45results about How to "Iteration is simple" patented technology

The invention discloses a face microexpression recognition method and relates to a method for recognizing graphs with electronic equipment. The method comprises the steps that after a face microexpression image sequence is preprocessed, the Birnbaum-Saunders distribution curve is used for building a regression model, then the CBP-TOP algorithm is used for extracting dynamic space-time texture features of the face microexpression sequence, and finally a classier is used for training and prediction. The defect that an existing face microexpression recognition method is sensitive to small features such as bright points, edges and white noise of a face microexpression image, thereby being low in recognition performance is overcome.

The invention discloses a distribution network power flow calculation method of a self-adaptive grid structure. The method includes performing the power flow calculation according to a calculation mode set by a user; verifying the electrical equipment parameters and the topological connection relations required by the power flow calculation; identifying an outgoing line switch and determining a control area according to the position of an interconnection switch; determining a calculation range according to the setting of the user; forming a network topology and a sub-network unit; determiningthe network type and determining the wiring mode of the sub-network unit; self-adaptively matching the corresponding power flow algorithm according to the type of the sub-network unit; performing theout-of-limit, heavy load or network loss calculation; and performing the three-phase power flow and the three-phase imbalance statistics. According to the method, the number of the network nodes is greatly reduced, the calculation speed and efficiency are improved; the advantages of the algorithms are maximized, the stability and the convergence of calculation are ensured; the requirement of a dispatcher on the on-line power flow can be well met, and the application prospect is good.

The invention relates to the field of computational intelligence, and relates to a multi-scale quantum harmonic oscillator concurrent optimization system and method. The method in the invention is composed of three embedded iteration processes: scale iteration, quantum harmonic oscillator iteration and energy level iteration. By adopting the method provided by the invention, sampling points to be compared at each time are fewer and are easier to fully iterate, so the result is accurate. According to the concurrency of the multi-scale quantum harmonic oscillator multimode optimization method, the Multi-scale quantum harmonic oscillator concurrent optimization method is proposed, and the method can be concurrently operated on a cluster. By adopting the concurrent method, the operational speed is effectively accelerated.

The invention provides an improved index encryption algorithm and system. The method comprises: indexing the first encryption array based on the first index data and the first data direction, sendingthe improved index data packet containing the first index data indication information to a receiving end computer; introducing a weight configuration function, an encrypted data pointing function andan encrypted data array feedback function into a computer cloud cluster; setting a special node for simultaneously processing a computer terminal data request needing to be encrypted in a cloud cluster, a first encryption array sent by an encryption array management node and a first data direction sent by a collaborative management node; separating the multi-source data and performing ternary combination, and obtaining different components of the encryption parameters from different nodes respectively, so that a receiving end computer can prompt corresponding first index data from the hidden pointer position of the index data packet.

The invention provides a weather particle rendering method and device, computer equipment and a storage medium. The device comprises: a script execution module which is used for triggering a weather situation change signal when a weather situation change condition is satisfied, and executing a target code corresponding to target weather in a particle generation script, so as to transmit particle data corresponding to weather particles of the target weather to a particle processing module; a particle processing module which is used for performing format conversion on the particle data after receiving the particle data transmitted by the script execution module, and submitting the particle data subjected to format conversion to a graphic processing module; and the graphic processing module which is used for generating and rendering a particle instance based on the real-time position information of the virtual camera and the particle data subjected to format conversion so as to present aweather scene corresponding to the target weather in the game scene. According to the method, the rendering execution logic of the weather particles can be iterated more easily.

The invention discloses an inverse treatment planning method of a treatment plan. The method comprises the following steps of: A, inputting a medical image of a patient; B, drawing tissue outlines of a body surface, a target body and injured organs of the patient; C, setting an inverse planning target; D, creating an initial treatment plan; E, setting iterative optimization parameters; F, generating an individual treatment plan; G, calculating a dose field of the individual treatment plan; H, selecting a current optimum plan according to the dose field; I, executing the step L if a difference value between the current optimum plan and the inverse planning target is smaller than a preset first threshold value, or executing the next step; J, executing the step L if the number of times of evolution is equal to the number of frequencies of population evolution, or executing the next step; K, evolving generation to new-generation population by mating and/or variation, and executing the step G; and L, stopping the population evolution and outputting the current optimum plan. The invention also discloses a treatment planning system. By adoption of a population evolution method based on a gene algorithm, iteration is carried out by a mating or variation method, so that the whole iterative process is simple and efficient.

The present invention provides an SDK for displaying and editing three-dimensional models. At least, that SDK comprises the following step: a process design of an SDK state mechanism; Data structure design, including individual model data for objects or scenes, and work data for scenarios, decomposing logical relationships between models and scenes; The server data management module ServerData encapsulates the loading process and encapsulates the interaction logic. Based on model presentation and model interaction, the SDK designs modular model data and work data, which is independent and easyto reach, and can realize the rapid display and editing of three-dimensional model in mobile terminal.

The invention discloses an optimization method of an implicit curved surface porous structure, which comprises the following steps of: inputting an optimization implicit curved surface function expression, the implicit curved surface porous structure relative density, the particle swarm size and the maximum iteration number; initializing particles, each particle position coordinate including curvature parameter and wall thickness parameter of the implicit curved surface porous structure; calculating the constraint fitness of the iterative particle; Updating the local optimal position and the global optimal position in the particle swarm; calculating the update position of each particle by linear weighting; the curvature parameter in the particle position being the inverse number when the catastrophe requirement is satisfied; when the maximum number of iterations is exceeded, the iteration being stopped, the optimal particle position being output, and the implicit curved surface porousstructure corresponding to the optimal parameters being generated. In the method, the particle is iterated by linear weighting, and the catastrophe probability is added to ensure the random behavior of the particle. The iteration speed of the method is fast, and the curvature parameter and the wall thickness parameter are obtained by iteration, which can generate the implicit curved surface porousstructure satisfying the multi-functional requirements.

The invention discloses a GaN compatible driving circuit, and belongs to the technical field of semiconductor devices, the GaN compatible driving circuit comprises a driving chip IC, an MOS driving module and a GaN driving compatible module, the driving end of the driving chip IC is connected with the input end of the MOS driving module, the GaN driving compatible module comprises a voltage limiting unit, an acceleration unit and a voltage dividing unit, the input end of the voltage limiting unit is connected with the output end of the MOS driving module, the output end of the voltage limiting unit is connected with the input end of the voltage dividing unit and the grid electrode of the N-type field effect transistor, the output end of the voltage dividing unit is connected with the source electrode of the N-type field effect transistor and the driving chip IC, and the acceleration unit is connected with the voltage limiting unit in parallel. The method has the effect that the GaN and the silicon-based MOSFET can be driven in a compatible mode.

The invention provides a path planning control system for an easily extended stereo garage, and the system comprises a control execution module, a basic control module, a movement control module, an instruction switching module, a path planning module and a plugboard interface. The control execution module is connected with the basic control module, the movement control module, the instruction switching module and the path planning module, and the control execution module and the basic control module are solidified on one circuit board. The control execution module sequentially transmits loading code instructions to the basic control module, the movement control module, the instruction switching module and the path planning module. The basic control module stores the basic codes of the whole operation. The system can provide abundant extending possibilities for a stereo garage through the control execution module, the basic control module, the movement control module, the instruction switching module, the path planning module and the plugboard interface. Only if circuit boards inserted at the plugboard interface are increased or replaced, can a manufacturer carry out the updating of a product easily and can a user carry out the upgrading and updating of the control function of the product.

The invention discloses a modeling method for a radar complex electromagnetic environment effect mechanism. The method provided by the invention comprises the following steps: firstly, performing Bayesian network structure learning on effect data of each subsystem of a radar in an electromagnetic environment by using a K2 algorithm, and establishing a Bayesian network topological structure of a radar system from an input element to an output element; then, under the condition that the network structure is known, calculating Bayesian network parameters by adopting a maximum likelihood estimation algorithm; then, after the Bayesian network model is obtained, performing sensitivity analysis based on the conditional probability change degree on the radar electromagnetic environment effect; andfinally, performing probabilistic reasoning on the model of the radar complex electromagnetic environment effect mechanism by adopting a connection tree reasoning algorithm. According to the method,the radar complex electromagnetic environment effect mechanism can be effectively modeled, the modeling precision is 91%, and theoretical support is provided for improving the combat ability of the radar in the complex electromagnetic environment.

The invention relates to a distributed power flow optimization method based on splitting a large network into a small network, which comprises the following steps of: selecting any line as a broken line on an annular topological structure in a power system, splitting a network topological structure of the power system into a plurality of radiation type networks, and generating breakpoints at the broken line positions; convex relaxation is carried out on the mathematical optimization model of the radiation type network power system, and a distributed optimization model is constructed in combination with coupling constraints between breakpoints on each radiation type network; the distributed optimization model is solved, when the convergence condition is met, the algorithm is terminated, the iteration process is ended, and the obtained variable value is the optimal solution. According to the method, the slack domain is gradually reduced through a distributed optimization model of dividing a large net into a small net and a solving method. According to the method, iteration is simpler, optimization solution can be rapidly achieved, meanwhile, good convergence performance is achieved, and the feasibility of understanding is guaranteed; meanwhile, the distributed architecture ensures privacy in the system, and the accuracy and solving efficiency of convex relaxation are ensured by splitting a large net into a small net.