16 results about "Exponential growth" patented technology

The invention belongs to the technical field of nuclear reactor physical numerical calculation, and particularly relates to a method for improving precision and convergence of solving a diffusion equation based on a PINN, and the method comprises the following steps: S1, building a neural network algorithm model combining a physical information neural network PINN and source iteration; s2, aiming at a multi-region smooth transition neutron diffusion problem with small cross section difference between materials, optimizing the neural network algorithm model by adopting an optimization strategy of a training level; and S3, aiming at the multi-region neutron diffusion problem of large neutron flux gradient at the boundary due to large cross section difference between materials, improving the neural network algorithm model by adopting a domain-divided neural network architecture. According to the method, the problem that the calculation cost is exponentially increased in a high-dimensional or complex geometric scene due to the fact that a traditional finite difference method and a finite element method depend on grid division is avoided.

The exponential growth in deep learning models is challenging existing computing hardware. Optical neural networks (ONNs) accelerate machine learning tasks with potentially ultrahigh bandwidth and nearly no loss in data movement. Scaling up ONNs involves improving scalability, energy efficiency, compute density, and inline nonlinearity. However, realizing all these criteria remains an unsolved challenge. Here, we demonstrate a three-dimensional spatial time-multiplexed ONN architecture based on dense arrays of microscale vertical cavity surface emitting lasers (VCSELs). The VCSELs, coherently injection-locked to a leader laser, operate at gigahertz data rates with a 7T-phase-shift voltage on the 10-millivolt level. Optical nonlinearity is incorporated into the ONN with no added energy cost using coherent detection of optical interference between VCSELs.

The invention provides a high-throughput random number generation method and generator based on a quantum entanglement source, and relates to the field of quantum computation.The high-throughput random number generation method comprises the following core steps that firstly, in the entanglement debiasing process, a quantum sequence entangled with a random seed is constructed through inference, complete extension mapping of an anti-correlation entanglement quantum sequence is achieved, and the quantum sequence is generated; and generating a random bit sequence with doubled length and unbiased statistics. Secondly, in the full-permutation expansion process, the inter-domain correlation of the full-permutation sequence is eliminated through a shuffling method while sequence length exponential growth is achieved; and thirdly, in the layered asynchronous arrangement process, the intra-domain correlation is further eliminated through two-layer segmentation, asynchronous arrangement and shuffling processing, and finally a random bit sequence which meets the statistical quality requirement and has the length. According to the method, through removal of intra-domain and inter-domain correlation, the unpredictability of output random numbers is remarkably enhanced; and exponential-level high-throughput output is realized while the output random number is ensured to pass through all test items of the NIST SP 800-22.

The invention discloses a parallel cut set simplification method based on a fault tree logic embedded neural network, and relates to the field of fault tree analysis. In order to solve the problems that in the prior art, the cut set simplification calculation amount is exponentially increased, the parallelization degree is insufficient, the efficiency is low due to dependence on explicit comparison, and efficient implementation on a GPU architecture is difficult, the invention provides an efficient cut set simplification scheme utilizing fault tree logic and neural network structure fusion. The method comprises the following steps of: vectorizing an initial cut set, grouping according to orders, generating a corresponding subset basic matrix, and generating'order minus one 'subset tensors in batches through an index mask; inputting the subset tensor into a fault tree logic equivalent neural network to execute parallel Boolean judgment, and outputting a result of whether the subset triggers a top event or not; according to judgment result slice statistics, non-minimum cut sets are quickly removed, and a minimum cut set is obtained. The method is suitable for reliability analysis and safety evaluation of large complex systems such as aerospace, nuclear power, rail transit and petrochemical devices.

The invention discloses a multi-agent collaborative test signal fitting analysis system, and relates to the field of ship data fitting analysis, and the test signal fitting analysis system is characterized in that a plurality of agents collaboratively construct a workflow, and full-process automation of test signal fitting analysis is realized by using the agents. Each agent realizes adaptation and optimization of a fitting analysis workflow complex environment through a cycle process of perception, decision making, action, learning and reflection, the breadth and depth of automatic fitting analysis are expanded, manual participation can be reduced when fitting failure occurs, and particularly, when uncertain signal features are superposed, the accuracy of fitting analysis is improved. According to the method, signal fitting analysis can be achieved without manual participation, end-to-end processing from the test signal data file to the optimal fitting model is completed at a time, the fitting analysis efficiency of complex test signals is remarkably improved, and the requirement that the data size of the test signals is increased in an exponential level can be effectively met.

The application discloses an inner-cross type tire crane instruction scheduling method based on multi-dimensional dynamic priority and predictive scheduling, which comprises the following steps: step S1, candidate instruction set generation and hard constraint filtering; step S2, opportunity overtaking window evaluation; step S3, nonlinear dynamic sorting of regular instructions; and step S4, instruction issuing and rescheduling. The application quantifies and decides whether to execute the overtaking instruction which can bring higher net benefit through the opportunity overtaking window evaluation, so as to improve the work efficiency. The application introduces the instruction waiting time factor and the maximum tolerance threshold with exponential growth characteristics, so as to ensure that the priority of the instruction waiting for a long time can be systematically improved, and the scheduling fairness can be ensured. In addition, the local congestion factor is used to reduce the priority of the instruction in the congestion area, so as to relieve the traffic bottleneck.

This invention provides a method for predicting the spatial distribution of demand for low-altitude take-off and landing facilities in a target area, relating to the field of infrastructure planning technology. The method includes: S1, analyzing and predicting historical demand data for four scenario types based on historical demand data of the target area; S2, using Monte Carlo simulation for parameter calibration and improving the prediction accuracy of the modified exponential growth model and logistic growth model through Bayesian estimation; S3, determining the demand for low-altitude passenger take-off and landing facilities and the demand for low-altitude cargo take-off and landing facilities, and optimizing the total demand for low-altitude take-off and landing stations in the target area; S4, planning the low-altitude infrastructure in the target area to achieve spatial matching between low-altitude take-off and landing facility nodes and demand hotspots. This invention integrates multiple dynamic technological progress models and multi-scenario analysis mechanisms to improve the accuracy and scientific nature of data prediction; it introduces a station reuse coefficient to achieve efficient sharing of passenger and freight facilities and optimize the planning and layout of low-altitude take-off and landing facilities.

The application discloses a variable fractional order viscous acoustic wave equation attenuation compensation reverse time migration method and system, and the variable fractional order viscous acoustic wave equation attenuation compensation reverse time migration method comprises the following steps: a stable attenuation compensation viscous acoustic wave equation is constructed by introducing a regularization term; and based on the stable attenuation compensation viscous acoustic wave equation, a local cross-correlation imaging condition based on the Nyquist sampling theorem is used to carry out attenuation compensation reverse time migration of viscous acoustic waves. The application adopts the method of adding a regularization term, can effectively inhibit the exponential growth of high-frequency noise in seismic records in the compensation process, and ensures the numerical stability of the attenuation compensation process; the local cross-correlation imaging condition based on the Nyquist sampling theorem can effectively reduce the calculation and storage cost of Q compensation reverse time migration on the basis of adapting to complex structure imaging, and improve the practicability of the reverse time migration imaging of viscoelastic media.

A device for measuring an exponentially decaying process or an exponentially increasing process according to one embodiment is provided. The device comprises a measurement window determination unit (110) for determining a current measurement window, wherein the current measurement window defines a time range determined by the first time point after excitation of a signal source and the second time point up to which the signal is to be measured. Furthermore, the device comprises a sensor (120) for performing a current measurement of the signal precisely during the time range defined by the current measurement window. The device also comprises an evaluation unit (130) for determining a current measurement result of the current measurement.The measurement window determination unit (110) is configured to determine the current measurement window depending on a previous measurement result of a previous measurement, wherein the previous measurement was carried out exactly during a time range defined by a previous measurement window.

The application provides a hyperbolic trajectory decoupling action generation method and system, and relates to the field of artificial intelligence and robot continuous control. The method comprises the following steps: adaptively estimating the curvature of the hyperbolic manifold based on a data-driven method of an expert action sequence; incrementally scaling the robot continuous physical action sequence with time through a time geometric embedding mechanism, and projecting it to the hyperbolic space; isolating long-term prediction uncertainty by using the exponential expansion of the boundary; learning the optimal transport geodesic flow in the hyperbolic space, and optimizing the network through a loss function with conformal factor weighting; performing first-order Riemann Euler single-step integration based on the geodesic flow with zero covariant acceleration characteristics during reasoning, and finally inversely scaling and back-projecting back to the Euclidean space. The application naturally decouples the conflict trajectory by using the volume advantage of the exponential growth of the hyperbolic space, avoids the mode average problem, and realizes high-precision single-step robot action generation with extremely low delay.

A device for measuring an exponentially decaying process or an exponentially increasing process according to one embodiment is provided. The device comprises a measurement window determination unit (110) for determining a current measurement window, wherein the current measurement window defines a time range determined by the first time point after excitation of a signal source and the second time point up to which the signal is to be measured. Furthermore, the device comprises a sensor (120) for performing a current measurement of the signal precisely during the time range defined by the current measurement window. The device also comprises an evaluation unit (130) for determining a current measurement result of the current measurement.The measurement window determination unit (110) is configured to determine the current measurement window depending on a previous measurement result of a previous measurement, wherein the previous measurement was carried out exactly during a time range defined by a previous measurement window.

The application discloses a kind of flow field real-time prediction methods of combining Latin hypercube sampling, it is related to the field of foundry production.Provide a kind of fast prediction model of continuous casting three-dimensional crystallizer flow field combining Latin hypercube sampling and machine learning technology.The model aims to fully cover the parameter space by Latin hypercube sampling method, excavate the potential mapping relationship between key process parameters and the distribution of molten steel flow field in crystallizer, realize the fast and accurate prediction of flow field form.The model has the characteristics of high precision, low computing resource demand and fast response, can effectively overcome the shortcomings of long time-consuming and large resource consumption of traditional numerical simulation technology.Meanwhile, it makes up for the defects of traditional sampling methods such as poor representativeness of randomly sampled samples and exponential growth of sample size in high-dimensional space under uniform sampling.

The application discloses a V-shaped hollow sill arrangement method for a lateral water distribution pool, and comprises the following steps: determining an inlet flow water level and a flow velocity, and calculating an inlet Froude number Fr; through numerical simulation of multiple flow conditions, an outlet comprehensive uniformity is calculated according to a lateral water distribution outlet section of the water distribution pool; data scatter point curves are fitted to obtain a unimodal function curve of the lateral outflow uniformity changing with the lateral sill angle under each flow condition, and a curve peak value is recorded; an exponential growth curve function of the lateral outflow uniformity peak value changing with the Froude number under each flow condition is obtained; and based on the determined Froude number, a diffusion angle of the lateral water distribution side of the sill is obtained and adjusted. According to the application, the V-shaped sill with a variable angle is set according to different flow conditions, so that the V-shaped sill is suitable for various flow conditions, and the engineering cost is reduced, and the hub benefit of the water distribution gate station is improved.

The invention provides a Botnet attack simulation-oriented knowledge enhanced large model time series data generation method. A simulation instruction is received through an interaction interface, a semantic enhancement module is communicated with a knowledge graph and threat intelligence to complete intention recognition, graph reasoning, template matching and consistency verification, and structured scene features with confidence are output; the environment perception adaptation module calculates a scale scaling factor in combination with asset surveying and mapping, vulnerability risks and a historical utilization success rate; the generation engine generates an exponential growth-cosine platform-exponential decay parameter template by using Transform in a unified normalized time sequence space, renders a curve, and introduces OU noise synthesis flow and an active host time sequence; the self-adaptive optimization module verifies the peak value, the accumulated value and the increment under the asset / IP / port / bandwidth boundary and updates the template or the low-rank adaptation layer in a closed loop mode, high simulation of form-scale decoupling is achieved, and simulation data can be restrained.

The invention relates to the technical field of non-linear energy collection in a smart power grid, a sensor and an actuator arranged in the smart power grid are used for monitoring the running state of the power grid, but a large number of nodes from power generation to terminal equipment are in exponential growth, and under the background that spectrum resources are scarce, the power grid running state cannot be monitored. And a great challenge is brought to ensuring reliable and low-delay communication access. The invention provides a power grid CR-NOMA system throughput improving method considering nonlinear energy harvesting, and the method comprises the steps: obtaining a power and time distribution coefficient of secondary equipment in each time slot through an intermediate variable by employing a convex optimization tool for the throughput maximization of the secondary equipment employing a nonlinear energy harvesting model; according to the method, a deep deterministic strategy gradient framework combining a long short-term memory and an attention mechanism is constructed, the optimal energy residue is searched in all time slots, the convergence speed is improved, the signaling overhead for searching the optimal energy residue in each time slot is reduced, and the method has practical guiding significance for deployment of green operation of the smart power grid.

This invention relates to a battery fault detection method based on quantile and exponential weighting adaptive normalization, belonging to the field of battery fault detection technology. The method includes the following steps: S1: Collecting electric vehicle operating data, including voltage data and time parameters of each cell in the battery pack, and preprocessing it; S2: Performing Kalman filtering on the cell voltage data; S3: Normalizing the filtered cell voltage data, including using quantiles to determine the normal value range, applying an exponential decay function to normal values, and applying an exponential growth function to outliers; S4: Calculating the DTW distance between each pair of normalized cell voltage time series and calculating the standard score of the distance; S5: Setting a fixed threshold and determining whether the standard score of the cell voltage exceeds the set threshold, thereby determining whether each cell battery has failed.