37 results about "Structural decomposition" patented technology

The invention discloses a schedule optimization method based on flexible scheduling. The schedule optimization method comprises the steps of: taking CR values of a customer order as a convergence formula of a particle swarm for searching an optimal solution on the basis of a time decomposition method; embedding an optimized particle swarm into an artificial immune algorithm based on a working process decomposition method and a mathematical model structural decomposition method to obtain an initial production schedule sequence; establishing a multi-objective optimization model; calculating a production bottleneck procedure and a highest-efficiency working procedure; matching product models with the complementary bottleneck working procedure and the highest-efficiency working procedure, andmutating original solutions in combination with artificial immunity to generate a plurality of schedule sequences; and calculating the affinity of the plurality of schedule sequences so as to screen the solutions and finally generate an optimal schedule sequence. The schedule optimization method based on flexible scheduling is combined with the actual workpiece production process, the scheduling plan is reasonably generated, and the effects of shortening the idle production time, reducing the production energy consumption and reducing the production cost are achieved.

The invention relates to a thin-walled beam design method, aiming at the problems of complex shape, poor economical and technological feasibility, and poor manufacturability of the existing single-component thin-walled beam structure designed by using the topology optimization method, and a multi-component thin-walled beam structure considering the manufacturing cost. Design method; including: 1. Establishing the initial design space finite element model of the multi-component thin-walled beam; 2. Establishing the multi-component thin-walled beam multiphase penalty material interpolation model; 3. Determining the multi-component thin-walled beam structure optimization goal; 4. Establishing the multi-component thin-walled beam Material cost constraints of thin-walled beam parts; 5. Establish a multi-component thin-walled beam topology optimization model considering manufacturing cost constraints; the present invention takes into account manufacturing constraints in the topology optimization process and realizes multi-component structure decomposition in the topology optimization process, improving topology optimization. The economic benefits brought by and the feasibility of the optimization scheme.

The invention discloses an incomplete multi-view clustering method based on missing graph reconstruction and self-adaptive neighbors. The method realizes the clustering of incomplete multi-view data by learning consistent non-negative features. The present invention considers the incomplete graph structure on different views and decomposes it into one view consistent feature and multiple view specific features, wherein the view consistent feature is used to preserve the neighbor graph structure information of multi-view data. The innovation of the present invention lies in rethinking the incomplete multi-view clustering problem from the perspective of graph structure decomposition of incomplete views, and learning the consistent non-negative features and common graph structures of missing multi-view data at the same time, wherein the consistent non-negative features satisfy the common Graph structure constraints. The incomplete multi-view clustering framework of the present invention is composed of a matrix decomposition model and an adaptive nearest neighbor model, and the training target and the derivation process are deduced at the same time; the method of the present invention can handle various incomplete multi-view scenarios without filling in missing views. .

The invention discloses a tree species classification method based on a LiDAR (Light Detection and Ranging) false-vertical waveform model. The method comprises a step 1 of collecting data using an airborne small-light spot full-waveform LiDAR sensor, in order to obtain complete waveform information, wherein the sampling interval is 1ns; a step 2 of preprocessing LiDAR waveform data; a step 3 of establishing a false vertical wave through waveform structural decomposition and comprehensive waveform recombination; a step 4 of carrying out single tree segmentation and information integration; a step 5 of gathering point cloud attributes within in a single tree range and calculating a comprehensive waveform characteristic parameter; and a step 6 of carrying tree species classification with a support vector machine classifier. With the tree species classification method, acquired LiDAR energy signals can be enhanced; on the basis of single tree segmentation, comprehensive waveform characteristic variables are extracted from multiple dimensions, and thus classification accuracy of tree species in the subtropical forest is obtained and improved by a single data source; spatial and temporal variation of main tree species of a forest type can be well embodied; through experimental verification results, the method provided by the invention is improved by 10% in terms of overall precision compared with other LiDAR-employing tree species classification methods; and the Kappa coefficient is improved by 0.1.

The invention discloses a LED automobile lamp failure location detection method based on a binary tree model. First, the structure of the LED lamp model is decomposed, a binary tree structure is constructed, the initial information of the automobile lamp model and the failure data packet are stored, and then the binary tree is sequentially traversed. Compare the initial information stored in the binary tree with the current analysis data to determine whether the current analysis data is consistent with the initial information, and finally determine the failure location and failure type of the sample, and complete the failure location detection. The method of the invention is easy to implement, fast in classification speed and higher in classification accuracy than conventional failure analysis methods, and is applicable to single or combined failure location detection of automobile lamps.

The invention discloses a light detection and ranging (LiDAR) waveform comprehensive feature-based individual tree identification method. The method includes the following steps of: acquiring data by means of an airborne small-light spot full-waveform LiDAR sensor; preprocessing LiDAR waveform data: positioning individual trees and extracting canopy spread; correcting the LiDAR waveform data based on emitted energy and distance information of the sensor and ground objects; constructing a voxel framework and performing structural decomposition of the LiDAR waveform data; extracting waveform feature variables of the individual trees; extracting point cloud feature variables of the individual trees; and screening out optimal feature variables by using a random forest method, and classifying tree species. As indicated by verification results, compared with others method in which remote sensing is adopted, the overall accuracy of the method is improved by about 15%, and the Kappa coefficient of the method is improved by about 0.13.