256 results about "Accurate estimation" patented technology

The invention discloses a compressed air energy storage optimization operation method based on multi-parameter cooperative control, and the method comprises the steps: multi-source data collection and preprocessing, thermodynamic state multi-dimensional parameter dynamic estimation, three-dimensional reconstruction and evolution prediction of a temperature field in a gas storage, and system loss analysis and efficiency evaluation based on an entropy generation theory. And multi-time scale energy flow dynamic tracking and state evaluation, and operation parameter optimization suggestion generation based on state evaluation. According to the method, accurate estimation of non-ideal gas parameters is realized through particle swarm-Bayesian hybrid inference, a temperature field in a gas storage is reconstructed, interlayer available energy transfer items are introduced to quantify temperature stratification influence, the operation efficiency and reliability of the system are improved, the service life of equipment is prolonged, and the operation cost is reduced.

The invention belongs to the technical field of image processing, and particularly relates to an object six-degree-of-freedom pose estimation method based on three-dimensional geometric information registration, which comprises a target detection module, a three-dimensional point cloud generation model and a geometric information registration pose estimation module. According to the method, geometric reconstruction, a geometric information high-precision registration technology and an object pose estimation task are combined, a geometric reconstruction module introduces object geometric feature extraction and feature decoding to realize deep understanding and modeling of an algorithm on object geometric features, the whole process accords with geometric disorder, and the accuracy of object pose estimation is improved. And the accuracy and scene adaptability of the algorithm are further improved. According to the method, accurate estimation of the six-degree-of-freedom pose of the object can be achieved, and powerful support is provided for the fields of industrial automation, robot vision, AR / VR, automatic driving, medical robots and the like.

The invention discloses a railway box girder full-section data monitoring method and system based on a three-dimensional scanner, and particularly relates to the technical field of data monitoring, and the method comprises the steps: collecting the data of the full section of a railway box girder through the three-dimensional scanner, scanning each part of the box girder for multiple times, and determining the data collection process of the full section of the railway box girder. Collected railway box girder full-section data are processed, point cloud data are spliced through an aerial triangulation calculation process based on dense point cloud data, key point circle center position accurate estimation and key plane fitting and optimization, and an operator uploads the spliced data and special mark point coordinates to an intelligent measurement integrated platform system. The point cloud information and the contour information of the prefabricated box girder are collected, the acceptance assessment model of the prefabricated box girder is constructed, the possible significant deviation between the box girder model and the design standard is quantified, and the railway box girder full-section acceptance efficiency can be improved.

The invention belongs to the technical field of highway tunnel ventilation control, and provides a highway tunnel ventilation control method and system. The method comprises the following steps of multi-source heterogeneous data acquisition, multi-source data fusion, state estimation, traffic flow prediction, dynamic pollution index calculation and multi-target weight generation, rolling optimization with constraints, control execution, model updating and feedback correction. The state sensing precision is improved through multi-source data fusion, real-time accurate estimation and error reduction of a pollutant concentration field and a traffic flow state are realized through a heterogeneous sensor and a Rayleigh fusion technology, the problem of sensing deviation caused by single data in a traditional method is solved, future traffic flow is predicted based on a graph neural network, and the traffic flow sensing accuracy is improved. The fan power is optimized through model prediction control, the ventilation intensity is adjusted in advance, pollutants are prevented from exceeding the standard, the response speed is increased compared with threshold value control, and energy consumption is reduced.

The invention discloses a security operation and maintenance management method and system based on a network security event, and relates to the technical field of network security, and the method comprises the steps: collecting and preprocessing multi-dimensional data, and calculating an attack state based on the preprocessed multi-dimensional data; the attack influence degree is calculated by using a weighted propagation model based on the attack state, and a fitness function is constructed and optimized based on the attack influence degree to obtain an optimal protection strategy set; and converting the optimal protection strategy set into a control instruction, executing the control instruction, and storing, collecting and analyzing generated multi-dimensional data. A non-linear ordinary differential equation is constructed through a non-linear reaction diffusion model, the predictability and the coping capacity of network security management and operation and maintenance are improved, the attack influence degree is calculated through a weighted propagation model, accurate estimation of the wide propagation effect of network security threats is achieved, and through construction and optimization of a fitness function, the network security threats can be accurately estimated. The flexibility and adaptability of the protection system are improved, and the requirement for manual intervention is greatly reduced.

The invention belongs to the technical field of aero-engine health management, and discloses an aero-engine remaining service life prediction method based on a diffusion model and a space-time attention mechanism. The invention provides a novel learning framework which is used for RUL prediction of an aero-engine. The framework fuses a diffusion process and an attention-based time module to achieve end-to-end accurate estimation from sensor signals to RUL. Aiming at the problem that the scarcity problem of aero-engine data limits the application of a deep learning network in the aspect of improving the RUL prediction precision, the method adopts a diffusion process to enrich the data diversity and improve the data quality. In order to improve the accuracy of RUL prediction, an attention mechanism is used to extract time features in samples and between the samples, and time information is effectively utilized, so that the capture capability of the aero-engine performance degradation process is enhanced. The RUL prediction method provided by the invention provides more reliable support for health management and maintenance decision of the aero-engine.

The invention belongs to the technical field of automatic driving, and particularly relates to a 4D millimeter wave radar and camera fusion three-dimensional target detection method based on surface parameter estimation guidance. According to the method, information of an image and point cloud is effectively combined through an explicit target surface depth fitting model, the target surface depth is fitted, then accurate estimation of the three-dimensional space position of the target is achieved, and dense three-dimensional space information is obtained based on surface depth estimation in a subsequent module, so that accurate estimation of the three-dimensional space position of the target is achieved. And visual angle conversion of image features and feature extraction of false point cloud spatial position information are guided, so that the performance of a target detection algorithm is greatly improved.

The invention provides a wind power prediction error estimation method based on a meteorological prediction error, and the method comprises the steps: firstly analyzing the correlation between data, and determining main influence factors of the wind power prediction error according to the magnitude of the correlation; a multi-dimensional random variable kernel density estimation method is adopted to construct a weather forecast-forecast error condition probability distribution model; constructing an input feature data set based on the meteorological prediction error data set and additional feature correlation analysis, constructing a multi-neural network mixed error estimation model, and training by using the input feature data set and the wind power prediction error data set; and finally, randomly generating a large-sample weather forecast error sample according to the input weather forecast based on a Monte Carlo simulation method, constructing an input feature data set, inputting the input feature data set into the mixed error estimation model to obtain an error estimation sample data set, and finally fitting probability distribution of errors by using a kernel density estimation method. And error probability distribution estimation is completed. According to the invention, accurate estimation of wind power prediction error estimation is realized.

The invention discloses a forest land ecosystem carbon reserve accounting method and system, and the method comprises the steps: obtaining and screening feature variables, such as optical remote sensing data, vegetation indexes, texture factors, satellite-borne laser radar feature parameters, so as to determine features having the most influence on a target variable, and removing redundant and irrelevant variables; the method comprises the following steps of: screening characteristic variables, reserving variables which contribute to the model prediction effect, improving the precision and stability of the model, constructing a canopy height inversion model by adopting a random forest method based on the screened characteristic variables, realizing accurate estimation of the canopy height, and finally calculating the biomass of vegetation of the forest ecological system by using a different-speed growth equation. And finally, summarizing the biomass of the vegetation of the forest ecological system to obtain the total biomass of the vegetation, and multiplying the total biomass of the vegetation by the carbon-containing coefficient to obtain the vegetation carbon reserve of the forest ecological system, so that wide-area and efficient forest parameter monitoring is realized, the monitoring cost is reduced, the environmental adaptability is improved, and the problems of data continuity and standardization are solved.

The invention relates to a multi-parameter estimation algorithm combining a disturbance quantification model and a rooting method in interference measurement. In order to solve the problems that in the prior art, the disturbance degree of a sequence is difficult to quantify, and harmonic key parameters are influenced by spectrum leakage, the invention provides the following innovative scheme: firstly, a sequence multi-classification measurement entropy quantification model is constructed, and quantitative evaluation of the signal noise degree is realized; secondly, through a Hankel matrix construction and rooting parameter estimation method, realizing accurate estimation of the harmonic frequency of the real-value signal; meanwhile, an amplitude correction algorithm based on window function characteristics is designed, and the signal sampling condition is effectively judged; and finally, through matrix weighting operation of a phase demodulation function, efficient reconstruction of an optical surface shape is realized, the problems of spectrum leakage and disturbance quantification in a traditional method are solved, and the method has relatively high measurement precision.

The invention relates to the technical field of data acquisition, and more specifically discloses a data acquisition and analysis system and method for indoor decoration, and the method comprises the steps: extracting the decoration scheme information of a target indoor decoration project and pre-estimating the characteristics of parameter information, constructing a related feature map, and decoding the related feature map, according to the method, the material consumption of the indoor decoration engineering is accurately estimated, so that the problems of low accuracy, material waste or construction period delay existing in a single means based on manual estimation at present are effectively solved.

The invention discloses a transmission conductor three-dimensional attitude estimation method, system and equipment based on RTK and inertial measurement coupling and a medium, and belongs to the technical field of three-dimensional state monitoring, and the method comprises the following steps: fusing RTK absolute positioning and IMU inertial measurement data, and realizing accurate estimation of the transmission conductor three-dimensional attitude by using extended Kalman filtering; and abnormal identification and risk early warning are carried out based on an attitude calculation result, and finally real-time monitoring and intelligent warning of the state of the power transmission line are realized through a three-dimensional visual platform. According to the method, RTK and IMU data are fused, accurate sensing and continuous tracking of the three-dimensional attitude of the power transmission line are achieved through extended Kalman filtering, the defect that a single technology is insufficient in reliability in a complex environment is overcome, and finally a complete monitoring closed loop of the safety state of the power transmission line is achieved through intelligent early warning and three-dimensional visualization.

The invention belongs to the technical field of underwater integrated navigation. The invention provides a multi-navigation-state unmanned system underwater integrated navigation method based on global information fusion. According to the embodiment of the invention, different navigation combination modes under different navigation states are selected, and an SINS / DVL / GPS combined navigation mode is adopted offshore, so that the calculated amount is reduced, and the efficiency is improved; and the deep sea sneaking adopts an SINS / DVL / USBL combined navigation mode. Measurement deviation accumulative divergence caused by navigation state change in maneuvering characteristics of the underwater unmanned system is considered, t distribution is introduced in a federated Kalman filtering architecture for trailing characteristics to replace conventional normal distribution, thick-tail measurement noise is modeled, approximate solution of a posterior probability density function is carried out by using a variational Bayesian theory, and the probability density of the underwater unmanned system is calculated. Accurate estimation of the motion state of the unmanned system is realized, and updating and data fusion of a plurality of sub-filters are realized by using a federal structure.

The invention belongs to the technical field of flight control. The invention provides a predetermined time robust control method for an aircraft attitude angle under unknown disturbance. The method comprises the following steps: establishing an attitude angle and angular rate kinetic equation of an aircraft according to wind tunnel experiment data; in consideration of uncertainty and unknown disturbance existing in the kinetic equation, modeling the kinetic equation into a comprehensive disturbance form, and converting the kinetic equation into a ray-imitated form containing disturbance; a preset time disturbance estimator is designed to achieve accurate estimation of comprehensive disturbance, an attitude angle preset time backstepping controller is designed, disturbance is compensated in the controller, and preset time control over the attitude angle of the aircraft is achieved. The method disclosed by the embodiment of the invention has the advantages of simple structure and strong robustness, the parameter adjustment requirement for convergence time is simplified, and the preset time control on the attitude angle of the aircraft under the condition of unknown disturbance can be realized.

The invention discloses a rut depth estimation method and system based on pavement crack curvature feature inversion, and the method comprises the steps: collecting a high-resolution pavement image, carrying out the preprocessing and crack detection, extracting a crack contour, calculating the minimum curvature radius, curvature change rate and other parameters of a crack through difference geometry, and calculating the rut depth. And constructing a rut crack geometric response index R-CGRI to characterize the response strength of the crack to sinking deformation. An image is divided into M * N sub-regions, and the R-CGRI value of each region is calculated to form a two-dimensional response matrix. And performing regression fitting on each row of R-CGRI values, extracting a minimum point of a fitting curve as a rut depth estimation value, and finally reconstructing a rut longitudinal section form. The method does not need to depend on high-cost three-dimensional measurement equipment, accurate estimation and form modeling of the rut depth can be achieved only through two-dimensional image analysis, and the method is suitable for intelligent road inspection and deformation risk assessment.

The invention relates to the technical field of high-precision anti-interference satellite navigation, and particularly discloses a high-precision anti-interference navigation antenna array attitude autonomous estimation method, which comprises the following steps: designing a multi-antenna carrier tracking loop, and measuring a carrier phase in an interference environment; clock difference and ionosphere and troposphere delay are eliminated by means of carrier phase double difference between array elements. Solving a unique solution of each baseline vector from a carrier phase double-difference equation by utilizing the constraint that the array element spacing does not exceed the carrier half-wavelength; rough estimation of the array attitude is obtained according to the least square solution of the equation set satisfied by each baseline vector; compensating the phase error of each array element by utilizing rough attitude estimation; according to the sequence of inter-array element carrier phase double-difference calculation, multiple baseline vector solution, array plane attitude rough estimation and array element phase error compensation, performing iteration for three times to obtain array plane attitude accurate estimation; an attitude sensor, an INS antenna and a GNSS antenna do not need to be additionally installed, the method does not depend on other navigation sources, calibration is not needed, estimation errors are not accumulated, and the initialization time is short.

The invention belongs to the technical field of geotechnical engineering test and investigation, and provides a method for selecting an optimal shear wave velocity prediction model according to field demand evaluation, which can quantitatively evaluate the accuracy and rationality of a plurality of shear wave velocity prediction models and select the optimal prediction model. A reference value is provided for accurate estimation of the shear wave velocity and geotechnical engineering test and investigation; according to the technical scheme, the method comprises the following steps: evaluating and selecting a shear wave velocity optimal prediction model according to field requirements; the method comprises the following steps: determining a performance angle target area alpha'according to field requirements, respectively calculating mean values of center rates i and deviation angles alpha of various types of models for multiple types of shear wave velocity prediction models, performing comparative analysis to evaluate the accuracy and rationality of the shear wave velocity prediction models, and selecting an optimal shear wave velocity prediction model.

The invention relates to the technical field of atmospheric science detection, in particular to a lightning radiation source positioning method based on multi-station synchronous observation. Comprising the following steps: synchronously acquiring lightning broadband electromagnetic signals by multiple stations and performing time-frequency analysis; by calculating indexes such as cross-station coherence and interference level, high-quality frequency points are intelligently screened out to suppress noise and interference; reconstructing a narrow-band signal for each high-quality frequency point and performing high-precision time difference (TDOA) estimation by adopting an improved generalized cross-correlation method; and carrying out weighted fusion on TDOA measurement values of a plurality of high-quality frequency points, carrying out optimization solution by adopting a Grid Search algorithm, and finally obtaining a three-dimensional coordinate of lightning. Through intelligent screening of high-quality frequency points, accurate time difference estimation, optimization of a positioning algorithm, multi-band information fusion and intelligent processing, the defects that a traditional single-band positioning method is prone to interference and limited in precision are effectively overcome, the lightning positioning precision, the anti-interference capability and the reliability are remarkably improved, and the method is suitable for popularization and application. And the lightning positioning requirements in various environments are met.

The invention discloses a 3D human body posture reconstruction method and a lighting system based on feedback, and the 3D human body posture reconstruction method comprises the steps: carrying out the inverse projection of a predicted 3D position of a human body key point at a t moment, associating the predicted 3D position with an actual 2D position at the t moment to form a 2D joint point set of the human body key point, and calculating a plurality of 3D joint points corresponding to the human body key point according to the 2D joint point set; and fusing the 3D joint point clouds by using a fusion algorithm to obtain fused 3D positions of the key points of the human body, and optimizing the fused 3D positions by using a multi-view geometric parameter optimization method to obtain estimated 3D positions of the key points of the human body at the t moment. The 3D position estimation is more accurate and more reliable and is close to the real situation, the shielding problem can be relieved to a certain degree, and a more stable and more accurate estimation effect can be generated when the number of cameras is small.

The invention provides a failure spacecraft relative pose estimation method based on dual algebra, and belongs to the field of spacecraft navigation, and the method comprises the following steps: building a failure spacecraft six-degree-of-freedom relative motion and dynamics model under dual algebra description; then, through a matrix vectorization operator, determining analysis forms of a state equation and an observation equation, and establishing a multiplicative extended Kalman filter; finally, unscented transformation is introduced, a process noise covariance and measurement noise covariance adaptive law is designed, the problems that parameter adjustment of a complex system is tedious and convergence is slow are solved, and accurate estimation of the near-operation real-time pose of the failed spacecraft is achieved. According to the invention, successful execution of the capture maintenance task can be ensured.

The invention discloses a power distribution network state parameter joint estimation method based on double extended Kalman filter, relates to the technical field of power grid state prediction, and effectively avoids the influence of parameter variation on a state estimation result and the influence of state deviation on the parameter estimation result by introducing the double extended Kalman filter. For the situation that state data and parameter data in a power distribution network are inaccurate and have mutual influence, prior estimation and posterior estimation of two processes are set to interact to construct a state-parameter iterative estimation model, and the model converts a traditional state and parameter segmentation model into a unified estimation model. The state estimation model and the parameter estimation model of the power distribution network are jointly solved to solve the problem that the estimation precision is limited due to the fact that the models are mutually independent in a traditional method. Therefore, the real-time and accurate estimation of the operation state and parameters of the power distribution network is realized, the observability of the power distribution network is more comprehensively improved, and powerful data support is provided for fault diagnosis and operation optimization.

To provide a technique capable of accurately estimating heat energy consumption of a heat utilization facility without requiring an expensive flowmeter and without stopping a factory.SOLUTION: An estimation device (1) includes an estimation unit (11) configured to acquire a first measurement value from each of a plurality of heat utilization facilities (21 to 22) and calculate an estimation value of thermal energy consumption of each of the plurality of heat utilization facilities (21 to 22) from the first measurement value, a correction value calculation unit (12) configured to calculate a correction value from an analysis value of a drain flow rate of an entire plant (2) including the plurality of heat utilization facilities (21 to 22) calculated by a simulator (3) and an actual measurement value of the drain flow rate of the entire plant (2), and an estimation unit (13) configured to estimate the thermal energy consumption of each of the plurality of heat utilization facilities (21 to 22) by correcting the estimation value using the correction value.SELECTED DRAWING: Figure 4

The present invention relates to a method for accurately estimating the external water inflow rate and inflow points of a sewage pipe network based on model prediction. By establishing a sewage pipe network model, it is possible to accurately estimate the river water inflow rate and inflow points entering the sewage pipe network, and obtain the dynamic changes in the external water inflow volume, avoiding the influence of the pumping station operation conditions when interpreting the readings of the pipe network flow meters, and providing a quantitative management tool for the external water investigation, operation, and maintenance of the sewage pipe network. Reducing the external water inflow is an important guarantee for ensuring the normal operation of the sewage pipe network. After reducing the external water in the sewage pipe network, the liquid level of the sewage pipe network can be lowered, preventing sewage overflow during the rainy season and reducing the occurrence of water environment pollutant events caused by sewage overflow. In addition, reducing the external water entering the sewage pipe network can also increase the actual sewage collection capacity of the sewage pipe network, increase the sewage pollutant concentration, and improve the sewage treatment efficiency of the urban sewage treatment plant. After the external water is blocked, the flow rate of the sewage pump station decreases significantly, and the sewage concentration increases significantly.

The invention provides a school express multi-node credible collaborative distribution system based on a parallel chain technology, and relates to the technical field of logistics distribution, and the system achieves the bidirectional credible collaborative connection between a primary client and a secondary client through employing a module based on the parallel chain technology, and guarantees the safety and real-time performance of information interaction. The to-be-taken information is dynamically obtained and processed within the cooperative waiting time, and the response speed and accuracy of the distribution request are improved; a weight factor is dynamically adjusted based on real-time feedback, and optimization of a distribution strategy and accurate estimation of pick-up time delay are realized; the overall response capability and the prediction precision of the system are improved, the security and the convenience of the package receiving process are enhanced through the application of the trusted personnel white list, and the blank of the prior art in the aspects of multi-node collaborative distribution and dynamic optimization is filled.

The invention discloses an auxiliary direction of arrival estimation and communication method based on a dynamically rotatable space-based intelligent reflecting surface, and belongs to the field of wireless communication. The method specifically comprises the following steps: firstly, a user periodically sends a positioning pilot signal, and the positioning pilot signal is reflected to a base station by an IRS to preliminarily and roughly estimate a direction of arrival from the user to the IRS; then, the IRS is adjusted to turn to the maximum reflection efficiency, and the theoretical rough estimation incoming wave angle of the user is calculated; the IRS direction and the value of a measurement matrix are adjusted according to the rough estimation result, the signal-to-noise ratio of a receiving end is improved, and a more accurate estimation result is obtained; the fine estimation is repeated a plurality of times until the error reaches a specified threshold. After the direction of arrival of the user is estimated, the direction of the IRS is adjusted according to a final estimated angle value so that the reflection efficiency of the system can be maximized, the beam coverage range is reduced to a direction of arrival estimation error threshold value, and the optimal phase shift matrix theta of the IRS is solved by taking the worst signal-to-noise ratio of the maximized beam coverage area as an optimization target. According to the invention, the performance of the intelligent reflecting surface in assisting the wireless communication system is improved.

The invention relates to the technical field of ocean engineering construction automation, and discloses a hinge sinking mattress construction real-time monitoring control system based on multi-sensor fusion, which inverts a near-field hydrodynamic environment acting on a hinge sinking mattress through a hydrodynamic field inversion module by using strain measurement data and a parameterized forward dynamical model, and performs real-time monitoring and control on the hinge sinking mattress. The state fusion estimation module fuses sparse absolute position measurement data and high-density form measurement data in an extended Kalman filtering framework, introduces form manifold projection correction, and accurately estimates the position, attitude and form of the sinking mattress. And the predictive control decision module takes the estimated system state and the inverted near-field hydrodynamic environment as input, and solves an optimization problem to generate an optimal control sequence under multiple constraints of considering structural safety and physical limitation of an execution mechanism. According to the method, active suppression of hydrodynamic disturbance is achieved, the robustness and precision of state estimation are improved, and the automation level and efficiency of hinge sinking construction are remarkably improved.

The embodiments of the present application belong to the field of financial technology and relate to a vehicle loss estimation method, apparatus, computer equipment, and storage medium. The method comprises the following steps: identifying, classifying, and extracting features from sample data to obtain vehicle image features, vehicle description features, and vehicle information features, and calculating vehicle loss estimation pricing; combining vehicle image features, vehicle description features, and vehicle information features to obtain vehicle loss features, vehicle accident features, vehicle configuration features, and vehicle insurance policy features; inputting vehicle loss features, vehicle accident features, vehicle information features, vehicle insurance policy features, and vehicle loss estimation pricing into a preset learner for training, and performing integrated learning on the trained learner to obtain a vehicle loss estimation model; and inputting repair report data into the vehicle loss estimation model to obtain loss estimation data. The present application can effectively realize automated and efficient accurate estimation of vehicle losses.

The invention provides a distributed inertia-based relative navigation method based on a data link, and relates to the technical field of navigation. Firstly, a relative navigation error state equation based on speed error reconstruction is adopted, and a specific force term of high-frequency change is eliminated, so that the degree of dependence on inertial navigation data sharing between nodes is reduced; secondly, by designing and calculating an efficient information fusion method, the calculation complexity of the whole method is reduced; and finally, designing equivalent GNSS measurement for rapidly diffusing high-precision absolute navigation information. According to the invention, two core functions of accurate estimation of the relative state of the cluster node and rapid coverage of absolute information can be realized; through a simulation experiment, high-precision group inertial navigation error state propagation can be realized by sharing own inertial navigation data by cluster nodes at second-level intervals.

The invention relates to a Lamb wave frequency dispersion characteristic sparse reconstruction method based on non-convex optimization, and belongs to the technical field of high-end equipment perception and test.The method comprises the steps that excitation is conducted on a U-shaped suspension arm, and time domain sampling signals at the junction of a U-shaped suspension arm structure are obtained; the time domain sampling signals are preprocessed; constructing a non-convex sparse frequency dispersion curve estimation model based on the preprocessed signal; and converting the non-convex sparse frequency dispersion curve estimation model into an optimization problem, searching a convex preserving condition of a target function, selecting a forward and backward splitting algorithm for solving, obtaining frequency-wave number distribution, and further extracting a frequency dispersion curve. The method solves the problems that a traditional convex regularization method is limited in induction sparse capability and underestimates signal amplitude. Through a non-convex optimization method, the purpose of accurate estimation of the frequency dispersion curve based on fewer measurement points is achieved.

The invention relates to the technical field of new energy power generation, and discloses a power prediction deviation estimation method based on an intelligent combination model. The method comprises the steps of collecting historical generation power data and prediction data, calculating a power deviation sequence and analyzing error characteristics; introducing a subtraction average optimizer algorithm with an elitist strategy, and performing iterative optimization on the bandwidth constraint factor, the modal decomposition number and the initial learning rate by taking the minimum average absolute error as a fitness function; decomposing the error sequence into a plurality of intrinsic mode functions (IMF) by using variational mode decomposition; for each IMF component, a CNN-BiGRU-AM neural network model is constructed, and modeling and prediction are carried out; and finally, reconstructing the prediction result of each sub-component to obtain the accurate estimation of the overall error. By fusing error decomposition, deep modeling and parameter optimization, the estimation precision and stability of the power prediction deviation are remarkably improved, and the method is suitable for error correction and scheduling optimization of new energy power generation systems such as wind power, photovoltaic and the like.