50 results about "Joint constraints" patented technology

The invention relates to a cooperative control method for full-state constraint mechanical arm trajectory tracking, and belongs to the technical field of control engineering. The method combines ABLF-based thrust-reversing control with Hamilton energy theory-based control, and builds a cooperative controller: establishing a mechanical arm system dynamics equation of n-joint constraint, and performing coordinate transformation on the equation to obtain a state-space equation of the system; designing an ABLF-based thrust-reversing controller, so as to improve system initial response speed, and prevent constraint from being damaged at the same time; designing the Hamilton energy theory-based controller, so as to improve tracking stability of a later period of system response; and designing the cooperative controller to realize rapid and stable control of a full-state constraint mechanical arm system. ON the premise of not damaging a full-state constraint condition, the cooperative controlmethod provided by the invention can improve dynamic-state and steady-state performance of the system, ensure fast mechanical arm system response speed and high positioning precision, and can effectively improve working performance of the mechanical arm in a robot system.

The invention discloses a one-dimensional range profile recognition method based on self-adaptive locality sparsity preserving projection. According to the method, actually measured one-dimensional range profile signal samples are preprocessed; a sparsity coefficient matrix is obtained through sparsity preserving projection (SPP), and a locality similarity matrix is obtained by locality preserving projection (LPP); sparsity preserving projection equations, locality preserving projection equations and a self-adaptive maximum margin criterion are fused, a joint constraint equation set is established, and a self-adaptive locality sparsity preserving projection matrix is obtained; and training samples and test samples are projected into lower-dimensional space through the projection matrix, and a support vector machine is used to carry out training and classification thereof. Based on the sparsity preserving projection, the locality preserving projection and the self-adaptive maximum margin criterion, the method makes full use of sparse reconstruction of the samples and recognition information contained in neighbor relations and combines with the self-adaptive maximum margin criterion to extract lower-dimensional features of the samples, the recognition accuracy of one-dimensional range profile signals is improved, the feature dimensionality is reduced, and the noise immunity is enhanced.

The invention discloses an establishing method for a three-dimensional Q body of shallow, medium and deep layers based on multi-information joint constraint. The method comprises the following steps: constraining through surface layer Q value measured data, and establishing a relation curve of a Q value and a speed value on the position of a measurement point; acquiring a shallow three-dimensional speed model through high-precision constraint chromatography inversion under the constraint of the speed value on the position of the measurement point; utilizing the shallow three-dimensional speed model and the relation curve of the Q value and the speed value on the position of the measurement point to acquire a three-dimensional Q body of a shallow layer in a depth domain; utilizing seismic data to establish a three-dimensional speed model of medium and deep layers; getting the three-dimensional Q body of medium and deep layers under the constraint of the three-dimensional speed model; and performing time depth or depth time conversion, converting the three-dimensional Q body of the shallow layer and the three-dimensional Q body of the medium and deep layers to a domain, utilizing a sine cosine matching constraint three-dimensional fusion technique to effectively fuse the three-dimensional Q body of the shallow layer with the three-dimensional Q body of the medium and deep layers, and establishing a near-surface factor-containing three-dimensional Q body of the shallow, medium and deep layers. According to the invention, the problem of splitting of the Q body of the shallow layer and the Q body of the medium and deep layers can be effectively solved.

The invention discloses a high-precision human body posture reconstruction method and system, and the method comprises the following steps: obtaining a human body video, and selecting a key frame of the human body video; building a deep neural network to obtain human body contour information and two-dimensional joint information of the key frame; constructing a parameterized human body model, andconstraining the parameterized human body model by utilizing the human body contour information and the two-dimensional joint information to obtain initialized human body posture parameters; and optimizing and initializing human body posture parameters by utilizing the two-dimensional joint information and the human body contour information, selecting all frames of a human body video, constructinga joint constraint, optimizing and initializing the human body posture parameters again, and iterating two optimization processes by utilizing a Gauss-Newton formula until an energy difference valueof two adjacent iterations is smaller than a preset threshold value. The human body model can be better reconstructed by utilizing a few angles of human body RGB information, so that the reconstructedhuman body model more accurately reflects the real human body figure.

The invention discloses a PET image reconstruction method based on non-local features and total variation joint constraint. The method comprises the following steps: establishing a mathematical modelof a reconstruction problem, adding the low-rank constraint based on non-local features and adding the total variation constraint to perform PET image reconstruction. The low-rank constraint filters noises in the structural information of an image according to the non-local similarity feature of the image. Meanwhile, the defect that a step effect exists in the total variation method is overcome. The total variation serves as a constraint for edge protection and noise reduction based on pixel points. At the same time, the defect that the edge of an image exists on the basis of an image block isovercome. The convergence speed is accelerated. Therefore, the constraint joint solving means is effectively utilized, and the defects of two methods are overcome. The problems that the result is lowin resolution and the noise interference is generated in the PET image reconstruction process can be further solved. Compared with an experiment of an existing reconstruction method, the reconstruction method disclosed by the invention has the advantage that a good reconstruction effect can be obtained.

The invention belongs to the technical field of indoor positioning, particularly an indoor positioning method based on global and local joint constraint transfer learning. According to the method provided by the invention, through minimization of inter-domain marginal and conditional probability distribution discrepancy and maximization of a sample variance in a potential subspace, consistency ofa global structure is constrained. Through minimization of an intraclass variance and maximization of an interclass variance, dependency of each class and corresponding samples is kept. Through manifold regularization, a local neighborhood relationship is kept. Further, the consistency of local structures is constrained. The problem that an existing transfer learning method is insufficient in knowledge transfer can be solved. According to knowledge obtained from a source domain through transfer, positioning precision of a target domain can be effectively improved. The problem of RSS (ReceivedSignal Strength) fluctuation resulting from environment change is solved. The indoor positioning method based on the global and local joint constraint transfer learning provided by the invention is anew high precision positioning method applicable to a complex indoor environment.

The invention discloses a robust space-time adaptive processing method for super-low-altitude target detection by a moving platform radar, which mainly solves the problem of performance loss caused as a training sample is polluted by a mirror false target generated by the super-low-altitude target in a multipath environment. The method comprises steps: 1, optimized frequencies of four constraint points are designed; 2, a Doppler domain constraint condition is built; 3, an airspace constraint condition is built; 4, an optimal weight vector is solved; and 5, output data are obtained. Through replacing an equality constraint in the previous method by an inequality constraint in the Doppler domain, an amplitude-phase joint constraint in the airspace is used for replacing an amplitude constraint in the previous method, main lobe shape preserving is realized on the premise of not losing degree of freedom, robustness of the space-time adaptive processing method is realized, the performance of super-low-altitude target detection by the moving platform radar is improved, and the method of the invention can be used for detecting the super-low-altitude target by the moving platform radar.

The invention, which belongs to the field of fine characterization of oil and gas geophysical reservoirs, provides a seismic section enhancement method based on the compressed sensing theory. The method comprises the following steps: step one, carrying out reflection coefficient inversion calculation under norm L1-L0.8 joint constraints on to-be-processed data; step two, synthesizing a reflectioncoefficient by using well data and selecting an optimal wavelet capable of distinguishing a target stratum; and step three, carrying out convolution on the reflection coefficients in the previous twosteps and the wavelet to obtain a new section. Therefore, noises are removed by using the compressed sensing; and a problem that an accurate wavelet matrix can not be established because the observation matrix establishment way of the existing compressed sensing technology is not mature is solved. The method has the following beneficial effect: a problem that a thin layer can not be tracked accurately because of low resolution in the seismic data is solved rapidly and efficiently by combining the well data.

The invention discloses a method used for measuring structure damages and safety conditions of Ying county buddha tower. The method comprises following steps: primary testing is carried out so as to establish dynamic characteristic standards of Ying county buddha tower, and qualitative determination of structure damages and safety conditions is carried out based on change situation of dynamic characteristic parameters; 2th to n-th testings are carried out, the results of the n-th testing are compared with the result of the primary testing and the result of the (n-1)-th testing, qualitative determination of structure damages and safety conditions is carried out based on change situation of dynamic characteristic parameters; and then examination analysis is carried out based on environment change and structure conditions. According to the method, comprehensive analysis based on practical boundary conditions, material performance, construction quality, and joint constraints is carried out, testing results are accurate and credible, and qualitative determination of structure damages and safety conditions of Ying county buddha tower can be realized based on comparative analysis of the change of frequency, vibration modes, and damping ratio obtained via testing.

The invention discloses a remote sensing image super-resolution reconstruction method based on adaptive joint constraint. The method mainly comprises the following steps of: carrying out up-sampling on a low-resolution image by using bicubic interpolation; principal component analysis and K- The means method learns a dictionary of the high-resolution image; constructing a non-local sparse prior byutilizing the non-local self-similarity of the image; proposing a local structure filter based on an image local gradient, and then constructing a structure to keep local priori; fusing non-local priori and local priori, adaptively selecting parameters according to the noise level, and finally forming an adaptive joint priori; constructing a cost function, and then using an alternating iterationalgorithm for solving; finally, obtaining a high-quality image. The single remote sensing super-resolution reconstruction method provided by the invention has good performance in the aspect of keepingimage details, and has higher objective evaluation indexes. Therefore, the single remote sensing image super-resolution reconstruction method is an effective single remote sensing image super-resolution reconstruction method.

The invention relates to bi-modal depth-image analysis. A depth-image analysis system calculates first mode skeletal data representing a human target in an observed scene if a portion of the human target is observed with a first set of joint positions, and calculates second mode skeletal data representing the human target in the observed scene if the portion of the human target is observed with a second set of joint positions different than the first set of joint positions. The first mode skeletal data and the second mode skeletal data have different skeletal joint constraints.

The invention relates to a joint list detection and decoding method for fixing the complexity of a multiple input multiple output (MIMO) wireless communication system adopting block codes. The method comprises the following steps of: producing a candidate input code word list of a channel decoder from a candidate list of an MIMO detector by using a list generator so as to avoid iterative processing of a conventional detection and decoding algorithm; replacing soft decoding by hard decoding to reduce the working complexity of a receiving end; and joint processing is performed by using the list generator and the encoder to realize a uniform candidate character sequence space structure under joint constraint of an MIMO signal space and a code word space so as to improve system performance. The invention provides a new joint detection and decoding processing technical scheme which is used for the receiving end and designed by adopting the block codes in a channel coding scheme in the MIMO system. The method reduces calculation complexity and improves detection performance.