42 results about "Adaptive randomization" patented technology

An adaptive random access system and method with learned query optimization for compacted data files that enhances random access performance through machine learning and pattern recognition. The system incorporates a query pattern learning module that analyzes historical access patterns and user behavior to build statistical models of data usage. An adaptive estimator module improves location estimation accuracy by incorporating learned patterns rather than relying solely on mathematical calculations. A predictive boundary detector uses learned codeword patterns to more accurately identify boundaries in compacted data, reducing misalignment errors. An intelligent search engine coordinates optimization strategies including context-aware search string parsing and encoding strategy selection based on learned performance data. A dynamic codebook optimizer reorganizes sourceblock layout based on access frequencies and co-occurrence patterns to improve retrieval speed. An enhanced search cache implements predictive caching algorithms that anticipate user queries and proactively load relevant data.

The invention discloses a mobile payment encryption system based on a data random algorithm, and the system comprises a collection module which is used for collecting multi-modal data, and constructing a cross-modal entropy source pool; the random seed generation module is used for extracting feature vectors from the cross-modal entropy source pool and generating initial random seeds; the time sequence self-adaptive random chain module is used for generating a random factor sequence based on the initial random seed recursion and forming a random chain structure; the encryption module is used for jointly encoding the random factor sequence and transaction data to be encrypted, generating a dynamic encryption parameter set and encrypting the transaction data; and the chain type updating module is used for collecting network environment parameters, updating the time sequence self-adaptive random chain and generating a new dynamic encryption parameter set so as to form a chain type evolved encrypted data sequence. According to the method, the cross-modal entropy source and the time sequence random chain are utilized to realize dynamic encryption of the mobile payment data, and the method has the advantages of high entropy randomness, strong attack resistance and high transmission security.

The invention discloses a historic building cylindrical structure inclination measurement method considering irregular surfaces and measurement errors, and the method comprises the steps: measuring the three-dimensional coordinates of measurement points at different section heights of a historic building wood structure cylinder with a cutting or irregular surface; determining an optimal solution after a certain error is removed from the coordinates of the circle center of the section by adopting a self-adaptive random sampling consistency algorithm, then obtaining the central axis of the cylinder according to principal component analysis, and calculating the included angle of the earth normal vector of the central axis of the cylinder, namely the inclination angle of the historic building wood structure cylinder. By using the algorithm, the influence of cutting or irregular surfaces and measurement errors on historic building wood structure cylinder inclination measurement can be eliminated, and the reliability and the accuracy of a measurement result are effectively improved.

The invention discloses a power system random sampling control method based on attack compensation. The method comprises the following steps: constructing an adaptive random switching model, establishing a lightweight DoS attack detection and event trigger compensation mechanism, and optimizing a PD controller fusing compensation data, the method combines the random switching model, a network security defense mechanism and an optimization control strategy, and aims at a dynamic multi-mode power system containing an electric vehicle EV. The technical bottleneck of traditional load frequency control in the problems of random disturbance, network attack and mode switching dependency is solved, the method is suitable for improving the frequency stability and the anti-attack ability of the power system, and the anti-interference ability and the operation efficiency of the power system can be improved.

The invention discloses a satellite hyperspectral remote sensing image-based adaptive random forest land utilization classification and identification method, which comprises the following steps of: data preprocessing: loading a vector boundary of any predetermined administrative region, the method comprises the following steps: acquiring surface reflectance data of a certain satellite hyperspectral image in any preset time period from a natural resource satellite remote sensing cloud service platform, and preprocessing; making a sample set; parameter definition and remote sensing index selection: calculating a group of representative remote sensing indexes by using a specific wave band combination of the hyperspectral image; a remote sensing index and an original wave band are used as input feature vectors together to form a complete feature space, and in the training process of the random forest, part of features are randomly extracted from the feature space during construction of each decision tree for node division; performing cross validation and parameter tuning; the visualization of the optimal parameter result is selected; according to the invention, adaptive parameter tuning based on random forest algorithm classification is realized.

The invention provides a high-precision precise point positioning method based on BDS-3, and relates to the technical field of wireless communication. The method comprises the following steps: firstly, introducing a traditional ionosphere-free combination model to eliminate the influence of ionosphere delay on positioning precision, and eliminating or weakening a satellite orbit error and a clock error error by using satellite orbit and clock error data provided by an IGS; secondly, constructing a troposphere correction model of adaptive weighted wet delay for eliminating troposphere delay errors; thirdly, establishing a self-adaptive random model, and reasonably weighting pseudo-range and carrier phase observed quantity; and finally, estimating the parameters to be estimated through extended Kalman filtering so as to obtain a positioning calculation result. The method has obvious advantages in the aspect of improving the positioning precision, can meet the requirement of high-precision positioning, can realize centimeter-level positioning precision no matter in a dynamic mode or a static mode, and can even reach millimeter-level positioning precision in the static mode in some observation stations and directions.

The invention discloses an unmanned aerial vehicle path planning method based on an improved snake optimization algorithm, and the method comprises the steps: introducing a self-adaptive random disturbance factor based on a sine function to improve an exploration mechanism of an exploration stage and a development mechanism of a development stage in a first half iteration stage, dynamically enhancing the randomness of position updating in a search process, and positioning a global optimal solution; in the latter half iteration period, the development mechanism of the development stage is improved by introducing a scaling factor and leader-based adaptive Levy flight strategy, the male snake leader is endowed with flight ability and the global exploration ability is enhanced, and the development mechanism of the development stage is improved by introducing an adaptive position updating strategy combining elite leader and Brownian motion; the female snake leader is endowed with the Brownian motion ability, the convergence speed is effectively accelerated while the precision is ensured, and the local search ability is enhanced. According to the method, the path which is low in cost, high in safety and short in arrival time can be quickly found, and the task execution efficiency of the three-dimensional unmanned aerial vehicle is improved.

The invention discloses a method, a device and equipment for acquiring a global initial pose of a loading machine and a medium. The method comprises the following steps: acquiring original point cloud data of an operation scene in real time by using a vehicle-mounted laser radar to generate a corresponding frame point cloud; the method comprises the following steps: constructing a real-time feature vector set based on a fast point feature histogram feature of a frame point cloud, and performing nearest neighbor search on the real-time feature vector set and a k-d tree index of a pre-established global point cloud map to form an initial matching point pair set; in each round of iteration, screening a high-precision inner point set meeting geometric constraints by adopting a self-adaptive random sampling strategy, and calculating a candidate pose transformation matrix and a corresponding inner point proportion score according to the high-precision inner point set; and a new iteration round is repeatedly executed to execute the operation until the total number of iterations reaches a target value, and the global initial pose of the target loader is screened out from the candidate matrix according to multiple rounds of scores. According to the embodiment of the invention, high-precision global initial pose acquisition of the unmanned loader is realized in an environment lacking satellite coverage, and the positioning precision and reliability are improved.

The invention discloses a heavy truck electric drive bridge fault diagnosis method based on a generative adversarial network, and the method achieves the precise analysis and diagnosis of the operation state of a heavy truck electric drive bridge through data collection and marking, data expansion, feature extraction model training, classifier model training and fault diagnosis. In the data expansion stage, a generative adversarial network based on adaptive stochastic disturbance is adopted, an entropy weight dynamic back propagation mechanism is introduced, the authenticity of generated data is improved by dynamically adjusting a learning rate, and distribution of the generated data is optimized by using a Gaussian mixture model and a feature dependency matrix; the feature extraction model is based on a full-connection neural network optimized by a bionic algorithm, a training rule is adjusted by adopting a self-correction mechanism, and the training efficiency and the model generalization ability are improved; the classifier adopts a high-order neural network of a reverse backtracking sparse strategy, the network structure and the feature weight are dynamically adjusted, and the sensitivity to abnormal data and the accuracy of fault classification are effectively enhanced. The method is high in fault diagnosis efficiency, high in precision and high in stability.

The invention particularly relates to a parameter adaptive hybrid disturbance serial cancellation (SC) decoding method for polarization codes. Aiming at the problems of disturbance parameter fixation and rigidity, lack of channel adaptability and dependence on off-line parameter search in the existing hybrid disturbance serial cancellation (HPSC) decoding method, the method comprises the following steps: firstly, establishing a linear mapping relation between disturbance noise power and channel noise power to realize automatic adaptation of a channel environment; then, the check characteristic of a cyclic redundancy check code is used as a detection indicator of a local optimal trap, intelligent switching is carried out between adaptive random disturbance and adaptive biased disturbance, and a bias adjustment strategy dynamically changing along with the number of iterations and the reliability of a received symbol is introduced in the biased disturbance stage; therefore, parameter-adjustment-free high-performance decoding is realized on the premise of keeping low complexity of SC. Simulation results show that compared with a traditional HPSC decoding method with fixed parameters, the method provided by the invention has better block error rate performance, and has lower average decoding iterations in a high signal-to-noise ratio region.

The application provides an underwater vehicle optimization cooperative detection array position method, a target optimization function of an AUV cooperative detection array position is constructed by coupling AUV detection error and AUV position error, and the optimization object is minimization of global average detection error, so that passive detection AUVs construct an array position according to maximum cooperative detection accuracy; an improved HS-DPSO fusion algorithm is adopted to iteratively optimize the target optimization function; in the local optimization process, a dynamic inertia weight adjustment strategy combining normal distribution randomness and an adaptive expectation mechanism is adopted, so that the inertia weight regularly and dynamically fluctuates in each iteration, and the ability of the algorithm to jump out of a local optimum is enhanced; and a random number is used to dynamically adjust the adaptive random weight expectation value, so that the limitation of a manually preset fixed random weight expectation value is avoided, and the adaptability of the algorithm to different detection scenes is improved. The method makes each passive detection AUV quickly obtain an optimal cooperative detection position, and improves the cooperative detection efficiency and detection accuracy of the AUV formation.

This invention provides a method for optimizing cooperative detection array positions for underwater vehicles (AUVs). It constructs a target optimization function for AUV cooperative detection array positions by coupling AUV detection errors with their own positional errors, aiming to minimize the global average detection error. This ensures that passively detected AUVs construct array positions with maximum cooperative detection accuracy. An improved HS-DPSO fusion algorithm iteratively optimizes the target optimization function. During local optimization, a dynamic inertial weight adjustment strategy combining normal distribution randomness and an adaptive expectation mechanism is employed. This allows the inertial weights to fluctuate dynamically and regularly in each iteration, enhancing the algorithm's ability to escape local optima. Furthermore, the adaptive random weight expectation value is dynamically adjusted using random numbers, avoiding the limitations of manually pre-setting fixed random weight expectation values ​​and improving the algorithm's adaptability to different detection scenarios. This method enables each passively detected AUV to quickly obtain its optimal cooperative detection position, improving the efficiency and accuracy of AUV formation cooperative detection.

The invention relates to the technical field of intelligent medical monitoring, and discloses a sleep disorder intelligent monitoring system and method based on a multi-modal physiological signal, and the method comprises the steps: collecting electrocardio, respiration and body movement signals, calculating a finite time Lyapunov index by using a phase-space reconstruction technology, taking the finite time Lyapunov index as a feedback control variable, and calculating the sleep disorder intelligent monitoring system based on the multi-modal physiological signal; noise intensity in a Langevin equation is dynamically adjusted, adaptive stochastic resonance enhancement is performed on the preprocessed signal, weak characteristics are improved by using noise energy, and causal coupling intensity between cross-modal symbol transfer entropy and net information flow index quantification systems is further calculated through phase-space symbolization mapping. And in combination with Lempel-Ziv complexity, a multi-dimensional judgment logic is constructed, and sleep staging and obstacle event results are output. According to the method, the problems that nonlinear physiological signal features are difficult to extract and a system coupling mechanism is unknown are solved, and the robustness and accuracy of sleep monitoring are improved.

The present application discloses a processing method, device and medium for question prompts of a large language model, and relates to the technical field of natural language processing. The method includes: obtaining the original text, extracting the entities of the original text, and extracting logical relationships based on the entities, the logical relationships including causal relationships and employment relationships; reconstructing the question according to the entities and logical relationships, and generating restated question prompts; based on the question prompts, using an adaptive random replacement strategy, calculating the semantic similarity and task type between the question prompts and the preset vocabulary, dynamically selecting a replacement mechanism, and generating multiple candidate prompts; calculating the global entropy value of the candidate prompts, and performing multi-source knowledge base verification on the candidate prompt with the smallest global entropy value, and outputting the smallest candidate prompt if the verification passes. The present application optimizes the generation process of a large language model when processing wrong premises or fuzzy problems through the above method, thereby effectively reducing the occurrence of hallucination output.

The invention provides a recurrent event clinical test sample size estimation method and device, equipment and a medium, and relates to the technical field of clinical test design and biostatistics. According to the method, test basic information is acquired, a statistical model integrating a subject falling mechanism is established based on the test basic information, then key bridge parameters and average exposure time are calculated, and a double-path closed-loop decision including a design stage and an execution stage is executed; calculating a target distribution probability based on the key bridge parameters and the average exposure time so as to calculate a total sample size; and dynamically updating parameters based on the current real-time data in the execution stage, and carrying out self-adaptive random allocation on new subjects. According to the method, the influence of falling of the subject on the exposure time can be explicitly modeled, parameter unified driving sample size estimation and dynamic randomization distribution are introduced, the benefit of the subject is optimized while the inspection effect is ensured, and integrated decision support from design to execution is provided.

The invention relates to the technical field of protection engineering, in particular to a facility system first-aid repair resource allocation method for a complex uncertain attack scene, which comprises the following steps of: S1, acquiring a facility system network structure and a facility number, and evaluating a facility initial function; s2, determining the level of an adverse event; s3, determining facility system first-aid repair resource preset planning constraint conditions, and constructing an improved facility system first-aid repair resource preset planning model; and S4, solving the first-aid repair resource preset planning model by adopting an adaptive random search algorithm (ARS-SVM) fused with a support vector machine to obtain an optimal solution of the first-aid repair resource configuration of the facility system. According to the method, the resource demand satisfaction degree expectation which maximally considers the conditional value-at-risk is taken as a target, the demand is regarded as a random variable, the probability distribution is obtained through historical data fitting, and the auxiliary variable is introduced to quantify the demand fluctuation, so that the urgent repair recovery accuracy can be improved, the cost effectiveness can be balanced, and the system adaptability can be enhanced.

The present invention discloses a digital-physical fusion test method for tool wear in the machining of integral bladed disks, belonging to the fields of mechanical engineering, computer science, and data analysis. The method includes: through physical monitoring, synchronously collecting vibration signal data during the machining process of a five-axis machining center and performing preprocessing to obtain effective vibration signal data; constructing a digital-physical fusion prediction model based on the adaptive random forest algorithm to predict the effective vibration signal data, the prediction results including the predicted value of the effective value of acceleration, and the adaptive random forest algorithm includes using Bayesian optimization to optimize the hyperparameters of the random forest algorithm; setting up a digital and physical operation integration interface to implement a dynamic threshold monitoring and real-time alarm mechanism for the predicted effective vibration signal data. The present invention provides a more accurate and efficient tool wear test method by combining digital technology and actual production data, improving production efficiency.

The present invention relates to an enterprise park meteorological fire risk assessment method based on an adaptive random forest model, which includes establishing a risk assessment model and a visualization warning platform. The risk assessment model collects alarm data, meteorological data, enterprise data, and park division data to form a risk assessment data set, and then integrates the internal characteristics and external data of the park into the data set. Relying on real-time data stream technology, new data is input into the adaptive random forest model to instantly update and evaluate the fire risk index value of the park, and optimize the model decision tree to automatically adjust the risk assessment model to adapt to changes in the environment and operations, provide accurate warning information to the visualization warning platform, display the park risk level in real time, and issue risk alarms in a timely manner.

The application discloses a kind of adaptive random finite set tracking methods based on low-cost spaceborne distributed radar, mainly solve the problem of high cost, poor adaptability of prior art.Its implementation scheme is: through auxiliary star forwarding mode to build equivalent one-to-many receiving model, realize low-cost recording echo data;According to echo data, form two equivalent long synthetic aperture;Through synthetic aperture imaging, clutter suppression and detection to obtain measurement point;Adaptively generate new target intensity using multi-step measurement points;Adaptively estimate the clutter rate at each step using multi-step measurement points;Adaptive random finite set target tracking is carried out using the intensity and clutter rate generated.

The present invention discloses a clinical trial allocation method, system, device and storage medium based on adaptive randomization. The method includes: randomly allocating subjects to each treatment group according to a preset ratio, and obtaining the current efficacy data of each treatment group; determining the posterior probability of each treatment group based on the current efficacy data of each treatment group, and adjusting the random allocation probability according to the posterior probability; determining the conditional test power of each treatment group, and performing a preset resource optimization strategy on each treatment group according to the conditional test power to obtain the test result; determining the current posterior probability of each treatment group according to the test result, and determining the optimal treatment group based on the current posterior probability. Since the present invention dynamically adjusts the random allocation probability of each treatment group during the trial process, performs a preset resource optimization strategy on each treatment group according to the conditional test power, and determines the target treatment group based on the test result, it optimizes the allocation of subjects in the trial and improves the reliability and accuracy of the test result.

The invention relates to the technical field of laser coherent synthesis, discloses a self-adaptive stochastic parallel gradient descent parameter optimization algorithm based on reinforcement learning, and solves the problems that a traditional parallel gradient descent algorithm (SPGD) is slow in convergence and the effect of an improved parallel gradient descent algorithm depends on manual appointing of hyper-parameters. The method comprises the following steps: inputting performance indexes measured after coherent combination of multiple beams of laser into a reinforcement learning parameter optimization algorithm to obtain SPGD parameters, then executing the SPGD algorithm according to the updated parameters to obtain the performance indexes of the current parameters, and repeating the process for multiple times until the reinforcement learning algorithm converges to obtain the performance indexes of the current parameters. The optimal SPGD parameters under different conditions can be stably output, so that the SPGD algorithm can quickly correct errors, and high-coherence synthesis output is realized. The adaptive stochastic parallel gradient descent parameter optimization algorithm is simple in structure, easy to implement and efficient, and has high robustness, real-time performance and capability of adapting to complex environment change.

The invention discloses an RGBT tracking method based on diversified semantic aggregation. According to the method, on the basis of an RGBT tracking framework of a single-branch Transform, local semantics and global semantics of 1-12 layers of features are aggregated through a designed diversified semantic aggregation method to be used for target positioning so as to improve tracking accuracy. Specifically, firstly, a shallow random feature is obtained from a plurality of shallow features by adopting an adaptive random selection method, and then the shallow random feature containing local semantics and a plurality of deep features containing global semantics are subjected to semantic fusion in a cascade mode. Wherein semantic fusion takes a semantic convergence template as a medium to learn semantic information from significance information of a previous layer semantic fusion feature, and the semantic convergence template and a current layer feature are fused based on a double-branch cross attention enhancement method to obtain a previous layer semantic fusion feature and a current layer semantic fusion feature. Diversified semantic aggregation results not only provide rich semantics for target positioning, but also can promote tracking accuracy.

The invention discloses a metamorphic test group generation method based on an isolated forest and an adaptive random test, which comprises the following steps of: firstly, generating a group of initial test cases, and judging whether defects of a to-be-tested program can be detected or not; if no program defect is detected, putting the program defect into a selected test case set; constructing an isolated forest model by using the selected test case set; thirdly, constructing a candidate test case set, judging whether a program defect is detected by the candidate original test case, and if the program defect is not detected, generating a candidate subsequent test case according to the metamorphic relationship; and calculating abnormal scores of the candidate test cases by using an isolated forest model, putting the test cases of which the abnormal scores are higher than an abnormal threshold value into a selected test case group set, and finally judging whether program defects are detected or not. Through the method provided by the invention, an effective method is provided for generating the test case of the high-dimensional input scientific calculation program in the metamorphic test, and the generation efficiency is improved while the boundary effect problem in the generation of the test case is relieved.

The invention discloses an ancient fresco element detection algorithm based on context semantics and adaptive augmentation, and belongs to the cross technical field of computer vision and cultural heritage protection. Comprising the following steps: constructing a Chinese ancient mural element target detection data set; designing an adaptive random erasure image augmentation model, and augmenting the constructed data set; constructing a lightweight multi-scale feature extraction backbone network, and performing feature extraction on the augmented image; constructing a semantic feature enhancement model, and performing semantic enhancement on the extracted features; the obtained enhanced semantic feature map is input into the full convolutional network, a detection result is output, the detection result comprises element categories, bounding box coordinates and confidence coefficients, experiments show that the detection precision of the method on a self-built data set reaches 87.5%, the single-map detection time is only 0.022 seconds, the mural elements can be efficiently and accurately recognized, and the method is suitable for popularization and application. And technical support is provided for ancient mural protection, historical research and digital inheritance.

The invention provides a federated learning communication method based on adaptive-random client selection and dynamic regularization. The method comprises the following implementation steps: a central server initializes federated learning communication parameters and sends the parameters; the central server adaptively and randomly selects an active client; the active client performs iterative training on the local model; and the central server obtains a federal learning result. According to the method, the active clients are selected by adopting a self-adaptive and random combined method, so that communication resources are prevented from being frequently consumed on the clients with relatively small data contribution at the initial stage of communication; after training enters a stable stage, active clients are randomly selected based on a fixed activation probability, so that overfitting is prevented, convergence speed and generalization ability are balanced more efficiently, coefficients of dynamic regularization items in loss values of a global model are calculated through weights of the active clients, and the robustness of the overall model is improved. Different clients are subjected to dynamic constraints of different degrees when executing local optimization, so that the convergence precision is improved while the training stability and robustness are ensured.

The application provides an arc fault detection method based on a completely self-adaptive random configuration network, which builds a network structure by self-adaptive learning of arc fault current characteristics, judges the redundancy of a newly added node to realize adjustment of the network width, judges the error level of the newly added node to realize adjustment of the network depth, and associates the network structure with the current signal, so that data driving is realized, the calculation complexity of the network is reduced, and accurate and rapid detection of the arc fault is realized.

The invention discloses a tool wear number-real fusion test method for blisk machining, and belongs to the field of mechanical engineering, computer science and data analysis. The method comprises the following steps: monitoring an entity, synchronously acquiring vibration signal data in the machining process of the five-axis machining center, and preprocessing the vibration signal data to obtain effective vibration signal data; a number-real fusion prediction model is constructed based on an adaptive random forest algorithm so as to predict effective vibration signal data, a prediction result comprises an acceleration effective value prediction value, and the adaptive random forest algorithm comprises the step of performing hyper-parameter tuning on the random forest algorithm by using Bayesian optimization; a digital and entity operation integrated interface is arranged, and a dynamic threshold monitoring and real-time alarm mechanism is carried out on predicted effective vibration signal data. According to the invention, a more accurate and efficient tool wear test means is provided by combining a digital technology and actual production data, and the production efficiency is improved.

A method for Bluetooth Low Energy (BLE) advertising includes determining, by a controller of a BLE device, a next transmission time for advertising content using a bounded mean-reverting random-walk process. The determination is based on a next drift variable (Δtn+1), a base interval value (T0), and a random jitter component (J). The next drift variable is determined based on a current drift variable (Δtn), a first parameter (λ), a random variable (ξn) and a scaling factor (σ) set to determine a standard deviation. The current drift variable and the next drift variable are constrained within boundaries (-Δmax, Δmax). The method further includes transmitting the advertising content, by a transmitter of the BLE device, autonomously at the next transmission time without external coordination. The bounded mean-reverting random-walk process may provide collision avoidance in dense beacon environments by enabling autonomous desynchronization between multiple BLE devices operating in proximity.

The invention discloses a dual-mode adaptive random model adjustment method, system and device and a medium, and the method comprises the steps: obtaining real-time state sensing data at each epoch of a Kalman filtering calculation process through monitoring a prediction error sequence of Kalman filtering and calculating a normalized prediction error square statistic; judging whether the monitoring point enters a high-dynamic mode or not; in response to the judgment of entering the high-dynamic mode, executing a model adjustment operation through a continuous adaptive control algorithm to obtain an adaptive filtering model for the high-dynamic mode; when the adaptive filtering model is adopted, judging whether a dynamic event is ended or not; in response to judging that the dynamic event is ended, recovering the filtering model to a state before adjustment; a filtering model in a current state is applied to original observation data for positioning calculation, a new prediction error sequence generated by calculation is fed back, and a continuous high-precision positioning result is obtained. According to the invention, timeliness and reliability of geological disaster monitoring and early warning are substantially improved.