66results about How to "Solving Convergence Problems" patented technology

The invention provides an underwater autonomous robot fixed depth control method based on reinforcement learning and belongs to the underwater robot control field. The method comprises steps that firstly, a Markoff decision process model for underwater autonomous robot fixed depth control is constructed, and expressions of a state variable, a control variable, a transfer model and a one-step lossfunction for underwater autonomous robot fixed depth control are respectively acquired; the decision network and the evaluation network are respectively established; through reinforcement learning, the decision network and the evaluation network can be continuously updated whenever an underwater autonomous robot moves forwards for each step during fixed depth control training till convergence; thefinal decision network for fixed depth control is acquired. The method is advantaged in that fixed depth control on the underwater autonomous robot under the condition of a completely-unknown underwater autonomous robot dynamic model is realized, and the practical value is high.

The invention discloses a hybrid-domain full wave form inversion method of central processing unit (CPU)/graphics processing unit (GPU) synergetic parallel computing. Compared with a traditional method, the method can be adopted to conduct the CPU/GPU synergetic parallel computing so as to obviously improve computing efficiency. The method enables a forward part of full wave form inversion to be placed to a time domain, namely forward is conducted in the time domain, the forward is converted to be conducted as the inversion in a frequency domain by the discrete fourier transform (DFT) being utilized, namely the DFT is adopted to pick wave field components corresponding to inversion frequency, and the inversion is conducted in the frequency domain from low frequency to high frequency. The method effectively resolves the problem of astringency of a standard time domain method, and avoids the problem that internal memory of wave form inversion of a standard three-dimensional frequency domain and a Laplace domain cannot be satisfied. The method is few in step, reduces computing expenses, and due to the fact the method can be adopted to conduct the CPU/GPU synergetic parallel computing, largely improves speed-up ratio of the computing method.

The invention discloses a load flow calculation method of a rectangular coordinate newton method. The load flow calculation method comprises the following steps: inputting original data and initializing a voltage; forming a node admittance matrix; calculating power and a voltage deviation; forming a jacobian matrix J; solving a modified equation and modifying a real part e and an imaginary part f of the voltage; and outputting nodes and branch data. The calculation formula of partial elements (i=j) of the jacobian matrix obtained in a manner of load flow calculation by utilizing the rectangular coordinate newton method is modified and improved, so that the problem of astringency of the load flow calculation of the rectangular coordinate newton method during the analysis of branch systems containing small impedance is solved. As compared with the fact that the astringency of the small impedance branch systems cannot be realized by adopting the load flow calculation of the conventional rectangular coordinate newton method, the reliable astringency of the small impedance branch systems can be realized by utilizing the load flow calculation disclosed by the invention; and the load flow calculation to normal systems can be carried out at the same time, so that harmful effects are avoided.

The invention discloses a crowd calculation quality control method and device. The crowd calculation quality control method includes partitioning crowdsourcing task set to be processed into a plurality of subtask sets; selecting test tasks from the subtask sets and determining difficulty coefficient and accurate answer of the test tasks; determining the first subtask set and the second subtask set; determining the difficulty coefficient of subtasks in the first subtask set according to the difficulty coefficient of the test tasks; assessing ability coefficient of workers according to the test tasks, determining subtask completion accuracy of the workers according to the ability coefficient of the workers and the subtasks in the first subtask set and acquiring completion results of the subtasks in the second subtask set. By the crowd calculation quality control method, completion accuracy in crowdsourcing tasks is improved and cost is controlled to some extent.

The invention discloses a rapid decomposing method load flow calculating method suitable for a branch system comprising small impedance. The method comprises the step of inputting original data and initializing voltage, the step of forming a node admittance matrix, the step of forming a coefficient matrix B' and a coefficient matrix B'' of a modified equation and decomposing a factor table, the step of setting iteration count and a convergence mark, the step of calculating the inactive power unbalance amount delta Q, the step of judging whether the absolute value delta Qmax is smaller than the convergence accuracy epsilon or not, the step of calculating the active power unbalance amount delta P, the step of judging whether the absolute value delta Pmax is smaller than the convergence accuracy epsilon or not, the step of calculating the balance node power and the inactive power of a PV node, and the step of calculating the branch power. Part of elements in the coefficient matrix B' of the modified equation of a rapid decomposing method are modified, and the convergence problem produced during analyzing the branch system comprising the small impedance with the resistance not equal to zero in rapid decomposing load flow calculation is solved. When a convention method is adopted to calculate non-convergence, the rapid decomposing method load flow calculating method can carry out convergence reliably.

The invention discloses a rectangular coordinate Newton method load flow calculation method for a small-impedance-branch-containing power grid based on a compensation method. Before performing load flow calculation, a small-impedance branch is processed firstly; the small-impedance branch is changed into two series branches, wherein the impedance of one branch is z1 which is equal to r<ij>+j(x<ij>+x<c>) while the impedance of the other branch is z2 which is equal to -jx<c>, wherein x<c> is a normal electric reactance average value x<av> of an electric power system. One small-impedance branch is changed into two branches with relatively high impedance through series compensation, so that the convergence of load flow calculation can be improved. By adoption of the load flow calculation method, the small-impedance branch of any impedance can be calculated, or even a branch with zero impedance also can be calculated; the small-impedance branch data is processed before the load flow calculation is performed; and the data is preprocessed, so that a load flow calculation program is not changed, so that the rectangular coordinate Newton method load flow calculation method is particularly suitable for transforming an old load flow calculation program.

The invention discloses a rectangular coordinate newton method power flow calculation method based on an improved jacobian matrix; a jacobian matrix calculation method which is different between the first time of iteration and the rest time of iteration process is adopted, so that the problem of convergence of the rectangular coordinate newton method power flow calculation in a small-impedance branch power system is solved. During the first iteration, a PQ node adopts a current vector real part and virtual part obtained by an injected power given value to calculate jacobian elements; a PV nodeadopts a current vector real part and virtual part obtained by an active power given value and a reactive power given value to calculate jacobian elements, without using a power supply reactive powergiven value, so that the adverse effect to the power flow calculation convergence caused by the PV node power supply reactive given value can be avoided. When the conventional rectangular coordinatenewton method is adopted for calculation, the flow calculation is not converged; and compared with the prior art, the method disclosed by the invention has the advantages that the convergence is reliable, and the number of iterations is less than that of the prior art. The method can also carry out power flow calculation on a normal electric power system without adverse effects.

The invention discloses a Newton method load flow calculation method based on a node type and a modified Jacobian matrix, and solves a problem of convergence when rectangular coordinate Newton methodload flow calculation is used for solving a branch power system comprising a small impedance through employing a Jacobian matrix calculation method difference from each iteration process in the firstiteration. In the first iteration process, a PQ node employs real and virtual parts of a current phasor calculated through a node injection power given value for calculating Jacobian elements, and a PV node does not use the real and virtual parts of the current phasor for calculating the Jacobian elements, thereby avoiding the adverse impact on the load flow calculation convergence from a reactivepower given value of the PV node. When the rectangular coordinate Newton method load flow calculation is not convergent, the method can achieve the reliable convergence. Moreover, compared with the prior art, the number of iteration times is reduced. The method can perform the load flow calculation of a normal power system, and caused no adverse impact.

The invention discloses a method for calculating cooperative power flow of an active power distribution network and a power transmission network based on a system of systems (SoS). The method comprises the steps of defining an electric power system to be the SoS, and building an SoS model; modeling the power transmission network and the active power distribution network to be two independent systems in the whole system; considering the relation between a distributed power supply and a load in the active power distribution network, and determining an operating state of the active power distribution network; respectively considering power flow calculation of the active power distribution system and the power transmission network system according to different operating states of the active power distribution network; taking boundary calculation results of the two systems as initial values of each iteration thereof until the boundary results converge, and acquiring a power flow calculation result of the whole electric power system. According to the invention, the SoS model for power flow calculation of the electric power system is built, the power transmission network and the active power distribution network are modeled to be two independent systems, and voltage and power of only boundary nodes are required to be interacted, so that complex power flow calculation of the electric power system is enabled to be simplified. In addition, distributed management is easy to be performed, and a certain basis is laid for optimization of a distributed electric power system.

The invention relates to a thermoelectric interface interaction method for integrated energy network dynamic hybrid simulation, and belongs to the technical field of integrated energy network digitalsimulation. The characteristics of different sub-network width time scales of the comprehensive energy network are fully considered, and the accuracy, effectiveness and convergence of comprehensive energy network hybrid simulation interface interaction of different calculation step lengths are achieved by utilizing a post-prediction correction method. Hybrid simulation electromechanical transientand thermal dynamic two-side thermoelectric interface interaction with different calculation step lengths can be effectively processed, effective interaction of thermoelectricity in hybrid simulationis ensured, and the wide time scale characteristic under system faults and disturbance is embodied. The method is easy to implement, is a practical and effective method for thermoelectric interface interaction of thermoelectric coupling equipment of an actual multi-energy system, is small in calculation amount, and is suitable for real-time and super-real-time dynamic simulation of an integrated energy network system.

The invention discloses a radio frequency amplifier circuit design method and a radio frequency amplifier circuit, and the method comprises the steps: obtaining a bias condition when a radio frequency amplifier works in a specified class, and controlling the radio frequency amplifier to be biased to the specified class; obtaining a radio frequency small signal S parameter of the radio frequency amplifier under a specified class through testing; obtaining an input equivalent circuit based on the radio frequency small signal S parameter; the input equivalent circuit comprises a first inductor, a first resistor and a first capacitor which are connected in series; the load-pull impedance of the radio frequency amplifier under different frequencies is obtained through testing; based on the load-pull impedance, an output equivalent circuit is obtained; the output equivalent circuit comprises a second inductor, a second resistor and a second capacitor; the second resistor and the second capacitor are connected in parallel and then are connected in series with the second inductor. According to the design method, based on the application characteristics of the amplifier, the circuit design process is optimized, the convergence problem is solved, and the amplifier design efficiency is improved.

The invention discloses a method for pre-evaluating the line loss of an electric power system. The method comprises the following steps of: collecting original data; preparing data needed by direct current power flows; carrying out direct current power flow calculation to obtain the power flow of each branch; calculating a line loss power of each branch according to a line loss calculation model provided by the method; and calculating the line loss of the system. The method provided by the invention is used for calculating on the basis of a direct current power flow model, the method is simple, the equation is completely linearized and the solving speed is rapid. The method is sufficiently exact for pre-evaluation of the line loss of a new project of a power grid, especially exact for pre-evaluation of the line loss of a new project of a main network with high voltage grade and high ratio of line reactance to resistance (X/R).

The invention discloses a load flow calculation method with a small-impedance power grid serial-connection compensation right-angle coordinate Newton method. Before load flow calculation, small-impedance branches except those of which nodes at two ends are both PQ nodes and a branch transformation ratio k is 1 are firstly processed, then the small-impedance branches can be transformed into a mode that two branches are serially connected, the impedance of one branch is z1=r+j(x+xav), the impedance of the other branch is z2=-jxav, and xav is a normal reactance mean value of a power system. Through serial-connection compensation, small impedance can be transformed into two branches with relatively large impedance, so that the astringency of load flow calculation can be improved. Meanwhile, as serial-connection compensation on small-impedance branches of which nodes at two ends are both PQ nodes and the branch transformation ratio k is 1 are not implemented, the number of new nodes and branches can be effectively reduced, furthermore the memory space can be reduced, and the calculation speed can be increased. The load flow calculation method is applicable to transformation of conventional load flow calculation procedures.

The invention discloses an improved constrained multi-objective optimization problem solving method. The method comprises the following steps: 1, processing equality constraint conditions and inequality constraint conditions of a constrained multi-objective optimization problem by adopting a constraint processing method based on a constraint violation index; 2, using an improved multi-target bacterial population taxis algorithm based on Pareto dominance to solve and obtain a Pareto optimal solution set and a corresponding Pareto front of a multi-target optimization problem; 3, selecting an optimal compromise solution for constraining the multi-objective optimization problem from the Pareto optimal solution set obtained by solving by adopting a multi-objective decision method based on the relationship between the objective satisfaction and the objective weight, According to the invention, the problems that constraint processing is complex, algorithm convergence, solution set diversity and time efficiency are poor, and preference of decision makers cannot be met are solved, and the requirement for solving the constraint multi-objective optimization problem is met.

The invention relates to a multi-objective optimization method and system for regenerative type electric heating to participate in wind power accommodation. A multi-objective optimization model for regenerative type electric heating to participate in wind power accommodation is constructed, wherein the multi-objective optimization model comprises a maximum wind power accommodation objective function, a minimum regenerative type electric heating operation cost objective function and a minimum system carbon emission objective function, and the problems that existing regenerative type electric heating participates in wind power accommodation research and mainly focuses on benefit evaluation and economy, and the objective functions are single are solved; and an NSGA-III algorithm is used for solving the multi-objective model for regenerative type electric heating to participate in wind power accommodation, and in the aspect of the algorithm, when a high-dimensional multi-objective optimization problem is processed, the problems that a traditional algorithm is insufficient in convergence and solving efficiency and the like are solved. According to the method and the system, the optimization efficiency and the wind power accommodation capability are improved, the carbon emission is effectively reduced, and the equipment operation cost is reduced.