30 results about "Coordinate descent method" patented technology

The invention discloses a UAV (Unmanned Aerial Vehicle)-assisted WSN (Wireless Sensor Network), as well as a method for designing node scheduling, route planning and power distribution of the same, and belongs to the technical field of wireless communication and IOT (Internet Of Things). The UAV-assisted WSN consists of a UAV aerial base station and K single-antenna ground terminal nodes. The method of the invention is characterized by aiming at minimizing system power consumption, considering constraint conditions such as a data transmission rate of the terminal node, a flight speed of the UAV and a transmitting power of the UAV, building a mathematical optimization model by taking a scheduling variable of the terminal node, a flying route of the UAV and a transmitting power variable of the UAV as parameters, respectively converting two sub-problems decomposed from a non-convex optimization problem of mixed integer variables into upper bound convex problems corresponding to the two sub-problems, alternately iterating sub-optimization problems through a block coordinate descent method and an interior point method, and obtaining a suboptimal parameter solution for scheduling of theterminal node, the flying route and the power distribution of the UAV.

The invention discloses a wireless communication method applicable to a communication network using a UAV to realize multi-hop relay, which comprises the following steps: acquiring channel informationand the number of UAVs, and creating a system throughput calculation model by using the channel information and the number of the UAVs; using a block coordinate descent method and a continuous convexoptimization method to carry out iterative optimization on transmit power and flight trajectory in the system throughput calculation model and obtaining optimized flight trajectory and optimized transmit power; and achieving communications by using the optimized transmit power and the optimized flight trajectory in the communication network. Thus, the degree of freedom of the UAVs can be fully utilized to achieve long-distance communication in the signal blind area that the base station cannot cover, and the system throughput can also be optimized. Correspondingly, the embodiment of the present invention further provides such a wireless communication device and equipment and a readable storage medium, which have corresponding technical effects.

The invention discloses a multi-view monitoring video behavior detection and recognition method based on rarefaction, structuralization and judgment under multiple constraints, and the method is used for realizing the intelligent analysis and management of a monitored video. The method comprises the following steps of (1) preprocessing the video; (2) detecting a target; (3) tracking the target; (4) extracting the time and space characteristics; (5) sliding a time window and normalizing the characteristics; (6) detecting and recognizing the behavior in the multi-view monitored video under multiple constraints; and (7) fusing results of two adjacent time windows. The method has the advantages that the intrinsic correlation characteristics of human body behavior characteristics in a multi-view scene can be adequately analyzed, and through the excavation of rarefraction, structuralization and judgment constraints and the compilation of corresponding regular terms, a multi-view human body behavior recognition target function can be established; and meanwhile, the human body target behavior can be recognized through a coordinate descent method.

The invention discloses a fast factorized back-projection SAR self-focusing method. A phase error optimization model is established by combining a fast factorized back-projection imaging algorithm (FFBP), and the optimization model is solved by using a coordinate descent method and a secant method so that the problem of phase error compensation can be effectively solved. Compared with the methods in the prior art, the phase error can be more accurately estimated and a great focusing SAR image can be obtained so that the problems of high computational burden, low accuracy and low adaptability of the existing FFBP self-focusing method can be solved, and accurate motion error estimation and great focusing combining FFBP imaging can be realized.

The invention discloses a method for reconstructing an original signal only through dispersed fractional order Fourier transform phase information and belongs to the technical field of signal processing. The method comprises the steps of converting a signal reconstructing problem into a convex optimization problem; performing dispersed fractional order Fourier transform on the original signal, and changing a transform matrix of the dispersed fractional order Fourier transform to obtain different pieces of phase information; storing or transmitting the obtained phase information; recovering the original signal through proper pieces of phase information by means of an amplitude recovery algorithm combining a coordinate descent method and an inner point method, namely, reconstructing the original signal. According to the method, the purpose of reconstructing the original signal with fewer pieces of phase information is achieved by means of the theoretical foundation that the phase information includes larger amount of information than amplitude information under the same number.

The invention relates to a turbulence-degraded image blind restoration method based on dark channel and Alternating Direction Method of Multipliers. The method includes the following steps: firstly on the basis of the multiple dimension theory, in each dimension, applying dark channel prior constraint on an image, applying sparse constraint and energy constrain on a point spread function, then using the coordinate descent method and conducting alternating iteration to estimate a fuzzy kernel and the image in current dimension, if the dimensions arrive at the maximum thereof, a final estimated fuzzy kernel is obtained, finally, in combination with a total variation model, using a derivative Alternating Direction Method of Multipliers to make details of the image restored quickly. According to the invention, the method, by using the dark channel prior information of a clear image as a constraint item, can help a cost function to converge to a clear solution in the iteration process, addresses the susceptibility of obtaining a fuzzy solution by using tapered prior information under the Maximum posterior probability in current blind restoration algorithm, such that the method herein can restore more image details, has less ring effect, and effectively increases restoring quality.

The invention requests for protecting a large-scale MIMO system energy efficiency optimization method based on wireless energy transmission under hardware damage. The invention builds a large-scale MIMO system energy efficiency optimization method based on wireless energy transmission under a hardware damage condition; an admissible control mechanism is introduced while the QoS requirement of a user is ensured; an original non-linear fractional programming problem is simplified via equivalent conversion and variable replacement and finally converted into a convex optimization problem, and computation is performed by using a block coordinate descent method. The method provided by the invention has the advantages of low computation complexity and strong channel adaptation, and meanwhile ensures QoS constraint of the sensor node, so as to be particularly applicable to the large-scale MIMO system based on wireless energy transmission under the hardware condition.

The invention provides a signal intensity prediction method and system based on wireless propagation model correction. The method includes the following steps: inputting the preset base station data to a preset wireless propagation model and building a least square model; solving the least square model through a coordinate descent method and obtaining an optimal correction coefficient; and adjusting the parameters of the wireless propagation model according to the optimal correction efficient and conducting wireless signal intensity prediction according to the wireless propagation model whose parameters are adjusted. The signal intensity prediction method and system based on the wireless propagation model correction can rapidly realize wireless model coefficient correction and accurately predicts the intensity of wireless signals.

The invention discloses a semi-supervised classification method capable of simultaneously learning an affinity matrix and a Laplacian regularized least square, which mainly comprises the following steps: firstly, a joint model capable of simultaneously learning the affinity matrix and the Laplacian regularized least square is established according to a training sample; secondly, the block coordinate descent method is used to optimize all kinds of variables in the model; and finally, the soft label of the sample is obtained by a Laplacian regularized least square classifier, and the dimension with the largest element in a label vector is selected as the category of the sample. The invention effectively fuses the sparse self-representation problem of samples and the Laplacian regularized least square classifier, and realizes the simultaneous optimization and mutual improvement of the sample affinity matrix and the Laplacian regularized least square classifier in the learning process. Theinvention has an explicit classifier function, so that the problem of an external sample can be effectively handled. Compared with other semi-supervised classification methods, the method has more accurate classification accuracy and good application prospects.

The invention discloses a fluorescence-molecular-tomography rebuilding method based on a random-variable alternating direction multiplier method. The mode of the alternating direction multiplier method is converted into a random parameter mode with the randomness and the decomposability of the random dual coordinate descent method, and then solving is carried out through parameter alternating updating of the random-parameter alternating direction multiplier method; the fluorescence-molecular-tomography rebuilding method includes the following achieving steps that (1) large-scale fluorescence data is collected; (2) the linear relationship between measurement data and target distribution is built; (3) the linear relationship is converted into the convex optimization problem; (4) the convex optimization problem is solved in an alternated mode with the random-parameter dual coordinate descent method and the random-parameter alternating direction multiplier method, and a target distribution diagram and rebuilding time are obtained. The rebuilding efficiency of fluorescence molecular tomography is effectively improved while the quality of rebuilt images is guaranteed, and the fluorescence-molecular-tomography rebuilding method has the important application value in the field of medical molecular images, the field of rebuilding methods and the like.

The invention discloses a sparse subspace clustering method based on selective coordinate descent optimization. The method comprises the following steps: 1, establishing a sparse subspace clustering model, substituting the sparse subspace clustering model into the Lasso formula, converting the sparse subspace clustering model into a quadratic programming problem, and solving a similarity matrix; 2, starting to solve the similarity matrix, traversing all features in the first time iteration by using a coordinate descent method, and using a calculated solution as an initial value; 3, starting from the second time iteration, traversing each feature, if the current solution item is non-zero, updating the feature of the coordinate position by using the coordinate descent method; otherwise, skipping the update of the feature of the coordinate position; repeating the process until the convergence of an objective function; and 4, after the similarity matrix is obtained, performing a spectral clustering process to obtain a classification serial number. According to the sparse subspace clustering method disclosed by the invention, whether the current solution item is zero or non-zero is quickly judged by using an infinite norm rule, thereby avoiding an unnecessary calculation process of a zero item result and shortening the time of optimization solution; and the effect on large-scale sparse problems is extremely obvious.

Along with the quick development of computer software and hardware technology and the wide application of the Internet, the information technology generates a large amount of data information in all fields, such as life, production, scientific research, army, and power supply and distribution. The mining and visualization of supply and distribution energy consumption data becomes a challenge under the condition of complex and redundant big data. The invention proposes a mining and visualization method for supply and distribution energy consumption data in a big-data environment, and relates to a sparse cording algorithm based on deep learning. The method comprises the steps: on the one hand, employing a dictionary learning method of a coordinate descent method to adjust the dictionary parameters in sparse coding; on the other hand, learning main correlation characteristics of supply and distribution energy consumption data through a conjugate gradient descent method, thereby achieving the dimensionality reduction and linearization of redundant data, and achieving data mining. The method employs a Weka data-mining working platform, is combined with an interface file of Weka software, integrates the above methods in Weka, and achieves the data mining and the visualization on a new interactive interface.

The invention belongs to the field of wireless communication technology, and specifically relates to a frequency offset detection method based on an interleave-division multiple access technology. Anorthogonal frequency division multiplexing-interleave-division multiple access (OFDM-IDMA) system will be affected by multiple frequency offsets caused by Doppler shift or different crystal oscillation precision of receiving and transmitting ends, and the conventional frequency offset suppression method for an orthogonal frequency division multiplexing (OFDM) system is not suitable for the OFDM-IDMA system, so a multi-frequency offset suppression method for the OFDM-IDMA system is required to be researched. The method suppresses frequency offsets by using a coordinate descent method (CDM) frequency offset detection algorithm based on the OFDM-IDMA system. After the frequency offset detection, the frequency is improved.

The invention discloses a neural network optimization method for solving and lifting an adjacent operator machine, and relates to the technical field of deep learning neural network optimization. In the training of a forward neural network, a block coordinate descent method is adopted to solve and improve an LPOM model of an adjacent operator, each sub-problem in the LPOM model has convergence, the weight and network activation of each layer of the neural network can be updated in parallel, and extra memory space is not occupied. By adopting the method of the invention, the parallelism, the applicability and the training effect of neural network training can be improved under the condition of using relatively less storage.

The invention discloses a multifunctional networking radar task planning method based on resource conversion and complementation. The method comprises the following steps: firstly, carrying out abstract modeling on tasks of multifunctional networked radars, then constructing an optimization model for multifunctional networked radar resource utilization by taking the sum of delay loss and compression loss as an objective function, and finally, obtaining an optimal solution through an algorithm based on nesting an interior point method and a coordinate descent method. According to the invention,the flexibility of the task duration time and the conversion and complementary relationship between the resources of the multifunctional networked radar system are considered; the problem that taskscannot be well completed due to unbalanced resource distribution when the number of tasks of the multifunctional networking radar is large and the task duration time conflicts with each other is effectively solved, and the task execution efficiency of the multifunctional networking radar is improved.

The fund bin is the proportion of funds invested into stock markets to assets that can be used by the funds, and is a reaction of market information. At present, warehouse location information of domestic funds is disclosed by taking seasons as units according to a security guard, and the wind direction of the market cannot be reflected in time. The method is mainly used for everyday estimation ofthe fund storage location. The method comprises the following steps: firstly, dividing stocks into 28 categories according to Shen000 first-level industry classification, and obtaining 28 industry index daily return rates and the daily return rate of a to-be-estimated fund; Establishing a regression equation by taking the industry index daily return rate as an independent variable and taking thedaily return rate of the fund as a dependent variable; Then, based on a lasso regression model, solving a regression coefficient through a coordinate descent method and a minimum regression loss function, wherein the sum of the regression coefficient is the proportion of the stock assets held by the fund to the fund assets, namely the fund location; And finally, comparing the actual bin position and the predicted bin position published in each season, calculating an average error after the maximum error and the minimum error are removed, carrying out model trimming, and improving the accuracyof the model. And a certain investment reference value can be provided for investors according to a result of the fund bin location prediction model.

The invention discloses a point-to-point iterative learning minimum energy control method for a double-rotor pneumatic system, and relates to the field of direct-current motor optimization control. According to the method, a dual-rotor pneumatic system which runs repeatedly is converted into an input and output matrix model of a time sequence based on a lifting technology, M tracking time points are selected as variables, a minimum energy problem design framework of point-to-point iterative learning control is provided, and on the basis of the framework, a control algorithm combining norm optimization point-to-point iterative learning control and a coordinate descent method is designed to solve the optimization problem. In addition, when the system model has uncertainty, the robustness of the designed minimum energy control algorithm is analyzed. The method can also be expanded to an input and output constraint system, and the purpose of minimum input energy is achieved while a point-to-point tracking task is achieved.

The invention relates to a seismic data random noise suppression method and system, and the method is characterized in that the method comprises the following steps: 1) obtaining original seismic data, and carrying out non-negative processing on the seismic data; 2) constructing a denoising model based on self-paced learning and non-negative matrix factorization according to the processed seismicdata; 3) solving the denoising model based on the self-paced learning and the non-negative matrix factorization by adopting an improved block coordinate descent method; and 4) performing inverse operation of non-negative processing in the step 1) on the obtained optimal solution to obtain denoised seismic data. The provided method can be widely applied to the technical field of seismic exploration.

The invention designs a phase coding sequence parallel accelerated descent optimization design method. For a non-convex optimization problem, a coordinate descent method is sensitive to an initial value and cannot guarantee that a satisfactory optimal solution is finally obtained. Aiming at the problems of the two optimization methods, the invention provides a phase coding sequence parallel accelerated descent optimization design method. The method comprises the following specific steps of generating a new solution by adopting a simulated annealing algorithm according to a set initial coding parameter, and performing variable updating by taking the new solution as an initial parameter of a coordinate descent method; and judging whether the target function increment corresponding to the updated solution meets an acceptance condition or not, and when the new solution meets a convergence condition, ending calculation. According to the method, the simulated annealing algorithm and the coordinate descent method are combined, the simulated annealing solution serves as input of coordinate descent, descent of the target function is accelerated, and the problem that the coordinate descent method is sensitive to initial variables is solved.

The invention provides an FBMC signal PAPR reduction method based on a punishment concave-convex process. The method comprises the following steps: firstly, constructing an optimization problem according to parameters of an FBMC system, namely minimizing a PAPR of a frame of FBMC signal by taking a sending symbol as a variable and taking symbol deformation on a data subcarrier and sending power ona free subcarrier as constraints; secondly, performing equivalent conversion on the optimization problem to obtain an equivalent problem meeting a penalty concave-convex process algorithm framework;secondly, solving an optimization problem according to a two-layer iteration algorithm based on a punishment concave-convex process, updating punishment parameters through outer-layer iteration, fixing the punishment parameters through inner-layer iteration, and solving the punishment problem according to the concave-convex process and a block coordinate descent method; and finally, outputting a calculation result to obtain an FBMC sending signal with a lower PAPR. The PAPR of the FBMC signal is optimized and reduced by utilizing the punishment concave-convex process, and compared with a traditional heuristic iterative algorithm, the method can remarkably reduce the PAPR of the FBMC signal under the condition that the bit error rate is slightly increased.