701 results about "Normal density" patented technology

The invention is suitable for the technical field of life evaluation, and provides an enhanced high-voltage circuit breaker life evaluation method, which comprises the following steps: constructing a dynamic evolution model of a contact material wear rate; constructing a nozzle degradation dynamic prediction model; constructing residual life probability distribution of the insulating material; according to the dynamic evolution model of the wear rate of the contact material, the dynamic prediction model of nozzle degradation and the residual life probability distribution of the insulating material, constructing a comprehensive life evaluation index; inputting the comprehensive life evaluation index into a preset layered prediction architecture for evaluation; wherein the first layer generates basic life distribution through a Bayesian network, the second layer outputs posterior life distribution correction parameters through a convolutional neural network, and the third layer optimizes posterior distribution through a KL divergence minimization algorithm to obtain a probability density function and a confidence interval of residual electrical life; according to the method, the error of life evaluation can be reduced on the basis of complex working conditions.

The invention relates to the technical field of structural damage identification, in particular to an unsupervised structural damage identification method based on CAE-LSTM. The method comprises the following steps: training a CAE-LSTM model by using a training set to obtain a trained model, and reconstructing unknown data including health data and damage data by using the trained model to obtain a reconstruction error of the health data and a reconstruction error of the damage data; determining a damage sensitivity factor of the acceleration response signal of each batch by combining a probability density function of health data, a probability density function of damage data and a reconstruction error in the acceleration response signal of each batch of the undamaged structure so as to determine a damage threshold and screen a damage position; acquiring a damage factor according to the health state data and the damage state data of the damage position sensor; and judging the damage degree based on the size of the damage factor. According to the invention, the accuracy and reliability of the structural damage identification result are improved.

The invention discloses a computer numerical simulation method, system and equipment for a composite material, particularly relates to the technical field of analog simulation, and is used for solving the problem of simulation distortion caused by neglecting the randomness of a microstructure in the prior art. The method comprises the following steps: acquiring microstructure image data of a composite material, extracting fiber arrangement direction and pore distribution parameters, and generating a fiber direction probability density function and a pore topological relation; constructing a non-uniform grid model constrained by a local coordinate system according to the fiber direction probability density function; identifying a weak connection area based on a pore topological relation and marking nodes; applying dynamic decoupling constraint to the weak connection area, wherein the dynamic decoupling constraint comprises initial rigidity and rigidity attenuation rate; applying an external load to the model and calculating a real-time stress coefficient of the weak connection area; dynamically adjusting the constraint parameter through threshold judgment and iterative correction until the stress coefficient is converged; and finally, outputting a corrected macroscopic mechanical property result.

The invention discloses a PINN-based high-precision hydrodynamic numerical simulation method and system, and the method comprises the steps: firstly building a computational domain, setting reasonable geometric parameters and boundary conditions, and constructing a dimensionless Navier-Stokes control equation set; then designing a deep neural network architecture with space-time coordinate input and flow field variable output, and adopting a loss function combining physical constraint and data driving; the core innovation lies in providing a timing sequence sensing RAR-D adaptive sampling strategy, dividing a time domain into a plurality of time frames, performing residual error evaluation in each frame, constructing a probability density function related to residual errors, and balancing priority sampling and overall coverage of a high residual error region; adam and L-BFGS optimizers are adopted to carry out network training, the weight of a loss function is dynamically adjusted, and a sampling point set is periodically updated; and finally, the solution precision is verified through multi-dimensional flow field visualization analysis. Therefore, the prediction precision of the complex flow field is effectively improved, and the calculation efficiency is remarkably improved.

The invention relates to an equipment residual life prediction method based on a physical-data model, and the method comprises the steps: building a degradation model based on a physical degradation mechanism and a standard Wiener process model; obtaining fixed parameters in the degradation model by using a maximum likelihood estimation method in combination with a nonlinear regression method; acquiring an equipment state observation value in real time, and updating random parameters in the degradation model by using a weight optimization particle filter algorithm; and utilizing the degradation model, the fixed parameters in the degradation model and the random parameters in the degradation model to obtain a probability density function of the residual life of the equipment under a random failure threshold value, and performing numerical integration on the probability density function of the residual life of the equipment to obtain an expected value of the residual life of the equipment, and outputting the expected value as the residual life of the equipment. Compared with the prior art, high-precision and high-reliability residual life prediction is realized by fusing a physical mechanism and a data driving method.

The invention provides a trajectory design method and system in an unmanned aerial vehicle hidden and sensing integrated network, and the method comprises the steps: firstly deducing a multi-listener cooperative receiving signal expression and a probability density function in the sensing integrated network with multiple cooperative listeners; then deriving a minimum detection error rate expression of multi-listener collaborative detection and an optimal power division ratio expression meeting covert communication constraints based on a maximum likelihood principle; and establishing an unmanned aerial vehicle track and scheduling optimization problem of anti-cooperative detection in the unmanned aerial vehicle hidden and inductive integrated network according to the deduced formula. And finally, iteratively solving the problem based on convex approximation and an ellipsoid method to obtain an optimal unmanned aerial vehicle trajectory and scheduling. According to the method provided by the invention, the covert communication requirement in the unmanned aerial vehicle communication perception integrated network can be well met, and the communication safety in the future unmanned aerial vehicle Internet of Things is improved.

The invention discloses a power system operation reserve quantification method, system and device based on photovoltaic probability prediction and a medium. The method comprises the following steps: calculating an Euclidean distance between a photovoltaic predicted value of a point to be decided and a historical photovoltaic predicted value, and searching a photovoltaic power generation historical data set similar to the point to be decided; performing quantile regression based on the similar historical data set to obtain quantiles corresponding to a plurality of tail end quantile levels, fitting a probability density function of photovoltaic prediction deviation by using a Gaussian mixture model, and further calculating a risk loss expectation of a point to be decided; and constructing a standby cost function, considering the reliability constraint of the standby, taking the sum of the minimum standby reserved cost and the risk loss expected cost as a target function, and finally optimizing to obtain the standby capacity of the power system. According to the method, the change of the reserve price along with the capacity is considered, the photovoltaic probability prediction information can be fully utilized, and a more economical and reliable power system reserve quantification result is obtained.

The invention discloses a real-time quality monitoring method based on a production process parameter dynamic optimization model, relates to the technical field of intelligent quality monitoring, and solves the technical problems of empirical parameter adjustment and insufficient pertinence of exception handling. The method captures time sequence association of process parameters, process variables and quality indexes through a dynamic optimization model, quantifies uncertainty in combination with a probability density function, avoids limitation of a fixed threshold value, realizes early warning accuracy through deviation degree grading, reduces false alarm and missing alarm, screens core influence parameters based on model feature importance, and improves early warning accuracy. Quantitative adjustment suggestions are generated in combination with historical cases and inversion calculation, and empirical operation is replaced; multiple parameters are ranked and adjusted according to influence degrees, coupling interference is avoided, a single-point problem and a linkage problem are distinguished through parameter association chain analysis, a processing flow is formulated in a targeted mode, the single-point problem focuses on local repair and rapid recovery, the linkage problem focuses on cutting off a conduction chain and radically treating the source, and invalid intervention is reduced.

The invention discloses an IGBT residual life prediction method based on variational Bayes and a mixed Wiener process, and belongs to the technical field of IGBT residual life prediction, and the method comprises the following steps: S1, giving out the definition of the mixed Wiener process, and building a degradation model to describe the degradation track of an IGBT; s2, performing off-line parameter estimation based on a variational Bayesian algorithm by using a prior sample; s3, selecting an optimal model in the degradation models according to an automatic correlation determination method; s4, selecting a prior sample which is most similar to the test sample at the current moment; s5, updating the mean value and variance of the weight vector of the test sample on line; and S6, solving a probability density function of the residual life of the mixed Wiener process in the sense of the first arrival time, and estimating the residual life value of the IGBT at each moment. According to the IGBT residual life prediction method based on the variational Bayes and the mixed Wiener process, the degradation mode can be automatically determined, the fitting precision of an IGBT degradation track is improved, the requirement for specialty is reduced, and the universality of the model is improved.

The invention discloses a new energy power grid probability harmonic load flow calculation method based on improved semi-invariant and Gram-Charlier series expansion. The new energy power grid probability harmonic load flow calculation method comprises the following steps: establishing a random variable probability model, a wind-light output model and a wind-light joint distribution model; secondly, considering the output correlation between wind and light fields, converting the wind and light combined distribution model based on a Copula function into independent variables by utilizing Rosenblatt conversion for calculation, then calculating each-order semi-invariant of load, photovoltaic output and wind power output according to the expected values of node voltage and branch power flow and a sensitivity matrix, and finally calculating the output correlation between the wind and light fields according to each order semi-invariant of the load, the photovoltaic output and the wind power output. The method comprises the following steps of: obtaining a state variable, further solving each-order semi-invariant form of the state variable and the branch power flow, and finally, obtaining a probability density function and a probability distribution function of harmonic voltage of each node by combining Gram-Charlier series expansion.

The invention discloses a method and a system for calculating energy consumption of a coupler system in the running process of a railway wagon, and relates to the technical field of crossing of railway vehicle dynamics and green energy technologies. The multi-source energy consumption dynamic coupling model comprises a car coupler gap collision area, a sliding friction area between car couplers, a car coupler-slave plate seat impact area, a car coupler-carrier wearing plate fretting wear area and a buffer damping area; based on the energy flow topology network, determining an energy transfer priority and efficiency matrix in the multi-source energy consumption dynamic coupling model; and generating an energy consumption probability density function based on a generalized polynomial chaos expansion method, and outputting a low-carbon design parameter set and total energy consumption. According to the method, high-precision calculation and low-carbon design optimization of the full-life-cycle energy consumption of the coupler system are realized, and a core theoretical support is provided for green transformation of railway freight.

The invention relates to the technical field of data processing, in particular to a multi-series short-duration rainstorm data frequency analysis method. The method comprises the following steps: obtaining rainfall original data of multiple series of short duration, and carrying out missing data interpolation and sequence extension processing, duration consistency and rainfall intensity rule check to obtain a rainstorm data set; respectively performing reliability, representativeness and consistency review and sample validity screening on the rainstorm data set; a Pearson III type probability density function model is constructed, probability distribution fitting is carried out, and a frequency distribution curve is generated; and carrying out extra heavy rain value analysis and empirical frequency adaptive line analysis adjustment, generating a corrected frequency parameter, further calculating a design heavy rain value of each design return period, and carrying out duration consistency and spatial rationality check to obtain a final design result output data set. According to the invention, through multi-dimensional data complementation, checking and optimization fitting, the system improves the accuracy, reliability and engineering applicability of short-duration rainstorm frequency analysis.

The invention provides a radar and optical observation segmental arc correlation method and system for space debris. A joint probability density function is obtained through a kernel density estimation algorithm. A radar observation arc section and an optical observation arc section are input. And converting the observation data into a geocentric celestial coordinate system. And acquiring system parameters of the observation equipment. Radar observation data and optical observation data are unified to the same dimension, and a distance measurement value weight and an angle measurement value weight are set. And calculating an initial orbit by the radar observation arc section. And calculating a joint objective function corresponding to the current orbit position speed. And if the convergence condition is met, outputting the minimum value of the joint objective function and the orbit position speed corresponding to the minimum value. Otherwise, updating the track position speed, and repeating until the convergence condition is met. And according to the minimum value of the joint objective function and the joint probability density function, judging whether association is carried out, and outputting confidence during association. The method is easy to implement and can be effectively applied to association of the space debris radar and the optical observation arc section.

The invention discloses a new energy joint output scene generation system based on data driving, and the system comprises a data preprocessing module which collects and preprocesses wind power and photovoltaic historical output data; the joint probability distribution model construction module obtains wind power and photovoltaic edge distribution through probability density function fitting, and performs correlation analysis on wind power and photovoltaic through a Kendall rank correlation coefficient and a tail dependency coefficient to obtain a matching error of each connection function; the Euclidean distance between each connection function and the preprocessed wind power and photovoltaic historical output data is calculated, and an optimal connection function is obtained by combining the matching error, so that a wind power and photovoltaic joint probability distribution model is established; the scene generation module uses a Markov chain to form a joint probability distribution model with time correlation, and generates a random scene set through Markov chain Monte Carlo algorithm sampling. The stability of the power system is improved, and the operation cost is reduced.

The invention discloses a small fault detection method based on a prediction error sum of squares and JS divergence, and relates to the technical field of engineering detection, the algorithm of the application replaces a traditional algorithm for calculating JS divergence based on projection data after PCA dimension reduction, and a probability density function of the prediction error sum of squares of projected data is fitted by using kernel density estimation, so that the detection accuracy is improved. And calculating the JS divergence between different data distributions. The sum of squares of prediction errors reflects parts, which cannot be well reconstructed in the PCA model, of original data points, and the parts may contain abnormal or fault information. According to the method, the fault extraction capability of the prediction error sum of squares in the JS divergence is utilized, so that tiny faults can be better detected. By introducing the time sliding window, the method can more sensitively detect the local abnormal condition causing the increase of the JS divergence value, thereby timely and accurately capturing the process change, providing the early warning about the fault occurrence, and ensuring the safety and reliability of industrial production.

The embodiment of the invention provides a method for determining a generative model. The method includes embedding a white box watermark and a black box watermark in a generative model. Black box watermarks are first embedded into probability density functions of data abstraction in different layers of the generative model. The method also includes embedding the white box watermarks into different layers for generative model output after the embedding of the black box watermarks is completed. Model data is generated by the generative model based on predetermined trigger data. The predetermined trigger data comprises a predetermined trigger text or a predetermined trigger image. An identity associated with the generative model is determined based on the model data. By using the method, white box and black box attacks can be resisted by embedding two complementary and independent watermarks, so that double-layer protection is provided for the generative model.

The invention provides a hidden backscatter communication method based on multicarrier randomized continuous waves. The method comprises the following steps: acquiring multicarrier randomized continuous wave signal characteristic parameters, backscatter signal characteristic parameters and channel parameters between a transceiver and a listener and between the transceiver and each user; according to the characteristic parameters of the multicarrier randomized continuous wave signal and channel parameters between the transceiver and the listener and between the transceiver and each user, determining relative entropy between probability density functions of the listener receiving signal when the user generates a backscattering signal and when the user does not generate the backscattering signal; determining the total reachable rate between the transceiver and all users according to the multicarrier randomized continuous wave signal characteristic parameters, the backscattering signal characteristic parameters and the channel parameters; and maximizing the total reachable rate through a target optimization algorithm by taking the condition that the relative entropy meets the concealment condition as the constraint, and obtaining the optimized statistical characteristic parameters of the multicarrier randomized continuous waves. According to the invention, hidden backscatter communication can be realized.

The invention discloses a Fisher information matrix-based deformation monitoring sensor optimization arrangement method. The method comprises the steps of defining observation variables and unknown parameters; establishing a relation matrix; calculating a probability density function and a quality function of the observation variable; calculating a Fisher information matrix according to the deformation reconstruction equation; the traces of the Fisher information matrix are maximized to achieve optimal arrangement of the sensors. According to the method, the relation between the Fisher information matrix and the deformation reconstruction precision is deduced from the basic principle of an inverse finite element, optimal arrangement of the sensor is completed by maximizing the trace of the Fisher information matrix represented by the overall stiffness matrix, and efficient optimal arrangement of the sensor is achieved.

The invention relates to a wind power distribution prediction method and device, electronic equipment and a medium. The method comprises the following steps: constructing multi-dimensional feature data, preprocessing the multi-dimensional feature data to obtain preprocessed multi-dimensional feature data, extracting time sequence features, spatial features and meteorological features, and constructing a multivariate condition feature vector. And further inputting the current multi-dimensional feature data into the trained diffusion generation model, and outputting a wind power probability distribution sample. And carrying out probability density function fitting based on the wind power probability distribution sample to obtain multi-dimensional joint probability distribution of the wind power. According to the method of the embodiment, the complex characteristics of the wind power are effectively captured by constructing the multi-dimensional characteristic data, the prediction accuracy is remarkably improved, the time-space characteristics of the data are fully mined through characteristic extraction, the depiction ability of the model to complex wind conditions is enhanced, and the reliable probability distribution prediction is provided by utilizing the excellent distribution learning ability of the diffusion model.

The invention discloses a box girder bridge whole system performance degradation probability evaluation method based on DBN and Gamma distribution, and belongs to the technical field of box girder bridge performance degradation probability evaluation. Determining the staying time of each component of the box girder bridge; calculating related parameters of a Gamma distribution probability density function; calculating an accumulated probability density function and an accumulated survival probability function of each component of the box girder bridge; modeling performance reduction of the box girder bridge to obtain a transition probability; calculating state probability distribution of each component; a probability degradation curve of each component and the overall structure is formed through a dynamic Bayesian network; and evaluating the influence of the component on the probability degradation curve of the affiliated component and the overall structure. By adopting the DBN and Gamma distribution-based box girder bridge whole system performance degradation probability evaluation method, the performance degradation of the box girder bridge from components to parts and from the parts to the whole can be more scientifically predicted, so that the service life of the box girder bridge can be predicted from the probability level.

The invention discloses a power distribution network optimization scheduling method and system based on multi-target deep reinforcement learning, and relates to the technical field of power distribution network optimization operation, and the method comprises the steps: building a power distribution network operation model and a probabilistic load flow model considering the photovoltaic output and load demand uncertainty, completing the probabilistic load flow calculation under the uncertainty, and obtaining a power distribution network optimal scheduling model; analyzing to obtain a probability density function of node voltage and line power flow of the power distribution network; introducing a utility function based on preference, establishing voltage out-of-limit and line overload risk indexes considering severity weight, and constructing a risk-economic collaborative power distribution network multi-objective optimization operation problem; the decision process of the problem is modeled into a multi-target Markov decision process, a decomposition-based multi-target deep reinforcement learning algorithm is adopted, learning and training of the decision process are performed on reinforcement learning agents, a Pareto strategy set is obtained, an optimal operation strategy is screened out, and optimization regulation and control are performed on the power distribution network. And collaborative optimization of the operation risk and cost of the power distribution network system is realized.

The invention discloses a JSD-based field distribution test method and device in a complex electromagnetic environment, a medium and equipment. The method comprises the following steps: acquiring sample data of an electromagnetic signal in a microwave reverberation chamber; the method comprises the steps of obtaining a preset zero hypothesis and a preset alternative hypothesis; converting discrete sample data into a continuous empirical probability density function through kernel density estimation; calculating the Jensen-Shannon divergence between the empirical probability density function and the probability density function of the assumed distribution as the statistical magnitude of the JSD test; if the JSD test statistic is lower than the critical value, the assumed distribution corresponding to the preset zero hypothesis is not rejected, otherwise, the assumed distribution corresponding to the preset zero hypothesis is rejected, and the alternative hypothesis is accepted. The method can recognize the distribution form of the electromagnetic field in the reverberation chamber, and solves the limitation of a conventional hypothesis testing method when the number of independent samples is small or candidate distribution is similar.

The invention provides an electric power system power supply capacity insufficiency evaluation method considering new energy uncertainty, and the method comprises the steps: building the probability distribution of electric quantity prediction results at different time scales in a to-be-evaluated time period through Gaussian mixture distribution, and obtaining the probability distribution of the electric quantity prediction results in a total time period; constructing a joint probability density function according to the prediction error distribution of the generated power of the plurality of wind power plants and the prediction error distribution of the generated power of the plurality of photovoltaic power stations, and obtaining a space joint probability density function; establishing the probability density distribution of a wind-solar joint prediction error band in the to-be-evaluated time period according to the probability distribution of the electric quantity prediction result in the total time period and the spatial joint probability density function; calculating to obtain load shedding probability distribution through probability density distribution of a day-ahead operation domain and a wind-solar combined prediction error band; therefore, the risk of insufficient power supply is determined. And flexible assessment of the short-term scale power supply capacity insufficiency risk in the power system with short-term uncertainty in the new energy is realized.

Embodiments of the invention disclose a slow node detection method and apparatus, a device, a storage medium and a program product. The method comprises the steps of obtaining respective first duration data of a plurality of processing nodes participating in distributed model training; the first duration data represents the calculation time consumption of the corresponding processing node in the current iteration; based on a parallel strategy trained by the distributed model, grouping the first duration data to obtain a first duration set of multiple groups; the parallel strategy represents an association relationship of the plurality of processing nodes in a forward and backward propagation process and training tasks undertaken by the plurality of processing nodes; and based on the respective probability density function and a first threshold value of the plurality of groups, respectively analyzing and processing the first duration sets of the plurality of groups to obtain slow nodes in the plurality of processing nodes. Thus, the slow nodes in distributed model training can be intelligently and accurately detected, the efficiency and accuracy of slow node detection are improved, and the method has high compatibility.

A nuclear power plant passive system reliability analysis method based on meta-model adaptive sampling comprises the following specific steps: 1, extracting a sample to establish a model, obtaining a response value through an optimal estimation program, and training an initial response surface; 2, extracting a sample, mapping by using a response surface, and selecting a response value as an intermediate event critical value; 3, selecting a design point from the samples within the critical value range as a new sampling center, generating a new sample, supplementing training data, and constructing a new response surface; 4, obtaining a sample obeying important sampling probability density function distribution by using new response surface mapping, obtaining sample points falling in a critical value range, and re-selecting a critical value; and 5, repeating the steps 3 and 4 until the critical value is less than 0, ending hierarchical calculation, and obtaining an estimated value of the failure probability. The method aims at reliability analysis and development of the passive system of the nuclear power plant, the precision of the response surface model is improved through the adaptive sampling algorithm, the calculation cost is low, and the calculation precision is high.

The invention discloses a processing process fault prediction method for essential oil production, and relates to the technical field of essential oil processing, and the method comprises the steps: constructing an environment equipment coupling model, and drawing an electrical equipment insulation resistance attenuation curve through a salt mist deposition experiment; environment disturbance variables are introduced into the generative adversarial network, and an anomaly detector capable of recognizing salt spray corrosion characteristics and humidity condensation characteristics is trained; capturing time-space correlation between temperature mutation and quality fluctuation through a self-attention mechanism; calculating a probability density function of the residual service life of the equipment in real time, and automatically correcting a temperature over-limit threshold value based on the residual service life distribution; calculating the contribution degree of each node to the fault through intervention simulation; and constructing a digital twinborn model containing thermodynamic and hydrodynamic characteristics. Through environment dynamic modeling, causal reasoning and digital twinning methods, the key problems of low fault prediction precision, poor traceability efficiency and lack of quantitative support in decision making in essential oil production are solved.

The invention belongs to the technical field of mechanical truss structure reliability quantitative analysis, and relates to an evaluation method and system for quantifying the reliability of a mechanical truss structure based on nested integration, and the method comprises the steps: generating an initial moment integral of a reliability parameter of the mechanical truss structure; establishing a continuous nested integral model for quantifying the reliability of the mechanical truss structure, and outputting a nested integral rule; and calculating a numerical integration form of the reliability parameters of the mechanical truss structure according to a nested integration rule to obtain reliability evaluation of the mechanical truss structure. According to the method, continuous nested integral nodes and weights of any probability density function for mechanical truss structure reliability parameters are generated, unknown variable values such as nested integral nodes, nested integral weights and integral nodes are obtained by solving a minimization problem, and small singular values and numerical error influences are reduced by using regularization; and a nested integral rule with the highest polynomial algebraic precision under any probability density function can be obtained, and the stochastic analysis problem in the mechanical truss structure response can be effectively solved.

The invention belongs to the field of temperature regulation, and particularly relates to a carburant drying equipment temperature regulation method and system to solve the technical problems that an existing regulation mode is insufficient in prediction precision and poor in reliability. Establishing a joint probability distribution model of process disturbance and measurement noise by adopting a Gaussian mixture model; s2, obtaining a probability density function of a state trajectory in a future prediction time domain; s3, setting probability constraints represented by conditional value-at-risk; and S4, taking the first element of the optimal heater power sequence as the heater power control quantity of the current control period, taking the first element of the optimal working mode sequence as the working mode of the current control period, and jointly applying to the equipment. On the premise of ensuring a high safety standard, collaborative optimization of the power and the working mode of the heater is realized, and the energy consumption of equipment is reduced.

The invention discloses a current transformer error prediction method based on improved adaptive composite mode decomposition. The method comprises the following steps: preprocessing current transformer measurement data based on improved fully adaptive noise ensemble empirical mode decomposition (ICEEMDAN); performing feature selection and reconstruction on the complex components decomposed by the ICEEMDAN based on a multi-scale permutation entropy and a probability density function; further decomposing the reconstructed complex component based on variational mode decomposition (VMD); and constructing a depth prediction model based on CNN-BiGRU, optimizing sub-model network parameters by using a target optimization method, and finally superposing sub-results to obtain a final prediction result of the current transformer. According to the method, ICEEMDAN is comprehensively utilized for signal decomposition, MPE and PDF are utilized for feature selection and reconstruction, VMD is utilized for further decomposition of complex components, and CNN-BiGRU-MHA deep learning architecture is utilized for feature learning and prediction, so that nonlinear and non-stationary signals can be effectively processed, deep-level features can be extracted, a complex dynamic relationship in time sequence data can be captured, and the time sequence data can be obtained. And the prediction accuracy and the generalization ability of the model are improved.

The invention discloses a method and device for determining the reserve capacity of an electric power system, electronic equipment and a storage medium, and belongs to the technical field of reserve capacity management.The method comprises the steps that empirical mode decomposition is conducted on a predicted net load curve and an actual net load curve, and a high-frequency component, an intermediate-frequency component and a low-frequency component of the predicted net load curve and the actual net load curve are generated; calculating and predicting a high-frequency error sequence, an intermediate-frequency error sequence and a low-frequency error sequence; carrying out probability distribution modeling to generate a high-frequency error probability density function, an intermediate-frequency error probability density function and a low-frequency error probability density function; and determining high-frequency reserve capacity, intermediate-frequency reserve capacity and low-frequency reserve capacity of the power system according to the high-frequency error probability density function, the intermediate-frequency error probability density function and the low-frequency error probability density function based on a preset confidence coefficient. By implementing the invention, the problem that the reserve capacity of the power system is not accurately determined in the prior art can be solved.