system basic parameters and an uncertainty range, building a nonlinear power flow model containing a synchronous generator, an LCC-HVDC, and a ZIP load, and formalizing the nonlinear power flow model into a polynomial optimization problem; a sparse moment relaxation technology is adopted to convert the model into a semi-definite programming model; decomposing and solving the sub-problems through a maximum clique decomposition strategy; and introducing McCormick constraint and an optimality boundary tightening technology to iteratively optimize a variable domain, and finally outputting upper and lower bounds of a power flow solution. The system comprises a parameter module, a conversion module, a decomposition module, a solving module and an output module. According to the method, the uncertainty power flow problem in the alternating-current and direct-current hybrid system can be efficiently and accurately solved, and the method is suitable for power grid safety assessment and optimal scheduling.

The invention relates to the technical field of hydrological simulation, and particularly discloses an extra-small watershed layered three-dimensional hydrological-hydrodynamic full-coupling simulation method, which comprises the following steps: collecting multi-source data to construct a three-dimensional space discrete grid and dividing initial time units; the control equation set is converted into an algebraic system based on a Legendre polynomial and a Gaussian-Loabar node; adaptively adjusting a time unit length and a spectral basis function order according to the local time gradient change rate; a convergence spectral coefficient is obtained through spectral delay correction iteration solution; and finally, forcibly executing mass conservation constraint, controlling step recalculation according to a local time error, and gradually outputting a flow hydrograph and water depth and infiltration rate distribution according to time unit endpoints. According to the method, the dual self-adaption of the time step length and the polynomial order is realized, the numerical oscillation of the infiltration rate caused by the sudden change of the ponding depth can be effectively inhibited, the mass conservation is ensured, and the stability and the efficiency of the rainstorm flood simulation of the extra-small watershed are improved.

The application discloses a special section stiffness accurate calculation method based on feature adaptive sampling and a general analytic operator, and comprises the following steps: feature adaptive sampling and data semanticization; solving of an explicit spline equation based on topological decoupling; normalized parameter mapping and coefficient vector construction; general analytic calculation based on polynomial algebra; global assembly and stiffness matrix generation; the method retains topological sharp points through feature semantic sampling, unifies an integral domain through a normalized parameter equation, and converts a calculus problem into a linear algebra problem through polynomial algebra operation, so that fast and accurate solving of special section stiffness characteristics is realized.

The invention relates to the technical field of ejector structure optimization, in particular to an ejector structure optimization method based on a polynomial function, and the method comprises the steps: obtaining a preset geometric boundary condition of a nozzle diffusion section; generating candidate nozzle contours; obtaining a nozzle injection ratio and a nozzle total entropy product corresponding to the candidate nozzle contour; determining an optimal nozzle contour; a preset pump body starting half angle and a preset pump body ending half angle of the pump body are obtained; acquiring a preset curve starting point coordinate and a preset curve ending point coordinate of the pump body; constructing a pump body contour curve function; generating a candidate pump body contour; constructing a complete ejector model; the model injection ratio of the complete ejector model, the model total entropy production of the complete ejector model and the model maximum allowable backpressure of the complete ejector model are obtained; obtaining a preset engineering application requirement; determining and outputting a final optimization scheme; according to the invention, a break angle caused by geometric mutation is eliminated, and a flow channel is continuous and smooth, so that the generation of shock wave and flow separation is inhibited.

The invention relates to a tough power grid, in particular to a convex hull solving method for tough looped network reconstruction, which is used for solving the problems that the solving efficiency of a reconstruction model of an existing annular structure power grid is low, and particularly, a mixed integer linear programming method for mainstream reconstruction is difficult to solve when facing a large-scale power distribution network reconstruction problem. According to the scheme, when the power grid of the annular structure is reconstructed, a completely equivalent convex hull solving method is deduced by adopting multiple rounds of mathematical changes and simplification of dynamic programming, bilevel programming and strong dual transformation, the method has polynomial complexity, repeated calculation is reduced, calculation performance is improved, and the calculation efficiency is improved. The reconstruction solving efficiency of the existing large-scale power distribution network is effectively improved, the result is ensured to be optimal, and technical support is provided for safe supply of a tough power grid.

The application discloses a method and system for expanding a vehicle stability domain estimated by a sum-of-squares programming algorithm, and belongs to the technical field of computers, and comprises the following steps: a rational polynomial-based whole vehicle polynomial model is constructed by combining a two-degree-of-freedom dynamics model of a vehicle and a rational polynomial nonlinear tire model; a vehicle stability region is estimated by a sum-of-squares programming algorithm based on the whole vehicle polynomial model, and a stability region boundary curve is obtained; the stability region boundary curve is subjected to discrete processing, and an initial point set is obtained; a convex hull of a stability domain expansion is obtained by trajectory inversion according to the stability region boundary curve and the initial point set; and the convex hull of the stability domain expansion is taken as an expanded stability region. The application can realize maximum estimation of a stability domain boundary, greatly improve the conservativeness of the stability domain, make the judgment of vehicle stability more accurate, and be favorable to improving the reliability of vehicle stability control.