45results about How to "Fast convergence rate" patented technology

The invention discloses a bearing fault diagnosis method. The method comprises the following steps: carrying out fault feature extraction on a rolling bearing vibration signal; constructing an SVM classifier model; updating penalty parameters and kernel function parameters of a classifier of the SVM according to a training set and an IWOA algorithm; constructing a test model of the SVM according to the obtained optimal penalty parameter and the optimal kernel function parameter; and according to a test set, optimal penalty and the optimal kernel function parameters, adopting the longest optimization time, the shortest optimization time, the average optimization time, the average accuracy and the standard deviation as evaluation criteria to determine a fault result of a bearing after ten-fold cross validation. The method is high in bearing fault diagnosis capability and high in recognition accuracy.

The invention discloses a medical image registration method based on a superpixel guide, comprising the following steps: 1) rigidly pre-registering a reference image and a floating image by adopting an FFD registration algorithm based on a B-spline to obtain a pre-registered reference image and a floating image; 2) downsampling the pre-registered reference image and the floating image to form a multi-resolution pyramid with a series of l; 3) the image registration of the multi-resolution scheme starts from the coarsest resolution, and gradually registers from coarse to fine. Under the currentresolution l, after iteration, the obtained spatial transform ul is up-sampled to obtain ul-1, and then use the space transformation ul Pairs of Pyramids 1-1 level image is transformed, and the obtained image is iterated again until the end of the lowest layer of the pyramid image; 4) the final obtained spatial transformation u is applied to the floating image after rigid registration to obtain the final registration result image. The invention obtains the registration accuracy and robustness superior to the current mainstream DIR registration models such as the FFD model and the DD model based on the B-spline.

The invention discloses a mixed coherent polarization method based on phase control and polarization control. According to the method, through cascaded phase control and polarization control, high-efficiency coherent polarization synthesis of three input light beams is realized by means of an optical fiber and an optical fiber device, thereby outputting linearly polarized light. A device required for realizing the method comprises an optical fiber phase compensator, an optical fiber polarization beam combiner, an optical fiber polarization controller, an optical fiber polarization analyzer, a photoelectric detector and a synthesis controller. Based on the method, the number of the input light beams can be increased to above three. Compared with a coherent polarization synthesis method based on phase control, the mixed coherent polarization method can prevent an influence of unmatched input light beam strength and realizes high-efficiency coherent polarization synthesis.

The invention discloses a robot calibration method based on an improved differential evolution algorithm, and the robot calibration method is used for six-freedom-degree industrial robot geometric parameter calibration and establishment of a robot kinematics model. The robot calibration method comprises the steps that a robot moves in the workspace, the joint corner of the robot at each road pointposition is read and recorded, and meanwhile, the actual position coordinate of a robot tail end actuator at each road point position is measured and recorded by utilizing a laser tracker; the theoretical position coordinate of the robot is calculated; the theoretical distance and the actual distance between the road points are calculated, and an objective function is established to determine structure parameters to be identified; the structure parameters are identified by utilizing the improved differential evolution algorithm; and identified structure parameter errors are substituted in forvalidation. By means of the robot calibration method based on the improved differential evolution algorithm, a simulated annealing algorithm and a population diversity evaluation function are integrated, the defect of poor global convergence of a traditional differential evolution algorithm is overcome, the accuracy and efficiency of robot structure parameter error identification are improved, and meanwhile, the characteristic of the fast convergence rate of the differential evolution algorithm is also absorbed.

The invention discloses a nonuniformity correction method of interframe registration based on time-domain high-pass filtering. The method comprises the following steps: carrying out A / D (Analog to Digital) conversion on an original analog image acquired by an infrared focal plane detector to obtain an original digital image, then multiplying the original digital image with a corresponding gain to obtain an image, passing through a low-pass filter to obtain a filtered image, then correcting the image by a high-pass filter, correcting the image which is subjected to the time-domain high-pass filtering through a nonuniformity correction algorithm with minimum inter-frame errors, finally obtaining a corrected image. The nonuniformity correction method of the interframe registration based on the time-domain high-pass filtering disclosed by the invention has the characteristics of simplicity, high efficiency and fast convergence rate, can correct images having massive nonuniformity all around when the nonuniformalty is strong, and has good correction effects for different scenes and wide applicability.

The invention discloses a reactive equal partitioning method of alternating current / direct current mixed micro power grid based on finite time control. The method comprises following steps of step 1:establishing an alternating current / direct current mixed power grid control framework considering alternating-direct current mutual support; step 2: establishing a reactive control model of the alternating current / direct current mixed micro power grid; step 3: acquiring reactive power change quantity of the alternating current / direct current mixed micro power grid; and step 4: based on finite timeconsistency, providing an alternating current / direct current reactive equally partitioning strategy. According to the invention, by constructing the mixed power grid distributed control model considering the alternating current / direct current interaction, a one-time control model of reactive control is established, requirements of the plug and play of a distributed power supply are met and reliability of system control is improved.

The invention discloses an output feedback trajectory tracking control method with predetermined performance and dead zone input constraints for an electromechanical system. The method comprises the following steps: firstly, establishing an electromechanical system model containing dead zone input, carrying out linear processing on a dead zone model, and giving a reference tracking trajectory; secondly, defining a tracking error, introducing a predetermined performance function, and converting an original system into a system with predetermined performance constraints; in order to solve the problem that the angular position, the angular speed and the motor current in the electromechanical system are not easy to measure or the measurement precision is not high, designing a neural network state observer to estimate the state of the electromechanical system; and finally, solving the problem of complexity explosion in the traditional backstepping method by adopting a command filter technology, and constructing adaptive output feedback control in combination with an adaptive backstepping method and a neural network state observer under predetermined performance control and dead zone input constraints. The method effectively improves the transient and steady-state performance of a system with an unknown state, accurately realizes trajectory tracking, and is suitable for electromechanical system control with accurate positioning.

The invention discloses a reactive power sharing method based on limited-time control of an AC-DC hybrid microgrid, comprising the following steps: Step 1: Establish an AC-DC hybrid microgrid control framework considering AC-DC interactive support; Step 2: Establish The reactive power control model of the AC-DC hybrid microgrid; step 3: collecting the reactive power change, quantity of the AC-DC hybrid microgrid; step 4: based on the finite-time consistency AC-DC reactive power equalization strategy; The DC-interactive hybrid microgrid distributed control model establishes a primary control model for reactive power control, which satisfies the plug-and-play requirements of distributed power sources and improves the reliability of system control.

The present invention provides a hierarchical codebook structure design method of a joint method used for the wireless communication technology. The codebook designed by the invention satisfies two design principles that all code words in a same layer needs to cover a whole space, the space coverage range of any code word in any layer of a codebook must be the set of the code words in a next layer; at the same time the lobe width of a kth layer is 2 / 2k, and the codebook of a last layer uses an expression form of a channel pointing vector. In the time of generating the code words of each layer, firstly the first code word of each layer is generated, then the code word is rotated to point to different directions, and all code words of the layer can be constructed. The hierarchical codebook designed by the invention is employed to carry out searching, the convergence rate of a search method is fast, the search success rate is high, the gain of beamforming is high, the receiving power is large at any stage, the utilization rate of emitting power is high, and the robustness is good in a condition of multiple paths.

The invention relates to a hypersonic aerocraft tracking control method with an interference observer. The problem that the fact that the observer is bounded in the interference process of an observing system cannot be proved in the prior art is solved. The method provided by the invention comprises the steps that 1 a second-order system model with system interference is established according to the longitudinal input and output linearization model of a hypersonic aerocraft; 2 according to the second-order system model with system interference in the step 1, a finite time terminal sliding mode controller is designed based on a sliding mode control theory; and 3 system stability proving is carried out on the finite time terminal sliding mode controller designed in the Step 2. According to the method provided by the invention, the finite time of the sliding surface of the system is stable, and the system state is asymptotically convergent. The method provided by the invention is applied to the field of hypersonic aerocraft control.

The invention relates to a distributed and robust multi-objective power distribution and control system for isolated island microgrids, belonging to the field of smart grids. The system is composed of equipment layer, control layer and communication layer; the communication layer is composed of communication transceivers and communication connections between transceivers, which are used for information exchange between power generation units; the control layer utilizes local measurement information and information exchanged by communication networks, According to the designed controller, the control signal of the distributed power generation device is generated; the equipment layer is the physical power infrastructure, and the distributed power generation device in it responds to the control signal to realize the transmission of electric energy. Compared with the existing finite-time and fixed-time control schemes that only achieve proportional active power distribution control, the designed fast and robust control scheme realizes active power economic distribution control under multiple optimization objectives, which is conducive to improving the overall benefit. In addition, the controller can make the convergence rate of the control system faster by using the exponential function.

The invention discloses an inter-frame registration non-uniformity correction method based on subspace projection. An original digital image is obtained after the original analog image acquired by an infrared detector is subjected to A / D conversion. Then, rows and columns are subjected to preprocessing in which row mean values and column mean values are enabled to be zero. A row matrix and a column matrix are projected onto the respective subspaces. Row displacement and column displacement of two adjacent frames are obtained in respective subspaces. Finally, the obtained row (column) displacement is substituted into an inter-frame registration formula to achieve non-uniformity correction in order to obtain a corrected image. The inter-frame registration non-uniformity correction method is simple and high in efficiency, can process the non-uniformity image with a large amount of noise and obtains a better correction effect.

The invention discloses a large-scale MIMO signal detection method based on Jacobi iteration, and belongs to the technical field of wireless communication. The method comprises the steps: converting amatrix inversion process into an iteration process of matrix multiplication and matrix addition, providing the search direction for a Jacobi algorithm through a gradient algorithm and an overall correction acceleration method, and determining a correction coefficient of an iteration equation. According to the invention, the method performs the estimation of a high-dimensional matrix inversion process through an improved Jacobi iteration method, and converts the matrix inversion process into the iteration process of matrix multiplication and matrix addition, thereby greatly reducing the computing complexity. The gradient algorithm and the overall correction acceleration method are used for providing the search directions for the Jacobi algorithm and determining the correction coefficient of the iteration equation, thereby enabling the iteration convergence to be better, and enabling the convergence speed to be higher.

The invention belongs to the technical field of communication, and relates to a massive MIMO signal detection method based on improved Newton iteration. The massive multiple-input multiple-output (MIMO) system needs to deal with a lot of matrix inversion operation in the signal detection process, many researches are performed to avoid matrix inversion in recent years, and the methods can be roughly divided into two kinds: the approximation method and the iteration method. Firstly the relationship between the two kinds of methods is introduced and then the improved Newton iteration method is put forward on the basis thereof. The Newton iteration method is superior to the Newman series expansion method and the present Newton iteration method. Besides, the matrix-matrix product is transformedinto the matrix-vector product in the iteration process so as to greatly reduce the computational complexity. Finally, numerical simulation verifies the advantages of the improved Newton iteration, and the performance of the MMSE method can be achieved by a small number of iterations.

The invention discloses an infrared image non-uniformity correction method based on scene inter-frame registration. The method comprise following steps: obtaining an original digital image by performing A / D conversion to an original simulation image acquired by an infrared detector; then performing correction to the original digital image to obtain an output pixel matrix Yn-1(i, j) of a corrected (n-1)th frame and an output pixel matrix Yn(i, j) of a corrected nth frame; performing statistics to the output pixel matrix Yn-1(i, j) of the corrected (n-1)th frame and obtaining its average value averagen; finally, performing updating to o<n+1>(i, j) and g<n+1>(i, j) through the self-adjusting of the output pixel matrix Yn-1(i, j) of the corrected (n-1)th frame, the threshold value ERRn(i, j), and the convergence step length step to achieve non-uniformity correction of the whole image. The method is simple and efficient; convergence rate is high; highly non-uniform images can be corrected by the method; good correction effect can be achieved by the method in different scenes; the method has broad application.

The invention discloses a guidance law design method for unmanned aerial vehicle path tracking control. The method includes the following steps that S1, the flying speed of an unmanned aerial vehicleis V, projection is conducted on a preset track from the initial position A of the unmanned aerial vehicle, a projection point is marked as O, and the distance between the flying path of the unmannedaerial vehicle and the preset track is marked as d; S2, a circle with the radius being r is obtained with the projection point as the circle center, the intersection point of the circle and the presettrack in front of the unmanned aerial vehicle speed direction is marked as B, the length of AB is marked as L1*, and a L1* guidance law is designed with the point B as a reference point; S3, a reference circle O1* with the radius being R* is selected, and an unmanned aerial vehicle lateral acceleration instruction as* is calculated by using centripetal acceleration and motion geometrical relationship. The method has the advantage that the problem that the tracking speed is too low in tracking control in the prior art can be solved.

The invention relates to a distributed robust island microgrid multi-target power distribution and control system, and belongs to the field of intelligent power grids. The system is composed of an equipment layer, a control layer and a communication layer. The communication layer is formed by communication connection between communication transceivers and is used for information exchange between the power generation units; the control layer generates a control signal of the distributed power generation device according to a designed controller by using the local measurement information and theinformation exchanged by the communication network; the equipment layer is a physical power infrastructure, and the distributed power generation device responds to a control signal to realize transmission of power energy. Compared with the existing finite time and fixed time control scheme which only realizes proportional active power distribution control, the designed rapid robust control schemerealizes active power economic distribution control under various optimization objectives, and is beneficial to improving the total benefit. In addition, the controller can enable the convergence rate of the control system to be higher by utilizing an exponential function.

The invention discloses a nonuniformity correction method of interframe registration based on time-domain high-pass filtering. The method comprises the following steps: carrying out A / D (Analog to Digital) conversion on an original analog image acquired by an infrared focal plane detector to obtain an original digital image, then multiplying the original digital image with a corresponding gain to obtain an image, passing through a low-pass filter to obtain a filtered image, then correcting the image by a high-pass filter, correcting the image which is subjected to the time-domain high-pass filtering through a nonuniformity correction algorithm with minimum inter-frame errors, finally obtaining a corrected image. The nonuniformity correction method of the interframe registration based on the time-domain high-pass filtering disclosed by the invention has the characteristics of simplicity, high efficiency and fast convergence rate, can correct images having massive nonuniformity all around when the nonuniformalty is strong, and has good correction effects for different scenes and wide applicability.