324 results about "Descent algorithm" patented technology

The invention discloses a dynamic self-adaptive recommendation strategy optimization method for a business handling failure scene, and relates to the technical field of business handling recommendation and intelligent decision making, and the method comprises the steps: firstly collecting various types of data of a whole business handling process, and guaranteeing the integrity and real-time performance at a frequency of 100 milliseconds per time; a decision tree and Bayesian network fusion algorithm is used for attribution, and direct and indirect reasons are clarified; integrating data to construct a user portrait, and mining potential and subsequent demands; generating a recommendation scheme set based on attribution and portraits, and adjusting priorities and forms in combination with scene features; feedback data is introduced, a strategy weight is optimized by using a gradient descent algorithm, and the scheme is updated regularly; a multi-dimensional index weighted evaluation effect is set, and emergency optimization is carried out if the evaluation result does not reach the standard; and establishing a distributed strategy library, and reusing the optimal strategy of the similar scene by using a K-nearest neighbor algorithm. According to the method, failure reason accurate positioning and personalized recommendation are realized, the recommendation effect is continuously optimized along with data accumulation, and the method is adaptive to multiple service types and user groups.

The invention provides a slope displacement monitoring data processing system based on an unmanned aerial vehicle laser radar, and relates to the technical field of data processing, and the system comprises the steps: carrying out the spatial interpolation processing of a topographic feature data set, and constructing a digital topographic surface model; the displacement field calculation module is used for performing iterative optimization based on a digital terrain surface model through spatial similarity analysis and fusion with a gradient descent algorithm, calculating a slope surface displacement vector field, identifying a potential sliding surface and a deformation abnormal region, and generating a displacement field calculation result; and the evaluation module is used for inputting a displacement field calculation result into a risk evaluation model and carrying out slope stability quantitative evaluation through a multi-source data fusion analysis platform. According to the invention, the practicability and operability of the monitoring result are improved.

The present invention relates to the technical field of natural ventilation in buildings, and in particular to a computationally assisted decision-making method and system for climate-adaptive building cavity design. The method comprises: using a Delaunay triangulation method to generate an initial mesh, using a Laplace operator-based mesh refinement method to adaptively refine the initial mesh, so as to obtain an adaptive mesh system, wherein the adaptive mesh system is used for dynamically adjusting the mesh density of conditional PINNs; constructing a multi-task learning framework within the conditional PINNs, wherein the multi-task learning framework is used for jointly predicting a plurality of physical field variables within the conditional PINNs; and iteratively training the conditional PINNs, and using a gradient descent algorithm to minimize a loss function until a predetermined number of training iterations or loss convergence is reached, thereby generating a key physical field variable prediction model in building cavity design. The present invention can implement efficient and accurate prediction of key physical quantities in building cavity design, and can be adapted to different building layouts and functional space characteristics.

The invention discloses an automobile part assembly precision intelligent compensation method and a self-adaptive regulation and control system, and relates to the field of intelligent compensation, and the method comprises the steps: synchronously collecting part geometric parameters, tool poses and environment data, and constructing an associated data matrix; analyzing data by using an improved random forest-attention model, and outputting a deviation factor contribution degree sequence; a compensation calculation model is designed accordingly, initial compensation amounts are generated in a segmented mode in combination with a precision margin threshold value, and correction is conducted through historical data similarity matching; selecting an execution path according to the compensation amount and the core deviation type; an actual precision value is obtained through laser detection after assembly, and a compensation error is calculated; based on an error triggering model optimization mechanism, model parameters are iteratively updated by using a gradient descent algorithm, and the deviation identification and compensation precision is improved. The method has the advantages that the model attribution deviation is improved, the precise compensation amount is calculated in combination with historical data, flexible execution, real-time monitoring and model self-optimization are matched, and the assembly precision and the production efficiency are efficiently improved.

The invention relates to the field of twitter recognition, in particular to a twitter classification method based on harmonic enhancement and time-frequency semantic joint modeling, which comprises the following steps: collecting twitter samples and carrying out noise reduction and standardized preprocessing, carrying out multi-scale convolution operation on Mel spectrograms by utilizing a layered acoustic encoder, extracting time-frequency features in combination with a channel attention mechanism, and classifying twitter classification results. The method comprises the following steps of: generating adaptive position codes through a dynamic time-frequency joint coding module, carrying out time-frequency mode modeling by combining a global-local interaction mechanism, introducing a semantic fusion module which comprises a frequency band pyramid unit, a harmonic enhancement unit and a time-frequency gating unit, realizing dynamic weighted fusion of multi-layer features, and carrying out time-frequency mode modeling through a global-local interaction mechanism. And inputting the fusion features into a classification layer, training a network by adopting a cross entropy loss function and a gradient descent algorithm, and outputting bird categories through a full connection layer, thereby solving the key problems of insufficient description of a non-stationary time-frequency mode, insufficient modeling of a harmonic structure, reduction of recognition performance in a complex noise environment and the like in the prior art.

The invention discloses a multi-dimensional training method and device for a support vector machine, electronic equipment and a computer readable storage medium, and the method comprises the steps: carrying out the discretization of a training sample data set, and obtaining a discretized data set, the discretized data set comprises a plurality of different attributes, and each attribute corresponds to a plurality of feature vectors; calculating a classification contribution parameter of each attribute feature vector to obtain a plurality of classification contribution parameters; performing data mapping on the plurality of classification contribution degrees by using a kernel function to obtain a target function; and optimizing and training the objective function by using a gradient descent algorithm to obtain a support vector machine model. According to the method, different dimension data are mapped through the kernel function, the data gain weight can be determined, the linear separable effect of the mapped dimension data can be improved, and then the classification and analysis capability of the SVM can be improved.

The invention relates to the technical field of antenna control, and discloses a control method for a leaky-wave antenna and a communication device. The method comprises the following steps: measuring the dispersion characteristic of the leaky-wave antenna, establishing a second-order nonlinear mapping relation between a propagation constant and a control voltage, and obtaining a dispersion control parameter group; reversely deducing the target beam directional angle to obtain a target propagation constant, and calculating an initial voltage value of each control unit in combination with the parameter group to form a voltage gradient control sequence; correcting the propagation constant of each unit by adopting a distance compensation algorithm, and recalculating to obtain distance self-adaptive voltage distribution data; and detecting a beam quality factor in real time, updating voltage distribution data through a gradient descent algorithm when the beam quality factor deviates from a target value, and outputting a control voltage sequence to drive each control unit. The leaky-wave antenna control method and device solve the problems that in an existing leaky-wave antenna control technology, the beam quality is reduced and the data rate is sharply attenuated during long-distance transmission, and the beam control precision and the long-distance transmission performance of the leaky-wave antenna in the terahertz frequency band are improved.

The invention discloses a numerical control machine tool dynamic error cooperative compensation control method adaptive to high-speed machining, and relates to the technical field of numerical control machine tool error control. According to the method, a multi-source dynamic error sensing system comprising a grating displacement sensor, a six-dimensional force sensor and the like is constructed to acquire data; after wavelet threshold denoising and Kalman filtering preprocessing, inputting a three-layer LSTM error coupling prediction model combined with an attention mechanism, embedding a servo motor load characteristic curve in the model, and outputting three types of error compensation amounts; through servo-level compensation and machining-level compensation, the position of a feed shaft, the rotating speed of a main shaft, the cutting feed rate and the behavior of a micro-displacement actuator are corrected, and machining errors caused by deflection and vibration conduction of the main shaft are counteracted. And iteratively updating model parameters by using a gradient descent algorithm. According to the method, through multi-source error synchronous sensing, error coupling modeling and hierarchical cooperative compensation, dynamic error cooperative control more adaptive to a high-speed processing scene is realized, and the method has a wide application value.

The invention discloses a cross-scale self-supervised spatio-temporal trajectory basic model training method, which comprises the following steps of S1, preprocessing original GPS (Global Positioning System) data to obtain trajectory data of different spatial granularities; s2, encoding the trajectory data of the spatial granularities to obtain trajectory representations corresponding to the spatial granularities; s3, on the basis of the trajectory representation of each spatial granularity, randomly selecting a part of elements to execute mask operation, and calculating the difference between the predicted trajectory data and the real trajectory data as mask reconstruction loss; s4, performing comparative learning training on the positive sample pair and the negative sample pair to construct comparative learning loss; s5, performing dynamic fusion on the trajectory representation of each spatial granularity and the importance weight to obtain a unified representation after trajectory fusion; and S6, performing weighted summation training on the mask reconstruction loss and the contrast learning loss through weight hyper-parameters to obtain total training loss of the model, updating model parameters to convergence by adopting a gradient descent algorithm, and generating a pre-training model.

The invention relates to the technical field of information processing, and discloses a method for dynamically regulating and controlling production parameters of a plastic film by utilizing program process scheduling. The method comprises the following steps: converging a real-time parameter sequence to a digital twinborn body to establish virtual mapping, and constructing a multi-dimensional data model; analyzing a material state evolution trend in the digital twinborn body, synchronizing with a physical production line state, fusing historical data to predict a potential influence range, and generating a regulation and control demand report; generating an adjustment scheme draft, performing priority ranking on simulation tasks in the adjustment scheme draft by adopting a program process scheduling algorithm, dynamically allocating computing resources, executing the simulation tasks in parallel, and determining an optimization parameter set in combination with a gradient descent algorithm; and iteratively correcting the parameters in the digital twinborn body through feedback circulation to generate a final coordination control instruction, inputting the final coordination control instruction into the digital twinborn body, confirming synchronism and outputting a report. According to the invention, the precision of dynamic regulation and control of the production parameters of the plastic film is improved.

The invention discloses a water supply pipe network water supply amount prediction method based on deep learning. The method comprises the following steps: analyzing a convolutional neural network CNN and a long short-term memory network LSTM; constructing a water supply prediction model based on LSTM, analyzing the water supply prediction model, and setting training parameters; further fusing the CNN and the LSTM, and constructing a CNN-LSTM combination model to obtain an optimized water supply prediction model; and carrying out verification and comparative analysis on the optimized water supply prediction model. According to the method, the CNN-LSTM model is constructed, the model adopts an Adam gradient descent algorithm to optimize a traditional algorithm of the model, and the CNN-LSTM model strengthens the time sequence modeling capability of the LSTM through CNN feature extraction, so that the model not only can identify the microscopic fluctuation of water supply amount data, but also can grasp the macroscopic evolution rule of the water supply amount data; the problem that a traditional model is poor in stability and low in reliability is solved.

The invention discloses a method and a system for detecting a jailbreak attack of a large language model, which are applied to discovering potential security risks of the large language model, and the method comprises the following steps: constructing a jailbreak attack-malicious query data set of the large language model; an automatic antagonism suffix generation method fusing cluster search and a gradient descent algorithm is provided; the detection system of the prison break attack large language model is constructed by training, testing and evaluating the validity of the resistant suffixes based on a prison break attack data set, and comprises a malicious query data set display unit, a detection validity verification unit, a comparison and analysis unit before and after an attack and a model defense strategy unit. The method is helpful for discovering defects of a large language model alignment mechanism, so that a more powerful large language security model is designed and constructed.

The invention discloses an ocean engineering structure extended Kalman filter parameter estimation method based on attitude angle virtual observation, and relates to the technical field of ocean engineering, and the method comprises the following steps: S1, initializing a system process noise covariance matrix and an observation noise covariance matrix, and obtaining an extended Kalman filter attitude angle estimation value; s2, establishing a high-precision angular velocity-attitude angle dynamic conversion model, generating attitude angle virtual observed quantity and using the attitude angle virtual observed quantity for online evaluation of a Kalman filtering result; s3, setting a multi-dimensional performance evaluation index system, and quantitatively evaluating the amplitude error and trend stability of the extended Kalman filter estimated attitude angle; s4, adjusting scale factors of a system process noise covariance matrix and an observation noise covariance matrix by using a gradient descent algorithm based on the attitude angle virtual observation quantity and the error measurement of the multi-dimensional performance evaluation index, and carrying out real-time iteration on the scale factors of the system process noise covariance matrix and the observation noise covariance matrix; the problem that attitude parameters cannot be accurately measured and predicted in real time in the prior art is solved.

The invention proposes a steel slag asphalt interface adhesion work prediction method based on a physical neural network, and relates to the technical field of deep learning, and the method comprises the steps: obtaining chemical component parameters and environmental impact factors of a steel slag asphalt mixture interface sample; creating an original neural network prediction model based on a diffusion item, a reaction source item, a multi-mechanism dissipation item and a state variable corresponding to spatial-temporal evolution of the adhesion work; inputting the chemical component parameters and the environmental impact factors into an original neural network prediction model, and generating interface adhesion work prediction parameters in combination with physical constraint terms; training the original neural network prediction model by using the interface adhesion work prediction parameters and a gradient descent algorithm to obtain a target neural network prediction model; the physical constraint items comprise data loss, PDE residual error loss, state equation loss and boundary condition loss; and inputting the chemical component parameters and the environmental impact factors of a to-be-tested material into the target neural network prediction model to obtain target prediction parameters.

This invention belongs to the field of CAD manufacturing information technology and relates to an efficient interference inspection method for complex structural CAD models. It employs a collision detection optimization algorithm using a layered bounding box assembly tree and synchronous recursive descent, comprising the following steps: Step 1: Establishing quantitative evaluation indicators; Step 2: Optimizing the layered bounding box assembly tree algorithm; Step 3: Recursively applying the results of the optimized layered bounding box assembly tree algorithm using a synchronous recursive descent algorithm. This invention uses a collision detection optimization algorithm combining a layered bounding box assembly tree and synchronous recursive descent, and uses synchronous recursive descent to calculate the nodes to be compared, thus solving the problems of high time complexity and inability to effectively balance efficiency and accuracy in existing methods.

The embodiment of the invention provides a ship automatic berthing control method and related equipment, and belongs to the technical field of ship berthing control. The method comprises the following steps: acquiring ship states, control input and path and speed expectation plans of a task ship in a current period and a previous period; acquiring a sensitivity matrix in the previous period; calculating based on the ship state and control input of the task ship in the current period and the previous period to obtain an updated sensitivity matrix; calculating according to the updated sensitivity matrix, the ship state and control input in the current period and the previous period, the path and speed expectation plan and a preset multi-target cost function to obtain a control input gradient; and based on a preset gradient descent algorithm and the control input gradient, updating the control input and outputting a control instruction. According to the embodiment of the invention, the problems of strong model dependence, poor real-time performance, insufficient steady-state precision, poor robustness and difficulty in giving consideration to both the navigational speed and the navigational direction in the prior art can be solved.

Disclosed in the present invention are an intelligent fermentation tank group management method and system based on the optimal fermentation time of soy sauce. The method comprises: collecting production data of a fermentation tank group and storing same as historical production data of the fermentation tank group; preprocessing the historical production data of the fermentation tank group; on the basis of the preprocessed historical production data, calculating a correlation coefficient between fermentation process data and product quality indexes; on the basis of the preprocessed historical production data, establishing a neural network model for a mathematical relationship between the fermentation process data and product quality index data; using the preprocessed historical production data as sample data, and using a gradient descent algorithm to train the neural network model; on the basis of the fermentation process data of the fermentation tank group collected in real time, using the trained neural network model to output predicted product quality index data of the fermentation tank group; and on the basis of the product quality index data, establishing a dynamic management mathematical model, with the objective of optimizing the optimal fermentation time of soy sauce and the overall production quality of the fermentation tank group, and using a genetic algorithm to calculate a solution of the optimization objective, so as to realize intelligent management of the fermentation tank group. In the present invention, big data, a neural network model, and a genetic algorithm are combined, and the overall optimization of quality indexes of products from fermentation tank groups is achieved by calculating the optimal fermentation time of soy sauce, so as to realize intelligent management of fermentation tank groups.

The invention relates to the technical field of Internet of Things security, in particular to a network attack early warning method based on an abnormal behavior map. Comprising; constructing a subject-behavior-resource three-layer heterogeneous graph, establishing a low-activity baseline model based on historical data, calculating weak association strength between nodes by adopting wavelet decomposition and a signal enhancement algorithm, and identifying a latent attack chain; the attack chain features are converted into multi-dimensional vectors, and an attack collaboration time window and intensity are predicted through a vector convergence angle and a distance change rate; the vulnerability of the graph topology is quantified, the edge weight is dynamically adjusted by using a gradient descent algorithm, and high-risk nodes are isolated; the convergence parameters are fed back to the baseline model and correlation strength calculation to form closed-loop optimization; and calculating a comprehensive threat index based on three-dimensional weighting of the latent threat intensity, the convergence urgency degree and the topology anti-attack capability, and realizing layered early warning. According to the method, the APT attack detection rate is remarkably improved, and the early warning response time is shortened.

The invention discloses an optimization method of an integrated micromodule based on complex power distribution, and relates to the technical field of integrated circuit thermal management, and the method comprises the steps: building a power distribution characteristic matrix through multi-physics field coupling analysis based on a structured data set, and generating micro-channel geometric parameters through a gradient descent algorithm; constructing a structure heat conduction efficiency mapping model, obtaining heat resistance characteristic change data of a flow channel section in combination with micro-flow channel geometric parameters, and outputting a heat conduction efficiency optimization matrix; thermal resistance distribution characteristics in the heat conduction efficiency optimization matrix are extracted, and a collaborative optimization parameter instruction is generated in combination with flow channel section thermal resistance characteristic change data; and executing the collaborative optimization parameter instruction to obtain thermal power distribution characteristics, calculating an equivalent thermal resistance gradient, and obtaining a heat dissipation efficiency regulation and control instruction through dynamic mapping. By constructing the structure heat conduction efficiency mapping model, the real-time prediction from the geometric parameters to the thermal performance of the micro-channel is realized, the efficiency is improved, and the overall performance is ensured to be optimal.

The embodiment of the invention provides a robot process path data processing method and device based on geometrical characteristics, and the method comprises the steps: obtaining a process template according to the geometrical characteristics of a workpiece and a set process association rule base, and the process template comprises a process sequence and associated process parameters; extracting process sequence keywords, performing algorithm matching with a process path algorithm library to obtain a path generation algorithm, generating an initial robot motion path according to the path generation algorithm and geometric features, and performing path constraint based on process parameters; constructing a hierarchical bounding box tree structure of an obstacle, performing route interference condition detection on an initial robot motion path and the hierarchical bounding box tree according to a ray projection method, if a collision condition occurs, iteratively calculating an obstacle avoidance path offset based on a gradient descent algorithm, and recording a normal vector at a collision point to obtain a correction direction; and correcting the initial robot motion path according to the obstacle avoidance path offset and the correction direction. The robot process path planning method and device can improve the efficiency and accuracy of robot process path planning.

The invention relates to a dynamic light beam shaping system and method based on a spatial light modulator. The system comprises a laser light source module, a beam expanding collimation light path, a light beam modulator, a focusing optical assembly, a detection and feedback module and a control and optimization module. According to the method, the uniformity of the output flat-topped beam can be corrected by virtue of a random parallel gradient descent (SPGD) algorithm and combining a Zernike polynomial as a control variable. Meanwhile, a composite performance index comprehensively considering energy distribution and shape features is constructed, the correction process of the flat-topped beam is accelerated by adopting a mode of gradually improving Zernike polynomial orders in multiple stages, the convergence trend of the performance index is used as a judgment basis of stage switching, stage independent variable dimension expansion is implemented, and the correction precision of the flat-topped beam is improved. And the convergence speed is obviously improved. Besides, adaptive selection of disturbance amplitude and learning rate parameters in the SPGD algorithm is realized by using noise estimation and a local linearity index, and the robustness of the system to noise and environmental disturbance is enhanced.

The invention provides a single-bit digital beam forming method and device, a terminal and a medium, and relates to the technical field of radar detection and wireless communication, and the method comprises the steps: constructing a target single-bit low-rank matrix recovery model of a uniform linear array comprising a plurality of sensors, and an optimization problem of the target single-bit low-rank matrix recovery model, and converting the target single-bit low-rank matrix recovery model into a processable problem, solving and determining a low-rank matrix by using a near-end block coordinate descent algorithm and a gradient descent algorithm, calculating an estimated covariance matrix of a target and an interference signal based on the matrix, performing eigendecomposition to calculate an interference subspace, and modeling a single-bit digital beam forming problem according to a base vector of the interference subspace to obtain a single-bit digital beam forming matrix. And calculating a closed-form solution based on the principle that the optimal solution is located in the orthogonal complement space of the interference subspace. According to the method, the resolution requirement of the analog-to-digital converter can be reduced by fully utilizing the single-bit quantization characteristic, and the beamforming performance of single-bit digital beamforming is close to that of high-bit digital beamforming when the hardware complexity and the power consumption are reduced.

The invention relates to a multi-modal combined image retrieval method fusing fine-grained semantic positioning and optimization generation features, which comprises the following steps: acquiring a reference image, predicting a bounding box of a target object based on the reference image and a positioning keyword, and dividing a background retaining region and a foreground editing region; semantic divergence prompt words of description information are obtained, text features in the semantic divergence prompt words are extracted to serve as guide targets to be used for establishing a composite optimization target function, a gradient descent algorithm is adopted to conduct iterative updating on learnable random noise vectors, and potential visual feature vectors are generated; the foreground editing area is filled with the image data, a background-foreground mixed feature map is generated, the fine-grained visual similarity between the background-foreground mixed feature map and the dense visual feature map is calculated, and the global text feature similarity between global text feature vectors in description information and global visual feature vectors is calculated; and carrying out weighted fusion on the fine-grained visual similarity and the global text feature similarity, and outputting a retrieval result.

The invention discloses a self-adaptive air pressure regulation and control system of quartz sand unpowered powder selecting equipment, and relates to the technical field of automatic control. According to the method, wind pressure and airflow disturbance signals are collected, a high-resolution map is generated through Hampel filtering and noise reduction, features are extracted through the convolutional neural network, key influence areas are identified, the problem of unstable wind pressure caused by airflow disturbance in a mountainous area is solved, and the particle separation precision is improved; a natural potential energy distribution matrix is constructed based on building floor height difference, velocity field reconstruction and airflow channel optimization are driven, technological parameters are adjusted in combination with a gradient descent algorithm, precise utilization of natural potential energy is achieved, energy consumption is reduced, uneven efficiency is improved, airflow distribution is simulated through a fluid dynamic model, and sedimentation behaviors are analyzed and predicted through particle trajectory. A support vector machine is used for classifying tracks and generating regulation and control instructions, PID control is used for stabilizing air pressure to form a closed loop, the problems that fine powder is mixed with coarse powder, coarse particles are left and the like are avoided, the equipment blockage risk is reduced, and the product purity stability is improved.

The invention relates to a port area illumination energy saving and visual performance comprehensive evaluation method based on multi-objective optimization, and the method specifically comprises the following steps: deploying an illumination sensor and an intelligent electric meter at a key position of a port area, and collecting illumination and energy consumption parameters to form a training data set containing a performance label; defining a sensitivity coefficient and a normalization index for each data feature and completing normalization calculation; constructing a comprehensive evaluation model, inputting a normalized value, and then realizing classification prediction through a non-linear interaction kernel illumination-energy consumption feature fusion enhancement module and a double-branch attention feature extraction deep neural network backbone module; fusing multiple constraints to construct a composite multi-objective loss function to calculate model loss; iteratively optimizing the model by adopting a gradient descent algorithm in combination with the training data set until a stop condition is met; and normalizing new data, inputting the normalized new data into the trained model, and selecting the highest probability category as an evaluation result. According to the invention, the energy-saving and visual performance of harbor lighting can be accurately balanced, and effective guidance is provided for optimization of a harbor lighting system.

The invention discloses a task processing method and device based on quantum computing, equipment and a storage medium. The method comprises the steps of determining a maximum cut problem corresponding to a target task and a weighted undirected graph of the maximum cut problem; performing community detection division on the weighted undirected graph to obtain a plurality of community sub-graphs; mapping the community sub-graph into a sub Hamiltonian and constructing a parameterized quantum circuit; updating parameters of the parameterized quantum circuit by adopting a gradient descent algorithm to minimize a global Hamiltonian expected value, and outputting a binary string of a quantum state corresponding to the global Hamiltonian expected value as an initial solution; generating a plurality of candidate solutions for the initial solution based on a neighborhood search algorithm, calculating cut values of the initial solution and the candidate solutions, and selecting a better feasible solution based on the optimal cut value; and applying a preset disturbance operator in the parameterized quantum circuit to construct a disturbance quantum circuit, updating parameters of the disturbance quantum circuit by adopting a gradient descent algorithm, and outputting a binary string of a quantum state corresponding to the global Hamiltonian expected value as a final solution of the target task by minimizing the global Hamiltonian expected value after disturbance.

The invention discloses a federated learning method and system for heterogeneous data of the Internet of Vehicles, and relates to the technical field of intelligent transportation. The method comprises the following steps: a roadside unit sends a current global model and a global control variable to a vehicle participating in training; the vehicle executes local training combined with a client drift correction mechanism through local data and a differential privacy stochastic gradient descent algorithm, and generates local model update; the vehicle calculates a noise multiplier of the next round of training through a personalized self-adaptive differential privacy strategy according to the loss value of the current round of training; the vehicle judges whether the current round of model updating is effective or not based on selective uploading; and if the update is valid, the vehicle uploads the model increment and the local control variable increment to the roadside unit, and the roadside unit updates the global model and the global control variable after aggregation. According to the invention, systematic innovation is carried out from three levels of privacy protection, model optimization and communication strategies, and a solution is provided for constructing efficient, safe and reliable federal learning in an Internet of Vehicles scene.

The invention relates to the technical field of industrial control systems, and particularly discloses a selenium-rich water equipment water yield control system based on neural fuzzy control. The system comprises a data acquisition module, a neural fuzzy control module, an actuator driving module and an online learning and parameter adjustment module, fuzzy rules and membership parameters are dynamically optimized through a neural network, a gradient descent algorithm and a working condition memory unit are combined, high-precision self-adaptive control over the water yield is achieved, and the response speed and stability of the system are improved.

The invention relates to the related field of unmanned aerial vehicle communication, and discloses an unmanned aerial vehicle relay communication link rapid establishment and optimization method in a three-network full-disconnection scene, and the method comprises the following steps: S1, carrying out the dynamic spectrum sensing and anti-interference channel distribution based on cognitive radio; s2, adaptive power-energy consumption joint optimization control is carried out; s3, quickly reconstructing the distributed intelligent route; according to the method, a new-generation anti-interference architecture is constructed based on a Conv-LSTM dynamic spectrum sensing system, the system has the capability of autonomously identifying and avoiding complex electromagnetic interference through real-time spectrum analysis driven by deep learning, an intelligent power regulation and control mechanism guided by a link stability index is adopted, and the reliability of the system is improved. Dynamic optimization of communication energy consumption is achieved, an innovative node dormancy strategy is combined with a gradient descent algorithm, the problem of energy waste in a traditional scheme is effectively solved, and the energy utilization efficiency of the system is remarkably improved.

The invention provides a rotational symmetry point cloud registration method and device based on fuzzy clustering and a storage medium, and the method comprises the steps: extracting inner points through employing an algorithm combining the minimum median square and random sampling consistency, and estimating a rotating main axis through principal component analysis; realizing point cloud global coarse registration based on a rotating main shaft; fuzzy clustering is carried out on the fixed point cloud to construct local geometric description, a progressive constraint mechanism and a directional search strategy based on a main shaft are introduced, and fine registration is completed in combination with branch and bound and gradient descent algorithms. According to the embodiment of the invention, the registration precision and efficiency of point clouds with rotational symmetry, such as a computer-aided design (CAD) model, can be effectively improved.