82 results about "Augmented matrix" patented technology

The invention discloses a fault moment determining and fault region location method based on a random matrix theory, and the method comprises the steps: obtaining the PMU data and signal to noise ratio of each node in a power system in a time period T, obtaining an original data matrix according to the PMU data of each node, carrying out the standardization processing of the original data matrix,and then obtaining the mean spectral radius at each moment of the time period T through a monocyclic theorem; obtaining an augmented matrix and a reference augmented matrix of each node according to the signal to noise ratio and the original data matrix, obtaining the mean spectral radius difference and mean spectral radius integral of the augmented matrix and the reference augmented matrix of each node in the time period T through the monocyclic theorem, wherein the moment when the mean spectral radius at each moment of the time period T is less than a normal operation value of the mean spectral radius is determined as a fault moment, and a node with the largest mean spectral radius in the nodes with the difference of the mean spectral radiuses of all nodes in the time period T being greater than a critical value is determined as a fault region. The method cannot be affected by bad data.

The invention provides a machine learning method and a machine learning device. The machine learning method includes the steps: constructing an original input matrix and an original output matrix according to an acquired training sample set; constructing a mapping feature node matrix based on the original input matrix by using a first random weight and a first random number; constructing an enhanced node matrix based on the mapping feature node matrix by using a second random weight and a second random number; and according to an augmented matrix composed of the mapping feature node matrix andthe enhanced node matrix and the original output matrix, determining to connect the weight matrix. In the machine learning method and the machine learning device of the present invention, not only the factor of direct connection between the input layer and the output layer in the neural network is added, but also more factors of the hidden layer between the input layer and the output layer is added, so that the machine learning method and the machine learning device have the advantages of satisfying the computational efficiency and guaranteeing the learning accuracy of a wide learning method.

Banding artifacts in steady state free precession (SSFP) magnetic resonance imaging (MRI) are reduced by acquiring and combining multiple SSFP images in an augmented matrix where an acquisition vector in k-space is equal to a Fourier matrix of the trajectory on a known distortion operator for each trajectory in k-space.

The invention discloses a network link performance measurement method. A uniform performance analyzing linear model is established by analyzing the characteristics of network link performance; in the linear model, when rank of a route matrix is not equal to that of an augmented matrix and optimal solution of nonhomogeneous linear equation system can not be obtained, the problem is transformed to a multiobjective optimal problem by mathematic transformation, and genetic algorithm is used for solving sub-optimal solution of the nonhomogeneous linear equation system; and finally the network link performance is obtained by statistic probability distribution of sub-optimal solution in discretization interval. The invention uniforms the network link performance analyzing model as a linear analyzing model, transforms the link performance measurement problem to the multiobjective optimal problem, utilizes the genetic algorithm to solve the sub-optimal solution of the interior network link performance, and obtains link performance parameters by analyzing, such as link delay time and loss rate.

The invention discloses an MIMO detecting preprocessing device and method. The device comprises a channel transfer matrix input cache module, an MMSE-SQRD preprocessing core module, an upper triangular matrix R cache module, a sorting information cache module, a unitary matrix Q cache module, a product computing module and a product computing result cache module. According to the MIMO detecting preprocessing device and method, QR decomposition of an augmented matrix of a channel transfer matrix received by an MIMO receiving end is achieved, sorting operation is added in the matrix QR decomposition process, a unitary matrix and an upper triangular matrix after QR decomposition are obtained, the sorting information vectors of the matrixes are obtained, the device and method are suitable for fields such as MIMO detecting, and the obtained unitary matrix and the obtained upper triangular matrix after QR decomposition can be effectively reused in next-level MMSE and K-Best detecting.

The invention discloses a linear minimum mean square error (LMMSE) detection method for multiple input multiple output-orthogonal frequency division multiplexing (MIMO-OFDM). The method comprises the following steps of: presetting a detection systolic array by adopting an estimated noise level value obtained from a channel estimation module, performing Givens rotation on a multi-antenna channel matrix, simultaneously performing unitary transformation on data received by a plurality of antenna, and finally solving linear equations by using a back substitution method to obtain an estimated value of a space division multiplexing signal. The processing method is implemented according to a sub-carrier sequence. Compared with the conventional methods, the LMMSE detection method for the MIMO-OFDM has the advantages that: a calculated amount is far less than that of the conventional systolic array Givens rotation-based quick response (QR) decomposition performed on the multi-antenna channel augmented matrix, and the processing time is relatively shorter.

The invention provides a minimum miscible phase pressure computing method of low-permeability reservoir CO2 flooding. The minimum miscible phase pressure computing method of low-permeability reservoirCO2 flooding comprises the following steps that minimum miscible phase pressure and underlying parameters of four or more samples are obtained by adopting a slim-tube displacement experimental method; the underlying parameters of the four or more samples are substituted through a theoretical formula method to calculate and obtain the minimum miscible phase pressure ; the minimum miscible phase pressure error of a same sample is obtained by comparing the slim-tube displacement experimental method and the theoretical formula method, and empirical constants are obtained by substituting the underlying parameters of four selected samples into a logarithm formula and establishing an augmented matrix; and a computational formula model is obtained by fitting the minimum miscible phase pressure ina low permeability reservoir and the underlying parameters. According to the minimum miscible phase pressure computing method of the low-permeability reservoir CO2 flooding, the minimum miscible pressure determined by the slim-tube displacement experimental method is more accurate, the theoretical formula method is more convenient and fast, and the cost is lower; at the same time, calculation requirements of the low-permeability reservoir in the Northern Jiangsu Basin can be met, a new derivation computational formula is more accurate in computational accuracy, the theoretical basis is higher, the labor intensity is low, and the minimum miscible pressure computing method of the low-permeability reservoir CO2 flooding is suitable for the use of low-permeability reservoir in the Northern Jiangsu Basin.

The invention discloses a dynamic process monitoring method based on a weighted dynamic distributed PCA model. The method is to solve a problem how to effectively describe the dynamic characteristics of each measured variable for the complex dynamic characteristics of modern industrial process data and to establish a dynamic distributed monitoring model on this basis. The method weights each variable in a augmented matrix by using a correlation coefficient between each measured variable and other different delayed measurement values so that the weighted training data can better reflect the dynamic relation of the corresponding measured variable. The PCA model established on this basis can better excavate the hidden information related to each measured variable, and the interpretability of the model can be further improved. Compared with a traditional method, despite establishing a PCA fault detection model by using all different delay variables, the method assigns larger weights to the variables with large correlation and assigns smaller weights to the variables with small correlation, which not only prevents information loss to the utmost extent, but also highlights the process variables with strong correlation because of different weight values while suppressing the interference of irrelevant variables. Therefore, the dynamic process monitoring method based on a weighted dynamic distributed PCA model can obtain a superior fault detection effect.

The invention discloses a batch process weak fault monitoring model establishment method based on tendency turning points, comprising the following steps: 1, acquiring I batches of normal operating data; 2, sectioning the batch process; 3, extracting turning points by adopting a method based on the tendency turning point, and ranking the turning points according to the orthogonal distance; 4, checking the turning points extracted with variables among the batches; 5, aggregating the tendency turning points of all variables in the batches and verifying the truth or falsity; 6, constructing a time sequence augmented matrix by using a time lag expanded analysis data array, and performing PCA (Principal Component Analysis) on the time sequence augmented matrix; 7, clustering all established models according to the similarity of the load direction at each phase; and 8, calculating T2 and SPE (Synergistic Processing Element) statistics and a control limit of each type. The invention further provides a batch process weak fault monitoring method based on tendency turning points.

The invention discloses an array antenna directional diagram fitting method based on a weighted normal regression method. Quick integration of a directional diagram is realized based on a weighted normal regression technology. The array antenna directional diagram fitting method mainly comprises the following steps: firstly, simulating an antenna array by using electromagnetic full-wave simulation software, extracting an array element active directional diagram, and building an array manifold: for an array requiring quick integration, only a middle array element is extracted, and a wave path-difference is introduced according to mutual positions of the array elements in the array to build the array manifold; building a desired directional diagram vector and a weighted matrix according to an integration demand; constructing an augmented matrix by using the weighted array manifold and the desired directional diagram vector, performing singular value decomposition, and extracting a feature vector of the minimum singular value; and finally, obtaining an exciting vector I of an array element port according to the feature vector to complete an array directional diagram integration. According to the array antenna directional diagram fitting method based on the weighted normal regression method, the problem that the conventional direct solving technology for directional diagram integration has difficulty in completing integration of irregular arrays and the problem that a bionic optimization algorithm is long in time consumption for large-size array integration are solved.

The invention discloses a power network weak point assessment method based on the high-dimensional random matrix theory. The method comprises the following steps: a node is randomly selected as a disturbed node from the normal operation power network system; a preset time disturbance is applied to the disturbed node in the test time, and voltage data of the power network system operation in the test time is collected; a preset sliding time window is used to truncate the voltage data into a plurality of original data matrices, and based on the original data matrices, an augmented matrix corresponding to each node of the disturbed power system except the disturbed node is constructed. Based on M-P rate, ring rate and linear eigenvalue statistic, the augmented matrix is processed to obtain the empirical spectral distribution and average spectral radius. According to the empirical spectral distribution and average spectral radius value, the comprehensive evaluation index of weak points ofthe disturbed node power network is calculated. By analogy, the other nodes of the power system are calculated as the disturbed nodes to evaluate the weakness of the power system. The method has highaccuracy and can avoid misjudgment.

The invention provides a modeling method capable of improving the accuracy of qualitative near-infrared spectroscopic analysis. The modeling method comprises the following steps: dividing acquired near infrared spectra and corresponding class target values into a training set and a prediction set; decomposing the spectra in the training set by using empirical mode decomposition (EMD); dividing IMFs and residual error matrixes obtained through decomposition into a high-frequency matrix and a low-frequency matrix according to frequency; expanding the two matrixes into augmented matrixes along the direction of a variable; converting a single row of class target values of various classes into a plurality of rows of target value matrixes of two classes; respectively building models for the augmented matrixes and a plurality of rows of target values; processing the spectra and target values of the prediction set as the spectra and target values of the training set and then substituting the processed spectra and target values into the models for prediction; and converting the obtained multiple rows of predicted target value matrixes of two classes into a single-row multi-class predicted value vector. The method expands the spectra into the augmented matrixes through empirical mode decomposition, so the accuracy of qualitative analysis is greatly improved. The modeling method is applicable to the field of analytical chemistry.

The invention discloses a meter-wave radar low elevation estimating method based on a minimum redundancy linear sparse submatrix. The meter-wave radar low elevation estimating method based on the minimum redundancy linear sparse submatrix mainly solves the problem that errors of estimation of meter-wave radar low elevations are large in the prior art. The meter-wave radar low elevation estimating method based on the minimum redundancy linear sparse submatrix comprises the implementation steps of (1) structuring a minimum redundancy linear sparse submatrix meter-wave radar, (2) extracting target signals from radar echoes, (3) calculating auto-covariance matrixes of submatrixes and cross covariance matrixes among the submatrixes, (4) structuring an augmented matrix of whole array data covariance matrixes, (5) restoring the rank of the augmented matrix by applying a spatial smoothing algorithm of distributed submatrixes, (6) carrying out characteristic decomposition on the covariance matrixes to obtain signal subspaces, (7) obtaining direction cosine non-fuzziness coarse estimation, (8) obtaining direction cosine fuzziness fine estimation, and (9) solving the fuzziness of the fine estimation by using the coarse estimation to obtain low elevation estimation with high precision and without fuzziness. According to the meter-wave radar low elevation estimating method based on the minimum redundancy linear sparse submatrix, the aperture of the meter-wave radar is expanded, the threshold of the signal to noise ratio is lowered, the precision of the lower elevation estimation is improved, and the method can be used for positioning and tracking targets.

The invention provides a method for quickly solving a nodal impedance matrix of a power system, and relates to the field of analytical computation of the power system. The method mainly comprises the following steps of inputting data of a nodal admittance matrix Y; establishing an augmented matrix B by the nodal admittance matrix Y and an identity matrix E together; normalizing the augmented matrix B and carrying out a Gauss-Jordan elimination method on the augmented matrix B for n times; obtaining an inverse matrix Z. At present, traditional methods for solving the nodal impedance matrix comprise an LDU (Logic Data Unit) triangular decomposition method and a Gauss elimination method, and compared with the two traditional methods, the novel method for quickly solving the nodal impedance matrix by utilizing the Gauss-Jordan elimination method, provided by the invention, has the advantages that the principle is simple and easy to understand, the computation time is reduced, the programming is convenient, and the like; compared with the traditional LDU triangular decomposition method and the Gauss elimination method, by utilizing the method for verifying systems such as an IEEE-57 node, an IEEE-118 node and an IEEE-300 node, the computation speeds can be respectively increased by about 25%-50%.

The invention belongs to the field of power system analysis and computing and discloses a method for solving a node impedance matrix of an electric system on the basis of a Gaussian elimination method of a sparse symmetric matrix technology. The method mainly comprises the following steps that a node admittance matrix Y is formed; the matrix Y and a matrix En form an augmented matrix Bn=[YEn]; elimination is carried out on the matrix Bn according to the spare symmetry to obtain Bn(n-1)'=[Y(n-1)'En(n-1)']; according to Y(n-1)'Zn=En(n-1)', sparseness and symmetry, elements above and on the left of a diagonal element Znn of a matrix Zn are solved; a matrix Y(k-1)' is obtained according to the Y(n-1)'; elements above and on the left of a diagonal element Zkk of the matrix Zk are obtained according to Y(k-1)'Zk=Ek(k-1)', sparseness and symmetry. By the utilization of the symmetric sparseness, all invalid computation of the previous generation process is avoided, and computation of about 50% of nonzero elements is reduced; by the utilization of the characteristics of the E matrix element structure and the sparseness of upper triangle elements, the elements of the matrix Zk are obtained in a back substitution mode according to a symmetry mode, and back substitution computation is greatly accelerated. The method can check IEEE-30, -57 and -118 node systems and the like, and the computation speed for the IEEE-118 node system can be improved by 96-97% compared with a traditional Gaussian elimination method and an LDU triangular decomposition method.

The invention discloses a spectrum measurement device based on the electro-optical effect. The spectrum measurement device based on the electro-optical effect comprises a first polarizing film, an electro-optical effect crystal, a second polarizing film and an optical detector all of which are sequentially arranged in the direction of incident light, wherein the polarizing direction of the first polarizing film is not parallel with the induction spindle direction of the electro-optical effect crystal under an external electric field. The invention further discloses a spectrum measurement method using the spectrum measurement device based on the electro-optical effect. The method comprises the steps that firstly, luminous power detected by the optical detector when different external impressed voltages are applied on the electro-optical effect crystal is measured, the obtained data of the luminous power serve as an augmented matrix, the augmented matrix is combined with a coefficient matrix formed by detectivity of the incident light of different frequencies from the spectrum measurement device under the different external impressed voltages, and then a linear system of equations is built; the linear system of equations is solved, the luminous power of each frequency component in the incident light to be detected is obtained, linear fitting and radiometric calibration are carried out on the luminous power, and the spectrum of the incident light to be detected is obtained. The spectrum measurement device based on the electro-optical effect and the spectrum measurement method have the advantages of being high in vibration resistance, high in resolution ratio, wide in spectrum measurement range and the like.

The invention discloses a uniform channel decomposition method for a vertical layered space-time coding communication system which mainly solves the problem that the traditional uniform channel decomposition method is greatly influenced by channel errors. The decomposition method is as follows: (1) a transmitting terminal carries out singular value decomposition on an obtained channel matrix*with errors; (2) the transmitting terminal and a receiving terminal respectively build an augmented matrix J by utilizing a unitary matrix*decomposed by the singular value; (3) the transmitting terminal and the receiving terminal respectively carry out geometric mean decomposition on the augmented matrix J; (4) the transmitting terminal designs a pretreatment matrix F according to a diagonal matrix*obtained according to the singular value decomposition and a unitary matrix P<J>, and the pretreatment matrix pretreats transmitting signals; (5) the receiving end obtains column vectors of an after-treatment matrix W according to results decomposed the geometric mean; (6) the receiving end decodes vectors y of the receiving signals by adopting a successive interference cancellation method. The method has the advantages of low bit error rate and high system volume and is used for the design of a transceiver of a multi-input multi-output communication system.

The invention belongs to the technical field of flotation foam industrial ore identification, in particular to a flotation foam working condition identification method based on a co-occurrence augmentation matrix in a dual-tree complex wavelet domain. The method comprises the following steps: carrying out dual-tree complex wavelet transform on an image to extract high-low frequency sub-images of the image; carrying out calculation of a gray level co-occurrence augmentation matrix of each subimage based on the dual-tree complex wavelet transform; calculating eigenvalues of each augmented matrix; building the identification model of flotation industry and ore; taking the eigenvalues of augmented matrix as input eigenvectors of the flotation identification model to identify flotation conditions. The method can accurately and quickly realize the working condition identification of the flotation foam image, avoids the subjectivity and randomness of the manual observation, provides the possibility for the optimization control of the flotation production, ensures the economic benefit and production efficiency of the enterprise, and guarantees the sustainable development of the mineral resources.

The invention relates to a near-infrared brain function signal processing method, in particular to a near-infrared brain function signal processing method based on differential pathlength factor estimation. The invention is intended to solve the problem of the prior art that since a big difference exists between a differential pathlength factor reference value used in modified Lambert-Beer's law and a real differential pathlength factor of an actual measurement object and measurement error disturbances also occur in time series signals of light intensity variation acquired by a light source detector, measurement and extraction of continuous wave near-infrared brain functional activity response signals are low in precision. The method of the invention includes: acquiring a time signal of light intensity variation under the fact that near-infrared rays of different wavelengths have an equal distance away from the detector; using modified Lambert-Beer's law to construct equations for thesignal; modifying the equations in matrix form; subjecting an augmented matrix to singular value decomposition so as to obtain a total least squares solution to concentration variation time signals ofoxyhemoglobin and reduced hemoglobin at the detector. The method of the invention is applicable to the field of brain function signals.

Banding artifacts in steady state free precession (SSFP) magnetic resonance imaging (MRI) are reduced by acquiring and combining multiple SSFP images in an augmented matrix where an acquisition vector in k-space is equal to a Fourier matrix of the trajectory on a known distortion operator for each trajectory in k-space.

The invention belongs to the field of process monitoring, particularly relates to a detection method for abnormal state of a multi-mode industrial process, a system and equipment, and aims to solve the problem of insufficient detection precision of an abnormal production state reflected by an abnormal change trend of a monitored multi-dimensional variable in a complex industrial production process. The method comprises the following steps: acquiring state monitoring data; preprocessing the state monitoring data to obtain preprocessed data; segmenting the preprocessed data, and constructing an augmented matrix used for describing the state of the production process at the t moment as a first matrix; calculating detection results under set abnormal indexes set and sbt based on errors between the matrixes; carrying out filtering processing on the obtained detection result; and if the value corresponding to the filtered detection result is greater than any one of two pre-acquired alarm thresholds, performing early warning. According to the invention, the detection precision of abnormal state detection in the multi-mode industrial process is improved.

The invention discloses a tracking controller design method for packet loss network control systems. According to the method, a discrete linear system model with observation is used for describing network control systems which have packet loss caused by equipment failure, network congestion, routing information error and the like; an augmented matrix method is used for establishing an augmented system model related to a reference trajectory; under a limited time domain, an optimum output feedback tracking controller design scheme related to both a state and the reference trajectory is put forward on the basis of a new maximum principle, a differential Riccati equation, a Calman filtering theory and a Lyapunov theory, and optimum estimation of the state is given; and in addition, under a limitless time domain, an optimum tracking controller design method capable of stabilizing systems is put forward according to the Lyapunov theory and an algebra Riccati equation. According to the method, a controller design method with an output feedback form is put forward, so that the stability of systems is ensured and favorable tracking performance is provided.

The invention discloses a method for diagnosing a metering fault of an electric energy meter in a limited range based on high-frequency current sampling, relates to the technical field of electric energy metering, and solves the problem that the electric energy meter error estimation precision or hit rate is seriously limited by loss estimation precision in the prior art. According to the technical scheme, the method comprises the following steps of: collecting current instantaneous values of a general meter and branch meters; obtaining a total sample space after multiple times of acquisition; establishing a coefficient matrix and an augmented matrix, and judging whether the ranks of the coefficient matrix and the augmented matrix are equal or not; if the collected data is invalid, collecting at least one group of data again, and discharging the same group number of data to obtain a new overall sample space until the ranks of the coefficient matrix and the augmented matrix are equal; and establishing an error calculation model, solving the error calculation model, and comparing the result with a precision grade threshold of the electric energy meter to obtain a fault diagnosis result. Accurate calculation of the error of an electric energy meter in operation can be realized, the electric energy meter error estimation precision and hit rate can be effectively improved, and human factor interference can be effectively eliminated.

The invention discloses a method of quickly solving a node impedance matrix of an electric power system based on Gaussian elimination method, relating to the field of analytical calculation of the electric power system. The method comprises the following main steps of inputting data of a Y array of a node admittance matrix; forming an augmented matrix Bn=[YEn] by utilizing the last column of En arrays of the Y array and the E array; carrying out the (n-1)th normalized Gaussian elimination on the Bn array to obtain Bn(n-1)'=[Y(n-1) 'En']; solving the Zn array in the Z array of the node admittance matrix according to Y(n-1)'Zn=En'; obtaining all elements at the left side of Znn according to symmetry; solving diagonal elements of the Zk array of (n-1)th column to first column of the Z array and the elements according to Y(k-1)]'Zk=Ek' in a back substitution manner; obtaining all elements at the left side of Zkk according to symmetry. The method has the advantages of simple principle and clear principle, the Z array is solved by the Y array of IEEE-57, IEEE-188 and IEEE-300 node systems, and in comparison with that of an LDU (Logical Data Unit) triangular decomposition method and a method without the normalized Gaussian elimination, the computation speed is greatly improved.

The invention discloses a gear pair meshing frequency generation method for gear box fault diagnosis and research. The conventional gear box fault diagnosis has the defects of excessively long preparation time and excessively low efficiency. The gear pair meshing frequency generation method comprises the following steps of: firstly, confirming the types and number of gear pairs in a gear box and the connective relationships among the gear pairs; secondly, generating an augmented matrix of equation sets related to rotating frequencies of all rotating elements in all the gear pairs according to the types and number of the gear pairs and the connective relationships among the gear pairs as well as initial conditions; and thirdly, solving the equation sets through a Gaussian elimination method to obtain the rotating frequencies of all the rotating elements in all the gear pairs, and then generating the meshing frequencies of the gear pairs according to the relationships between the meshing frequencies of the gear pairs and the rotating frequencies of the related rotating elements forming the gear pairs. The gear pair meshing frequency generation method can be used for rapidly generating the meshing frequency of each gear pair in any complex gear system, thus laying a foundation for the gear box fault diagnosis and research.

The invention discloses an operation state prediction and evaluation method for a high-power integrated fuel cell system, and particularly relates to the field of state evaluation of a proton exchangemembrane fuel cell system. The method specifically comprises the steps: firstly, conducting characteristic signal conversion according to a monitoring signal set of the high-power integrated fuel cell system, and establishing an original random matrix of the system running state; then, carrying out original random matrix dimension expansion based on random tensor augmentation to further mine feature information; on the basis, achieving efficient recursive updating of characteristic values of the covariance matrix corresponding to the tensor augmented state matrix on the basis of the rank-to-one transformation principle; and finally, constructing a performance index based on a nonlinear test function, and judging and realizing a state prediction based on a system tensor augmented matrix and a state evaluation verification process based on a system original characteristic matrix according to a statistical threshold, thereby realizing effective prediction and evaluation of the operationstate of the high-power integrated fuel cell system.

The invention discloses a drilling process abnormity early warning model based on dynamic principal component analysis. The method comprises the steps of obtaining original data;, performing standardized preprocessing on the original data; forming an augmented matrix according to the standardized and preprocessed original data; forming an initial model according to a dynamic principal component analysis method and the augmented matrix, using the initial model to monitor abnormal data; if the detected data is normal, updating the initial modelaccording to the moving window principle, and if thedetected data is abnormal, analyzing and judging the fault cause according to the residual contribution rate. According to the technical scheme provided by the invention, the accuracy of abnormity detection is improved, and the early warning time delay is reduced, so that the problem of low abnormity early warning precision in the drilling process in the prior art is solved, and the effective early warning of abnormity in the drilling process is realized. Moreover, according to the technical scheme provided by the invention, real dynamic detection is realized, the method has adaptability, andthe abnormal detection effect is improved.

The invention provides a system modeling method for a small rotor unmanned aerial vehicle on the basis of closed loop subspace identification. On the basis of the characteristics of a subspace identification algorithm and a closed loop control system of the small rotor unmanned aerial vehicle, the closed loop control system of the small rotor unmanned aerial vehicle is divided into a deterministicpart and a random part by means of orthogonal decomposition; by decoupling the correlation between input signals and noise of the closed loop control system of the small rotor unmanned aerial vehicle, a closed loop control system model of the small rotor unmanned aerial vehicle is accurately identified. According to the system model, an augmented matrix gamma of the system is estimated and acquired on the basis of a Hankel matrix constructed by input data and output data of the small rotor unmanned aerial vehicle, and then the least squares algorithm is used for obtaining a state space matrixof the system. The uncertainty problem caused by modeling errors of system or caused by change of system parameters and noise is solved, and the asymptotic stability and robustness of the closed loopsystem can be ensured.

The invention discloses a design method of a tracking controller of a random system with multiplicative noise and time delay. A discrete linear system model with an observation function is used for describing a random system which is changed due to sudden environmental disturbance, equipment faults and the like and has multiplicative noise and input time delay, and an augmented system model related to a reference track is established by using an augmented matrix method. Under the condition of a limited time domain, an optimal tracking controller design scheme related to states and reference tracks is provided based on a new maximum value principle, a differential Riccati equation and a Lyapunov theory. In addition, under the condition of infinite time domain, an optimal tracking controllerdesign method capable of enabling the system to be stable is provided according to the Lyapunov stability theory and the algebraic Riccati equation, so that the stability of the system is ensured, and good tracking performance is provided.