42results about How to "Guaranteed convergence" patented technology

The invention discloses a non-linear non-strict feedback multi-agent system fixed time fault-tolerant control system, and belongs to the technical field of agent system cooperative control. Comprising a multi-agent system model, a coordinate transformation and command filtering module, an error compensation signal module, a backstepping recursion module, a neural network module, an adaptive law module and an adaptive fault-tolerant controller module. The method is used for compensating the influence of sensor faults on system consistency and ensuring that the system converges within fixed time, so that when the system has sensor deviation, gain and other faults, the system can still realize output consistent tracking within a predetermined time upper limit irrelevant to an initial state, the reliability and safety of the system are remarkably improved, and the system reliability and safety are improved. The technical problems of state feedback information distortion and system cooperation precision and stability reduction caused by sensor faults of the intelligent agent in the prior art are solved.

This specification pertains to the field of refining and chemical production processing technology, and particularly relates to a method and apparatus for optimizing refining and chemical production plans for a target secondary processing unit. The method includes: obtaining parameters of the secondary processing unit under a refining and chemical production optimization scenario; constructing a nonlinear constraint equation for product output based on the parameters of the secondary processing unit; performing an equivalent transformation on the nonlinear constraint equation to obtain an equivalent transformation equation; extracting the nonlinear constraint terms and performing a Taylor expansion; transforming the Taylor expansion formula into a linear approximation equation based on the proportion of the feed material to the output material of the blending tank, the feed material quantity to the blending tank, and the feed property values ​​of the blending tank, according to the property balance constraints of the blending tank; and then substituting these into the nonlinear constraint equation for product output, thereby transforming the nonlinear constraint equation for product output into a linear approximation equation containing only material information variables. This method effectively ensures model convergence and shortens the solution time.

The application discloses a radar complementary sparse frequency waveform sequence set design method based on a complex circle flow surface algorithm, and specifically comprises the following steps: according to a monitored spectrum environment, delimiting a usable spectrum range and an unusable spectrum range; converting a weighted integral sidelobe level of a waveform sequence set into a quadratic function form with a waveform sequence set vector as a variable, constructing an objective function to describe the weighted integral sidelobe level of the waveform set; calculating a power spectrum of the waveform set, controlling a weighting coefficient according to an expected spectrum, performing weighted summation on the power spectrum, and converting a weighted summation expression of the power spectrum into a quadratic function form with the waveform sequence set vector as the variable; and performing weighted summation on a WISL objective function and an EFS objective function to construct a joint objective function. The CCM optimization algorithm is applied to the complementary sparse frequency waveform sequence set design problem for the first time, effectively solving the constant modulus constraint problem of the waveform sequence, and compared with the existing cyclic iteration algorithm and the interior point method.

The application belongs to the technical field of optical three-dimensional measurement, and discloses a two-division iteration sub-pixel matching method and system for phase structured light three-dimensional measurement, which comprises the following steps: obtaining a phase map; performing legality marking and pixel-level initial matching on the phase map; determining a sub-pixel search area based on the pixel-level initial matching; and performing sub-pixel matching point search based on the sub-pixel search area by using a two-division iteration method to obtain a sub-pixel matching point. The application can improve the precision of binocular matching in phase structured light three-dimensional measurement, thereby improving the precision of three-dimensional reconstruction.

This invention discloses a lightweight joint estimation method for battery state of charge (SOC) in resource-constrained MCUs, comprising: extracting the weight matrices and bias terms of each layer of a trained LSTM time-series estimation network and flattening them into a one-dimensional static constant array; constructing a custom pure C language matrix operation library independent of a third-party AI inference framework on the target microcontroller, employing a global static memory pool management strategy, and defining a unified structure containing the number of matrix rows, columns, and one-dimensional floating-point data pointers at the C language level; the target microcontroller calling the custom pure C language matrix operation library to directly read the one-dimensional static constant array to complete the forward inference calculation of each gate unit of the LSTM network and output the prior SOC observation value; and calling the custom pure C language matrix operation library to perform state equation iteration of the Kalman filter and output the SOC estimation result. This invention enables the safe and efficient operation of complex coupled algorithms on a very low computing power platform.

The invention relates to a fixed-time sliding mode anti-lock braking control method and system based on fuzzy logic and a neural network. The method comprises the steps that a dynamic model of a vehicle anti-lock braking system is established; designing a fixed time sliding mode controller, and constructing a sliding mode surface containing a terminal attractor and an approaching law; a T-S fuzzy logic system is adopted to approach a non-linear tire-road surface friction function in an ABS system on line; an approaching law of the sliding mode controller is approached online by adopting an artificial neural network; carrying out online optimization on additional control item gains introduced by neural network approximation by adopting a particle swarm optimization algorithm; and the fixed time stability of the whole closed-loop system is monitored based on the Lyapunov stability theory. According to the invention, the braking efficiency, the stability and the robustness of the ABS system are effectively improved, and the braking distance and the braking time are reduced.

An automatic modeling method for damaged concrete structure belongs to the technical field of artificial intelligence and building structure health assessment, and comprises the following steps: (1) collecting point cloud information of the damaged concrete structure, and performing two-dimensional slice dimension reduction processing on the point cloud information; (2) constructing a SDS-YOLO segmentation network optimized for sparse discrete features; (3) adopting a U-Net model to detect and extract the outer contour edge line of the mask information of instance segmentation; (4) adopting a surface reconstruction method based on an implicit function to construct an entity unit; and (5) extracting internal steel information by using an LLM, and finally realizing finite element modeling. The present application utilizes the instance segmentation and edge detection technology in deep learning to extract the outer contour information of the point cloud information slice, simultaneously utilizes the large language model LLM to extract the internal steel information of the concrete structure, and finally realizes the reconstruction of the entity unit, and analyzes the fire resistance and static behavior of the damaged structure.

The invention relates to the technical field of electricity markets, in particular to a day-ahead-real-time rolling pricing method and system for an electricity selling company, and the method comprises the following steps: S1, constructing an evolutionary game replication dynamic equation of vehicle network interaction of the electricity selling company and an electric vehicle; s2, constructing an evolutionary game income matrix; s3, constructing a day-ahead two-stage pricing optimization model; and S4, constructing a real-time rolling optimization model. According to the method, the interactive behavior with the electric vehicle is simulated through the replication dynamic equation of the evolutionary game, and the economic benefits of the electricity selling company and the electric vehicle user under different cooperation strategies can be effectively predicted and adjusted; powerful theoretical support is provided for establishing a stable cooperation relationship through an evolutionary game income matrix; through constructing a day-ahead two-stage pricing optimization model and a real-time rolling optimization model, an effective price and load management strategy is provided for an electricity selling company in a complex and changeable spot market.

The application belongs to the technical field of computers and discloses a parallel segmentation learning method and device based on clusters, equipment and a storage medium. The method comprises the following steps: obtaining a plurality of to-be-learned user terminals and user communication information of each to-be-learned user terminal; performing cluster division on each to-be-learned user terminal according to each to-be-learned user terminal and the user communication information of each to-be-learned user terminal, and determining a plurality of target user clusters; and performing cluster serial segmentation learning according to an aggregated user terminal model and each target user cluster to obtain a target user terminal model, wherein the aggregated user terminal model is obtained by performing parallel segmentation learning on each target user cluster according to the target spectrum resources of each to-be-learned user terminal. Through the above method, the overall training time delay in the segmentation learning process is effectively reduced, the efficiency of segmentation learning is improved, the negative effects caused by network heterogeneity and dynamics are inhibited, and the convergence and accuracy of existing segmentation learning technology are ensured.

This invention discloses a method and system for adaptive allocation of preheating power for a target drone turbojet engine based on dynamic compensation of environmental parameters. It relates to the field of aero-engine control and thermodynamic process optimization technology, and includes: collecting flight environmental parameters and calculating the mass flow rate of air at the combustion chamber inlet; constructing a multi-condition compensation matrix to determine aerodynamic reference quantities; dividing the combustion chamber into regions based on the reference quantities; establishing a thermodynamic unsteady model and generating a combustion instability sensitivity matrix; calculating the dynamic instability index and performing thermoacoustic risk classification; constructing a preheating energy transfer matrix based on the risk level; introducing an inertia factor to achieve adaptive allocation of preheating power in each region; converting the allocation strategy into injector pulse modulation control commands; and performing dynamic iterative optimization by combining a spatiotemporal evolution model of thermoacoustic parameters and wall temperature. This invention can achieve refined allocation of preheating power, effectively suppress thermoacoustic instability, and improve engine preheating uniformity and operational reliability.

This application relates to an adaptive rapid positioning method and system based on GNSS synchronization status, belonging to the field of navigation and positioning technology. The method includes: entering the normal positioning process when all tracked satellite signals have completed message time synchronization; otherwise, entering the adaptive time-free positioning process, which includes first dynamically determining the minimum reliable time unit and time search interval based on the synchronization status level of the tracked satellite signals; then determining a relatively accurate local reference time as the local baseline time based on the synchronization status level and time search interval; and after correcting the signal transmission time of the tracked satellites based on the local baseline time and the minimum reliable time unit, performing rapid AGPS iterative calculation. This method enables rapid receiver positioning under a wide range of time errors.

The invention provides a mechanical arm trajectory tracking control method fusing model prediction and sliding mode control. The mechanical arm trajectory tracking control method comprises the steps that a discrete sliding mode controller is constructed, and a discrete sliding mode surface based on trajectory tracking errors is designed; calculating a sliding mode control law; constructing a dynamics prediction model based on an echo state network, and performing online learning and updating by taking historical state data and a sliding mode control law of the mechanical arm as input so as to predict a state track of the mechanical arm in a future time domain; constructing a model prediction controller embedded with sliding mode control, and introducing a prediction state and a sliding mode control law into an optimization objective function of the model prediction controller; and under the model prediction controller, an optimization problem is converted into a quadratic programming problem to be solved, a smooth optimization control quantity meeting physical constraints is obtained, and the control quantity acts on the mechanical arm system. The high-frequency buffeting problem of sliding mode control is effectively solved, and meanwhile high-precision trajectory tracking of the mechanical arm under strong nonlinear interference is guaranteed.

This invention discloses a learning-based fault-tolerant control method and system for quantifiable performance of nonlinear systems, belonging to the field of automatic control technology. The method designs a learning-based fault-tolerant control algorithm that embeds an RBF network to compensate for fault dynamics when a fault occurs. Based on deterministic learning theory, an RBF network is constructed by setting neurons along a desired reference trajectory. Secondly, by analyzing the convergence characteristics of the learning-tolerant control system through sampled data, a detailed quantitative analysis of the learning accuracy and learning speed of the proposed learning-based fault-tolerant control method is performed, and calculation formulas for learning accuracy and learning speed are given. This invention establishes quantitative expressions for learning speed and learning accuracy through theoretical derivation, realizing quantitative analysis and pre-setting of fault-tolerant control performance, and achieving quantitative controllability of learning speed and approximation accuracy while ensuring the closed-loop stability of the system.

The invention discloses a battery equalization control system and a control method, which are used for controlling an active equalization circuit and a passive equalization circuit to equalize at least two battery packs in the battery equalization system. The battery equalization control method comprises the following steps: acquiring single voltage and / or single residual electric quantity of each single battery in at least two battery packs; selecting to execute an active equalization algorithm or a passive equalization algorithm according to the difference value between the maximum single residual electric quantity and the minimum single residual electric quantity of the single batteries in each battery pack; and when the difference value corresponding to any battery pack is greater than a first threshold value, the active equalization algorithm is selected to be executed, and when the difference values corresponding to all the battery packs are smaller than or equal to the first threshold value, the passive equalization algorithm is selected to be executed, so that the problem of system oscillation caused by relatively large active equalization current can be reduced.

The application discloses an underwater image enhancement method and system based on dynamic multi-channel compensation. The method comprises the following steps: acquiring an underwater image to be processed, limiting the exclusive stretching intensity of a red channel, performing linear dynamic range stretching on a red-green-blue three-channel based on high and low quantile points, performing defogging based on inverse operation of an atmospheric scattering model, performing color correction on a blue-biased image and a green-biased image by using a green-channel-dominant differential color compensation strategy through an iterative loop, performing mean centering fine adjustment on a and b channels in a Lab space, performing strengthening processing on a brightness channel in an HSV space, extracting feature details of different levels and performing edge detection, fusing the details and edge information with the original brightness channel, and obtaining an enhanced underwater image. The application effectively recovers the lost information of the red channel by adaptively processing the blue-biased or green-biased image, significantly improves the visual effect of the underwater image, and enhances the detail presentation, color accuracy and overall clarity of the image.

The application provides a sparse communication method and system based on federal deep learning, which comprises the following steps: acquiring the aggregated gradient transmitted by the server participating in the federal learning and counting the transmission round; if the transmission round reaches the fusion round, training the preset client training model based on the aggregated gradient and using the preset client data set to obtain the global gradient of the client training model; uploading the global gradient to the server to participate in the gradient aggregation of the server; if the transmission round does not reach the fusion round, training the client training model based on the aggregated gradient and using the client data set to obtain the local gradient of the client training model; sparsifying the local gradient by using a sparse algorithm to obtain the local sparse gradient; and uploading the local sparse gradient to the server to participate in the gradient aggregation of the server. The application adopts the mode of local updating combined with asynchronous fusion, which can stabilize the model updating direction and ensure the convergence of the model.

This invention provides a method and apparatus for intelligent unmanned vehicle (UAV) platooning reconfiguration control based on an adaptive potential field, relating to the field of intelligent UAV platooning control technology. The method includes: acquiring environmental information and the state information of the intelligent UAV platoon; generating a critical adaptive potential field threshold through an adaptive potential field and calculating the desired position of each intelligent UAV in the platoon; generating continuous and bounded desired trajectory information using a fixed-time nonlinear filter; and using an anti-saturation fixed-time sliding mode controller to control each intelligent UAV to track the desired trajectory information within a fixed time period under actuator saturation, thereby achieving intelligent UAV platooning reconfiguration and fixed-time convergence control. This invention is applicable to scenarios such as intelligent transportation, environmental detection, and rescue operations, and can realize real-time platooning reconfiguration, obstacle avoidance, and fixed-time convergence control of intelligent UAV platoons in dynamic environments.

This invention belongs to the field of communication technology and discloses a federated learning method based on partial client participation and power constraints. The method includes: employing biased client selection, setting client addressability and continuously reporting to a central server, allowing the central server to understand the state of each client during iteration; and allocating power across communication time slots by adding a total power constraint, enabling devices to transmit their local updates to the server. This invention addresses the problems of existing technologies by reducing communication costs while maintaining model convergence performance. The proposed design can mitigate the negative impacts caused by imperfect channel conditions. Convergence of the learning process is slowed by power control, while biased client selection accelerates convergence, balancing model aggregation. This invention can guarantee fast convergence under harsh wireless conditions with low signal-to-noise ratios, and is cost-effective in terms of time and energy savings in wireless FL systems.

This invention discloses an adaptive control method for vehicle queuing based on specified performance and fixed-time sliding mode, relating to the field of distributed vehicle queuing control technology. The method includes: establishing a vehicle dynamics model; designing a piecewise continuous function based on a constant spacing strategy to constrain follower tracking errors; designing an error transformation function with specified time performance constraints to transform the constrained problem into an unconstrained problem and constructing a fixed-time sliding surface; constructing a controller by combining adaptive parameter estimation to compensate for unknown nonlinear terms and disturbances, estimating parameters, and achieving vehicle queuing stability within a fixed time; and integrating specified performance and fixed-time sliding mode control to achieve fixed-time convergence and transient and steady-state error performance, thereby improving the reliability of queuing control.

This invention relates to the field of robot skill learning technology, specifically to an efficient robot skill learning method, system, device, program product, and medium. The method includes: obtaining the robot's task policy equation in Cartesian space; obtaining the expected trajectory at each time step of the robot's task policy; constructing the cost function of the robot's task policy; and obtaining the robot's final expected trajectory. This invention utilizes interactive reinforcement learning to integrate human and machine intelligence, improving algorithm efficiency and overcoming the limitations of existing reinforcement learning methods in robot operation applications. It enables real-world robot skill learning, allowing robots to be applied in a wider range of scenarios.

The application relates to a UAV pilot signal identification method, device and computer equipment. The method comprises the following steps: obtaining a training sample, constructing a pilot signal identification model, performing multi-scale feature extraction on an input time-frequency spectrum through a feature extraction unit, dividing an initial feature map into non-overlapping local frequency blocks according to continuous frequency segments, independently completing channel dimension adaptive weight distribution in the blocks, and outputting an enhanced multi-scale feature map; performing bidirectional cross-scale feature interaction and local dynamic weighted fusion through a feature fusion unit, outputting a multi-scale fusion feature map, completing boundary box regression and classification prediction through a detection output unit, and outputting an identification result; using a training sample, training the pilot signal identification model by using a loss function, inputting a to-be-identified signal time-frequency spectrum into the trained model, and obtaining a UAV pilot signal identification result. The method can balance model lightweight and identification accuracy and robustness in a complex electromagnetic environment.

The application relates to a dynamic population multi-objective particle swarm optimization method based on crowding degree and contribution degree, and relates to the technical field of intelligent optimization algorithms, in particular to a dynamic population multi-objective particle swarm optimization method based on crowding degree and contribution degree. The method aims to solve the problem that the calculation efficiency and solution set quality of a traditional multi-objective particle swarm optimization (MOPSO) algorithm cannot be considered simultaneously when solving complex engineering problems. The method establishes grid division to realize solution space discretization, combines non-dominated sorting and Euclidean distance to calculate particle contribution degree, and defines a crowding degree index based on the total number of particles in the grid. Then, a particle deletion probability formula is constructed based on the contribution degree and the crowding degree, a dynamic deletion strategy and a quantity limitation mechanism are introduced, and low-value and high-redundancy particles in the population are adaptively removed according to the deletion probability, so that the population size is reduced and the calculation efficiency of the algorithm is improved under the premise of ensuring the diversity and convergence of the population. The method provided by the application has obvious advantages in efficiently solving multi-objective optimization problems in combination with non-dominated sorting, grid division and a dynamic deletion strategy.

This invention relates to a formation control method for a heterogeneous multi-agent system based on a time-defined observer. The method includes: establishing a dynamic model for each agent in the heterogeneous multi-agent system, treating each agent as a communication node, and constructing a communication topology; collecting state information from neighboring nodes to construct an error system to estimate the leader's state matrix and output matrix; setting an adaptive observer that estimates the convex hull of the leader's state based on the real-time state information of neighboring nodes, combined with the leader's state matrix and output matrix; constructing a time-varying formation tracking control protocol based on the output of the adaptive observer and a time scaling function; initializing the time-varying formation tracking control protocol according to the expected formation structure of the multi-agent system, and updating the state of each agent according to the time-varying formation tracking control protocol through a consensus controller, so that all agents reach a preset formation; this invention can achieve the desired formation shape within a specified time without introducing global information.

The invention relates to the technical field of optical element testing, and discloses an optical fiber connector butt joint position deviation detection method and system.The method comprises the steps that firstly, physical butt joint is carried out on a standard reference jumper fiber, and a static reference insertion loss value and a transient Fresnel reflection jitter waveform of a full butt joint period are collected; extracting an end face structure scattering entropy and constructing a multi-dimensional optical transmission feature reference library; then collecting real-time transient waveforms of field docking and calculating real-time scattering entropy, completing decoupling branch judgment of end face physical defects and mechanical geometric position deviation through feature matching, and generating a to-be-corrected optical link data packet for the position deviation; and finally, obtaining a connection point characteristic signal through iterative deconvolution signal reconstruction, analyzing the three-dimensional position deviation, and generating a correction instruction to execute closed-loop verification. According to the method, end face damage caused by misjudgment of end face defects is effectively avoided, and high-precision detection and self-adaptive correction of butt joint position deviation are achieved.

This invention discloses a convex optimization method for an integrated radiation pattern of a phased array-radome, relating to the field of radiation pattern technology. The method comprises the following steps: S1, obtaining the overall short-circuit active radiation pattern of the array based on a fast electromagnetic algorithm; S2, constructing upper and lower mask functions for the sum and difference radiation patterns, and synthesizing the overall array radiation pattern based on the short-circuit active radiation pattern; S3, constructing a multi-constraint radiation pattern optimization model that minimizes the excitation amplitude range while satisfying the upper and lower mask constraints of the sum and difference radiation patterns; S4, solving the multi-constraint radiation pattern optimization model using the alternating direction multiplier method, achieving simultaneous synthesis of the sum and difference radiation patterns sharing a feed network. This invention achieves accurate synchronous synthesis of the sum and difference radiation patterns under multiple constraints, significantly improving data acquisition efficiency and radiation pattern control accuracy, while also possessing wide applicability and engineering feasibility, and reducing system design and manufacturing costs.

This invention discloses a composite disturbance rejection control method and system for permanent magnet synchronous motors (PMSMs) based on recurrent neural network disturbance estimation and prediction. It constructs a dynamic model of the PMSM considering parameter uncertainties, slowly varying load disturbances, and impulsive load disturbances. A recurrent neural network with a self-feedback structure for historical disturbance estimation is designed. This network learns the dynamic characteristics of disturbances online through a recursive structure and estimates the current disturbance in real time. Simultaneously, it predicts the disturbance at the next moment while keeping the network weights constant, thus compensating for the computation and execution delays of the control system. A fixed-time non-smooth robust controller is designed, using the residual error after prediction compensation as input to rapidly suppress sudden impact disturbances. The predicted disturbance is introduced into the controller as feedforward compensation, constructing a predictive-feedback cooperative composite control law to achieve hierarchical suppression of multi-source disturbances. This invention improves the dynamic response performance, disturbance rejection capability, and control stability of PMSMs in complex operating environments.

The invention discloses a defect detection method and device based on multi-source data. The method comprises the following steps: acquiring data to be subjected to defect detection and source field sample data with defect labels; based on a pre-constructed mapping model, performing alignment mapping on the to-be-defect detection data and the source field sample data to obtain defect detection features and source field sample features; performing distance analysis on each defect detection feature and all source field sample features to determine a first pseudo tag; based on a pre-constructed label classification model, performing label updating on the first pseudo label and the source field sample data with the defect label to obtain a second pseudo label; and updating the mapping model and the label classification model based on the first pseudo-label and the second pseudo-label, repeating alignment mapping and label updating until a predetermined condition is met, and giving a defect label corresponding to the data to be subjected to defect detection. Through high-robustness defect detection models of two mechanisms, the precision of generating defect labels is improved.

The application discloses a kind of satellite-borne image compression method and system based on generative adversarial network, wherein the method comprises: obtaining compressed code stream and binary mask chart;Get logic gate control signal;Using neural network to train training data set obtains network weight parameter;The JPEG-LS lossy compressed code stream of real-time transmission on satellite is decompressed to obtain preliminary recovery image and satellite binary mask chart;Primary recovery image and satellite binary mask chart are input into neural network, and network weight parameter is used to repair the pixel value of the pixel value of satellite run-length encoding area to obtain intermediate recovery image data;Residual chart data is obtained;According to residual chart data and maximum allowable error value, satellite run-length encoding area optimization image data is obtained;According to satellite run-length encoding area optimization image data and preliminary recovery image, the final recovered image data is obtained.The application improves the quality of reconstructed image.

The invention discloses an RIS phase shift optimization method in an RIS-assisted cellular-free large-scale MIMO and ISAC system, and the method comprises the steps: building an optimization problem about an RIS phase shift matrix by taking a multi-user weighted communication rate and maximization as optimization targets, and taking a unit modulus value constraint of RIS phase shift and a condition that a sensing signal to interference plus noise ratio is not lower than a preset threshold value as constraints; a first auxiliary variable and a second auxiliary variable are constructed for the two constraints respectively, and the optimization problem is decoupled into three sub-optimization problems associated with one another; and the three sub-optimization problems are solved alternately and iteratively, the solutions of the RIS phase shift matrix, the first auxiliary variable and the second auxiliary variable tend to be consistent, and finally the optimized RIS phase shift configuration meeting the constraint is obtained. Compared with the prior art, the method has the advantages that auxiliary variable decoupling and alternate iterative optimization are carried out on constraints, so that the multi-user weighted communication rate is effectively improved while the sensing performance is ensured.

This invention discloses a seamless switching positioning method integrating RTK and PPP-RTK algorithms. This method achieves seamless switching between RTK and PPP-RTK algorithms by designing a dual-state model within a Kalman filtering framework. When network connectivity is available, the system automatically uses NRTK corrections to achieve high-precision RTK positioning; when the network is interrupted, the system seamlessly switches to satellite-based PPP-RTK corrections for PPP-RTK positioning without requiring a software restart. This invention, through the design of a state transition matrix and an adaptive noise adjustment mechanism, ensures positioning continuity and accuracy stability during the switching process, solving the problem of traditional methods requiring re-initialization during switching, which leads to positioning interruptions. This significantly improves the robustness and availability of GNSS positioning systems in complex environments.