68results about How to "Solve the problem that is easy to fall into local optimum" patented technology

The invention discloses a power distribution network route planning method based on tabu differential evolution and a GIS (Geographic Information System). Equal allotment capital recovery calculation is performed under the consideration of the time value of capital, a model is established specific to the aims of minimal annul investment and minimal running maintenance cost, a tabu differential evolution algorithm is put forward on the basis of the model, an obtained optimal solution is decoded, and a planned route is drawn on an electric power GIS platform according to the route number obtained by decoding. According to the method, a non-optimal solution is allowed to be accepted by using tobu search, the problem the differential evolution algorithm tends to falling into local optimum easily is solved by improving the capability of jump-out local optimum, and the planning process of a power distribution network line has the advantages of interactivity, more intuitive planning result and higher flexibility in adjustment of a planning scheme by using the electric power GIS system.

The invention discloses a three-dimensional face reconstruction method, which comprises the following steps: obtaining point clouds respectively corresponding to two frames to be registered, and taking the point clouds as a first group of point clouds and a second group of point clouds; determining attribute vectors of feature points in the first group of point clouds and the second group of pointclouds, the feature points being concave points or convex points in a curved surface formed by the pointing point clouds, and the attribute vectors of the feature points comprising average curvaturesand Gaussian curvatures corresponding to the feature points; performing coarse registration on the first group of point clouds and the second group of point clouds according to the attribute vectorsof the feature points in the first group of point clouds and the second group of point clouds to obtain coarse registration initial values; performing fine registration on the first group of point clouds and the second group of point clouds based on the coarse registration initial value through a point cloud accurate registration algorithm; due to the fact that the rough registration initial providing value provides a good iteration initial position for fine registration, the problem that registration is trapped in local optimum is avoided, registration precision is improved, and then precision of the finally-built three-dimensional face model is improved. The invention further discloses a corresponding device, equipment and a medium.

The invention discloses an unmanned aerial vehicle formation reconstruction system and method based on an ant colony algorithm and an artificial potential field method. The system comprises a target distribution module, a path planning module and a ground station module. A swarm intelligence optimization algorithm is adopted to improve a selection strategy of a standard ant colony algorithm, and the global search capability and search precision of the algorithm are improved; and when the scale of an unmanned aerial vehicle group is large, a global optimal solution can be searched with higher probability. An improved artificial potential field method is adopted in the path planning process, a problem that collision is possibly caused because a gravitational field is too large in the initialmoving stage of the unmanned aerial vehicle is solved through improvement of a gravitational field formula, and a problem that a target cannot be reached in the process that the unmanned aerial vehicle gets close to the target point is solved through improvement of a repulsion field. If the unmanned aerial vehicle sinks into the local minimum point, an escape force is additionally applied to thecurrent unmanned aerial vehicle to help the current unmanned aerial vehicle get rid of the local minimum point, and a problem that the artificial potential field method sinks into local optimum easilyis effectively solved by adopting a local minimum point escape strategy. The system and method are high in calculation speed.

The invention discloses a workpiece pose estimation method based on point cloud, which is mainly used for solving the problem of workpiece pose estimation in the field of automatic assembly, especially in the field of small part assembly. The method comprises the steps of point cloud data preprocessing, point cloud virtual view extraction and pose estimation. Wherein the pose estimation step further comprises the steps of performing iterative operation by adopting a particle filter algorithm based on a dynamic model of an iterative closest point algorithm, outputting the pose of the effectiveparticle in a weighted mean mode if an iterative stop condition is met, and calculating the pose of the target workpiece relative to the camera coordinate system through inversion operation. The invention also relates to an apparatus comprising a memory and a processor implementing the above method steps when the processor executes a program stored in the memory.

The invention discloses a mechanical equipment fault diagnosis method based on deep learning. The method specifically comprises the following steps of S1, carrying out the data collection and preprocessing of a main data source and a secondary data source of mechanical equipment, and obtaining a data set; S2, a five-fold cross validation method being adopted to divide the data set into a trainingset, a validation set and a test set; and S3, establishing a fault diagnosis model based on the CNN and the BD-LSTM, inputting the training set into the fault diagnosis model, extracting hidden features, performing training, and outputting a diagnosis result. According to the method, BD-LSTM is adopted to perform smooth tracking and result prediction, and uncertainty caused by operation and environmental interference is processed, sensor monitoring data adopts CNN and BD-LSTM to extract hidden features in parallel, output of two irrelevant paths can influence prediction, and each parameter inthe network can be corrected according to predicted errors.

The invention relates to an intelligent control method for an aircraft steering engine electro-hydraulic servo system, which comprises the steps that an improved artificial bee colony algorithm moduleand a PID controller module form a controller; system error information output by a force sensor and a displacement sensor is obtained in real time by using the improved artificial bee colony algorithm, the fitness is calculated, and an optimal food source is searched to serve as PID controller parameter output; the PID controller module outputs a loading force instruction signal to an electro-hydraulic servo valve by using the system error information output by the force sensor and the displacement sensor and PID controller parameters outputted by the improved artificial bee colony algorithmmodule so as to drive a valve-controlled hydraulic cylinder to move and generate a loading force, the loading force is loaded to an aircraft steering engine via a buffer spring and the force sensor,and the aircraft steering engine carries out corresponding motion according to the loading force instruction signal. The control method effectively improves the loading accuracy, response speed, tracking effect and stability of the aircraft steering engine electro-hydraulic servo system, and realizes effective suppression for the redundant force interference of the system.

The invention discloses a method for reducing the loss of a micro power grid by establishing a reactive power optimization model of the micro power grid and calculating the reactive power compensation of a capacitor bank in the micro power grid by adopting the optimization algorithm when the current network has the lowest active loss. The reactive power compensation is calculated by adopting the optimization algorithm comprises the following steps: (1) generating an initialization group; (2) calculating the network loss of individuals in the group, searching the individuals with the lowest active loss, saving as the optimal values, and recording the position; subjecting all the individual positions to iteration update according to a update formula, searching the individuals with lowest active loss, saving as the new optimal values and (3) comparing the new optimal values with the original optimal values, returning the individuals of the new optimal values to the last-time iteration position if some optimal values are updated, and outputting the final optimization result after finishing the iterative computation. Compared with the prior art, the method has the advantages of reducing the loss of the micro power grid and improving the overall utilization rate of the electric energy. The selected optimization algorithm is not liable to lead to the local optimization; the equation parameters used for iterative computation are few and convenient to adjust and are more stable.

The invention discloses a weak supervision target detection method and a weak supervision target detection system, which are used for training a target detector to detect a target in a picture under the condition of only annotation of an image category, and can save a large amount of manpower, material resources and financial resources consumed by annotation information. In the prior frame generation part, a selective search algorithm and a gradient weighted class activation mapping method are combined to generate a better prior frame, and meanwhile, in the optimization iteration process of a detector, supervision information of low-level features is added, and the concept of likelihood is introduced to measure the degree that a target in the prior frame is a complete target. The problem that a current weak supervision target detection method is prone to falling into a local optimal pain point, so that a network tends to select a prior frame covering a whole target under the condition that no target bounding box information supervision exists is solved. The network improves the performance of weak supervision target detection, and can be used in the fields of automatic driving, intelligent security and protection and the like; experimental results show that the method has good competitive performance.

The invention provides an ultra-short-term prediction-based smooth new energy power generation control method for an energy storage system. The method comprises the following steps: reading related operation data of new energy and the energy storage system; building a target function on the basis of an ultra-short-term prediction power and a charged state of the energy storage system; optimizing six control variables in a control strategy by an adaptive chaotic particle swarm optimization algorithm according to the target function; obtaining a power command value of the energy storage system on the basis of the optimal solution of the control variables and carrying out power limitation on the power command value of the energy storage system; updating the control variables in a rolling manner according to the characteristic that ultra-short-term prediction forecasts once every 15 minutes; and outputting the power command value of the energy storage system to an energy storage control system to execute control on the energy storage system, and achieving a smoothing function of new energy output. By the ultra-short-term prediction-based smooth new energy power generation control method for the energy storage system, the charged state of the energy storage system is kept in an appropriate level; the continuous charging and discharging capabilities of the energy storage system are improved; and cooperative optimization of the smoothing capability and the performance index of the energy storage system is achieved.

The invention discloses a DG grid connection optimization configuration method based on the improved genetic algorithm. The method comprises the following steps: Step 1, establishing an objective function based on system losses, voltage deviations and economical performances; Step 2, establishing equality constraints according to the objective function; Step 3, establishing inequality constraints;and Step 4, using the reciprocal of the objective function as a fitness function, and calculating the optimal solutions through the genetic algorithm to obtain an installation position and an installed capacity of a DG. The method disclosed by the invention has the advantages that the problem in the prior art that voltages of nodes in a power distribution system cannot be effectively improved after adoption of an installation position and an installed capacity of the DG because the installation position and the installed capacity of the DG are determined mostly based on empirical equations and human judgments is solved, the system losses are reduced, and the system operation stability and the power supply safety are improved.

The invention discloses an intelligent identification method for a hydroelectric generating set model. The intelligent identification method comprises the steps of: creating a corresponding identification system model according to a water turbine speed regulation system, acquiring an actual response signal outputted by the water turbine speed regulation system under the excitation of a given inputsignal, and acquiring an analog response signal outputted by the identification system under the excitation of the given input signal; defining a difference value between the actual response signal and the analog response signal as a target function, and performing iterative optimization on to-be-identified parameters by adopting a whale optimization algorithm to minimize the target function to obtain optimal identification parameters of the hydroelectric generating set; and increasing the global search probability by balancing random search and optimal search in the iteration process. According to the intelligent identification method, the global search probability is increased in the traditional whale optimization algorithm, immune operators are fused, the search space is adjusted by adopting a self-adaptive correction method, the optimization efficiency is improved, the intelligent identification method has the advantages of high convergence speed, short calculation time and high efficiency, and the identification precision is effectively improved.

The invention provides an economic optimization method of a microgrid containing wind power and photovoltaic power generation. The method comprises the following steps: constructing a microgrid operation data set; establishing an operation optimization target function; establishing a constrain condition of the operation optimization target function; constructing an improved particle swarm optimization algorithm model, and outputting an optimal location; and obtaining the least cost of the optimal solution according to the optimal location and the operation optimization target function, and accomplishing the microgrid economic optimization. Through the microgrid economic optimization method provided by the invention, the diversity of the particle is increased, the capacity of searching theglobal optimum is increased, and the method is hard to trap into the local optimum. And meanwhile, the searching capacity on the optimal solution by each particle is further improved by adopting self-adaptive inertia weight and learning factors, better optimization effect can be acquired when performing economic optimization on this cooling-heating-power cogeneration microgrid, the problem that the economic optimization problem is easy to trap into the local optimum is effectively solved, and the better economic optimization effect is acquired.

The invention provides an optimization configuration method of an active filter. The optimization configuration method comprises the following steps of setting a target function according to influence on configuration of the active filter; setting a constraint condition conforming to optimization configuration of the active filter of an intelligent power distribution network; and performing optimization configuration on the active filter of the intelligent power distribution network by a configuration algorithm combined with a mode analysis method and a genetic algorithm. According to the optimization configuration method of the active filter, provided by the invention, a candidate position node of the active filter is determined by introducing the mode analysis method, the calculation quantity of the genetic algorithm for determining a configuration position of the active filter is reduced, and the problem that the genetic algorithm is easy to fall into local optimum is solved; and moreover, the operational speed of the optimization configuration method is fast, rapid convergence of the algorithm is facilitated, a better optimal solution can be found out, the optimal access position of the active filter can be effectively calculated, and the optimization configuration method has a good effect on improving the running economy of the system and improving the electric quality.

The invention discloses an MPPT control method based on improved quantum particle swarm algorithm. Based on the quantum particle swarm algorithm, an improved quantum particle swarm algorithm is proposed. In the QPSO algorithm, the chaotic search strategy is added to form the JQPSO algorithm to realize the adaptive search of the algorithm. This improved strategy not only improves the convergence precision of the algorithm but also improves the convergence speed of the algorithm. In practical applications, due to the phenomena that a photovoltaic panel often experiences partial overshadowing and that the output voltage from a photovoltaic array output are likely to have multiple peaks, a traditional algorithm cannot track the maximum power point correctly under these conditions. Therefore, the JQPSO algorithm adopted by the invention tracks the maximum power point of the PV panel; through the adjustment of the duty cycle d of the switching power tube and through the use of the MATLAB/Simulink experiment, the simulation results show that the JQPSO is better to seek the optimization and the proposed JQPSO algorithm can find the maximum power point in the shortest possible time under the condition that the search accuracy is guaranteed.

The invention discloses a distribution network reactive power optimization method oriented to multiple random uncertainty. The method comprises the following steps: acquiring power grid data; substituting the data into a pre-built reactive power optimization model; and solving the reactive power optimization model through a particle swarm optimization algorithm to obtain an optimal solution so asto achieve distribution network reactive power optimization, wherein the reactive power optimization model takes the lowest active loss as a target, and equality and inequality constraints are introduced. The randomness of distribution network loads, the randomness of distributed generation output and the randomness of a reactive compensation device are considered at the same time, so that the actual condition of a distribution network is more accurately reflected.

The invention discloses a precise splicing method of large complex curved surface multi-view scanning point cloud, and belongs to the field of three-dimensional point cloud registration. According tothe invention, a hybrid optimization algorithm of a fruit fly optimization algorithm and an improved nearest point iterative algorithm is adopted to realize fine splicing of large complex curved surface point clouds, and the problems of low convergence rate and easy falling into local optimum in the prior art are solved by combining the local high-efficiency optimal search capability of an ICP algorithm and the global optimal search capability of an FOA algorithm. The precise splicing precision of the large complex curved surface multi-view scanning point cloud is effectively improved. The roughly spliced point cloud is processed through an FOA-ICP algorithm and then a multi-view cloud spliced in a high-precision mode is output, and the surface topography of a measured object is truly reflected. According to the method, the FOA splicing algorithm is improved, the optimal target parameter is simplified into the three-dimensional translation vector from the six-dimensional vector, the search of the abandoned three-dimensional rotation vector is compensated through the fused ICP algorithm, and the search efficiency is improved.

The invention discloses a multi-target interval value fuzzy clustering image segmentation method based on double membership driving, which mainly solves the problems of being sensitive to noise and easy to fall into local optimum in image segmentation. The scheme includes: inputting an image to be segmented, and setting initial parameter values; constructing an interval value blurred image; constructing a global interval value fuzzy compactness function JLN driven by double membership degrees and an interval value fuzzy separability function SLN driven by double membership degrees, and performing multi-objective evolution on the two objective functions to obtain a non-dominated solution set P; calculating an interval value selective solution index W driven by double membership degrees, and selecting an optimal chromosome from the non-dominated solution set P by using the index to decode the optimal chromosome to obtain an optimal clustering center; and updating the joint membership matrix by using an optimal clustering center, and obtaining a classification result of the pixel points according to a maximum membership principle. According to the method, noise can be effectively inhibited, local optimum is prevented, the segmentation accuracy is improved, and the method can be used for natural image recognition.

The invention discloses a welding seam recognition device and recognition method suitable for flat plate butt welding seams, the welding seam recognition device comprises a sliding device and an adjusting device, the end of the sliding device is provided with a driving device, the rear side face of the sliding device is provided with a fixing device, and the top of the sliding device is provided with a horizontal regulator and an industrial personal computer. A fixing device is arranged at the end of the adjusting device and used for fixing an industrial camera and a laser pen which are used for recognition. According to the recognition method, aiming at the limitation that segmentation of a current FCM clustering algorithm is easy to fall into local optimum, the FCM clustering algorithm is optimized by searching a clustering center in combination with a mucous flora intelligent algorithm, so that the image quality is obviously improved, and details and contours of an image edge region are clear. The method can be suitable for flat plates with different weld lengths and is high in adaptability; the field installation is rapid, the stability of device operation during identification is good, the identification method is high in identification speed, good in reliability and robustness and high in precision, and the welding seam identification efficiency of a construction site is greatly improved.

The invention discloses a dangerous chemical stacking type storage cargo positioning optimization method, which comprises the following steps: establishing a dangerous chemical stacking type storage cargo positioning monitoring scene, and obtaining the time difference of arrival between a to-be-positioned label and each base station; establishing a TDOA-based positioning model in combination withthe obtained time difference of arrival, thereby determining an objective function of the position of the to-be-positioned tag; and searching an optimal solution of the objective function by utilizinga PSO algorithm for dynamically adjusting the inertia weight and the acceleration weight so as to obtain the position of the to-be-positioned tag. The method aims at the defects that an existing PSOalgorithm is prone to falling into local optimization and search stagnation occurs, the algorithm is improved, and inertia weight and acceleration weight are dynamically adjusted; The problem that theparticle swarm optimization algorithm is liable to fall into local optimum is effectively solved, so that the algorithm can be quickly converged to a globally optimal solution, and the algorithm is stable in performance and high in positioning precision; And a good technical means is provided for government departments to carry out dangerous chemical storage and mixed storage monitoring.

The invention relates to a CCHP microgrid operation optimization method. The method comprises the steps of S1, constructing a microgrid system model; s2, constructing a microgrid operation data set; s3, establishing a target function of the microgrid economic operation optimization model; s4, establishing constraint conditions of the microgrid economic operation optimization model; and S5, solvingan optimal solution of the economic optimization problem of the CCHP microgrid based on a Q learning and quantum particle swarm optimization algorithm. According to the method, population initialization is optimized by utilizing a chaos optimization algorithm, so that the diversity of particles can be increased, the global optimum searching capability is improved, and the probability of falling into global optimum is low. Meanwhile, the search capability of each particle for the optimal solution is further improved by adopting a crisscross algorithm, a better optimization effect can be obtained when economic optimization is carried out on the cold-heat-electricity combined supply microgrid, the problem that the economic optimization problem is likely to fall into local optimization is effectively solved, and a better economic optimization result is obtained.

The invention discloses a subway train delay adjustment method considering passenger flow influence and regenerative braking energy utilization, and the method comprises the steps: building a passenger flow prediction model based on an LSTM (Long Short Term Memory) in consideration of the constraint of a passenger flow factor on a dwell time adjustment upper limit after a train is delayed; in consideration of absorption and utilization of regenerative braking energy in the delay train adjustment process, a direct-current traction network energy consumption calculation model is established; a station dwell time model is established by simulating the getting-on and getting-off process of passengers; then, a subway train delay adjustment model is established in combination with the submodels, the minimum energy consumption variable quantity of a transformer substation in the adjustment process serves as an optimization target, and a solving process is designed for the model through a particle swarm optimization algorithm CDPSO based on center-discrete learning; and finally, performing simulation verification by using actual subway line data. The method is scientific, reliable and efficient, provides data and theoretical support for subway train delay adjustment, and has high use value and application prospect.

The invention discloses an image segmentation method based on strong and weak joint semi-supervised intuitionistic fuzzy clustering, and mainly solves the problems that the existing image segmentation is sensitive to an initial value, is easy to fall into local optimum, and is linearly inseparable to low-dimensional data. According to the scheme, a to-be-segmented image is input, initial parameters are set, and manual lineation is carried out; carrying out intuitive fuzzy processing on the image; designing a strong and weak combined semi-supervised strategy to obtain a strong supervised membership degree, a weak supervised membership degree and an initial clustering center; introducing the kernel function, the strong supervision membership degree and the weak supervision membership degree into an intuitionistic fuzzy clustering objective function to obtain a strong and weak combined semi-supervised kernel intuitionistic fuzzy clustering objective function; minimizing the objective function by adopting a Lagrange multiplier method to calculate a clustering optimal solution; and obtaining a classification result of the image pixel points according to a maximum membership degree principle. According to the method, the sensitivity to an initial value is improved, local optimum is prevented, the segmentation accuracy of linear inseparable data is improved, and the method can be used for natural image recognition.