45 results about "Random search" patented technology

Embodiments of the present disclosure disclose a vehicle escape method and device, electronic equipment and storage medium, the method comprising: in the case where the vehicle is in a blocked state, determining a virtual end point according to the starting point, target lane, each obstacle and preset search distance of the vehicle; generating a sampling map based on the Frenet coordinate system according to the starting point, virtual end point, current lane width, other lane width and each lane line; determining an initial path based on the fast random search tree according to the starting point in the sampling map, virtual end point and each obstacle; determining a plurality of target escape points according to the starting point in the initial path, each initial escape point in the initial path and each obstacle; and performing escape processing on the vehicle according to the plurality of target escape points. The present disclosure can flexibly determine a plurality of escape points when the vehicle is in a blocked state, improve the rationality of the escape points, and improve the success rate of vehicle escape.

The invention provides an operation path planning method and system integrating environmental perception and occupational health, and the method specifically comprises the steps: obtaining a three-dimensional map of a farm, and building a harmful gas prediction model related to an exhaust state; taking a personnel starting and ending point as a root node, performing bias sampling on a low-concentration region according to the model, and respectively constructing a forward and backward random search tree, a path cost comprehensive geometric distance and a predicted average concentration; performing reconnection optimization on the double trees, and updating father nodes of lower-cost adjacent nodes; connecting the optimized dual trees to generate an initial path; accumulated harmful gas exposure of the path is calculated, the gas exposure is the integral of the product of the retention time and the concentration, and target operation parameters of all exhaust equipment along the line are determined accordingly; and updating the prediction model by using the new parameters, repeatedly executing path search and exhaust parameter optimization until the total cost change of adjacent iteration paths is smaller than a convergence threshold value, and outputting an operation path and exhaust cooperation scheme.

The application provides a pulse eddy current pipe wall thickness detection method based on an RSBO optimized LSTM network, pulse eddy current signals are used to estimate the thickness of the pipe, a data sample length evaluation criterion based on a least square fitting algorithm is used to determine the interval of the retained measurement signals, the processed signals are input into an LSTM model to detect the thickness of the pipe, and a combination optimization based on random search and Bayesian optimization is used to set the hyperparameters of the LSTM model.

This invention belongs to the technical field of railway tracks, specifically disclosing a method for identifying rail corrugation in rail transit based on the composite characteristics of train vibration and noise. The method includes the following steps: acquisition and processing of train vibration, noise, and related signals; enhancement and fusion of vibration and noise signals; mapping of vibration and noise composite data to corrugation relationships and establishment of a sample set; design and training of a convolutional neural network structure for the vibration and noise composite sample set; and identification of rail corrugation status. This invention uses a one-dimensional convolutional neural network to adaptively extract features from the sample set of vibration fusion data based on vibration and noise fusion data. Simultaneously, a random search method is used to determine the optimal parameters, shortening the sample training and identification time, and meeting the requirements for accuracy and timeliness in rail corrugation detection and monitoring.

The invention provides a substation project multi-objective optimization method and device. The method comprises the following steps: acquiring a process set, a process preposition and postposition logic relation, selectable construction modes of each process and corresponding parameter information; introducing the logical relationship as a constraint, taking a construction mode selection result as a discrete decision variable, and constructing a multi-objective optimization model including a construction period, a cost present value, quality and a carbon emission objective; an improved swarm intelligence algorithm is adopted for iterative solution and layered sampling to generate an initial population and establish an external file, a target function is calculated in iteration and feasibility is judged, non-dominated sorting and congestion degree are executed after merging, the external file is updated, and a leader individual is selected; and candidate positions are generated based on nonlinear convergence control and hunting, random search and spiral bubble net updating, discretization and constraint restoration are performed after fusion, and an external file non-dominated solution set is output. According to the technical scheme, multi-target scheme optimization is achieved, and scheme generation efficiency and evaluation quality are improved.

This invention relates to the field of computing power scale prediction technology, specifically disclosing a method and system for predicting urban computing power scale based on a sled dog-optimized MLP. This invention constructs a parameter optimization architecture for an MLP driven by a sled dog optimization algorithm, encoding the number of hidden layer neurons, truncation quantiles, and year-specific switch variables as decision vectors. It designs a multi-objective fitness function that integrates training set error, validation set error, generalization gap penalty, model complexity penalty, and stability penalty. This simulates the dynamic selection, movement, obstacle avoidance, disorientation, training, and retirement behaviors of a sled dog population through iterative optimization. This solves the problems of traditional grid search and random search easily getting trapped in local optima, and the reliance on human experience for key MLP parameter configuration. It also overcomes the shortcomings of insufficient fitting of statistical models and overfitting of conventional neural networks in small sample scenarios, achieving improved accuracy in urban computing power scale prediction and enhanced model generalization performance.

The present application discloses an encoding method for low-density parity-check (LDPC) codes. The method comprises: acquiring an information bit sequence; and on the basis of a check matrix, performing LDPC encoding on the information bit sequence to obtain an encoded bit sequence, wherein the check matrix is determined by a base matrix, the base matrix comprises a fundamental sub-matrix, and the fundamental sub-matrix is formed by concatenating at least two basic unit blocks with cyclic or translational properties. Compared with existing 5G new radio (NR) matrices of which shifting values are all obtained by random search without guidance from a fixed algebraic structure, the base matrix having cyclic block structures or translational block structures designed in the present application can be guided by a fixed algebraic structure, thereby solving the technical problems of the lack of practical code design for LDPC and the difficulty in finding a cycle-free exponent matrix by means of random search.

A computing system generates directives of a program. Beam nodes are selected one level at a time from multiple nodes in the tree structure. Each node represents a subset of operations in the program. A first number of the beam nodes are selected at a given level of the tree structure. The selection of the first number of the beam nodes uses a cost model that is based on a neural network. A second number of the beam nodes are selected using a random search. The ratio of the first number to the second number is determined based on a search completion percentage at the given level. A path is identified that passes through respective beam nodes at multiple levels of the tree structure. The path represents a schedule for executing the program on a target machine. Then the directives corresponding to the schedule are generated.

The invention provides a large-scale farmland scale design method based on double-target constraint, and relates to the technical field of farmland standardized construction, and the method comprises the steps: carrying out the fitting of a farmland actual irrigation sample and an agricultural machinery operation actual sample, and obtaining an irrigation performance prediction model and an agricultural machinery operation efficiency prediction model; judging actual constraint conditions, and if the actual constraint conditions do not exist, performing simulation by utilizing the paddy field specification prediction model to obtain a paddy field specification prediction data sample so as to obtain a paddy field specification suitable range; and if the actual constraint condition exists, based on the irrigation performance prediction data sample and the agricultural machine operation efficiency prediction data sample, performing multi-objective optimization and global random search by using the paddy field specification optimization model to obtain a suitable paddy field specification, and analyzing a paddy field specification suitable range and the suitable paddy field specification to obtain a farmland scale design result. The problem that an existing farmland design technology lacks large scale, whole-process mechanization and high intensification is solved.

The present application relates to the technical field of pipeline structure reliability analysis, in particular to a pipeline structure reliability analysis method under interval uncertainty, comprising: collecting basic parameters affecting the vibration fatigue life of the pipeline structure; establishing a function function in reliability analysis; standardizing the input vector and the function function; establishing a reliability model of the pipeline structure under interval uncertainty with the standard function function as a constraint; introducing an intermediate variable for equivalent conversion; solving by using the dichotomy combined with the random search feasibility judgment method to obtain the reliability index of the pipeline structure. The present application solves the problems of difficulty in solving the black box function and difficulty in ensuring global convergence in the vibration fatigue life analysis of the engine pipeline structure by using the interval reliability analysis framework of the dichotomy combined with the random search feasibility judgment.

This invention relates to the fields of artificial intelligence and medical clinical decision support, specifically a method and system for predicting postoperative delirium risk in elderly patients based on machine learning. The method includes: acquiring perioperative data of the patient to be predicted, including clinical indicators from the preoperative, intraoperative, and postoperative stages; preprocessing and preliminary feature screening of the data to obtain a structured feature set; constructing and optimizing a machine learning-based postoperative delirium prediction model, forming a modeling pipeline by combining multiple feature selection methods with a classifier, determining the optimal hyperparameters using random search and k-fold hierarchical cross-validation, and selecting the best pipeline based on feature stability assessment and multiple evaluation indicators; training and evaluating the performance of the final model; outputting the postoperative delirium risk prediction results and providing model interpretation. This invention is applicable to scenarios such as perioperative risk assessment of elderly patients, early warning of high-risk patients with postoperative delirium, individualized intervention plan formulation, and clinical auxiliary decision systems, providing reliable technical support for reducing the incidence of postoperative delirium and optimizing the allocation of medical resources.

This invention proposes a method for predicting tree fall specifications based on historical meteorological data and UAV post-disaster inspection surveys. It integrates historical typhoon data, UAV inspection data, and geographic information data to form a comprehensive and multi-dimensional dataset. Through in-depth analysis of various influencing factors, key features are identified, and advanced machine learning methods such as regression and deep learning models are used to construct the model. Simulation optimization is employed to handle complex data relationships, improving prediction accuracy and reliability. Scientific evaluation indicators such as accuracy, recall, F1 score, and mean squared error are set, and k-fold cross-validation is used to comprehensively evaluate the model's generalization ability. Grid search and random search methods are used to fine-tune the model parameters. The constructed real-time monitoring and early warning system, along with post-disaster response and recovery strategies, effectively reduces tree fall damage and plays a positive role in protecting urban greening and the ecological environment.

The invention discloses a method for extracting parameters of a three-diode model of a photovoltaic cell, and the method comprises the steps: building a three-diode equivalent circuit model of the photovoltaic cell, and determining a plurality of to-be-recognized parameters of the photovoltaic cell according to the three-diode equivalent circuit model; output current-voltage I-V curve data of the three-diode equivalent circuit model are obtained, and the curve data at least comprise three feature point data, namely open circuit point voltage, short circuit point current and voltage and current of the maximum power point; and solving the three-diode equivalent circuit model based on the obtained I-V curve data, and obtaining an optimal solution of the to-be-identified parameters. The whole parameter extraction step has no matrix inversion operation and no random search process, the calculation complexity is low, continuous automatic model monitoring can be realized, and a parameter feasible region does not need to be manually set.

The invention relates to the technical field of motor optimization design, in particular to a motor optimization design method and system based on a multi-objective myxomycete algorithm, and the method comprises the steps: determining an optimization objective and a design variable; constructing a target function vector; iterative optimization is carried out based on an improved multi-target myxomycete algorithm, and the position of a myxomycete individual is updated based on a guiding individual and a self-adaptive exploration mode. According to the method, the elitist strategy is introduced and the elitist pool is updated on the basis of the traditional multi-target myxomycete algorithm, so that directional retention and utilization of the historical high-quality solution are realized, the loss of the high-quality solution in random search is avoided, a stable reference direction is provided for algorithm evolution, the optimization efficiency is improved, and the method is suitable for large-scale popularization and application. And the robustness of the optimization result and the engineering practicability are enhanced.

A system and method are disclosed for training and optimizing machine learning models for computer vision using mixed activation functions across model layers. A baseline model is received and a search space of candidate activation functions is defined. For each candidate substitution, a zero-cost accuracy score is computed without full training, and latency and memory costs are benchmarked across target hardware devices. Using this information, an optimization process such as random search, integer linear programming, or Local Zero Cost Maxima selects a layer-specific configuration of mixed activation functions that satisfies application constraints including accuracy, latency, and memory budgets. The selected model is then trained or fine-tuned to produce an optimized model. Experimental results on YOLO architectures demonstrate improved mean Average Precision, lower latency, and reduced memory usage relative to baseline models. This approach enables efficient deployment of computer vision models across CPUs, GPUs, and neural processing units.

The application provides an optimization design method of additive manufacturing process parameters based on machine learning, which can realize rapid and accurate prediction of product forming quality under any process parameters for a wide range of material systems by establishing an additive manufacturing process parameter-material performance gradient boosting regression tree (GBDT) model and double optimizing the model hyperparameters by using random search (RS) and K-fold cross validation (K-CV) algorithms, so as to quickly and accurately determine the best process parameters, and solve the problems of high calculation and test cost and long cycle in the optimization of the process parameter window of laser additive manufacturing.

The invention provides a manual assembly wiring harness modeling and crosstalk prediction method based on equal-outer-diameter wires, relates to the technical field of locomotive system research, development and assembly, and solves the problems of low calculation efficiency and limited use range in random wiring harness modeling in the existing method. The method comprises the following steps: segmenting a manual assembly wire harness based on a cascading method, abstractly constructing all numbered wires on the cross section of the wire harness into a topological graph form for each cascading segment, randomly generating the number of wires needing to be exchanged, randomly selecting a starting point, and using an optimization algorithm to optimize the number of wires needing to be exchanged. Randomly searching all lead numbered paths which meet the adjacent node exchange conditions and start from the starting point, and executing lead exchange operation; and after the wire exchange operation is repeated for preset times, taking the obtained wire harness cross section as a new cascade section, and solving crosstalk corresponding to a plurality of cascade sections by using a chain parameter method, thereby completing manual assembly wire harness modeling and crosstalk prediction. The method has the advantages and characteristics of high efficiency, simplicity, convenience and easiness in implementation.

The invention provides a multi-thread path planning method and device for a flexible cable type picking mechanical arm, and relates to the technical field of picking mechanical arms. The core step of the multi-thread path planning method is that at least three optimization threads with different optimization targets are triggered in parallel based on an initial geometric path generated by a fast random search tree; respectively generating candidate paths paying attention to dexterity, tension smoothness and obstacle avoidance robustness, and finally screening out an optimal to-be-executed path through multi-target evaluation in a sliding window. According to the parallel multi-thread optimization step, path optimization potentials of different performance dimensions can be synchronously explored and fused, and the comprehensive quality of the planned path is remarkably improved. Compared with an existing single-thread or serial optimization technology, the problem of local optimum or performance shortness possibly caused by a single optimization target is effectively avoided. The whole balance of the final execution path in the aspects of length, safety, stability and reliability is systematically promoted, and efficient and stable picking in the complex orchard environment is achieved.

This application discloses a pre-assessment method for earthquake-damaged buildings, relating to the fields of artificial intelligence and disaster assessment. By synergistically employing three strategies—historical and random transitions, multi-information social learning, and collision tracking—it can more effectively search for the globally optimal solution in a complex hyperparameter space. Compared to traditional grid search, random search, or single intelligent optimization algorithms, it can find a set of hyperparameters with superior performance. Applying this optimal set of hyperparameters to the pre-assessment model for earthquake-damaged buildings significantly improves the accuracy and generalization ability of identifying building damage levels. Deploying the trained pre-assessment model on a drone allows for rapid access to disaster areas that are difficult or dangerous for humans to reach, enabling image acquisition and identification, and outputting the pre-assessment results for earthquake-damaged buildings, thus improving assessment efficiency.

The invention relates to an intelligent auxiliary decision-making method and system for a ship in a complex operation scene, and the method comprises the steps: obtaining the navigation, ship body, cabin and energy efficiency data of the ship in historical operation, and forming a multi-source heterogeneous data set; state judgment is carried out based on a parameter preset range, and a structured data set with binary tags is generated; carrying out standardization processing on the structured data set, and carrying out category balance processing on the training set by adopting a mixed oversampling strategy; a CART decision-making tree is used as a base learning device, hyper-parameters are optimized in combination with a random search optimization and cross validation method, and a random forest aided decision-making model is trained and evaluated; and inputting data acquired in real time into the random forest auxiliary decision-making model after the performance reaches the standard, outputting auxiliary decision-making information including the category of the ship operation state, the safety risk level and the abnormal early warning information, and converting the auxiliary decision-making information into a visual graph or chart through a feature importance visualization technology. And real-time intelligent auxiliary decision making of the ship in a complex operation scene is realized.

The invention discloses an intelligent networked vehicle collaborative optimization control method and system under multi-vehicle communication, and the method comprises the steps: carrying out the vehicle distance adjustment through employing a queue control mode, and enabling a following vehicle to carry out the vehicle distance adjustment based on an error weighting mode considering the multi-source information delay; in other words, the expected acceleration of the vehicle is determined in the mode that the distance error, the speed error and the acceleration error between the vehicle and the front vehicle or between multiple vehicles in front are multiplied by corresponding control parameters, and then the weighted sum or the weighted cumulative sum is obtained; wherein each error takes into account an information tracing time and a state prediction time between vehicles to cope with perception delay, communication delay information and execution delay. According to the invention, an intelligent network connection vehicle queue control multi-objective optimization model is further constructed, and adjacent random search and a nonlinear decreasing weight strategy are introduced to improve a particle swarm optimization algorithm to solve the multi-objective optimization model so as to select values of control parameters. The method is simple and easy to implement, and the control performance and stability of the intelligent network connection vehicle queue can be improved.

The invention relates to a cattle water drinking state intelligent identification method based on rumen temperature characteristics and XGBoost, and belongs to the technical field of animal behavior identification. According to the method, a rumen capsule type sensor is implanted to collect temperature data, characteristic variables including temperature first-order difference, three-point rolling average temperature and the like are constructed, and accurate recognition of the drinking water state is achieved in combination with an XGBoost algorithm and random search hyper-parameter optimization. The method overcomes the limitations of strong scene dependence, complex system, low identification precision and the like in the prior art, and has the advantages of stable data acquisition, flexible deployment, high calculation efficiency, strong environment robustness and the like; the method can effectively adapt to various scenes such as captivity and grazing, and provides an efficient technical path for behavior monitoring and health early warning of animals in a smart pasture.

The application provides a kind of flexible cable type picking mechanical arm multi-thread path planning method and device, it is related to picking mechanical arm technical field, the core step of the present application is based on the initial geometric path generated by fast random search tree, at least three optimization threads with different optimization objectives are triggered in parallel, respectively generate candidate paths focusing on dexterity, tension smoothness and obstacle avoidance robustness, and finally the optimal path to be executed is screened out through multi-objective evaluation in sliding window. The parallel multi-thread optimization step can simultaneously explore and integrate the path optimization potential of different performance dimensions, significantly improving the overall quality of the planned path. Compared with the existing single thread or serial optimization technology, the local optimal or performance short board problem caused by a single optimization objective is effectively avoided. The present application systematically promotes the overall balance of the final execution path in length, safety, stability and reliability, and realizes efficient and stable picking in complex orchard environment.

The invention relates to the technical field of robot motion planning and autonomous navigation, and discloses a bidirectional RRT * path planning method based on target offset guidance, which comprises the following steps: initializing a path planning environment, and creating two random search trees of a starting point tree and an end point tree; a target bias strategy is adopted, and linear expansion from a current node to a target point is tried; on the basis of the distance field gradient calculation normal and tangential direction, expanding new nodes in a mixed manner along the equidistant line slippage direction; constructing an artificial potential field resultant force direction, and guiding the tree to extend in combination with target gravitation and obstacle repulsive force; executing RRT * optimization operation, and reselecting a father node and rewiring in a new node neighborhood; detecting node pairs meeting a distance threshold and a security condition in the two trees, and generating a complete path through bidirectional backtracking after connection; and optimizing the complete path to generate a continuous curve. Compared with the prior art, an efficient and reliable solution is provided for autonomous navigation of the intelligent mobile platform through a multi-strategy collaborative intelligent expansion mechanism and a three-level optimization process.

The invention specifically relates to a fixed-wing unmanned aerial vehicle cluster coverage reconnaissance path sensor joint optimization method, belongs to the technical field of unmanned aerial vehicle cluster collaborative area coverage task planning, and is suitable for emergency search and rescue, resource inspection and other scenes. The method aims at solving the problems that in the prior art, track planning and sensor parameter configuration are decoupled, and dynamic response is insufficient. The method comprises the steps of task initialization and environment modeling, determining an optimal coverage direction of a region, and establishing a sensor and unmanned aerial vehicle kinematics model; constructing a regional decomposition and joint optimization model, dividing operation strips, and establishing three types of track models and a mixed integer programming model; solving based on a random search heuristic algorithm, and obtaining an optimal track and sensor parameters through random construction and reconstruction inversion; and dynamic event response and re-planning are carried out to quickly cope with emergencies such as unmanned aerial vehicle exit and new areas. According to the invention, joint optimization of the track and sensor parameters is realized, redundant voyage is reduced, coverage efficiency is improved, and high robustness and real-time performance are realized.

The present application relates to the technical field of fault early warning, in particular to a wind turbine generator fault early warning method based on adaptive double control strategy, which comprises the following steps: S1, obtaining historical operation data of the wind turbine by using a data acquisition and monitoring control system, and preprocessing the historical operation data; S2, importing the data set into a training model, setting the input and output of the training model, training the model, and saving the trained model as a category gradient boosting benchmark model; S3, finding the optimal hyperparameter variable of the category gradient boosting benchmark model by using random search; S4, calculating the residual error between the predicted value and the actual value, and constructing an adaptive double control strategy to dynamically determine the change trend of the residual error; S5, determining the fault point by using the adaptive double control strategy. The present application can be used for real-time early warning and alarm of offshore wind turbine groups and land wind turbine groups, and has the advantages of robustness, universality, accuracy and high efficiency in fault early warning.

The invention provides a program simplification method and system based on multi-granularity random optimization, and the method comprises the steps: obtaining a user portrait input set and a target function input subset of a to-be-simplified program, respectively inputting the user portrait input set and the target function input subset into the to-be-simplified program, obtaining a corresponding candidate initial simplification program based on code coverage information, selecting a candidate initial simplified program with a higher objective function value as an initial sample; the objective function is a linear weighting result of the code deletion amount and the program universality; performing variation and search operation on the initial sample based on a Markov chain Monte Carlo random search method to obtain a simplified program; wherein the variation and search operation comprises three stages, namely a code coverage granularity variation stage, a coverage section granularity variation stage and a statement granularity variation stage.

The invention relates to a communication method, communication equipment, a communication device and a computer readable storage medium. The method comprises the following steps: determining a lifting factor from a lifting factor set based on the length of an information bit sequence and a basic matrix; based on the lifting factor, the prime number corresponding to the lifting factor and a first coefficient, a translation value set is determined, translation values in the translation value set correspond to non-zero positions in the basic matrix, and the first coefficient is a non-negative integer; and determining a check matrix for encoding or decoding the information bit sequence based on the translation value set and the basis matrix. Therefore, a translation value set independent of random search can be realized, and good ring properties under different coding lengths and coding rates are ensured. In addition, a wider lifting factor set can be supported, and more stable coding and decoding performance can be realized.

The invention discloses a big language model inference service performance optimization method and system based on Bayesian optimization. The method comprises the following steps: constructing a parameter space covering inference service starting parameters and client request parameters; designing a weighted objective function fusing throughput and a delay index; executing sequential search based on a Bayesian optimization framework, and gradually approaching an optimal parameter combination through initial sampling, probability model construction, acquisition function selection and iterative updating; and performing stability, service adaptability and comparison verification on the obtained optimal parameter combination to ensure the reliability of parameter configuration. According to the method, traditional grid search or random search is replaced by an intelligent search strategy, the optimal parameter combination can be efficiently positioned within the limited number of experiments, the computing power cost is remarkably saved, the performance and the resource utilization rate of big language model reasoning service are improved, and the method is suitable for parameter optimization in the big language model deployment and debugging stage.