165 results about "Matrix estimation" patented technology

The invention discloses a multi-unmanned ship cluster cooperative tracking control method based on data driving, and belongs to the field of unmanned ship cluster intelligent control. The method comprises the following steps: firstly, constructing a communication topology for cluster distributed cooperative control of multiple unmanned ships; constructing a single unmanned ship speed and course system dynamic linearization data model with interference terms based on unknown external interference and nonlinear uncertainty; constructing a smooth anti-interference trajectory tracking algorithm of a single unmanned ship tracking dynamic target point; constructing a pseudo-Jacobian matrix estimation and distributed disturbance compensation controller for distributed model-free adaptive control of the multiple unmanned ships; constructing a reset algorithm and a controller of an actual unmanned ship; and finally, multi-unmanned ship cluster cooperative tracking control is realized. The method breaks through the limitation of a traditional control method, does not need to depend on complex system model information, only utilizes input and output data of the system, and achieves a complete data-driven intelligent control strategy.

The invention relates to the technical field of pollution traceability, in particular to a harbor storage yard dust pollution traceability method and system, and the method comprises the following steps: collecting the real dust concentration and dust diffusion influence factors of different monitoring points in a harbor storage yard, and constructing a dust diffusion evaluation model; when the position of the pollution source is not determined, obtaining the position and the source intensity of the pollution source by using an improved sparrow search algorithm based on the real dust concentration, the dust diffusion influence factors and the dust diffusion evaluation model; obtaining a matrix estimation formula for defining an inverse Gaussian function according to a least square method estimation method, and further defining the inverse Gaussian function according to the dust diffusion evaluation model; when the position of the pollution source is determined, the source intensity of the pollution source is obtained based on an inverse Gaussian function according to the real dust concentration and the dust diffusion influence factors. According to the invention, a large amount of monitoring equipment which is high in cost and complex in installation and maintenance does not need to be installed, and the position and the source intensity of the pollution source of the port storage yard can be traced intelligently, accurately and efficiently.

A precoder index is derived from a precoder codebook using a low computationally complex process. A plurality of uplink channel matrix estimates are received at a Base Station (BS) from a User Equipment (UE). The plurality of uplink channel matrix estimates are processed to determine an indicator of the precoder having a highest spectral efficiency estimate. The indicator of the precoder having the highest spectral efficiency estimate is selected for transmission by the BS.

The invention discloses a GNSS (Global Navigation Satellite System) signal quality adaptive integrated navigation method, and relates to the technical field of vehicle navigation. IMU and GNSS positioning information and positioning states are read, integrated navigation is realized through an ESKF loose combination algorithm, a measurement covariance matrix is changed in real time according to the GNSS positioning state and a DOP factor, and finally, an error state value is estimated and fed back to an integrated navigation system. Position and speed information errors and attitude Euler angle errors under a navigation coordinate system are fed back to INS mechanical arrangement, angular acceleration zero offset errors of the IMU are fed back to the IMU, and updating of the state quantity is completed. A measurement covariance matrix is changed in real time according to a GNSS positioning state and a DOP factor by aiming at IMU and GNSS data through a loosely combined ESKF fusion algorithm, an error state is estimated and fed back to an integrated navigation system, and the positioning precision can be effectively improved.

The invention belongs to the field of space manipulators, and relates to a vibration suppression control method and system for a flexible base space manipulator, and the method is based on a Lagrangian method and considers multiple coupling effects to completely establish a dynamic model of the flexible base space manipulator system. An inertia matrix estimation item is constructed based on a parameter identification method to realize motion decoupling of the space manipulator, and a control law is designed based on a fast and slow subsystem theory to realize high-precision control of an assembly task of a space manipulator module and vibration suppression of a flexible base. And mechanical arm fine assembly operation control and base vibration suppression control for the flexible base space mechanical arm are achieved. The problem of how to complete a high-precision assembly task on the basis of restraining vibration of the base of the double-space mechanical arm is solved.

The invention discloses a time domain channel estimation method and device based on a multi-antenna diversity technology, and the method comprises the steps: mapping a bit stream into a symbol stream to construct a transmitting signal matrix, carrying out the STBC coding, inserting a time domain pilot signal, obtaining a matrix after insertion, selecting a vector in the matrix after insertion, and carrying out the time domain channel estimation. The method comprises the steps of obtaining a frequency domain receiving signal by combining with relevant parameters of Gaussian white noise and a Rayleigh fading channel, then performing matrix transformation to obtain an STBC decoding matrix, introducing a weight matrix to update an established MIMO-OFDM system input and output model to obtain a time domain channel estimation value, and finally performing FFT transformation on the time domain channel estimation value to obtain a frequency domain channel matrix estimation value. And substituting the frequency domain channel matrix estimation value into the STBC decoding matrix and then carrying out ML signal detection to obtain a final estimation value. A channel estimation algorithm is introduced to improve the estimation of a dynamic channel, a space-time block coding technology is combined to improve the demodulation signal precision of a MIMO system, and the full-state channel of the MIMO system is estimated.

The invention relates to the technical field of attitude control, in particular to a moving object attitude control method based on an improved LKF. Predicting prior attitude estimation at the next moment; calculating a process error covariance matrix; calculating a measurement attitude; calculating a quaternion measurement error covariance matrix; constructing feature vectors of acceleration interference and geomagnetic interference amplitude levels, and calculating an interference amplitude level value by using a Gaussian function model; updating the process error covariance matrix and the measurement error covariance matrix by using the interference amplitude level value; calculating a priori attitude estimation error covariance matrix by using the updated process error covariance matrix; calculating a Kalman gain matrix by using the priori attitude estimation error covariance matrix and the updated measurement error covariance matrix; calculating attitude posteriori estimation at the next moment; and performing attitude estimation at the next moment. According to the method, the problems of insufficient estimation precision and poor interference suppression capability of the attitude observation error covariance matrix of the existing LKF are solved.

The invention relates to the technical field of inspection management, in particular to a wharf storage yard global visual video fusion intelligent inspection method and platform, and the method specifically comprises the steps: S1, video fusion processing; s2, video preprocessing; s3, extracting and matching feature points; s4, estimating a homography matrix; s5, coordinate transformation; s6, performing image fusion processing; s7, carrying out post-treatment; according to the system and the method, the refinement level of storage yard management is improved, multiple functions such as business collaboration, commanding and dispatching, monitoring and early warning, risk control and equipment maintenance are integrated, powerful technical support is provided for safety production and management of a storage yard, and the comprehensive application technology innovation point of a video fusion technology based on a three-dimensional scene is fully embodied.

The invention discloses an electroencephalogram signal source space function connection estimation system and method based on deep learning, and the system comprises a training data set construction module which is used for constructing a training data set containing N training sample pairs, and each training sample pair comprises a sensor space function connection matrix and a source space function connection matrix corresponding to the sensor space function connection matrix; the signal preprocessing module is used for acquiring a real electroencephalogram signal and preprocessing the real electroencephalogram signal; the sensor space function connection matrix calculation module is used for calculating a corresponding sensor space function connection matrix true value; and the deep learning mapping module is used for learning a mapping relation from the sensor space function connection matrix to the source space function connection matrix, receiving a true value of the sensor space function connection matrix and outputting a predicted value of the source space function connection matrix. According to the method, the source space function connection matrix can be efficiently and accurately estimated.

Apparatuses and methods in a communication system are disclosed. A signal is received. The signal comprises demodulation reference information. The apparatus compute estimates of the channel and of interference-plus-noise vectors. The interference-plus-noise vectors are used as input for choosing a tapering matrix and for determining shrinkage target matrix and parameter. An interference plus noise covariance matrix estimator is computed and used in an equalizer, which has as input an estimate of the interference-plus-noise covariance matrix.

The present invention discloses a direction of arrival estimation method, direction of arrival estimation device, and system. The method comprises: using antennas to form a uniform linear array, establishing a received signal model based on array signal input signals, and determining a received signal matrix; performing low-rank matrix approximation on the received signal matrix using a low-rank matrix estimation algorithm to obtain a low-noise, low-dimensional matrix; updating the variable step size of a least mean square algorithm (LMS) using statistical parameters of the low-noise, low-dimensional matrix to update and iteratively calculate weight vector coefficients; and calculating the antenna array pattern and spatial spectrum based on the obtained weight vector coefficients to estimate the direction of arrival of the signal. The present invention uses the low-rank matrix approximation method to obtain a low-noise, low-dimensional matrix from the received signal matrix. Utilizing the instantaneous prediction error of the low-noise, low-dimensional matrix and the variable step size of the noise reduction signal power update algorithm, the method can ensure accurate DOA estimation even under unfavorable conditions such as low signal-to-noise ratio, a small number of antenna array elements, and a small amount of signal sampling data, due to prior noise reduction preprocessing.

The invention discloses a navigation signal blind separation method, system and equipment in an alpha pulse interference environment based on Gaussian kernel filtering, and a medium, and aims to improve the signal processing performance in a complex interference environment. According to the method, alpha pulse interference in aliasing signals is suppressed by adopting a Gaussian kernel filtering technology, and effective suppression of the alpha pulse interference is realized by reasonably designing filtering parameters; two correlation matrixes are subjected to joint diagonalization, and a joint diagonal matrix is used as an initial value of a left aliasing matrix and a right aliasing matrix in a blind separation algorithm, so that the convergence speed of aliasing matrix estimation is effectively improved; then, separating the signals after the pulse interference is suppressed by using a blind separation algorithm of joint diagonalization and an initial value of a left and right aliasing matrix; experimental results show that the method has excellent performance in the aspects of suppressing impulse noise influence and signal separation, and the problems that a traditional method is insufficient in alpha impulse interference suppression and poor in signal separation performance are effectively solved; systems, devices, and media are used to implement the methods.

The invention discloses a low-complexity channel estimation method and device based on an MIMO system, and the method comprises the steps: recovering a complete matrix through an approximate low-rank matrix, carrying out the singular value decomposition of the complete matrix, building a large-scale spatial multiplexing MIMO system model, carrying out the conversion processing of a received signal on a kth antenna, and obtaining a transmission signal estimation value, and performing channel matrix recovery on the transmission signal estimation value by using an approximation algorithm to obtain a channel matrix estimation value, finally performing signal detection on the channel matrix estimation value by using a zero-forcing technology, and performing iterative updating to obtain an optimal transmission signal estimation value, thereby effectively reducing the algorithm complexity and effectively reducing the calculation amount. The estimated channel matrix is effectively equivalent to the complete channel matrix, and the channel matrix is updated through iterative updating, so that the dynamic change of the channel can be effectively tracked, and the final demodulation signal precision is ensured.

The invention provides a joint broadband DOA and frequency band estimation method based on sparse Bayesian learning, and the method comprises the steps: calculating a covariance and a mean value of signal parameters based on obtained array receiving data through a broadband signal model defined based on sparse Bayesian learning, and calculating an intermediate parameter estimation value based on a variance vector; obtaining a binary matrix estimation value and a potential feature matrix estimation value based on the intermediate parameter estimation value, and calculating a variance vector estimation value based on the covariance and the mean value; judging whether an iteration ending requirement is met or not, if not, updating the covariance and the mean value, and further updating the binary matrix estimated value and the potential characteristic matrix estimated value until the iteration ending requirement is met so as to obtain a target binary matrix estimated value and a target potential characteristic matrix estimated value; obtaining DOA preliminary estimation results and frequency band information of the plurality of broadband signals based on the target potential feature matrix estimation value and the target binary matrix estimation value; and obtaining a DOA final estimation result of the corresponding broadband signal by adopting a maximum search criterion.

The invention discloses a large-scale MIMO multi-satellite positioning and direction estimation method and system based on angle information. The method comprises the following steps: firstly, obtaining channel matrix estimation between each satellite and a user by sending a pilot signal; then, obtaining estimated values of channel parameters by using a tensor-ESPRIT algorithm, wherein the estimated values comprise an angle of departure and an angle of arrival; according to the estimated departure angles, direction vectors between the user and all satellites are obtained, an optimization problem on spherical manifold is established, altitude constraints are considered, and a Riemann conjugate gradient method is used for solving and obtaining a user position estimation value; and finally, according to the estimated user position and arrival angle, establishing an optimization problem on the SO (3) manifold, and solving by using a Riemannian gradient descent method to obtain an estimated value of a user array rotation matrix. According to the method, positioning and direction estimation are performed by combining channel information between a plurality of satellites and users, and altitude constraints are considered during positioning, so that compared with single-satellite and non-constraint conditions, the positioning performance can be remarkably improved, and meanwhile, the method has good direction estimation performance.

The invention discloses a health degree estimation method for a pneumatic actuator. According to the invention, system identification is carried out at a stable working point and a discrete state space model is obtained. And comprehensively considering disturbance and excitation signals and the health condition of the actuator, and constructing a discrete state space expression form suitable for the augmented Kalman filter. The health information is modeled as an unknown input disturbance variable and is taken as part of a state variable. And an augmented Kalman filter is adopted to carry out combined prediction on the system state quantity and the unknown input interference quantity. And maintaining the acquired control input sequence and the unknown input disturbance variable sequence output by the filter by adopting a sliding window mechanism, and obtaining an estimated value of a health matrix of the pneumatic actuator by adopting a least square method for a data sequence in a window. And finally, performing feature extraction and analysis on the health matrix estimation value to realize real-time evaluation and dynamic tracking of the health state of the pneumatic actuator. According to the method, accurate, stable and real-time health degree information can still be provided under the condition that the pneumatic actuator has faults, disturbance or model mismatch, and powerful technical support is provided for fault-tolerant control and predictive maintenance of a complex system.

The invention discloses an apple recognition method, system and device and a storage medium, and relates to the technical field of apple recognition and detection, and the method comprises the following steps: calculating feature points on each edge contour in an original infrared thermal image and a visible light image based on the curvature of the edge contour; determining the main direction of the feature point based on the position information of each feature point on the edge contour; performing parametric coding on the position and the main direction of each feature point to generate a feature vector of each feature point; matching based on the Euclidean distance of the feature vector of each feature point on the two images to obtain a plurality of matching pairs; determining a transformation matrix through the coordinates of the plurality of matching pairs; performing image registration on the original infrared thermal image and the visible light image based on the transformation matrix to obtain a registered image; and carrying out image fusion and target identification on the registered image and the visible light image to obtain the position, the category and the confidence of the apple. According to the invention, the accuracy of transformation matrix estimation and the precision of registration are improved.

The invention discloses an inverter predictive control method based on dynamic linearization. The method comprises the following steps: representing an inverter system as a local partial format dynamic linearization linear data model; estimating a linear part of the local partial format dynamic linearization linear data model through an improved projection algorithm to obtain pseudo partial derivative matrix estimation; constructing a partial format dynamic linearization disturbance observer, estimating the gain of the partial format dynamic linearization disturbance observer based on pseudo derivative matrix estimation to obtain an observer gain matrix, and updating the partial format dynamic linearization disturbance observer; and based on a partial format dynamic linearization disturbance observer, in combination with a finite set model prediction framework, predicting the output of an inverter system in a switching tube state, and selecting the switching tube state with the minimum cost function to control the inverter. According to the invention, the method can achieve the precise prediction control of an inverter system, reduces the dependence on the load parameters of the inverter, and effectively resists the negative impact caused by the drift of the load parameters.

The application discloses a kind of sewer silting disease intelligent diagnosis method, data is collected using full-scale test, one-dimensional flow characteristic index is upgraded to two-dimensional image data with spatial information using data processing module based on gram angle and field. Data set containing pipeline silting information is constructed. Convolutional neural network module is constructed through convolution layer and pooling layer to extract spatial information in image. Time characteristics of one-dimensional data are extracted using gated recurrent unit. Adaptive matrix estimation optimizer is used to train sewer silting disease intelligent diagnosis model, and hyperparameters are automatically optimized using adaptive mutation particle swarm optimization algorithm during training process. The sewer silting disease intelligent diagnosis model is trained using optimized hyperparameters, and a diagnosis method is proposed that can intelligently predict whether pipeline silting occurs and its silting degree. The application can improve the efficiency of water pipeline silting inspection and meet the needs of modern city management.

The invention provides a water surface micro-amplitude wave detection method based on nonlinear reconstruction and a high-dimensional decoupling technology. The method is used for a water surface micro-amplitude wave detection task. According to the method, a time-frequency domain constraint module and a blind source separation precision feedback module are introduced into a U-NET adversarial network to carry out nonlinear reduction on a micro-amplitude wave signal, then feature extraction is carried out based on a subspace translation covariance matrix, a non-metric scaling technology is introduced to improve hybrid matrix estimation, and a source signal is recovered through an optimal submatrix in a sparse space. And efficient and accurate water surface micro-amplitude wave detection is realized.

The invention discloses a motion sensing game linear acceleration anti-drifting method fusing space anchor points, and belongs to the technical field of motion sensing game pose tracking. The method comprises the following steps: synchronously acquiring IMU (Inertial Measurement Unit) original data and a space anchor point absolute pose; the method comprises the following steps: preprocessing IMU data, resolving an attitude matrix in real time by utilizing the angular velocity of a gyroscope, estimating attitude, velocity and displacement, and constructing a state vector; by taking the anchor point pose as observation, through fusion of extended Kalman filtering and state estimation, the optimal estimation is corrected, and the sensor error is calibrated on line; the six-degree-of-freedom pose and the linear acceleration are output to drive a game role to interact with an object, meanwhile, the correction state is fed back to a filter to form a closed loop, integral drift is remarkably restrained, and long-term precision and robustness are improved.

The invention relates to the technical field of broadband radar signal processing, in particular to a target adaptive fusion detection method based on interference subspace suppression. Aiming at the problem that an existing broadband radar distance extension target adaptive detector is difficult to consider CFAR characteristics, detection performance and calculation complexity under a partial uniform Gaussian background, and the problem that pure clutter training data is difficult to obtain due to actual clutter heterogeneity, how to fully excavate clutter covariance matrix structure information, and how to fully excavate clutter covariance matrix structure information, the invention provides a clutter covariance matrix detection method. The requirement for the training data volume is reduced, the estimation precision of an unknown clutter covariance matrix is improved, the target adaptive fusion detection method based on interference subspace suppression in a closed form is constructed, and the detection precision is improved while the CFAR characteristics are ensured. The multi-aspect requirements of intelligent anti-interference, calculation complexity, detection performance and the like of a distance extension target adaptive detection algorithm are considered, and the adaptive detection performance of a multi-channel broadband radar on a weak and small target in a complex interference environment is improved.

The present application relates to the field of audio processing, in particular to a sound separation method and system based on a vector microphone, the method comprising: constructing a fourth-order cumulant matrix of a sound signal received by the vector microphone, solving sound source direction angle estimation and sound source pitch angle estimation of each incident signal; constructing a spatial characteristic matrix of each incident signal at different frequencies; based on the spatial characteristic matrix, calculating a spatial correlation matrix estimation of the sound signal at different frequencies and different frame numbers, and constructing a spatial correlation matrix of the sound signal at different frequencies and different frame numbers; constructing an orthogonal MNMF model; adjusting the variables of the MNMF through the multiplication update method, so that the difference between the spatial correlation matrix estimation and the spatial correlation matrix is less than a preset target, to obtain an optimal spatial correlation matrix estimation; and separating the sound signal based on the optimal spatial correlation matrix. The present application effectively solves the sound separation problem when the distance between sound sources is close, and improves the accuracy of sound separation.

The invention discloses a registration method and device based on a multi-modal image, equipment and a storage medium, and the method comprises the steps: inputting a preprocessed target 3D-C arm image, a target preoperative CT image, a first mask and a second mask into a diffusion model module, and obtaining a high-resolution CT image and a third mask; inputting the preprocessed target 3D-SPACE MR image and the preprocessed high-resolution CT image into a multi-affine matrix estimation module to obtain a multi-scale feature image; and inputting the multi-scale feature image into an affine elastic fusion module to obtain a first rigid deformation field and a first elastic deformation field, and inputting the first rigid deformation field and the first elastic deformation field into a local rigid constraint module to obtain a target registration image. The rigid-elastic hybrid registration network and the diffusion model module are combined, the 3D-C arm image and the MR image are automatically registered, and compared with the scheme that operation navigation is conducted only through the 3D-C arm image or the CT image, the comprehensive and accurate lumbosacral intervertebral foramen 3D structure can be visually displayed in the operation.

This invention discloses a method and system for separating AIS aliasing signals using a four-circular array antenna. The steps are as follows: S100: AIS signals are received through multiple antennas to obtain the spatial baseband signal matrix X, and the phase error Δφ is calculated using interval calculation. k S200: Calculate the phase compensation matrix p; S200: Compensate the baseband signal according to the phase compensation matrix to obtain the baseband signal X. comp S300: For baseband signal X comp By expanding the spatial and temporal dimensions, we obtain the spatiotemporal snapshot matrix X. st (t); S400: The covariance matrix R is obtained by estimating the spatiotemporal snapshot matrix. xx S500: For the covariance matrix R xx Eigenvalue decomposition is performed to obtain the signal subspace and noise subspace; S600: DOA estimation is performed using the MUSIC algorithm to construct the spatial spectrum P(θ); S700: The beam w at the optimal solution is obtained using the space-time constrained minimum variance algorithm; S800: The separated signal is output through the sparse separation method of protocol features. This method can significantly improve the availability of AIS self-organizing network and provide more reliable protection for ship navigation safety.

The invention discloses a control method and device for an active folding arm deformation aircraft, a medium and a product, and relates to the field of aircraft control, and the method comprises the steps: obtaining a real-time output angle value of a servo motor through PID control based on expected deformation configuration; designing an inertia self-adaptive estimation mechanism based on a coordinate system represented by kinematics and kinetic equations and a variable theoretical value during attitude determination, determining a current inertia matrix estimation value by adopting a recursive least square method, adding a kinetic model, and setting a sliding mode backstepping control rate; damping and feed-forward compensation optimization are carried out by adopting a set control rate, a current rotating speed value of the power servo motor is determined by combining a time-varying distribution matrix, an actual position and an attitude angle are determined by considering a real-time output angle value, and the actual position and the attitude angle are fed back to a position and attitude controller. According to the method, the control rate and the feedback term related to the inertia in the controller design can be corrected in real time, and the sudden change problem of the inertia is solved, so that the adaptability to the uncertainty of the aircraft structure is effectively enhanced.

The present application belongs to the technical field of distributed speech enhancement, and particularly relates to a distributed speech enhancement system based on maximum likelihood. In order to expand the diversity of speech enhancement technology in WASN and complete good noise elimination performance, the system comprises a discrete Fourier transform module, a speech activity detection module, a steering vector estimation module, a data compression module, a result output module, a signal construction module, a weighted correlation matrix estimation module, a filter update module, and a discrete inverse Fourier transform module. The present application is a distributed speech enhancement technology which can be applied to a wireless acoustic sensor network without a data processing center. The technology estimates a weighted correlation matrix through a local signal constructed by a node and a variance of an output result, and updates a filter by combining the estimated weighted correlation matrix with a constructed local steering vector, so as to complete distributed speech enhancement.

The invention discloses a uniform circular array robust underwater target azimuth tracking method and system, which adopts a P-element uniform circular array with a radius of r to receive narrowband acoustic signals radiated by a target; each sensor on the uniform circular array converts the received underwater acoustic signal into an electrical signal, sets the time step k=0, initializes the target state estimation value, the root mean square error matrix, the degree of freedom estimation value, and the inverse scale matrix estimation value; calculates the target state estimation value of the kth time step according to the one-step predicted target state of the kth time step, the one-step predicted root mean square error matrix of the kth time step, the degree of freedom estimation value, and the inverse scale matrix estimation value; takes the first target azimuth estimation value of the target state estimation value of the kth time step as the underwater target azimuth tracking result of the kth time step, sets the time step k=k+1, continuously tracks the underwater target azimuth, and completes the uniform circular array robust underwater target azimuth tracking. High-precision, robust and efficient target azimuth tracking is achieved.

The invention provides a graph neural network training method and device, a medium, equipment and a program product, and the method comprises the steps: obtaining sample graph data, and enabling a plurality of nodes of the sample graph data to comprise a plurality of nodes with noise marks; iteratively training the graph neural network and the matrix estimator based on the sample graph data and the noise marks; in any round of iteration, the sample graph data and the noise marks are input into the graph neural network, so that the graph neural network outputs original prediction marks of all the nodes in the multiple nodes, the original prediction marks are corrected based on the current noise transfer matrix, and corrected prediction marks are obtained; performing parameter adjustment on the graph neural network based on the corrected prediction mark and the noise mark; and after parameter adjustment of the graph neural network is completed, inputting the original prediction mark and the noise mark into a matrix estimator, so that the matrix estimator outputs a noise transfer matrix, and performing parameter adjustment on the matrix estimator based on the noise transfer matrix of the adjacent nodes.

The speech presence probability calculation method, system, speech enhancement method and headphones provided in this specification correct the speech presence probability and speech non-existence probability in the iterative process by comparing the entropy of the speech presence probability and the entropy of the speech non-existence probability to obtain faster convergence speed and better convergence results, thereby making the speech presence probability and noise covariance matrix estimation more accurate, thereby improving the speech enhancement effect of MVDR.