35results about How to "Avoid divergence" patented technology

The invention discloses an industrial robot motion pose error calibration method based on constraint solution. The industrial robot motion pose error calibration method comprises the steps that S1, joint angles, tail end coordinates and temperature data of multiple measurement points are collected; s2, correcting the kinematic model by using the thermal elongation and constructing a parameter identification model; s3, constructing a mixed integer nonlinear constraint pool according to mechanical limitation; s4, introducing a semi-definite programming slack variable to convert the non-convex constraint into a linear matrix inequality; s5, combining a primal-dual interior point method and an alternating direction multiplier method to iteratively solve the optimal kinematics parameter correction; s6, introducing a square root information filtering improvement mechanism to correct a state vector square root factor, and calculating an updated nominal kinematics parameter; and S7, based on the condition number of the error Jacobian matrix, judging observability and repeatedly calibrating. According to the method, the problem of precision reduction caused by thermal errors is solved, and the identification precision, the numerical stability and the self-adaptive calibration capability of the kinematics parameters of the robot are improved.

The invention relates to the technical field of inertial navigation, in particular to an inertial navigation data correction method based on a neural network, which comprises the following steps: firstly, collecting original data of an inertial measurement unit and preprocessing; secondly, generating an attitude error label by using a high-precision reference system, and constructing a training data set containing historical time sequence information in a sliding window mode; then constructing a hybrid neural network model, and performing iterative training by using the training data set; and finally, in real-time navigation, obtaining an inertial data sequence at the current moment by maintaining a historical data buffer area, inputting a trained model to predict an error compensation value, correcting an original calculation attitude, and outputting the corrected original calculation attitude. The method gets rid of dependence on an accurate physical model through a data driving mode, utilizes the hybrid neural network to deeply mine time sequence accumulation characteristics of inertial data, realizes long-endurance high-precision autonomous navigation, and remarkably inhibits divergence of attitude errors.

The invention discloses a multi-source fusion continuous positioning method in combination with a prior feature map, and aims to solve the problem that an existing vision-inertial odometer easily generates accumulative errors in a GNSS denial environment, the method comprises the following steps: constructing a prior feature map comprising key frame poses, feature point descriptors and 3D coordinates based on an open source algorithm; a track displacement difference alignment strategy is adopted to complete visual-inertial system and GNSS joint initialization, and ESIKF is used to replace EKF to improve fusion precision; robust alignment of a local coordinate system and a map coordinate system is achieved through a single-frame geometric constraint accumulation-batch manifold alignment strategy; based on the reverse PnP, the current 3D point cloud and the historical 2D observation constraint are utilized to resolve the relative pose, the map matching result is used as observation to be fused with the GNSS or replace GNSS observation, the ESIKF loose coupling is utilized to eliminate accumulative errors, and continuous high-precision positioning in the GNSS failure scene is achieved.

The utility model provides side slope concrete spraying equipment which comprises a slope top moving mechanism and a slope bottom moving mechanism which are arranged on the top of a side slope and the bottom of the side slope, a guide mechanism parallel to the side slope is arranged between the slope top moving mechanism and the slope bottom moving mechanism, and a spraying mechanism capable of sliding in the height direction of the side slope is arranged on the guide mechanism. A lifting mechanism connected with the spraying mechanism is arranged on the slope top moving mechanism, a concrete spraying machine is arranged on the slope bottom moving mechanism, and the conveying end of the concrete spraying machine is connected with the spraying mechanism through a telescopic hose. The moving mechanisms are arranged at the top and the bottom of the side slope and matched with the guide mechanism parallel to the side slope, so that the spraying mechanism can stably ascend and descend along the guide mechanism, ecological concrete is accurately sprayed to the side slope, mechanization and continuity of spraying operation are effectively achieved, and the working efficiency is improved. And the phenomena of concrete dispersion and waste caused by long distance and large height in manual operation are avoided, and the spraying uniformity and the construction quality are remarkably improved.

The application discloses a radar target stable tracking method under a shielding environment, and comprises the following steps: a northeast two-dimensional coordinate system is established, and a target ship state is mapped into a state vector; a Kalman filter is used for prediction, a residual error and an innovation covariance matrix are calculated in combination with actual observation, and an elliptical threshold is constructed; if the observation falls into the threshold, the state is updated; otherwise, the observation is judged as a mismatched observation, and short-time missed detection prediction is performed; a sliding time window is set, when the number of mismatched observations in the window reaches a threshold, shielding judgment is performed in combination with the geometric relationship among the predicted positions of the ship and the target and the boundary of a static obstacle; if the condition is met, a shielding maintaining mode is entered, and the target track is continuously estimated by adjusting filter parameters or a predicted track; when a plurality of continuous observation frames fall into the threshold and the geometric shielding is removed, normal tracking is recovered through smooth transition. The application can effectively solve the problem of tracking interruption caused by the shielding of a radar target, and improve the continuity and stability of tracking under a shielding environment.

The invention relates to the technical field of industrial robot load identification, in particular to a six-degree-of-freedom robot end load parameter identification method, which comprises the following steps: establishing a six-degree-of-freedom series robot dynamical model, reconstructing a load minimum parameter set, and determining that identification can be completed only by exciting a third joint, a fifth joint and a sixth joint. Designing four groups of symmetric excitation trajectories combined by Fourier series and quintic polynomials, and performing discrete sampling to generate an observation matrix; optimizing a load parameter set and a regression matrix, and reducing the condition number of an observation matrix; a load minimum parameter set is obtained through difference calculation of the no-load identification result and the on-load identification result; and based on the parameter set, solving the mass, the mass center coordinate, the inertia moment and the inertia product of the load. The method solves the ill-conditioned problem of a regression matrix by reconstructing the minimum parameter set, reduces the number of excitation joints, does not need to depend on a robot body kinetic model, and has the advantages of high identification precision, high stability, low experiment complexity and high efficiency.

The application discloses a communication-restricted multi-agent tracking control method for trains, which comprises the following steps: S1, for a high-speed train dynamics model, a train formation dynamics model is designed; S2, a non-singular sliding mode surface is designed, which is brought into a dynamics equation, an equivalent control law and a total control law are designed; S3, a fixed threshold strategy is designed based on a control input signal error, and an event trigger condition is constructed; S4, through information interaction of adjacent vehicles, unmeasured states are estimated by using local measurable states, full-state feedback is provided for a controller, and a distributed observer is designed; the application has the beneficial effects that an improved terminal sliding mode surface is designed by using a non-singular terminal sliding mode surface, wherein the non-singular terminal sliding mode surface is a coprime odd number, fraction exponent irreducibility is ensured, periodic oscillation is avoided, and the singularity problem of traditional sliding mode control in time is eliminated by the fraction exponent term.

The application belongs to the technical field of integrated circuits, and particularly relates to a reconfigurable photonic sampling analog-digital converter control method, system and device, which comprises the following steps: S1, obtaining a starting safety configuration, performing an initialization self-check, and when the initialization self-check results all satisfy preset safety starting conditions, the initialization self-check is passed; S2, after the initialization self-check is passed, a target analysis frequency band is determined based on a target working interval, an initial sampling configuration is obtained, calibration samples corresponding to each activated channel are collected, and a coarse calibration configuration is obtained; S3, continuously sampling an analog signal to be converted to generate a performance state frame, performing deep learning analysis based on the performance state frame, and outputting an error estimation result; and S4, generating a candidate control vector according to the error estimation result, performing safety constraint auditing on the candidate control vector, taking the candidate control vector that passes the auditing as a system control vector, and performing target parameter soft updating in the system control vector. The application can be used for data conversion circuit design in a high-performance measurement system.

The invention relates to the technical field of motor control, and discloses an electrical transmission system parameter self-tuning system based on a neural network, and the system comprises a state monitoring and gating module which is used for activating the system when the high-frequency energy characteristic of a torque instruction exceeds a threshold value; the homomorphic mapping and reference reconstruction module is used for constructing a homomorphic digital filter to convert a torque instruction into a homomorphic instruction aligned with a feedback speed time sequence; the momentum residual calculation module is used for calculating the difference value between the theoretical impulse and the actual momentum increment based on the momentum theorem; the morphological decoupling and reasoning module is used for calculating morphological similarity between the residual vector and the reference vector by using the neuron structure so as to output confidence; and the parameter updating module is used for adjusting the adaptive updating gain according to the confidence coefficient and iteratively correcting the inertia parameter. According to the method, through signal homomorphic reconstruction and waveform form reasoning, the influence of loop delay and load disturbance is effectively inhibited, and the accuracy and robustness of parameter setting are improved.

The invention relates to the technical field of hydrological monitoring and computer vision, in particular to a hydrological station flow monitoring system and method based on deep learning, and the method comprises the steps: synchronously collecting the water level time sequence data of a river section and a video key frame flow containing water surface texture; converting the water level time sequence data into a corresponding water passing area numerical value and a hydraulic radius numerical value; outputting a flow state feature vector representing the dispersion degree of the water surface flow velocity; outputting a physical parameter vector changing along with time; calculating a moving average value of the river course roughness coefficient in a preset time window; judging whether the moving average value is greater than a preset siltation alarm threshold value or not; if the judgment result is yes, generating a river deposition early warning data packet and sending the data packet to a remote monitoring terminal; otherwise, generating a state log of normal operation of the river channel and storing the state log into the local database; the limit that a traditional flowmeter can only output numerical values is broken through, and a key diagnosis basis is provided for operation and maintenance and flood control decision making of a hydrological station.

The invention relates to a power generation method for powering at least one electrical grid of an aircraft using a power generation system, the aircraft comprising two power supply paths, each power supply path comprising an electrical converter and a generator, the power system comprising two regulation units (21a) respectively associated with each power supply path, each regulation unit (21a) comprising a correction module (5), the correction module comprises an integral corrector (CI), the method comprising the following steps performed for the correction module (5): determining a power setpoint (PBP *) from a distribution voltage setpoint (VDC *) and a measurement of the distribution voltage (VDC); determining a divergence indicator (M3) from the gradient (GPBP *) of the power setpoint (PBP *); and resetting the integral corrector (CI) of the adjustment unit (21a) of the power supply path according to the divergence indicator (M3).

The application provides a signal type recognition method based on reinforcement learning, and relates to the technical field of artificial intelligence.The method comprises the following steps: performing zero-mean normalization processing and slicing operation on IQ data of training signals to generate IQ sample data; constructing a data set with one-hot labels; constructing a deep learning strategy network and a deep learning target network; iteratively performing a training process: extracting a sample data from the data set as a current state; selecting an action according to an epsilon-greedy algorithm, executing the action, and obtaining a reward and a next state according to the corresponding relationship of the data set; updating the parameters of the deep learning strategy network by a back propagation method according to the loss value calculation result and an optimizer, and outputting a deep learning strategy network model; inputting a signal to be recognized into the trained deep learning strategy network, and outputting a final signal type by a voting algorithm.The method can realize efficient recognition of communication signal types, and has strong signal recognition accuracy and generalization ability.

This invention discloses a robust aggregation method and system for federated learning based on multidimensional logs and deep reinforcement learning, applicable to federated learning systems including a server and multiple clients. The method includes: initializing a global model, a deep reinforcement learning agent, and an experience replay pool on the server side; sending the global model to each client and receiving local model updates; extracting multidimensional feature vectors and constructing a feature matrix through standardization and normalization; calculating the statistical moments of all client feature vectors, constructing a global state vector independent of the number of clients, inputting it into the agent's policy network to obtain an indicator weight vector; obtaining a corresponding comprehensive reputation score from the indicator weight vector and the feature vector of each client, introducing a temperature coefficient and normalizing it to obtain aggregation weights; weighted aggregation of the local model updates of all clients to update the global model; and updating the agent network parameters based on the accuracy of the new global model on the validation set as a reward signal.

The invention discloses a dynamic wear process simulation prediction method and device for inter-structure long-period large wear loss and a storage medium, and relates to the technical field of nuclear reactor fretting wear. The method comprises the following steps: constructing a finite element model among a plurality of to-be-analyzed components based on initial morphology parameters of the plurality of to-be-analyzed components; according to a grid reconstruction condition, based on the finite element model and the wear model, executing a preset operation in a segmented manner to obtain a wear depth set and a morphology parameter of each segment; determining wear process distribution based on the wear depth set and the morphology parameter of each segment; the preset operation comprises the following steps: detecting that a grid reconstruction condition is met; and stopping calling the finite element model and the wear model of the current time period, determining a wear depth set and morphology parameters of the current time period, updating the morphology parameters of the finite element model, calling the updated finite element model and wear model, and determining a wear depth set and morphology parameters of the next time period.

The embodiment of the invention relates to the technical field of information, and discloses a discontinuous perception area traffic flow estimation method, device and product, and the method comprises the steps: obtaining traffic flow observation data of a target area, and constructing a mask matrix of a missing value position in the traffic flow observation data; calculating a time delay matrix between observation time points; inputting the traffic flow observation data and the mask matrix into a time feature extraction network, and processing by using a long short-term memory network to obtain hidden variables; generating a time decay estimation value containing the time dimension feature; filling missing positions in the traffic flow observation data by using the time attenuation estimation value, and constructing pre-filled graph structure data; inputting the pre-filled graph structure data into a graph attention network, aggregating traffic flow information of adjacent vertexes based on a graph attention mechanism, and generating a spatial dependency estimation value; and fusing the time attenuation estimation value and the spatial dependence estimation value by using a tradeoff coefficient to obtain traffic flow complementation data, thereby making up for the deficiency of traffic situation index data.

The present application relates to the technical field of image processing, more particularly, the present application relates to an AOI defect detection method and system based on industrial vision, comprising: acquiring a gray scale image of an electronic screen to be measured; determining the physical period of pixel arrangement of the electronic screen; constructing a local period difference degree, which includes calculating the gradient difference value between any pixel point in the gray scale image and its adjacent pixel point along the physical period. The present application quantifies the degree of dirtiness of the image by calculating the local period difference degree and constructing a period confusion factor combined with statistical characteristics. On this basis, the ADMM acceleration coefficient is dynamically generated: a conservative step size is used for low-quality images to prevent algorithm divergence, and an aggressive step size is used for high-quality images to improve efficiency. This method effectively solves the problem that the traditional fixed step size strategy cannot balance detection speed and accuracy, significantly improving the robustness of the AOI system.

The application discloses a multi-source fusion continuous positioning method combined with prior feature maps, and aims at the problem that existing visual-inertial odometry is prone to cumulative error in GNSS denial environment. The application constructs a prior feature map containing key frame poses, feature point descriptors and 3D coordinates based on an open source algorithm. A trajectory displacement difference alignment strategy is adopted to complete the joint initialization of a visual-inertial system and GNSS, and ESIKF is used to replace EKF to improve fusion accuracy. A single-frame geometric constraint cumulative-batch manifold alignment strategy is adopted to realize robust alignment of a local coordinate system and a map coordinate system. Based on reverse PnP, relative poses are solved by using current 3D point clouds and historical 2D observation constraints, the map matching result is taken as observation and is fused with GNSS or replaces GNSS observation, and ESIKF is used for loose coupling to eliminate cumulative error, so that continuous high-precision positioning in a GNSS failure scene is realized.

This invention belongs to the field of radiative heat transfer technology, specifically relating to a method for obtaining the radiation characteristics of divergent suppressed particles. The calculation results of the absorption factor and scattering factor obtained by this method are in good agreement with those obtained by traditional Mie scattering calculations. When the shape factor increases to the point that traditional Mie scattering calculations diverge, the proposed divergent suppressed particle radiation calculation method does not diverge, effectively solving the divergence problem that occurs in traditional Mie scattering calculations when the shape factor increases. By introducing the divergent suppressed Riccati-Bessel formula and stepwise recursive calculations, the stable calculation of the absorption factor Qa and scattering factor Qs is ensured, avoiding divergence and thus improving the reliability and accuracy of particle radiation characteristic calculations.

The application discloses a kind of dual-mode scramjet internal flow field automation CFD solving method, comprising: using Python script to build AnsysFluent automation solving environment, load the grid file of dual-mode scramjet model, using Python script to configure the physical model and boundary condition of dual-mode scramjet, using Python script to complete the setting of numerical discrete format, solving method and the flow field initial value construction of calculation domain;Flow field initial value is used as the starting point of iteration, using Python script to execute the dynamic climbing calculation under first-order accuracy, obtain stable initial flow field distribution;Stable initial flow field distribution is used as the initial value of calculation, using Python script to switch second-order high-precision calculation and carry out convergence determination.This method effectively solves the problems of artificial intervention tedious, shock capturing difficulty and convergence determination subject in high Mach number strong compressible flow calculation.

The invention discloses an intelligent jig control method integrating multi-dimensional sensing detection and technological parameter closed-loop regulation and control, and particularly relates to the technical field of intelligent jig control. A space coupling description set reflecting the influence relation of local regulation on the global situation is constructed, and the independent working condition characteristics of each pinch point are extracted; real clamping change caused by local adjustment and passive response caused by elastic force transmission of a workpiece are effectively distinguished, and the system can sense the diffusion trend of local parameter change in a multi-station system and the potential clamping unbalance evolution direction in advance by introducing a cross-station adjustment influence matrix and a cooperative stability evaluation quantity. Therefore, unified constraint and coordinated correction are carried out on the process parameter adjustment instruction, and force chain oscillation and state divergence caused by simultaneous adjustment of multiple stations are prevented.

This invention discloses a method and system for predicting the operating status of a belt conveyor based on digital twins. The method includes: collecting operating and environmental data of the belt conveyor drive system; constructing a digital twin thermodynamic observation model based on lumped parameter thermal network theory; analyzing the model prediction residual characteristics to obtain an initial adaptive forgetting factor; combining load fluctuation characteristics with compensation and correction to obtain a final adaptive forgetting factor; updating the algorithm covariance matrix and equivalent heat dissipation coefficient; finally, using the optimized algorithm to deduce the trajectory of the equivalent heat dissipation coefficient; predicting the critical time for thermal damage through an extrapolation model; and automatically triggering spare parts inventory allocation when this time meets the warning conditions. This invention optimizes the adaptive forgetting factor mechanism of the recursive least squares method, improving the accuracy and robustness of status prediction, realizing intelligent collaboration between equipment operation and maintenance and spare parts scheduling, and reducing the operating cost of the equipment throughout its entire lifecycle.

The application provides a split smoldering device, which comprises a smoldering box body, a heat-conducting pipe and a heating pipe, the smoldering box body is internally provided with a reaction cavity for containing smoldering reactants, and the smoldering box body is externally provided with a first external interface and a second external interface; the bottom of the smoldering box body is hingedly connected with a discharging plate; the heat-conducting pipe is provided with a pipe cavity for inserting the heating pipe and a first end and a second end located at two ends of the pipe cavity; the first end and the second end of the heat-conducting pipe are fixed to two opposite side walls in the smoldering box body, respectively; and the first end of the heat-conducting pipe is fixedly connected with the first external interface; the heating pipe is pluggable into the pipe cavity of the heat-conducting pipe; the heating pipe is provided with a first end and a second end; the first end of the heating pipe is connected with a heating device; and the second end of the heating pipe is adjacent to the second end of the heat-conducting pipe. The smoldering box body and the heating pipe are detachable, the heating pipe can be directly inserted into the smoldering box body for smoldering reaction, the smoldering is performed from the middle to the periphery, the smoldering speed is accelerated, and the smoldering efficiency is improved.

The application relates to the technical field of tidal current energy utilization, and discloses a solution method for oscillating hydrofoil motion response and fluid-structure coupling deformation, which comprises the following steps: defining hydrofoil motion modes and key parameters of structures and fluids, and clearly defining the core features of given pitching motion, free heaving motion and bending-torsion coupling deformation; establishing a heaving motion dynamics equation based on Newton's second law, and constructing a bending-torsion coupling deformation control equation in combination with the principle of structural mechanics; building a three-dimensional numerical simulation model containing a block grid and a boundary layer encryption design, and meeting the calculation requirements through spanwise and time discretization processing; solving fluid distribution load and total lift based on the Navier-Stokes equation; solving the heaving motion response by using the fourth-order Runge-Kutta method, and solving the bending-torsion coupling deformation by using the Newmark- β time integration method and the finite difference method. In this way, the strong coupling characteristics of hydrofoil motion and fluid-structure coupling deformation can be accurately captured, thereby providing technical support for the design optimization and performance improvement of an oscillating hydrofoil type tidal current energy device.

The invention discloses a power state prediction method, device and equipment of a battery pack and a medium, and relates to the field of battery pack detection.The method comprises the steps that a to-be-detected battery pack is placed under a preset pre-charging pulse working condition, and pulse voltage and pulse current changes of the to-be-detected battery pack under the preset pre-charging pulse working condition are obtained; extracting the discharge DC internal resistance of the to-be-detected battery pack according to the pulse voltage and pulse current change under the condition that the online parameter excitation is satisfied; acquiring operation data of the to-be-detected battery pack, determining polarization internal resistance and capacitance of the to-be-detected battery pack under the operation data according to the operation data, and performing parameter adjustment on the polarization internal resistance and the capacitance according to the operation data; and predicting the battery voltage of the to-be-detected battery pack according to the future current load working condition, the polarization internal resistance and the battery voltage at the current moment. On the premise of ensuring the estimation stability, the method also has the capabilities of tracking the aging condition of the battery pack and eliminating the individual difference of the battery pack.

This invention discloses a TCN-assisted GRU-based integrated navigation method for GNSS interruption environments, belonging to the field of integrated navigation technology. First, a TCN-assisted GRU temporal prediction network model is constructed. The GRU master predictive encoder extracts global temporal features and outputs GNSS position increment prediction results. During the training phase, the TCN auxiliary feature encoder extracts local temporal features from subsequences, guiding the GRU to learn stable feature representations through consistency constraints. During the effective GNSS period, IMU data and INS estimation states are collected to train the model. When GNSS is interrupted, the current and several previous periods' data are input into the trained model to predict pseudo-GNSS position increments. These are recursively accumulated to obtain the pseudo-GNSS position, and a Kalman filter is introduced to correct the INS calculation results. This invention effectively improves the positioning accuracy and robustness of integrated navigation under GNSS interruption environments.

An adaptive active noise control system and an adaptive active noise control method. The adaptive active noise control system includes an active noise control circuit and a control circuit. The active noise control circuit generates an anti-noise signal and includes at least one adaptive filter. The control circuit receives a first input signal derived from a reference signal output by a reference microphone that collects ambient noise, receives a second input signal derived from an error signal output by an error microphone that collects residual noise after noise reduction, and performs a comparison operation based on a first characteristic value of the first input signal and a second characteristic value of the second input signal to control the at least one adaptive filter.

The invention relates to the technical field of data processing, in particular to a fault section positioning method and system for an intelligent power grid. The method comprises the following steps: acquiring a topological structure, a fault alarm state and auxiliary data, wherein the auxiliary data comprises communication signal strength and battery voltage; calculating an original vulnerability index based on the voltage difference and the signal intensity, and performing saturation truncation to obtain a terminal state vulnerability value; determining a credibility attenuation coefficient of negative correlation based on the terminal state vulnerability value; dynamically correcting the false alarm rate and the missing report rate to generate an adaptive condition probability table; constructing a Bayesian network which takes a line section as a father node, applying an adaptive conditional probability table and taking fault alarm as evidence reasoning, and selecting a section with the maximum fault probability; the method comprehensively evaluates the physical state of the power distribution terminal, dynamically adjusts the data credibility, adapts to terminal changes in real time, effectively avoids misjudgment and missed judgment caused by fixed parameters, and improves the accuracy and reliability of fault section positioning of a smart power grid.

The application relates to the display field and specifically discloses a lamp strip and a display device. The lamp strip comprises a substrate and a plurality of light emitting units arranged on the substrate. The plurality of light emitting units are arranged along the extension direction of the substrate. A dam structure is further arranged on the substrate. The dam structure is arranged around at least part of the light emitting units and at least part of the light emitting units are surrounded in the space surrounded by the dam structure. The dam structure is used for shielding part of the light emitted by the light emitting units.

The invention discloses a vehicle side slip angle estimation method and system fusing Gaussian process uncertainty and unscented Kalman filtering. The method comprises the following steps: firstly, establishing a nominal vehicle model based on nonlinear three-degree-of-freedom dynamics, and calculating tire force by adopting a longitudinal and lateral coupled magic tire formula; a Gaussian process in machine learning is trained offline by using multi-working-condition data so as to learn a residual error between a predicted value and a true value of a nominal vehicle model and uncertainty of the residual error; fusing the nominal vehicle model output with the Gaussian process prediction mean value to construct a hybrid vehicle dynamics model; designing an adaptive unscented Kalman filter, taking the hybrid vehicle dynamics model as a state prediction equation, and dynamically injecting a residual variance predicted by a Gaussian process into a prior error covariance matrix to realize adaptive adjustment of process noise; the longitudinal and transverse speeds are estimated through the filter, and the side slip angle is calculated in real time. According to the method, the robustness and the adaptive capacity of the filter are improved through uncertainty quantization.

The application discloses an unmanned ship formation control optimization method based on a double-channel reinforcement learning framework, aiming at solving the problems of existing adaptive dynamic programming control parameter setting difficulty, traditional genetic algorithm easy to fall into local optimum and multi-dimensional error physical quantity dimension not unified. The application constructs an unmanned ship formation mathematical model and a bottom online control channel, and constructs an upper offline optimization channel; a dimensionless fitness function based on state energy ratio is designed; the upper channel utilizes a reinforcement learning intelligent agent to dynamically decide a genetic operator selection strategy according to a population evolution state to generate a control parameter, the bottom channel utilizes the parameter to perform simulation and feeds back a performance index to the upper channel to update a decision model, and the optimal control parameter is output through double-channel closed-loop iteration. The application improves control parameter optimization efficiency and precision, eliminates the interference of order of magnitude difference on optimization, and significantly enhances the dynamic response speed, steady-state tracking precision and adaptive capacity of the unmanned ship formation system.