7 results about "Pattern search algorithm" patented technology

The invention discloses a satellite orbit thrust section inversion method based on pattern search, solves the problem of how to accurately invert parameters of a middle unknown thrust section by using two extremely short radar observation arc sections, and belongs to the field of satellite orbit thrust section inversion. Comprising the following steps: finding epoch moments of two observation arc segments; using a Laplace-LM algorithm to obtain an initial orbital element, and substituting the initial orbital element into POD to obtain a precise orbital element; constructing a satellite orbit dynamic model based on the initial dynamic parameters, and performing orbit propagation on a first precision orbit element of the first observation arc section from the first epoch moment to a second epoch moment of the second observation arc section to obtain a second propagation precision orbit element; constructing a target function according to the position difference between the second propagation precision orbit element and the second precision orbit determination result; introducing a pattern search algorithm to obtain a target dynamic parameter when the target function is minimum; according to the method, the thrust section can be accurately inverted, and the real propulsion strategy corresponding to the target dynamic parameter is summarized.

The invention relates to the technical field of power electronic devices, in particular to a flexible interconnection system control parameter optimization method and a related device, and the method comprises the steps: obtaining the response time of a flexible interconnection system; inputting the response time into a pre-trained first data-driven proxy model, and outputting to obtain a target control parameter; the first data-driven proxy model represents a mapping relationship between the target control parameter and the response time; inputting the target control parameter into a pre-trained second data-driven agent model, and performing optimization operation on the second data-driven agent model by adopting a mode search algorithm by taking the minimum overshoot as an optimization target and taking the condition that the frequency change rate is in a preset threshold range as a constraint condition, so as to obtain an optimal target control parameter; the second data driving agent model represents the mapping relation between the target control parameter and the overshoot and the frequency change rate. The technical problems that an existing flexible interconnection system control parameter setting method is low in efficiency and poor in effect are solved.

In various examples, query-based evaluation of error patterns in annotated training data for machine learning systems and applications is provided. An error pattern evaluator may include an error pattern extraction function and an error pattern search algorithm. An annotated validation sample associated with an inaccurate prediction from a machine learning model may be analyzed to extract an error pattern. A set of annotated training data used for training the machine learning model may be analyzed by the error pattern extraction function to generate an error pattern for individual samples of annotated training data to define a search dataset. An error pattern search algorithm may use the error pattern derived from the inaccurate prediction as a query to search the dataset to identify each occurrence of samples having an error pattern similar to the query error pattern. The set of annotated training data may be corrected based on the search results.

This invention discloses a deterministic real-time optimization modeling method, system, electronic device, and storage medium for high-dimensional process parameters, belonging to the field of intelligent manufacturing and process optimization technology. The method includes: constructing a neural network with initially optimized weights using a genetic algorithm as a discriminator; training a generative adversarial network as a generator to determine a deterministic starting point based on target material state data; analyzing the physical influence weights of each process parameter on the material state data from the discriminator, and using this to guide an improved pattern search algorithm for targeted local search and iterative optimization, quickly converging to the globally optimal combination of process parameters. The system includes corresponding functional modules and can be encapsulated as intelligent decision-making micro-units. This invention completely eliminates the randomness and unpredictability of traditional optimization methods, achieving second-level, stable, and high-precision process reverse design, and is particularly suitable for offline R&D and online real-time closed-loop control of integrated multi-stage material preparation and processing processes.

The invention discloses a sideband vibration suppression method and system for a common-mode voltage injection permanent magnet synchronous motor, and the method comprises the following steps: S1, calculating the injection range of a sideband common-mode voltage and the frequency component of the injected sideband harmonic voltage according to a sideband harmonic voltage model; s2, establishing a sideband vibration model under sideband common-mode voltage injection by combining the sideband electromagnetic force and the vibration transfer function; s3, taking the output of the sideband vibration model as a target function, taking minimization of a high-frequency vibration peak value of the motor as a target, and iteratively searching an optimal DC bias M0 and a third harmonic modulation ratio M3 based on a mode search algorithm; and S4, adjusting the distribution proportion of zero vectors SV0 and SV7 in space vector modulation SVPWM according to the optimal direct current bias M0 and the third harmonic modulation ratio M3 obtained by searching, and realizing common-mode voltage injection. According to the method, the common-mode voltage which has no influence on the fundamental frequency current is injected into the closed-winding permanent magnet synchronous motor, so that the high-frequency vibration of the motor caused by resonance is effectively inhibited.

The application discloses a satellite orbit thrust segment inversion method based on a mode search, solves the problem of how to accurately invert the parameters of an unknown thrust segment in the middle by using two extremely short radar observation arc segments, and belongs to the field of satellite orbit thrust segment inversion. The application comprises the following steps: finding the epoch time of the two observation arc segments; obtaining initial orbit elements by using a Laplace-LM algorithm and substituting the initial orbit elements into POD to obtain precise orbit elements; constructing a satellite orbit dynamics model based on initial dynamic parameters, performing orbit propagation on the first precise orbit elements of the first observation arc segment from the first epoch time to the second epoch time of the second observation arc segment to obtain second propagated precise orbit elements; constructing an objective function according to the position difference between the second propagated precise orbit elements and the second precise orbit determination result; and introducing a mode search algorithm to obtain target dynamic parameters corresponding to the minimum value of the objective function.

The application provides a dual-sector permanent magnet synchronous motor sideband vibration suppression method and system based on pattern search, a sideband vibration model is established according to the side frequency electromagnetic force and a vibration transfer function; carrier phase shift angles of two sectors are iteratively searched respectively, a minimum value of a corresponding system sideband vibration peak value is found by searching the carrier phase difference between the sectors as an optimization target; the global optimal carrier phase shift angle is determined after the comparison process through parallel search, and the optimal carrier phase shift angle is finally output. The application utilizes the characteristics of the independence of the carrier phase of each sector of the dual-sector motor, optimizes the carrier phase shift angle, changes the spatial order of the high-frequency side frequency electromagnetic force, makes the natural frequency of the corresponding order far away from the frequency of the high-frequency side frequency electromagnetic force, and realizes the suppression of vibration; for different motor vibration characteristics and working conditions, the optimal target carrier phase shift angle can be iteratively searched through the pattern search algorithm, and the effective suppression of motor high-frequency vibration is realized.