36results about How to "Global optimization" patented technology

The invention discloses a point cloud quality evaluating and unmanned aerial vehicle (UAV) track planning method for UAV scanning reconstruction, which includes the following steps: the original pointcloud model of a scene is acquired and reconstructed through scanning by a UAV; the points of the point cloud model are classified by using a kd tree; for each point in the point cloud model, the probability that the point is a hole boundary point is estimated based on three criteria respectively, and weighted averaging is carried out on the three probabilities to comprehensively judge whether the point is a hole boundary point; a hole boundary line is generated by searching for closest points, in order to distinguish between different holes; a road map with visibility information is constructed in a point cloud model scene through a ball filling method; an optimal rescanning track of the UAV is generated; and finally, a result is output, a generated path is fed back to the UAV, the UAV flies along the planned path, and the hole area of the point cloud model is repaired. The method can assist an UAV in quickly and automatically rescanning a scene to be reconstructed and generating a more accurate building model, so as to reduce the hole area with texture missing in a building point cloud model.

The invention relates to the technical field of hybrid electric vehicles, and discloses an energy control method and device based on driving conditions, a storage medium and a device. The method comprises the following steps: acquiring current driving characteristic information of the hybrid electric vehicle; analyzing the current driving characteristic information to obtain driving condition category information; determining a preset target relation function according to the driving condition category information; searching a corresponding target equivalent factor through a preset equivalentfactor comparison table according to the preset target relation function; determining use proportion information of an engine and a motor in the hybrid electric vehicle according to the target equivalent factor, controlling the energy of the hybrid electric vehicle according to the use proportion information, so that real-time adjustment is performed according to different driving conditions through the target equivalent factor, instantaneous and global optimization of energy management is realized, and the energy-saving potential of the PHEV is brought into full play.

The invention discloses a transient electromagnetic inversion method based on transverse smooth constraint and is applied to the earth exploration field. The method comprises steps that S1, a primary pulse magnetic field is emitted to the underground, a secondary magnetic field excited by induction eddy current generated by an underground conductive geologic body under the excitation action of the primary pulse magnetic field is measured, secondary magnetic field data acquired through measurement is taken as observation data; S2, an initial inversion model is constructed; S3, a mismatching functional between the observation data and forward theory data is constructed; S4, a vertical constraint functional for constraining electric parameter change of the conductive geologic body is constructed; S5, a transverse constraint functional for constraining electric parameter change of the conductive geologic body among measurement points is constructed; S6, an inversion target function is constructed on the basis of the mismatching functional, the vertical constraint functional and the transverse constraint functional; and S7, the initial inversion model is taken as initial conditions, an electric parameter corresponding to a smallest inversion target function value is solved, and the underground structure is reconstructed according to the electric parameter. The method is advantaged in that the underground structure acquired on the basis of inversion imaging results is more accurate and reliable.

The invention discloses an optimal control method based on a data-driven single network structure. By continuously updating training samples and adding to an LS-SVM model, the partial derivative of a system state variable x for the optimal performance index function is approximated on line, and then the optimal control law is directly obtained by calculation according to the optimal control law expression. In this way, the controller network is saved, so that the structure becomes simpler. In addition, the calculation procedures involved in each iteration of the traditional computing method are reduced, thereby increasing the execution speed.

The invention provides a genetic material specificity based tumor cell drug sensitivity evaluation method. The method comprises steps of: dividing a tumor cell sample set into a training sample subset and a test sample subset; performing training by using the training sample subset and obtaining an extreme learning machine prediction model set of tumor cell drug IC50 value under different parameter conditions; testing the prediction models one by one by using the training sample subset, and selecting the prediction model that has best performance as a tumor cell drug sensitivity evaluation model; and calculating a drug sensitivity evaluation value of the tumor cell to be tested. According to the method provided by the invention, tumor cell drug sensitivity evaluation on the biological genetic level is implemented; meanwhile, artificial subjective influence can be avoided by using drug IC50 value, thereby making drug sensitivity determination more objective and reliable; iteration is not needed in learning by using the extreme learning machine; the number of parameters to be adjusted is small, and global optimization can be ensured, thereby making tumor cell drug sensitivity detection more convenient, highly efficient and practical.

The invention relates to a control-oriented lithium battery thermal model establishing method, and belongs to the technical field of battery management. The method comprises the following steps of S1,performing order reduction on second-order partial differential heat conduction equations of cylindrical and square lithium batteries based on a spectrum-Galerkin method, and establishing control-oriented thermal models of the cylindrical and square lithium batteries; S2, selecting a cylindrical lithium battery and a square lithium battery, performing dynamic working condition test on the cylindrical lithium battery and the square lithium battery, and establishing a dynamic working condition data set of the batteries; S3, selecting a group of typical dynamic working condition data, and identifying unknown parameters of the lithium battery thermal model offline based on an optimal parameter identification algorithm; and S4, verifying the model precision and the working condition adaptability of the thermal model by adopting other working condition data in the dynamic working condition data set. Compared with the prior art, the modeling technology has the advantages that the modeling technology is not limited by the geometrical shape of the battery, the established thermal model is efficient in calculation and globally optimal, and detailed temperature information in the battery canbe extracted.

The invention belongs to the technical field of bioinformatics and discloses an RNA secondary structure prediction method for a quantum genetic algorithm based on multi-population assistance. According to the method, a stem pool and a stem compatibility matrix of an RNA sequence is established according to the RNA sequence; quantum bit vectors are used to initialize multiple chromosome populations; quantum measurement is performed on each population; optimal individuals are acquired according to measurement results; the optimal individual b in all the populations is obtained and used to replace worst individuals, nonhomologous to b, among the optimal individuals in other populations, then all the populations are updated by use of different rotational angles, and other populations not participating in replacement are updated by use of a fixed rotational angle; and the process is iterated till a stop condition is met. Through the method, the global search capability and search efficiencyof the quantum genetic algorithm are effectively improved, and the evolution algebra of the genetic algorithm is lowered. Meanwhile, all the populations suppress competition and cooperate mutually, so that the globality of the algorithm is improved, and prediction accuracy is substantially enhanced.

The invention discloses a combined forecasting method of urban water demand based on minimum sum of square error. At first, a water demand database of the water supply pipe network is established. Then the RBF neural network model, GRNN model and ARIMA model are trained and established. Finally, combined forecasting is carried out based on the minimum sum of square error. The invention combines the characteristics of RBF neural network, such as strong approximation ability and global optimization, with the characteristics of GRNN neural network, such as fast learning speed and easy convergence, and with the characteristics of ARIMA, such as flexibility and strong adaptability, and combines the rolling renewal strategy, so that the prediction method can dynamically adapt to the developmentand change of the environment.

The invention discloses a regularization-based RBF network multi-label classification method and belongs to the field of multi-label classification technique. According to the technical scheme of the invention, the method comprises the steps of (1) constructing an RBF network, namely, constructing the input layer nodes, the hidden layer nodes and the output layer nodes of the network; (2) training the RBF network based on training data; (3) predicting labels based on the RBF network. The multi-label classification method is designed based on the regularization technique and is fast in clustering speed and good in generalization performance. Therefore, the generalization performance of the RBF network can be effectively enhanced.

The invention discloses a fractionally-spaced multi-modulus algorithm (FS-MMA) based on a DNA genetic bat algorithm (DNA-GBA) (DNA-GBA-FS-MMA). The DNA-GBA-FS-MMA combines the DNA-GBA with the FS-MMA and fully utilizes the advantages of the FS-MMA and the DNA-GBA. The invention introduces a DNA genetic algorithm into a bat algorithm (BA) so as to obtain the DNA-GBA. A global optimal position vector of a bat group is found by the DNA-GBA and is used as the real part and the imaginary part of the initialized optimal weight vector of a MMA in order to increase rate of convergence and reduce excess mean square error. Compared with a BA-FS-MMA and a BA-MMA, the DNA-GBA-FS-MMA is fast in rate of convergence, low in mean square error, and global optimal performance, and has practical value in communication technology field.

The invention provides an EA (Evolutionary Algorithm)-based English text clustering method. The EA-based English text clustering method comprises the steps of firstly preprocessing English texts into vector space models, and then proceeding a clustering process, wherein the clustering process comprises the steps of step one, randomly selecting n clustering centers, carrying out clustering division on the clustering centers by utilizing Euclidean distance, enabling texts in the same type to be classified in the same cluster, and thus completing and obtaining local optimum clustering division; step two, carrying out EA processing, selecting clustering centers of a new generation by using an alliance thought and a gene crossover mutation process, carrying out the clustering division through an inter-text closet distance principle, and thus achieving global optimum. According to the EA-based English text clustering method provided by the invention, the English text can be effectively clustered, unnecessary clustering results can be removed, and thus a clustering process can be more quickly converged.

The invention discloses a tomato irrigation intelligent control system. The system is formed by a tomato growing environment soil moisture collection and control platform and a tomato growing environment soil moisture intelligent control system. Aimed at the characteristics of tomato growing environment soil moisture parameter variation of nonlinearity, large time lag, great inertia, and time dependence, and problems that the tomato growing environment soil moisture parameter variation is hard to accurately detect and control, the tomato growing environment soil moisture intelligent control system based on a wireless sensor network is designed. The system is an intelligent monitoring system which detects, intelligently controls and manages soil moisture of a controlled tomato growing environment through the tomato growing environment soil moisture collection and control platform and the tomato growing environment soil moisture intelligent control system, so as to realize high-quality efficient regulation of tomato growing environment soil moisture, and improve tomato quality and requirements of economic benefits of growers. The system has wide application prospect and great popularization value.

The invention discloses a distribution network state estimation method based on semi-definite programming. The method is characterized by comprising the following steps of step 1 establishing a mathematical description of a WLS (Weighted Least Squares) estimation problem; step 2 establishing a semi-definite programming model of the WLS estimation problem; step 3 determining a coefficient matrix of the semi-definite programming model; and step 4 solving. The distribution network state estimation method based on semi-definite programming has super-linear convergence and good precision, a computed result thereof is consistent to a computed result of the general WLS method, the obtained solution can be guaranteed to be globally optimal, the storage efficiency is improved and the computation complexity is greatly reduced.

The invention discloses a virtual potential field cooperative obstacle avoidance topology control method based on heterogeneous unmanned aerial vehicle formation. According to the method, the unmanned aerial vehicles are superposed with formation keeping virtual potential fields based on heterogeneous formation datum points according to heterogeneous cluster improved dynamic situation fields, cluster networking communication relations among the heterogeneous unmanned aerial vehicles are integrated, a heterogeneous unmanned aerial vehicle dynamic autonomous cooperative obstacle avoidance control algorithm is adopted, autonomous cooperative obstacle avoidance processing is carried out on multiple static / dynamic obstacles, and the unmanned aerial vehicles are effectively prevented from being damaged. The formation can be dynamically adjusted in real time to adapt to obstacle states of different numbers, different distributions and different dynamic characteristics, a predetermined basic formation is synchronously kept in the obstacle avoidance process, meanwhile, the unmanned aerial vehicle can smoothly bypass obstacles, the problem that the unmanned aerial vehicle falls into a local extreme value is solved, and the unmanned aerial vehicle obstacle avoidance efficiency is improved. And oscillation caused by approaching an obstacle can be avoided, and the global optimum of the system is ensured.

The invention discloses an intelligent cable automatic manufacturing machine based on a fuzzy support vector machine. The intelligent cable automatic manufacturing machine is characterized by comprising a wire inlet device (1), a peeling and cutting device (2), a stripping and twisting device (3), a wire clamping mechanism (4), a rotary conveying device (5), a needle pressing device (6), the wireinlet device (1), the peeling and cutting device (2), the stripping and twisting device (3), the wire clamping mechanism (4), the rotary conveying device (5), and the needle pressing device (6) construct an adjusting platform for adjustment and control through a single-chip microcomputer, a rotary drum rotating speed intelligent cascade control system based on the fuzzy support vector machine is designed in the adjusting platform to carry out the closed-loop control of the rotating speed on the pulling drum (14) of the wire inlet device (1). According to the invention, the problems that the automation degree of existing cable manufacturing equipment is not high, most procedures need manual participation, the working efficiency is low, and a certain safety accident can be caused are effectively solved, and the problem that the rotating speed precision of the pulling drum of the wire inlet device is poor is effectively solved.

The invention relates to a method for improving the power utilization reliability of a major user based on active power seeking, and belongs to the technical field of power grid management. The method comprises the following steps of acquiring a real-time stoppage probability P of each component in a generation / transmission / transformation / distribution system according to a real-time running state of the system; calculating a load loss probability PL of a nose where the major user is located in the generation / transmission system according to P; calculating load loss probabilities Pf of all outgoing lines of the transformation system supplying power to the major user according to P and PL; calculating a power outage probability P<*>k of the major user according to P and Pf; calculating P<*>k under all feasible transformation and distribution power supply schemes, and employing the scheme corresponding to a minimum value as a reliable power supply scheme. According to the method, the characteristic that the power supply reliability of the user is constantly changed along with the running state of the system is taken into account, and power supply schemes are dynamically provided for the major user according to local conditions by regulating transformation and distribution system network topology, so that the power utilization risk of the major user in any running state of the system is always kept at the lowest level.

The invention discloses a civil aircraft cabin door automatic drilling and riveting path planning method based on an improved ant colony algorithm, and relates to the technical field of civil aircraft cabin door automatic drilling and riveting path planning. A shortest path model for automatic drilling and riveting path planning in each area and an inter-area switching path planning model are established, the cabin door is divided into a plurality of areas according to process constraints, heuristic information of an ant colony algorithm is improved under the condition of guaranteeing reference hole correction precision constraints, and the accuracy of the cabin door is improved. And the problem that the distance between the to-be-drilled rivet holes is small and cannot be searched or even is stopped is solved. The method combines actual constraint division regions, improves heuristic information and pheromone updating strategies of the ant colony algorithm, enables the algorithm convergence speed to be higher, achieves global optimum in the regions, uses the genetic algorithm between the regions, achieves optimal switching selection between the regions on the basis of achieving the optimum in the regions, and improves the optimal switching selection between the regions. And finally, automatic drilling and riveting path planning is globally optimal, the path is shortest, and the automatic drilling and riveting efficiency is improved on the basis of ensuring the drilling and riveting quality.

The invention discloses a permanent magnet motor model prediction control method with a fixed switching frequency. The method comprises the following steps: building a mathematical model of a permanent magnet motor and a driving converter system of the permanent magnet motor, and obtaining the current of the motor, a DC bus side voltage value, and an inverter output voltage value; establishing a discretization stator current prediction model under a rotating coordinate system, and predicting the current value of the motor at the next moment after performing one-step compensation on the system time delay; establishing a switching period discretization model equivalent to the switching frequency of the IGBT, counting switching actions of the device, and predicting a switching period value; and establishing a total cost function, determining a switching state, and outputting and controlling on and off of an inverter switching tube. Switching signals are not generated through modulation, the control structure is simple, the operation load is low, the effects of constant switching frequency and neutral point potential balance are achieved at the same time, the control effect can be globally optimal, and the permanent magnet motor can have low torque ripple and high dynamic response at the same time during operation.

The invention provides an intelligent control method for a conveying system based on a particle swarm algorithm. The method comprises the steps: judging whether the current conveying system needs to be rescheduled or not according to the material storage condition of a warehouse, the fault condition of processing equipment and the change condition of a product order; if it is judged that rescheduling is needed, the data set of the current conveying system is called from the industrial personal computer and used for solving the mathematical model again, a new scheduling scheme is generated, andthe new scheduling scheme returns the new data set to the industrial personal computer and is executed by the industrial personal computer. According to the invention, the production line intelligentscheduling scheme can be optimized, and the method is suitable for multi-thread, multi-product and large-batch production line intelligent scheduling.

The invention discloses a calculation method for automatically zooming content sensitive image with prominent depth-of-field characteristics, includes A1 selecting a plurality of patches arbitrarily from 1000 high-definition images, A2 obtaining a fuzzy dictionary Dblur by training ((using Gaussian fuzzy of sigma=2), A3 each patch in the target image G being then decomposed into a weighted superposition sum of several basic atoms, A4 by calculating the number of basic atoms decomposed by the target patch, clear images having a lot of detail, more atoms in Dblur being needed to approximate, theblurred image being decomposed into fewer atomic superpositions, A5 describing the number of atoms fblur in the image decomposition as the fuzzy depth of field and proposing a depth of field estimation algorithm, an A0 depth of field estimation algorithm being obtained; A6 reducing the distortion of non-salient region by enhancing the protection formula of background in x-direction and y-direction for depth of field F, and A 7 keeping salient region and reducing the change of structure information of non-salient region, which can alleviate the distortion of non-salient region.

The invention relates to a forest illegal reclamation land parcel detection method based on deep learning target detection. The invention relates to the technical field of forest illegal reclamation land parcel detection, and the method comprises the steps of constructing a sample data set of deep learning training based on expert knowledge; inputting a sample image, and establishing a multi-scale feature pyramid recognition network; based on a continuous learning mode, training model fine tuning is carried out; and according to the fine-adjusted training model, applying the fine-adjusted training model to a test image, and detecting an illegal reclamation land parcel in a forest region. In order to improve the recall rate of detection model identification and greatly improve the precision of machine interpretation, the method of the invention adopts a continuous learning mode to carry out two-stage model training, and can prevent bare land parcels with similar spectrums and textures from being mistakenly classified.

The invention discloses an intelligent cable automatic manufacturing machine based on a fuzzy support vector machine. The intelligent cable automatic manufacturing machine is characterized by comprising a wire inlet device (1), a peeling and cutting device (2), a stripping and twisting device (3), a wire clamping mechanism (4), a rotary conveying device (5), a needle pressing device (6), the wireinlet device (1), the peeling and cutting device (2), the stripping and twisting device (3), the wire clamping mechanism (4), the rotary conveying device (5), and the needle pressing device (6) construct an adjusting platform for adjustment and control through a single-chip microcomputer, a rotary drum rotating speed intelligent cascade control system based on the fuzzy support vector machine is designed in the adjusting platform to carry out the closed-loop control of the rotating speed on the pulling drum (14) of the wire inlet device (1). According to the invention, the problems that the automation degree of existing cable manufacturing equipment is not high, most procedures need manual participation, the working efficiency is low, and a certain safety accident can be caused are effectively solved, and the problem that the rotating speed precision of the pulling drum of the wire inlet device is poor is effectively solved.

The invention provides an evolutionary neural network structure search method, which aims at solving the problem that an existing convolutional neural network is widely applied to the field of machinevision such as object detection, image recognition and classification, but the convergence speed of the convolutional neural network is low due to the fact that a training result is overfitted due toa huge network structure and parameters. The advantages of the particle swarm algorithm in the aspects of the search range and the convergence speed are utilized, the search breadth and the convergence speed can be considered at the same time, the structure discovered by the method can effectively solve the problems, and an interpretable and robust neural network method is established.

The invention discloses a multi-scale convolution feature fusion pedestrian re-identification method based on pose embedding, and the method comprises the steps: carrying out the preprocessing of an original pedestrian image through a random erasing mode, obtaining a pedestrian image, carrying out the baseline network optimization of a Resnet-50 network model, and extracting deep convolution features; extracting a saliency human body image from the original pedestrian image; performing posture extraction on the human body saliency image, and then extracting local semantic features from the body part image; carrying out weighted fusion on the deep convolution features and the local semantic features, carrying out distance measurement on weighted fusion features, and generating an initial measurement list; and reordering the images in the initial measurement list according to a reordering algorithm to obtain an image correct matching ranking, and outputting a pedestrian matching image to identify a specific pedestrian. The identification and positioning precision can be greatly improved.

The invention provides an intelligent distributed line selection system and a line selection method. The system includes the following parts: model generation component, information collection component, attribute weight prediction component, attribute action component, screening component, intelligent line selection component, evaluation Output component: the line selection model is generated by the model generation component, and the importance of different attributes can be obtained by calculating the weights of different attributes of the initial control point, so as to obtain the influence of the control point in different aspects, which is convenient for a comprehensive understanding of the control point and Further analysis ensures the accuracy of line selection.

The invention discloses a combined prediction method of urban water demand based on the minimum error square sum. The invention first establishes the water demand database of the water supply pipe network. Then train and build RBF neural network model, GRNN model, ARIMA model. Finally, combined forecasting is performed based on the minimum sum of squared errors. The invention combines the characteristics of RBF neural network with strong approximation ability and global optimality with the characteristics of GRNN neural network such as fast learning speed and easy convergence with the characteristics of ARIMA flexible and strong adaptability, and combined with rolling update strategy, so that the prediction method can dynamically adapt The environment develops and changes.

The invention relates to a method for improving the reliability of key users' power consumption based on active power source searching, and belongs to the technical field of power grid management. First, according to the real-time operating status of the system, obtain the real-time outage probability P of each component in the relevant generation / transmission / transformation / distribution system; according to P, calculate the load loss probability p of the node where the key user is located in the power generation and transmission system L ; according to P, P L , calculate the load loss probability p of all outgoing lines of the substation system supplying power to key users f ; According to P, p f , to calculate the power outage probability of key users Calculate all feasible variable, power distribution and power supply schemes Take the scheme corresponding to the minimum value as the high-reliability power supply scheme. This method takes into account the characteristics that the user's power supply reliability changes from time to time with the system's operating status. By adjusting the network topology of the substation and power distribution system, it dynamically provides power supply solutions for key users according to local conditions, so as to ensure the key users' safety in any operating state of the system. Electricity risk is always kept low.

The invention discloses a transient electromagnetic inversion method based on transverse smooth constraint and is applied to the earth exploration field. The method comprises steps that S1, a primary pulse magnetic field is emitted to the underground, a secondary magnetic field excited by induction eddy current generated by an underground conductive geologic body under the excitation action of the primary pulse magnetic field is measured, secondary magnetic field data acquired through measurement is taken as observation data; S2, an initial inversion model is constructed; S3, a mismatching functional between the observation data and forward theory data is constructed; S4, a vertical constraint functional for constraining electric parameter change of the conductive geologic body is constructed; S5, a transverse constraint functional for constraining electric parameter change of the conductive geologic body among measurement points is constructed; S6, an inversion target function is constructed on the basis of the mismatching functional, the vertical constraint functional and the transverse constraint functional; and S7, the initial inversion model is taken as initial conditions, an electric parameter corresponding to a smallest inversion target function value is solved, and the underground structure is reconstructed according to the electric parameter. The method is advantaged in that the underground structure acquired on the basis of inversion imaging results is more accurate and reliable.