36results about How to "Guaranteed Global Convergence" patented technology

The invention discloses a robust controller of a permanent magnet synchronous motor based on a fuzzy-neural network generalized inverse and a construction method thereof. The construction method of the invention comprises the following steps of: combining an internal model controller and a fuzzy-neural network generalized inverse to form a compound controlled object; serially connecting two linear transfer functions and one integrator with the fuzzy-neural network with determined parameters and weight coefficients to form the fuzzy-neural network generalized inverse, serially connecting the fuzzy-neural network generalized inverse and the compound controlled object to form a generalized pseudo-linear system, linearizing a PMSM (permanent magnet synchronous motor), and decoupling and equalizing the linearized PMSM into a second-order speed pseudo-linear subsystem and a first-order current pseudo-linear subsystem; and respectively introducing an internal-model control method in the two pseudo-linear subsystems to construct the internal model controller. The robust controller of the invention has the advantages of overcoming the dependence and local convergence of the optimal gradient method on initial values and solving the problems of randomness and probability caused by using the simple genetic algorithm, obtaining the high performance control, anti-disturbance performance and adaptability of the motor and simplifying the control difficulty, along with simple structure and high system robustness.

The invention relates to an urban air quality grade predicting method based on multi-field characteristics. The method comprises a data preprocessing stage, a training stage and a predicting stage. In the data preprocessing stage, mesh generation is carried out to obtain a training sample, and a training data set is obtained. In the training stage, the optimal parameter of a conditional random field model is estimated, an optimal model shown in the specification is obtained, and the conditional random field model is output. In the predicting stage, the maximum output sequence, shown in the specification, of the conditional probability P(Y/X<f>) is obtained by combining the viterbi algorithm and the conditional probability model shown in the specification and obtained in learning of the training stage, and a prediction result is output. The AQI grade can be predicted more accurately, the AQI grade prediction in the following time quantum of all stations is carried out, and the requirement for selecting the time quantum with the high air quality to go out and take activities of people is met.

The invention provides a goods allocation optimization method applied to a Flying-V untraditional layout warehouse. The method is characterized by comprising the steps that S1, goods allocation relevant parameters of the Flying-V warehouse are set; S2, the goods allocation parameters are initialized; S3, a population is initialized according to an information list of to-be-stored goods; S4, an adaptive genetic algorithm is adopted to perform individual optimal selection on the population; S5, whether the number of algorithm termination iterations is reached is judged, if yes, the step S6 is performed, and otherwise the step S4 continues to be cycled; and S6, an optimal goods allocation scheme is output. The goods allocation optimization method is applied to the application occasion of "Flying-V untraditional warehouse layout", goods storage and delivery efficiency and a lowest gravity center of a goods shelf after the goods are placed on the goods shelf are optimized, and a multi-target optimization problem processing method with different dimensions is provided; by the adoption of the adaptive genetic algorithm, the crossover rate and the variation rate change along with adaptation values; and population diversity is maintained, and global convergence of the genetic algorithm is guaranteed.

The invention discloses a classification method for a multi-kernel support vector machine, which relates to the artificial intelligence field, in particular to the data mining technology, and comprises a data pretreatment section, a kernel function selection section, a support vector machine realizing section, and a human-computer interaction section. The work processing comprises that users submit classification request of data to the DPP, then KSP chooses the kernel function, an SILP solution module converts a multi-kernel support vector machine problem to an SILP problem and then solves the problem, a condition detecting module detects whether the condition is satisfied, and if the condition is satisfied, the HIP returns the result to users, otherwise, the parameter and the objective function are updated, and the SILP solution module is transferred to solve. The invention improves the capability of processing complex data of the support vector machine through multi-kernel functions, promotes the complexity of a module and the calculation, and converts the multi-kernel support vector machine problem to a semi-infinite linear program for avoiding the increasing of kernel functions simultaneously, and solves through a method of global convergence.

The invention relates to the technical field of underwater positioning, in particular to a single beacon positioning method of an underwater vehicle. According to the single beacon positioning methodof the underwater vehicle with global convergence, an underwater vehicle is provided with a hydrophone, a Doppler velocimeter, a depth meter, an attitude heading reference system and a GPS; and underwater acoustic beacons periodically broadcast underwater acoustic signals. According to the method, a nonlinear single beacon positioning model in a discrete state is converted into a linear time-varying model through state augmentation; when the underwater acoustic signal is not received, the relative speed and the attitude of the underwater vehicle and water by a device carried by the underwatervehicle are obtained to carry out dead reckoning; after the underwater acoustic signal is received, the underwater acoustic signal transmission time is collected through known underwater acoustic signal transmission time and is taken as an observation variable, and dead reckoning data and observation data of various sensors are simultaneously integrated to predict and update the single beacon positioning based on Kalman filtering. On the premise of meeting the observability of a positioning model, the method has global exponential convergence.

The invention discloses a speed sensorless control method of a permanent magnet synchronous motor, and belongs to the field of permanent magnet motors. A conventional assumption rotating coordinate method is adopted as a basis, on the basis of a model motor method, an estimation rotating speed is directly adopted as a control quantity through a sliding mode control method, then, a self-adaption law of the estimation rotating speed is obtained through a Lyapunov non-linear design method, an algorithm is greatly simplified, the system response speed is increased, meanwhile, forward and reverse rotation of the motor can be achieved, and the method can be used for the permanent magnet synchronous motor.

The invention discloses an attribute selection method based on a binary system firefly algorithm. The attribute selection method is characterized in comprising the following steps: 1: utilizing a fractal dimension box-counting method to calculate a fractal dimension of a high-dimension data set, and obtaining the number of attributes which need to be selected; 2: initializing a firefly population; 3: utilizing the binary system firefly algorithm to select a plurality of attributes of the high-dimension data set to obtain an optimal attribute subset; 4: outputting an optimal solution. The attribute selection method uses the binary system firefly algorithm as a search strategy of attribute selection, takes the fractal dimension as an attribute selection evaluation measurement criterion, and selects the optimal attribute subset from the plurality of index attributes of the high-dimension data set, so that the data processing complexity can be lowered, and the data processing efficiency is improved so as to meet the requirements of solving practical problems.

The invention belongs to the field of robot control and nonlinear systems, and particularly relates to a robot self-adaptive impedance control method based on a biological heuristic neural network. The problem that real-time accurate control of a robot in a complex nonlinear system cannot be realized in the prior art is solved. The robot self-adaptive impedance control method comprises the steps that initial control moment, expected impedance and movement trails of a system are obtained; a dynamic equation and an expected impedance model of n-degree-of-freedom mechanical arm system containingimpedance are built and a t-moment system real state and an expected state of the robot are correspondingly obtained; a self-adaptive controller is built and a (t+1) moment control moment is obtainedbased on full-state feedback and the biological heuristic neural network; and states are circularly obtained and self-adaptive impedance control and movement control are carried out, until a robot mechanical arm completes the movement trails. According to the robot self-adaptive impedance control method based on the biological heuristic neural network, by combining with a biological heuristic neural network structure and delay feedback, a Hebbian algorithm with an award value adjustment and a structure combining network estimation and full-state feedback are adopted, the system is stable, andthe control precision is high.

The invention aims to provide a full waveform inversion method and system based on a non-monotonic search technology. The method comprises the following steps: firstly, dividing multi-scale inversionfrequencies into N frequency groups according to a set frequency group range; secondly, determining a speed model of next iteration corresponding to the jth frequency group according to the seismic observation data and an initial speed model by adopting a non-monotonic search technology in combination with the initial step length; judging whether j is greater than N or not; if j is greater than N,outputting the speed model of the next iteration corresponding to the jth frequency group as an optimal speed model; and finally, performing seismic imaging based on the optimal speed model. According to the method, the step length is determined by combining the non-monotonic search technology with the initial step length, so that a high-precision inversion result can be obtained on the premise of ensuring the inversion calculation efficiency; and when the initial speed model is poor, the global convergence of inversion can be ensured to a certain extent, and a good inversion result can be obtained.

The invention discloses a node localization method of a distributed wireless sensor network based on the angle of arrival. In the method, each node does not require to be equipped with a device similar to a compass to unify the orientation of a local coordinate system, so the ability of an algorithm to adapt to the scene is increased, and the method is also applicable in a complex scene where no compass can be equipped. Meanwhile, the method of the invention is linear, the calculation is simple and convenient, and has an exponential convergence property with a high convergence speed. The algorithm can ensure global convergence, which shows that the algorithm will not fall into the local optimum when an initial estimated value is selected randomly. Besides, the invention requires only two anchor nodes, and avoids high cost brought by the installation of a large number of GPSs.

The invention discloses a distant-relative pointer genetic algorithm-based cross section size optimization method of a steel truss structure, which is used for cross section size optimization design and calculation of the steel truss structure so that the structure conforms to the economical and safe requirements. The method comprises the steps of building a mathematical model of a structural optimization design; mapping a target function and a constraint condition to an adaptive value function so as to further randomly generate an initial population; calculating an adaptive value, and judging whether the adaptive value conforms to a convergence criterion or not; arranging a distant-relative pointer to sequentially complete genetic operator operation and immunity operator operation if the adaptive value does not conform to the convergence criterion, and further completing population substitution; repeatedly calculating the adaptive value until the convergence criterion is satisfied; and outputting an optimization result. The distant-relative pointer can be used for effectively avoiding individual repeated occurrence, meanwhile, the distant-relative pointer has the capability of generating a new mode under the condition of no damage on excellent individual, the global convergence of the algorithm is ensured, and the method is simple to operate; and the local search capability of the algorithm can be improved by vaccination on an immunity operator, individual degeneration can be prevented, and the complete effect of an adaptive technology is ensured.

The invention relates to an identification method of random structural damage based on genetic algorithm and static force measurement data. The identification method of the random structural damage based on the genetic algorithm and the static force measurement data comprises the following steps: (1) preliminarily obtaining statistical properties of random structural damage indexes; (2) defining the probability of unit damage as the random rigidity before occurrence of the damage or the probability that an elastic modulus Kai is larger than Kdi; (3) introducing damage probability indexes, determining several units of the damage probability indexes as non-damage units, and conducting corresponding adjustment on the statistical properties of the damage indexes; (5) going back to the governing equation of the initial damage identification in the step (1), and obtaining the objective function of the structural damage indexes after rearranging; and (6) solving the minimum value of the objective function in the step (5) by means of the genetic algorithm, and obtaining the statistical properties of the damage indexes. The identification method of the random structural damage based on the genetic algorithm and the static force measurement data has the advantages that multiple damage identification of different damage degrees can be carried out due to the fact that the genetic optimized algorithm has little limit on the type, the quantity and the size of the parameter.

The invention relates to a super-large-scale integrated circuit layout method considering atomization and proximity effects. The method comprises the following steps: (1) establishing an energy distribution model for the atomization and proximity effects in electron beam lithography (EBL) and describing changes of the energy distribution model; (2) optimizing a model and a function in the global layout, and determining a smooth target function; (3) introducing a new variable to convert an unconstrained minimization problem into a separable minimization problem CSMP with linear constraints; (4)determining an adjacent group ADMM algorithm and an iterative formula of the separable minimization problem; (5) solving two sub-problems in adjacent group iteration; and (6) performing convergence analysis on the ADMM algorithm of the adjacent group. The method is beneficial to reducing adverse effects caused by atomization and proximity effect and improving the layout efficiency.

The invention provides an information fusion and adaptive coupling method of an extensively-related sparse matrix space and belongs to the technical field of artificial intelligence. The invention comprises steps of 1) establishing an extensively-related sparse matrix information space and achieving a fusion expression of complex heterogeneous information; 2) providing an establishing method of the extensively-related sparse matrix and designing multi-module sparse coding; 3) defining operation rules of heterogeneous elements in the extensively-related sparse matrix information space; and 4) accordingly achieving a multi-attribute intelligent matching and service discovery system. The problem that global elements cannot be fused because existing methods of the same kind do not have a unified system configuration and heterogeneous information fusing method and adopt classification and layered optimization is solved. The problem that the time resource cost in application level is too much and the movable end environmental application is difficult to achieve is solved. The problem that the global stability and convergence in technology level is hard to ensure is solved.

The invention discloses a multi-dimensional system contour error estimation method based on a simplified Newton method, relates to the technical field of numerical control machining and aims at solving problems that an existing contour error estimation method based on a Newton extremum search algorithm is large in calculated amount, long in solving time, singular in track peak and the like. Compared with a classical Newton extremum search algorithm, the simplified Newton contour error estimation algorithm provided by the applicant ensures high precision and does not need to calculate derivatives, singularity is avoided in the contour error estimation process, and meanwhile the calculated amount is reduced. According to the specific experimental effect, estimation precision and efficiency of a contour error of the multi-dimensional system are improved by about 30%-50% compared with those of a traditional scheme; a traditional contour estimation method based on a Newton extremum search method is a local convergence method, selection of an initial value of the contour estimation method determines convergence and accuracy of the algorithm, and global convergence is guaranteed by adopting the estimation value as an initial value of subsequent iteration.