30 results about "Modal matrix" patented technology

The invention discloses a wind field rapid prediction method and system based on optimized Latin hypercube sampling and POD-BPNN, and the method comprises the steps: carrying out Latin hypercube sampling to generate an initial sample point set, introducing a sensitivity weight, and carrying out the iterative optimization of sample distribution through a greedy strategy; assembling the CFD numerical simulation flow field data of all sample points in the sample point set into a flow field snapshot matrix, and determining a dynamic multi-target truncation order and a corresponding POD mode and coefficient; building and training a BPNN model, packaging the trained BPNN model, the POD modal matrix, the mean field and the flow field reconstruction logic into an FMU module, obtaining the FMU module which can be called in a cross-platform manner, and achieving the real-time prediction of a wind field. The invention relates to the technical field of wind field prediction, significantly improves the precision and calculation efficiency of wind field prediction, and provides an efficient and accurate solution for the rapid prediction of a wind field.

The invention belongs to the technical field of reduced-order model development, and particularly relates to a POD-SVD random forest reduced-order method and system for digital twin engines. Adjusting a plurality of simulation data generated by the digital twin engine under different working condition boundary conditions; extracting a physical quantity of the region of interest; combining and forming an original matrix required by PODSVD decomposition; decomposing an original matrix into three matrixes by using a PODSVD algorithm, and converting the three matrixes into a modal coefficient matrix, a modal matrix and an energy matrix through a calculation formula; training a random forest algorithm; inputting a boundary condition parameter corresponding to a working condition needing to be predicted into the trained random forest model, and generating a prediction modal coefficient; calculating and restoring a physical quantity needing to be predicted; and adding grid information for the obtained reduced physical quantity, and generating a. Dat file for post-processing. The method is used for breaking through the limitation of tradeoff between cost and model precision in traditional researches such as test / CFD, and meanwhile, the model precision and the evaluation speed are improved.

The invention relates to the technical field of signal processing, in particular to a pipeline guided wave signal processing method and system. The method comprises the following steps: acquiring a multi-channel ultrasonic guided wave signal through a sensor array circumferentially arranged around a pipeline, and converting the multi-channel ultrasonic guided wave signal into a frequency domain response matrix; calculating a circumferential amplitude distortion degree based on the power spectrum amplitude distribution of each channel; constructing a modal coupling matrix by combining the circumferential amplitude distortion degree and the theoretical wave number of the ideal pipeline; correcting the standard orthogonal modal matrix to generate a modal projection matrix; and decomposing the frequency domain response matrix by using the projection matrix, and separating a structure echo component and a defect signal component. According to the method, the influence of geometric defects such as ovality deviation is evaluated more accurately, energy exchange between modeling modes is accurately established, and orthogonality hypothesis is corrected; modal energy leakage is inhibited, and the decomposition accuracy is improved; signals are effectively separated, coherent noise interference is solved, and the accuracy and reliability of pipeline defect detection are improved.

The invention relates to the technical field of wind power plant power control, in particular to a wind power plant dynamic power distribution optimization method based on projection dimensionality reduction. The method comprises the following steps: collecting a high-dimensional original operation data stream of a wind power plant, and preprocessing to output a high-dimensional state vector; extracting a dominant mode matrix through an intrinsic orthogonal decomposition algorithm, and projecting a high-dimensional state vector to a low-dimensional space; future modal coefficient evolution is predicted through an autoregression prediction model, and a low-dimensional rolling optimization problem is constructed and solved through a sequential quadratic programming algorithm; reconstructing the optimal low-dimensional modal coefficient vector into a target power instruction of each fan, and issuing and executing the target power instruction; and updating the dominant mode matrix through an incremental singular value decomposition algorithm and correcting parameters of the autoregressive prediction model. According to the method, projection dimensionality reduction from high-dimensional data to a low-dimensional space is realized through intrinsic orthogonal decomposition, and the problem of low optimization solution efficiency caused by high data dimensionality of a traditional method is solved.

The invention is suitable for the field of portal frames, and discloses a portal frame dynamics analysis method which comprises the following steps: establishing a full-order finite element model of a portal frame, and obtaining a mass matrix and a stiffness matrix of the full-order finite element model; dividing the full-order finite element model into a plurality of substructures based on the modular structure characteristics of the portal frame, and decomposing the degree of freedom of each substructure into an internal degree of freedom and a boundary degree of freedom; constructing a fixed boundary main modal matrix and a constraint modal matrix of each substructure, and constructing a reduced-order basis matrix; performing modal polycondensation on a mass matrix and a stiffness matrix of the full-order finite element model by using the reduced-order basis matrix to generate a reduced-order mass matrix and a reduced-order stiffness matrix, and constructing a reduced-order kinetic equation; and solving the reduced-order kinetic equation to obtain reduced-order modal coordinates, carrying out inversion reduction in combination with the reduced-order basis matrix and the reduced-order modal coordinates to obtain a displacement response vector of the full-order finite element model, and retaining key kinetic information of the portal frame to the greatest extent through reduced-order efficiency improvement.

The invention discloses a transient signal detection and sorting method based on time-frequency image deep learning, and relates to the technical field of signal sorting. Based on a hypergraph model and core polymorphic sites, the hypergraph model is divided into a plurality of independent substructures by analyzing the distribution characteristics of hyperedge dynamic weights, independent modal analysis is performed on each substructure, and a transient signal detection and sorting result is obtained. Extracting a time-frequency distortion coefficient matrix and an interference response mode matrix; inputting the multi-modal data into a multi-modal network to generate a homogeneous transformation matrix; and based on the homogeneous transformation matrix, deriving attitude error parameters and translation error parameters, feeding back the attitude error parameters and the translation error parameters to a feature input layer and a decision layer of the multi-modal deep learning detection and sorting model, completing dynamic correction of model parameters, and detecting and sorting real-time signals again after model correction. According to the sorting method, the detection rate and sorting accuracy of transient signals are remarkably improved in real-time processing, and meanwhile, the false alarm and leak detection risks are reduced.

The invention discloses an order reduction method for aerodynamic reconstruction in a wallboard ground flutter simulation test, and relates to the technical field of aircraft design, and the method comprises the following steps: employing a hypothesis modal method to carry out spatial discretization on a wallboard structure of an aircraft, and obtaining a first expression; determining a second expression based on the intrinsic mode matrix; substituting the second expression into the first expression to obtain a third expression; substituting the second expression into the first aeroelastic equation to obtain a second aeroelastic equation under the intrinsic mode coordinates; physical aerodynamic force is determined based on the third expression and the second aeroelastic equation; the physical aerodynamic force is subjected to order reduction through the displacement interpolation matrix and the force interpolation matrix, and the physical aerodynamic force after order reduction is obtained. According to the method, the number of loading points is reduced, it is guaranteed that the limited number of loading points are equivalent to the distributed aerodynamic force, the method is not limited by the structure size, test implementation is facilitated, time domain fitting is not needed, and the calculation precision and efficiency are higher.

The invention discloses a traffic flow combination prediction method based on graph convolution. The method comprises the following steps: 1, abstracting a traffic network into a graph formed by nodes and edges; 2, taking the minimum total envelope entropy as a target function, and performing parameter optimization on a variational mode decomposition algorithm by using a crown porcupine optimization algorithm to obtain an optimal hyper-parameter; 3, obtaining a modal matrix of a corresponding traffic flow by using an improved variational modal decomposition algorithm, performing singular value decomposition on the modal matrix to obtain a singular value vector, and constructing a traffic flow characteristic matrix; 4, obtaining a correlation matrix by using cosine similarity, calculating a feature matrix of the convolutional neural network of the layer of graph, and updating the correlation matrix after each layer of graph passes through the convolutional neural network of the layer of graph; and 5, constructing a residual error-based graph convolutional network, and training the traffic flow to obtain a traffic flow prediction sequence. The method can improve the prediction precision of the traffic flow.

The invention discloses a storm surge forecasting method combining empirical orthogonal decomposition and deep learning, and belongs to the field of ocean storm surge forecasting. The method comprises the following steps: a, determining a wind field area according to the position of a forecast ocean station and the like; b, collecting and arranging a 10-meter wind distance flat data matrix from the reanalysis data set, and collecting observed storm surge data from the ocean station; c, performing empirical orthogonal decomposition on the data matrix to obtain a main modal matrix and a main component matrix; d, taking a principal component in the principal component matrix as a forecast factor, taking the storm water increase data as a forecast quantity, and constructing a sample library; carrying out model training to obtain a long-short term memory neural network forecasting model; and e, collecting a 10-meter wind distance flat data matrix for future forecasting, calculating a principal component value, and substituting the principal component value into the forecasting model to obtain future storm surge data. According to the method, the precision of storm water increase forecasting can be improved, and compared with the actual observation value of storm water increase, the calculated root mean square error is low.

The invention relates to a POD secondary manifold lower chamber flow field nonlinear model order reduction method based on anchor point optimization, and the method comprises the steps: obtaining a two-dimensional lower chamber computational fluid mechanics model as a full-order model, and building a snapshot matrix according to the full-order model; carrying out POD order reduction on the snapshot matrix, and calculating a POD reconstruction error matrix and a secondary manifold coefficient matrix of the snapshot matrix; determining an anchor point, and calculating a quadratic manifold modal matrix at the anchor point according to the POD reconstruction error matrix and the quadratic manifold coefficient matrix; according to the secondary manifold modal matrix at each anchor point, obtaining a secondary manifold modal matrix at the test point; and according to the quadratic manifold modal matrix at the test point, calculating an approximate snapshot of the POD quadratic manifold nonlinear reduced-order model at the test point. According to the method, the representative anchor points are introduced into the parameter sampling space, so that self-adaptive calculation of the secondary manifold modal matrix is realized, and the approximation precision of the reduced-order model is improved.

The present application relates to the technical field of signal processing, and particularly relates to a pipeline guided wave signal processing method and system. The method comprises the following steps: acquiring multi-channel ultrasonic guided wave signals through a sensor array arranged circumferentially around a pipeline, and converting the signals into a frequency domain response matrix; calculating a circumferential amplitude distortion degree based on the amplitude distribution of the power spectrum of each channel; constructing a modal coupling matrix in combination with the circumferential amplitude distortion degree and an ideal pipeline theoretical wave number; generating a modal projection matrix by correcting a standard orthogonal modal matrix; and decomposing the frequency domain response matrix by using the projection matrix to separate a structure echo component and a defect signal component. The present application more accurately evaluates the influence of geometric defects such as ellipticity deviation, accurately models the energy exchange between modes and corrects the orthogonality assumption, suppresses modal energy leakage, improves the decomposition accuracy, effectively separates signals, solves the interference of coherent noise, and improves the accuracy and reliability of pipeline defect detection.

The invention provides a dynamic system order reduction method suitable for improving dynamic mode decomposition of a wind power signal, and belongs to the technical field of wind power signals, and the method comprises the steps: obtaining a wind power time sequence signal, carrying out the preprocessing, constructing a time delay data matrix, carrying out the dynamic mode decomposition of the time delay data matrix, and obtaining a wind power signal. Obtaining a system order reduction model comprising a modal matrix, an amplitude matrix and a dynamic evolution matrix; arranging the modal matrix by using an improved discrete time coefficient weighting criterion, and obtaining a reconstructed low-order linear model by using an inversion formula and a strategy of increasing a reserved modal number; performing inverse time delay operation and inverse normalization processing on the reconstructed low-order linear model to obtain one-dimensional signal data, and completing original dynamic system order reduction; according to the method, the problems that an existing wind power signal data driving model cannot give consideration to interpretability and accuracy, and errors caused by an original dynamic mode decomposition method in the aspects of mode selection and mode number determination are large are solved.

The invention belongs to the technical field of stability evaluation of a new energy station grid-connected system, and discloses a participation factor calculation method of repeated and similar oscillation modes and related equipment, and the method comprises the steps: obtaining a linear state space model of a new energy station system; acquiring a linearized state matrix according to the linearized state space model of the new energy station system; calculating a characteristic value of the new energy station system and a characteristic vector corresponding to the characteristic value according to the linearized state matrix, and respectively establishing a modal matrix according to the characteristic value and the characteristic vector; identifying a repeated oscillation mode or a similar oscillation mode of the new energy station system, and selecting a key repeated oscillation mode or a similar oscillation mode; based on a key repeated oscillation mode or a similar oscillation mode, a group participation factor of the repeated oscillation mode or the similar oscillation mode is calculated according to a modal matrix of a characteristic value and a characteristic vector, only one summation operation is involved for calculation of the group participation factor, and the calculation process is simple and efficient.

The invention belongs to the technical field of steel structure floor design, and particularly relates to a transient response analysis method for converting structural damping into equivalent viscous damping in a full frequency band, which comprises the following steps of: 1, establishing a finite element model of a structure, and outputting characteristic matrixes of the structure, including a stiffness matrix and a mass matrix; 2, performing modal analysis based on the finite element model of the structure to obtain a modal matrix of the structure; 3, calculating a generalized stiffness matrix and a generalized mass matrix of the structure based on the stiffness matrix, the mass matrix and the modal matrix; step 4, calculating viscous damping of each order vibration mode of the structure based on the generalized stiffness matrix; 5, calculating the transient response of each order vibration mode of the structure based on the generalized stiffness matrix, the generalized mass matrix and the viscous damping of each order vibration mode of the structure; and 6, converting the transient response of each order vibration mode of the structure to a physical coordinate to obtain the transient response of the structure at each moment.

The invention relates to the technical field of cross-border logistics fraud detection, and discloses a cross-border logistics fraud detection method and device, electronic equipment and a storage medium, and the method comprises the steps: obtaining to-be-detected specified lading bill text data; obtaining corresponding multi-modal data, obtaining historical abnormal order data, constructing a multi-modal matrix, adjusting model parameters of each preset dimension detection model, obtaining a target detection model of each preset dimension, detecting the multi-modal matrix, and determining whether the specified lading bill text data has a fraudulent behavior according to a detection result. The fraud risk assessment method has the beneficial effects that the fraud risk of each dimension is comprehensively assessed by constructing the multi-modal matrix and comprehensively analyzing the lading bill text and related historical abnormal order data, in addition, the detection model parameters are adjusted based on historical data, and through dynamic and flexible multi-modal data analysis, the fraud behavior identification capability is improved.

The invention provides an aero-engine infield prediction method based on combination of POD dimensionality reduction and machine learning. According to the method, dimension reduction is carried out on a three-dimensional numerical simulation result of the whole aero-engine through a POD method, main characteristic modes are extracted, and a time mode matrix and a space mode matrix are generated. And then training the time modal matrix by using a machine learning model, predicting the time modal matrix of the test working condition, and reconstructing a predicted field value matrix in combination with the space modal matrix. The method has remarkable effects in the aspects of reducing the data dimension and improving the calculation efficiency and the prediction precision, and is suitable for rapid prediction of the Mach number field, the total temperature field and the total pressure field of the whole aero-engine.

The invention provides a PCA-based rear duct variable area ejector generation method and device, equipment and a storage medium. The method comprises the following steps: converting original coordinate data of a basic ejector shape into low-dimensional controllable parameters through a CST parameterization method; the method comprises the following steps: performing disturbance on CST fitting parameters to generate a large number of disturbance shape samples, compressing a high-dimensional parameter space to a low-dimensional modal coefficient space formed by a small number of main modals by extracting a main modal matrix of shape deformation, further screening the first K main modals according to a variance contribution rate, determining a value range of modal coefficients, and determining the value range of the modal coefficients; uniformly sampling in the K-dimensional modal coefficient space by adopting a Latin hypercube sampling method; and finally, reconstructing the sampled target modal coefficient into the shape coordinate data of the target ejector through inverse principal component analysis transformation. According to the method, on the premise of ensuring physical rationality of the shape of the ejector, uniform coverage and efficient exploration of a design space are realized with a small number of samples, and the shape generation efficiency of the ejector is remarkably improved.

The present invention discloses a splitting section selection strategy considering splitting fitness index, and the main steps include: S1. Linearize the system including load nodes and generator nodes to determine the regional oscillation mode; S2. Select the modal matrix of the system and construct the correlation matrix; S3. Classify the nodes of the system including load nodes and generator nodes; the nodes are divided into three categories: strongly correlated nodes, weakly correlated nodes and buffer nodes; S4. Construct the splitting applicability index, consider the minimum unbalanced power between regions, and form a comprehensive index; S5. Take the comprehensive index as the objective function for finding the optimal splitting section, and adopt a genetic algorithm to select the optimal splitting section. The splitting section selection strategy of the present invention classifies the system nodes, solves the problem of unclear classification of some original nodes, and reduces the splitting space; by proposing the splitting fitness index, it solves the adaptability of splitting section selection and avoids generator out-of-step after splitting.

The invention discloses a lightweight optimization method of a prestress self-balancing tension lattice structure based on buckling dependence, and relates to the technical field of structural engineering and computational mechanics. Base structure generation and balance analysis: generating a node set by adopting a discretization method according to a design domain, forming a base structure, and establishing a balance matrix based on a geometric topological relation; carrying out dimensionality reduction preprocessing of symmetry driving, carrying out singular value decomposition on the balance matrix, obtaining a self-stress mode matrix, judging a symmetric group, converting a symmetric self-stress mode matrix, obtaining an independent base, and representing a force density vector as a product of the independent base and an independent design variable; and based on LPSO topological optimization, decoding, adaptive component classification, dual buckling stability check, fitness evaluation and convergence judgment are carried out until iteration is terminated. Through a symmetry-driven dimension reduction strategy and a dual buckling constraint mechanism, high-stability lightweight structure design capable of being periodically assembled is realized, and the optimization calculation efficiency is greatly improved.

The invention relates to a state prediction method and device for a multi-grid-connected inverter system. The method comprises the following steps: based on a topological structure of a single grid-connected inverter system, obtaining an impedance model corresponding to the single grid-connected inverter system, constructing a power grid structure node admittance matrix, further determining an impedance matrix corresponding to a multi-grid-connected inverter system, performing eigenvalue decomposition on the impedance matrix to obtain a modal matrix corresponding to the impedance matrix, and determining the impedance matrix corresponding to the multi-grid-connected inverter system. Analyzing a modal amplitude boundary condition corresponding to any parameter change in the multi-grid-connected inverter system to obtain a safe operation boundary condition, and predicting the state of the multi-grid-connected inverter system based on the safe operation boundary condition; by adopting the method, the accuracy of predicting the state of the multi-grid-connected inverter system can be improved.

The embodiment of the invention provides a question recommendation method and device based on multi-modal matrix fusion and dynamic weighting, and the method comprises the steps: constructing a basic matrix through historical user related data, determining a plurality of corresponding recommendation algorithms through the basic matrix, setting a recommendation algorithm weight for each recommendation algorithm, comprising a basic weight and a user behavior weight, the basic matrix and the user behavior weight are updated in an incremental updating and momentum updating mode, recommendation calculation is conducted on the updated basic matrix through recommendation algorithms in parallel, and recommendation scores output by the recommendation algorithms are obtained. The method comprises the steps of updating user behavior weights, updating recommendation algorithm weights through the updated user behavior weights, performing weighted fusion on scores of recommendation algorithms through the recommendation algorithm weights, determining a final recommendation result, and collecting new user related data for performing new incremental updating and momentum updating. The recommendation accuracy and efficiency of the question recommendation system can be improved.

The invention discloses a pressure vessel lower chamber nonlinear model order reduction method based on POD and secondary manifold. The method comprises the steps that sample points of sampling parameters are determined, a lower chamber full-order fluid mechanics CFD model corresponding to the sample points is operated, internal velocity field and temperature field data of a lower chamber are obtained, and a snapshot matrix of the velocity field and a snapshot matrix of the temperature field are constructed; performing singular value decomposition on the snapshot matrixes of the velocity field and the temperature field to obtain a left singular value matrix, a diagonal singular value matrix and a right singular value matrix; determining an intrinsic orthogonal decomposition (POD) modal matrix and an intrinsic orthogonal decomposition (POD) coefficient matrix according to the singular value proportion; determining a mapping relation between the sampling parameter and the POD coefficient; and respectively calculating the POD error matrix, a Kronerkel matrix of the POD coefficient matrix and a quadratic manifold modal matrix to obtain a nonlinear reduced-order model of the lower chamber of the pressure vessel. The method can be used for rapidly obtaining field distribution data of internal flow parameters of the lower chamber.

The application provides a sensor optimization arrangement method and system based on structural vibration response, and belongs to the technical field of equipment safety detection. The application comprises the following steps: establishing a finite element model of a structure to be monitored; calculating an equivalent energy matrix and an average value of diagonal elements of the equivalent energy matrix; starting from all degrees of freedom, sequentially deleting the minimum value on the diagonal of the equivalent energy matrix, and simultaneously deleting the degree of freedom represented by the value; using the remaining degrees of freedom to construct an iteration matrix; sorting the main diagonal elements on the iteration matrix, selecting the smallest element, and deleting the degree of freedom vector corresponding to the element from the modal matrix; repeating the step until the number of remaining degrees of freedom is equal to the number of sensors to be arranged, wherein the sensors are arranged on the remaining degrees of freedom. The application not only considers high-energy measuring points and improves the noise resistance of the measuring point combination, but also fully guarantees the modal independence and modal information amount of the measuring point combination, and is more suitable for the application of sensor optimization arrangement in actual engineering practice.

The application belongs to the field of gantry cranes, and discloses a gantry dynamics analysis method, which comprises the following steps: establishing a full-order finite element model of a gantry crane, and obtaining a mass matrix and a stiffness matrix of the full-order finite element model; based on the modular structure characteristics of the gantry crane, dividing the full-order finite element model into a plurality of substructures, and decomposing the degrees of freedom of each substructure into internal degrees of freedom and boundary degrees of freedom; constructing a fixed boundary principal modal matrix and a constraint modal matrix of each substructure, and constructing a reduced-order basis matrix; using the reduced-order basis matrix, performing modal condensation on the mass matrix and the stiffness matrix of the full-order finite element model, generating a reduced-order mass matrix and a reduced-order stiffness matrix, and constructing a reduced-order dynamics equation; solving the reduced-order dynamics equation to obtain a reduced-order modal coordinate, and combining the reduced-order basis matrix and the reduced-order modal coordinate to inversely restore a displacement response vector of the full-order finite element model, which maximally retains the key dynamics information of the gantry crane through reduction.

The present invention belongs to the technical field of aerospace, and particularly relates to a modal-based method for reconstructing aircraft structural loads. This method uses a structural finite element model to obtain modal information at key locations of the aircraft structure, including nodal displacement modes and element stress and strain modal matrices. Based on the topological relationships of the finite element mesh, nodal resultant force modes and section resultant force / moment modal matrices are derived. Finally, based on these modal matrices, the stress / strain intensity at key locations and the resultant force / moment at key sections are reconstructed from the generalized displacements of the structure. This algorithm does not require processing the overall stiffness and damping matrices of the structure, facilitating practical engineering applications.

A power distribution network new energy carrying capacity evaluation method based on comprehensive similarity degree relates to the technical field of power distribution network new energy planning and operation evaluation, and comprises the following steps: collecting source and load time series data, power grid topology and equipment parameters and preprocessing, constructing a multi-scale modal matrix through CEEMDAN decomposition, calculating power fluctuation offset rate to obtain time dimension similarity, combining electrical distance weighted spatial distribution similarity to obtain two-dimensional time-space comprehensive similarity. Then a multi-objective optimization model of source and load matching, power grid safety and economic operation coordination is constructed, an improved NSGA-III algorithm is used to solve the Pareto solution set, and the optimal access scheme is selected through a four-dimensional evaluation system and an entropy weight-TOPSIS method. The application has high evaluation precision, can improve the new energy carrying capacity by more than 18%, reduce the power abandonment rate by more than 23%, and is suitable for power distribution networks containing energy storage and demand response.

The invention discloses an epidemic situation prediction method based on a modal decomposition network and deep learning, and relates to the technical field of deep learning, and the method comprises the steps: S1, collection and preprocessing: collecting various types of epidemic situation related data, and carrying out the data cleaning and normalization, and obtaining original data; s2, modal decomposition: constructing a modal decomposition network; inputting the original data into a modal decomposition network to generate a plurality of modal components, and forming a modal matrix by the plurality of modal components; s3, predicting the data: constructing a Transform model; inputting the modal matrix into a Transform model, and generating prediction development data; s4, data integration: integrating the multiple pieces of prediction development data to obtain a final result; and outputting a final result. Complex time sequence data are decomposed through MDN, and a long-time dependency relationship in a time sequence is captured in combination with a self-attention mechanism, so that high-precision prediction of nonlinear and non-stationary data is realized.

The invention discloses an excitation point optimization selection and aerodynamic force equivalence implementation method in a ground flutter test, and relates to the technical field of aeroelasticity tests of aircrafts. The method comprises the following steps: constructing a system modal matrix, introducing a weighting coefficient matrix based on a prior aeroelasticity analysis result, and constructing a weighting effective independent matrix; calculating diagonal elements of the weighted effective independent matrix, and gradually removing candidate measuring points which contribute the minimum linear independence of the modal vector by adopting an iterative deletion method until a preset number of optimal excitation points are left; and estimating modal coordinates of the structure based on response data of the optimal excitation point, and realizing aerodynamic equivalence through generalized force condensation. A traditional measuring point arrangement method is often difficult to give consideration to measuring point representativeness and global recognition precision, and according to the method, excitation points are optimally selected by introducing a weighting coefficient matrix, so that a limited number of excitation points can reflect a main vibration mode of a structure to the maximum extent, the measurement precision of a key flutter mode is ensured, and thus the estimation precision of modal coordinates is improved.

The invention discloses a short-time traffic flow prediction method of a traffic flow combination model based on graph convolution. The method comprises the following steps: 1, abstracting a traffic network into a graph formed by nodes and edges; 2, processing the traffic flow sample by using a fast ensemble empirical mode decomposition algorithm to obtain a modal matrix of the traffic flow, and performing singular value decomposition so as to construct a traffic flow characteristic matrix; 3, inputting the normalized traffic flow characteristic matrix into a short-time traffic flow prediction network formed by a simplified graph convolutional neural network, a Transform layer and a full connection layer for processing to obtain predicted traffic flow data; and 4, training the short-time traffic flow prediction network, and obtaining an optimal traffic flow combination prediction model for predicting future traffic flow data. According to the method, the characteristics of short-time non-stationarity, spatial correlation and time dependence of the traffic flow time sequence can be combined, so that the prediction precision and the prediction speed of the traffic flow can be improved, and intelligent management and control of a traffic system are realized.