381 results about "Calibration Error" patented technology

The invention discloses a tunneling equipment position calibration method and system adopting a cursor frame, and relates to the technical field of coal mine tunnel construction tagging. Geologic feature data in a coal mine tunnel are collected in real time through a dynamic environment data set, an environment adaptability index Szp is calculated, and the influence of the environment on equipment calibration is evaluated; a calibration error evaluation coefficient Gz and an error growth trend value Wczs are combined, a calibration strategy is dynamically optimized, adaptive filtering and a multi-modal fusion algorithm are adopted to improve positioning precision, and stable work in an environment with large geological change is ensured; meanwhile, calibration parameters are adjusted in real time by using a deep learning model and a self-adaptive path calibration algorithm, and the adaptability to complex geological conditions is enhanced; the interference compensation module processes electromagnetic interference through Kalman filtering, particle filtering and other algorithms, optimizes the error coefficient Ins of the inertial navigation system, corrects the deviation of the sensor, reduces the influence of external interference on the calibration precision, and guarantees the high precision and stability of the system.

The invention relates to the technical field of sensor calibration, and particularly discloses a high-precision automatic calibration system and method for a multispectral sensor. Comprising a multispectral signal acquisition module, a sensor response optimization module, a spectral feature matching module, a calibration parameter optimization module, an automatic calibration execution module, a calibration error evaluation module and a multi-node cooperative calibration module. The method comprises the following steps: acquiring an original spectral signal, and performing dynamic compensation to generate an optimized signal; extracting spectral feature points and matching the spectral feature points with a standard library to quantify feature deviation; fusing the environment data and the deviation value, and generating an optimal calibration parameter through optimization; real-time calibration is implemented based on parameters, and a cloud report is automatically generated; and constructing a calibration error evaluation model, evaluating the calibration error of the multispectral sensor, and dynamically adjusting a calibration strategy through cooperation of the distributed nodes and the central control unit. According to the invention, the automatic calibration process of the sensor is realized, and the calibration precision of the multispectral sensor in a complex environment is improved.

The invention relates to wide-range multi-component sensor calibration equipment and a wide-range multi-component sensor calibration method. The method comprises the following steps: firstly, checking equipment parts and setting a target temperature range and a compensation model of each temperature interval; then, dynamic temperature control is started, jig deformation is monitored in real time, a transient temperature gradient suppression technology is applied, multi-physical-quantity calibration is executed, and a compensation measurement value is output; then setting multi-environment coupling calibration test conditions to simulate actual working conditions, and monitoring and recording calibration data of each physical quantity; and finally, calculating a calibration error by combining a calibration result and a reference standard value. According to the process, by precisely setting parameters and combining dynamic control, real-time monitoring and multi-physical-quantity calibration, the transient temperature gradient influence is effectively inhibited, and the measurement precision of the sensor in a complex environment is ensured; the temperature control performance is improved through the micro soaking film and the intelligent temperature control strategy, the actual working condition is simulated through multi-environment coupling testing, the accuracy and reliability of the calibration result are remarkably improved through compensation and error calculation, and guarantee is provided for precise application of the sensor.

The invention particularly relates to a binocular vision-based three-dimensional reconstruction method, which comprises the following steps of: calibrating a binocular camera based on an improved Zhang Zhengyou calibration method to obtain internal and external parameters and a distortion coefficient of the camera; performing stereo correction on the image by using the internal and external parameters of the camera and the distortion coefficient obtained by calibration, so that the binocular image meets an epipolar constraint condition; a multi-strategy optimized semi-global stereo matching algorithm is adopted to process the image after stereo correction, and a disparity map is generated; based on the generated disparity map, generating a three-dimensional point cloud through a triangulation principle; carrying out anti-interference processing and registration optimization on the three-dimensional point cloud; and performing global splicing on the three-dimensional point clouds subjected to anti-interference processing and registration optimization based on a sequential registration error sharing strategy to complete three-dimensional reconstruction. According to the method, the key problems of large calibration error, weak texture matching failure, point cloud noise sensitivity and registration accumulative error in a traditional method are solved, and the reconstruction precision and stability are remarkably improved.

The invention discloses a three-dimensional shielding target tracking method based on multi-modal space-time interaction, and relates to the technical field of target tracking. The method comprises the following steps: acquiring a point cloud and an image and preprocessing to obtain global fusion features; obtaining an initial detection frame and region-of-interest features through region proposal network processing; projecting the non-empty voxel point cloud to the image features, and reconstructing shielded target features; convolution and neural network processing are utilized to obtain a refined detection frame; screening legal detection frames through distance calculation and legality judgment; the bipartite graph and the self-adaptive channel graph are adopted for convolution, and appearance correlation scores are calculated; and matching the detection frame and the trajectory based on a Hungary algorithm to realize whole-course tracking. The target identification accuracy and robustness are improved, the shielding problem is solved, the accuracy of the detection frame is ensured, the correlation accuracy is improved by using the bipartite graph and the adaptive convolution, the nodes are matched in combination with the geometric cost matrix, and whole-course tracking and error calibration are realized.

The invention relates to the technical field of electrical measurement, and discloses an electrical measurement instrument dynamic calibration error compensation method, which comprises the following steps: S1, multi-source data real-time acquisition: integrating a temperature sensor, a humidity sensor and an electromagnetic interference detection module in a measurement instrument, configuring voltage, current, frequency and other measurement channels, and carrying out multi-source data real-time acquisition at the frequency of not less than 100HZ; environmental parameters (temperature T, humidity H and electromagnetic interference intensity E) and measurement data (model characteristic parameters X1, X2,..., Xn such as voltage, current and frequency) are synchronously acquired. According to the electrical measuring instrument dynamic calibration error compensation method, through multi-source data real-time acquisition and composite error model construction, in combination with multiple linear regression and an improved BP neural network, environment and measurement signal changes are comprehensively captured, an error compensation value is accurately calculated, a measurement error can be effectively controlled within an extremely small range, and the error compensation accuracy is improved. Compared with a traditional method, the measurement precision is greatly improved.

The invention discloses a monocular and binocular cooperative localization and mapping method and device for underwater refraction compensation, and the method comprises the steps: extracting checkerboard angular points, compensating an underwater light path through employing a two-time vector correction model based on a Snell law, and precisely calibrating the internal and external parameters of a binocular camera. According to the method, an underwater real light path is effectively restored, and calibration errors caused by medium refraction are remarkably reduced. Then epipolar correction is performed on the corrected left and right images, so that the geometric consistency of stereo matching is ensured; and finally, monocular and binocular poses are fused, and seven-degree-of-freedom similarity transformation global alignment is applied, so that the problem of monocular scale uncertainty is effectively solved, and high-precision and high-robustness three-dimensional mapping is realized by means of binocular real scale introduction. Key links such as camera pre-calibration, refraction compensation, image correction, monocular tracking, binocular depth measurement, scale alignment and the like are connected in series to form an integrated process, so that the underwater vision positioning and mapping precision is improved.

The invention provides an intelligent electric meter calibration method and device, computer equipment and a storage medium, and the method comprises the steps: constructing a single meter-transformer area double-layer data system through a special transformer collection terminal, and collecting transformer area total intelligent electric meter data, user meter temperature and humidity, an electric energy observation value, transformer area line impedance and load unbalance degree data; constructing an energy conservation optimization function according to the user meter electric energy observation value and the total meter electric energy by combining the deviation with the observation value; determining line loss data according to total meter current and user temperature and humidity, determining power supply stability through total meter voltage, and determining optimization function constraint conditions according to corresponding data; and solving an optimal user table software compensation coefficient and a transformer area line loss correction coefficient by a generalized genetic algorithm, updating a humidity-error correlation coefficient and a line loss reference value by a feedback performance attenuation model to determine an adjusted single table calibration error range, and performing single table calibration by combining the user table temperature and humidity and an electric energy observation value. By adopting the scheme, the accuracy, the reliability and the environmental adaptability of the calibration of the intelligent electric meter are improved.

The invention relates to the field of microscopic imaging, and discloses a miniature microscopic system optical precision improvement and image calibration error optimization method, which comprises the following steps: establishing an optical distortion model and a mechanical disturbance model, and respectively describing a static error and a dynamic error; constructing a joint dynamic error model, and uniformly expressing a coupling relationship between optical distortion and mechanical disturbance; decoupling and separating the joint error through a space-time separation method to obtain a static error component and a dynamic error component; performing iterative optimization on the optical distortion parameter and the mechanical disturbance parameter based on an optimization algorithm to minimize a system calibration error; a real-time sensor is utilized to obtain dynamic disturbance data, and real-time dynamic compensation is performed on an image calibration error in combination with an optimization result, so that a high-precision calibration result is obtained. The method can effectively improve the calibration precision of the microscopic system in a dynamic environment, remarkably improves the imaging quality of the system, and has a wide application prospect.

The invention provides a chip test calibration device and method, and relates to the technical field of chip test. The chip test calibration device comprises a base, a cooling air supply device, a control device, a chip and two groups of manipulators. According to the invention, the chip is heated by setting different temperature environments in stages, and the multi-point synchronous temperature measurement of each temperature stage is combined, so that the heat distribution on the surface of the chip can be comprehensively captured, and the calibration error caused by non-uniform temperature or a detection blind area is avoided; the chip is automatically circulated among the heating station, the detection station and the cooling station, synchronous processing and continuous detection of multiple chips are achieved, the single-batch detection period is greatly shortened, the surface of the chip is directly blown by directional airflow, an efficient air return channel is matched, the temperature of the chip can be rapidly reduced after detection is completed, idle waiting for station switching is reduced, and the overall production efficiency is improved.

The invention discloses an MEMS and MHD combined gyroscope low-high frequency composite calibration method, which comprises the following steps: synchronously acquiring vibration response data of a combined gyroscope sensor, separating low-high frequency signals, calculating low-frequency static error parameters and high-frequency dynamic error parameters, constructing a static orthogonal matrix based on a mounting angle error, and calibrating the low-high frequency composite of the MEMS and MHD combined gyroscope. The method comprises the following steps: generating a normalized error transfer coefficient by utilizing a high-frequency dynamic error parameter cross response ratio and a rejection ratio, generating a dynamic error matrix according to the normalized error transfer coefficient, establishing an expression containing error output of a sensor by utilizing the dynamic error matrix, and inversing the dynamic error matrix to obtain a dynamic error transfer matrix, and the static orthogonal matrix and the dynamic error transfer matrix are combined to construct a composite error compensation model, error parameters after dynamic compensation are evaluated, a residual feedback signal is generated in combination with an error evaluation result, calibration error parameters are iteratively corrected until convergence, and the calibration parameters are output.

The invention provides a calibration method and device of motion capture equipment, equipment and a medium, the motion capture equipment is worn on a limb of a human body, the method is applied to head-mounted display equipment, the head-mounted display equipment comprises a camera, and the method comprises the following steps: collecting an image of the motion capture equipment through the camera; according to the motion capture equipment image, determining target attitude information of an equipment coordinate system relative to a world coordinate system; and calibrating the motion capture equipment according to the target attitude information. The motion capture equipment is calibrated by using the image of the motion capture equipment, so that the calibration difficulty of the motion capture equipment can be effectively reduced, the calibration time can be shortened, the calibration error can be reduced, and the calibration precision of the motion capture equipment can be improved.

The invention relates to the field of trajectory tracking, and discloses a gun trajectory tracking system and method fused with multi-source localization, the system comprises a satellite positioning module, an RFID node network, an inertial measurement unit (IMU) and a processing unit, the processing unit determines an open area, a semi-shielded area or a full-shielded area in real time through an environment classification sub-module, based on a dynamic weight distribution strategy, the fusion proportion of the satellite data, the RFID data and the I MU data is adjusted; synthesizing multi-source positioning data by adopting a weighted fusion algorithm, and outputting a real-time gun trajectory; and when the RFID node is located in a full-shielding area, the topological optimization calibration sub-module is combined with the RFID node topological relation to correct the accumulated error of the I MU, and weight distribution is dynamically fed back and adjusted based on the calibration error. According to the method, the problems of track interruption and drifting caused by failure of a single positioning source or inertial navigation accumulative errors in a complex shielding scene are solved, and high-robustness continuous tracking of gun tracks in multiple scenes such as cities, fields and indoors is realized.

Disclosed is a method and device for determining uncertainty measure of a classification model. The method includes: obtaining a sample data set for evaluating the uncertainty of a classification model, each sample data including an original evaluation sample and a corresponding classification label; inputting the original evaluation sample into the classification model to obtain a first predicted classification result, which includes a first predicted category and a first predicted probability; calculating the first prediction loss of the classification model according to the difference between the first predicted probability and the classification label to obtain a first uncertainty estimate; calculating the first calibration error of the classification model according to the difference between the first predicted category and the classification label and the first predicted probability to obtain a second uncertainty estimate; determining the uncertainty measure of the classification model based on at least the first uncertainty estimate and the second uncertainty estimate. The classification model uncertainty measure determination method according to the present application can achieve applicability in a wide range of application scenarios.

The invention discloses a target positioning method and system based on machine vision, and relates to the technical field of machine vision and target positioning, and the method comprises the steps: carrying out the model selection of machine vision key hardware, and collecting image information; carrying out image preprocessing, and automatically detecting key points and feature points of the image by using an SIFT algorithm; establishing a coordinate system to generate an internal reference matrix, establishing a camera coordinate system to generate an external reference matrix, and performing coordinate transformation by combining the internal reference matrix and the external reference matrix; through a camera calibration method, target positioning calibration is carried out. The SIFT algorithm adopted by the method has very strong robustness for scale change, rotation, illumination change and a certain degree of view angle change of the image, and is suitable for a complex environment in an industrial scene. In the calibration process, a chessboard calibration plate and a least square method are utilized to improve the calculation precision of internal and external parameters, the calibration error is effectively reduced, and the positioning precision is improved.

The invention discloses an evidence deep learning method for double-layer dynamic uncertainty calibration based on a meta-strategy. A double-layer optimization architecture is adopted. An inner layer optimizes an evidence deep learning model to execute a pixel-level segmentation task and estimate uncertainty; and the outer layer optimizes a state-aware meta-policy network. The meta-policy network receives state information reflecting training dynamics in real time and generates key hyper-parameters used for configuring an inner-layer model loss function according to the dynamic state; and a multi-target reward signal is formed through performance in multiple aspects of prediction accuracy, calibration error, misclassification uncertainty and the like of a periodic evaluation model on a verification set. According to the method, hyperparameter dynamic adaptive adjustment is carried out by introducing a state-aware meta-strategy, the limitation that a traditional method depends on static setting is overcome, prediction precision and uncertainty calibration can be better balanced, and the reliability and generalization ability of a deep learning model in high-risk application scenes (such as medical image analysis) are remarkably improved.

The invention relates to the technical field of drainage plate positioning, and discloses a pile hole positioning method and system for a drainage plate pile pipe based on a multi-source sensor. Synchronous data acquisition of a multi-source heterogeneous sensor is achieved through a GNSS receiver and an IMU, a data basis is provided for subsequent fusion calculation, and then pile pipe root coordinates and pile pipe three-dimensional attitude angles are dynamically calculated through Kalman filtering fusion according to pile driver top coordinates and pile pipe motion data. Root posture resolving errors caused by static calibration errors and structural deformation in motion are avoided; then, the coordinates of the tip end of the pile pipe are dynamically calculated through coordinate rotation transformation, machine body motion data are introduced to compensate deviation calculation, and therefore the problem that in the prior art, due to the fact that the three-dimensional posture change of the pile pipe in the construction process is neglected, drainage plate pile hole positioning is not accurate enough is solved; and it is ensured that each pile hole can meet the design precision requirement, and the problem that the construction perpendicularity is uncontrollable is fundamentally solved.

The invention provides an error modeling method and system based on a CCD camera alignment assembly system, and relates to the technical field of high-precision electronic manufacturing assembly, and the method comprises the steps: obtaining internal parameters and external parameters of a CCD camera through calibration, building a camera and mechanical coordinate system conversion relation, and constructing a collaborative calibration error tree; controlling the motion platform to move, synchronously acquiring pose and environment temperature data, decomposing vibration and thermal deformation error components through time-frequency analysis, and superposing the vibration and thermal deformation error components into dynamic errors; extracting a positioning error based on the pose data, and constructing a mechanical-optical coupling error spectrum in combination with the distortion parameters; and finally, fusing the collaborative calibration error tree, the dynamic error and the coupling error atlas to generate a global error atlas, and completing alignment assembly system error modeling based on the CCD camera. According to the invention, the precision and applicability of error modeling are improved.

The invention relates to the technical field of data error correction, in particular to a calibration error correction method and system for a falling weight deflectometer, and the method comprises the steps: obtaining deflection values of a displacement sensor of the falling weight deflectometer to form a deflection program sequence, calculating deflection abnormal values of the deflection values, obtaining deflection rightward sequences of the deflection values, and calculating deflection error sequences of the deflection values; the method comprises the following steps: acquiring a deflection value of each deflection value, further acquiring a deflection continuation characteristic value of each deflection value, acquiring a load transfer weight of each deflection value through a correlation degree between local deflection sequences of deflection values of the same bit sequence in a program sequence during adjacent deflection, and acquiring a road looseness influence degree of each deflection value in combination with the deflection continuation characteristic value corresponding to each deflection value. Further acquiring deflection vibration characteristic values of the deflection values and smooth window scales of the deflection values, and smoothing the program columns during deflection so as to correct deflection data errors of the drop weight deflectometer. According to the application, the calibration error correction precision of the drop weight deflectometer can be improved.

The invention discloses a thunder-vision fusion multi-sensor collaborative sensing system along a railway and an installation and calibration method, and belongs to the technical field of intelligent traffic and environment sensing. Aiming at the problems of low target detection precision, poor multi-source data fusion timeliness, dependence on manpower on sensor installation and calibration and the like in a complex scene along a railway, the system provides an innovative architecture of'layered sensing-dynamic fusion-autonomous calibration '. The multi-modal sensor array of a laser radar, a millimeter-wave radar, a visible light camera and a thermal infrared imager is deployed; autonomous optimization of sensor pose parameters is realized by using track geometric constraint and a deep learning model; an edge computing node and a cloud collaboration platform are integrated, and real-time target tracking, intrusion early warning and equipment state diagnosis are supported. Experiments show that the target detection accuracy rate of the system is greater than or equal to 98.5%, the false alarm rate is less than or equal to 0.3% and the self-calibration error of sensor installation parameters is less than 0.05 degree in the scenes of rain and fog, night, high-speed movement and the like, which are improved by more than 40% compared with the traditional scheme.

The invention belongs to the technical field of automatic welding, and particularly relates to a welding seam detection method based on surface structured light. Comprising the following steps: (1) three-dimensional reconstruction of a workpiece: performing three-dimensional scanning on the workpiece by using a surface structured light three-dimensional scanner to obtain a coded image II, a phase diagram and three-dimensional data of the workpiece, and reading the coded image II, the phase diagram and the three-dimensional data through a computer; and (2) identifying welding seam feature points: performing welding seam feature identification or manual welding seam marking on the coded image II to obtain a two-dimensional coordinate set of a welding seam on the coded image II. Two-dimensional information and three-dimensional information of a surface structured light scanner are utilized at the same time, an efficient welding seam detection method integrating two-dimensional feature recognition and three-dimensional coordinate mapping is achieved through triangulation mapping, the problem of welding seam detection without obvious structural features is solved, meanwhile, compared with a split type multi-sensor fusion method, the welding seam detection method has the advantages that the welding seam detection efficiency is improved, and the welding seam detection cost is reduced. Calibration errors between devices are eliminated, the welding seam detection precision is improved, and the detection process is simplified.

The invention discloses a multi-galvanometer cooperative laser processing splicing calibration method, and relates to the technical field of additive manufacturing, and the method comprises the steps: dividing a to-be-processed image into a plurality of sub-unit images through an image segmentation processing technology, then completing the splicing processing, and carrying out the calibration through a nominal parameter calibration matrix of each galvanometer, the overall processing breadth requirement can be met, meanwhile, the optimal processing area of each galvanometer and overall rotation and translation calibration of a single galvanometer are considered, and then a global error threshold value obtained through conversion of a processing result of an evaluation breadth based on a virtual large field lens method is used for processing quality evaluation so as to adjust a splicing calibration method. According to the technical scheme, a plurality of galvanometers can use a calibration strategy of a single equivalent large field lens, and then the splicing problem and the calibration correction problem between the adjacent galvanometers are considered one by one, so that the dependence of the splicing precision of the whole system on single calibration is reduced, the calibration error in the whole processing breadth can be eliminated, and the processing effect is improved.

The invention provides a camera calibration error analysis method and system based on Kinect V2. The method comprises the following steps: calculating internal parameters and external parameters of a camera; applying a calibration result to a new image, calculating a distance between a re-projection point and an actual observation point, evaluating the calibration result, carrying out statistical analysis on re-projection errors of all images, and calculating statistical indexes such as an average error and a maximum error; the method comprises the following steps: adjusting input parameters such as a calibration plate position and a camera attitude, observing the change condition of a calibration result, determining the influence degree of each input parameter on the calibration result, carrying out eigenvalue decomposition on a covariance matrix of a calibration error, and analyzing an error source and the influence degree of the error source on the calibration result; if some error sources are found, calibration is carried out again by adding calibration points or changing calibration conditions. The system comprises a parameter calculation module, an error analysis module and a calibration optimization module. The accuracy and reliability of the calibration result are improved.

The invention relates to a line laser camera hand-eye calibration method comprising the following steps: S1, determining a triangular pyramid calibration block, and selecting a fixed calibration point; s2, based on the triangular pyramid calibration block and the fixed calibration point, TCP position calibration and TCP attitude calibration are carried out on the center point of the tail end tool, and a calibrated TCP point is obtained; s3, selecting calibration parameters of the line laser camera, and verifying an internal reference matrix of the line laser camera in combination with the triangular pyramid calibration block; s4, on the basis of the calibrated calibration parameters and the calibrated TCP points, fixed calibration points are measured multiple times at different poses, feature values of the fixed calibration points are extracted, and then a hand-eye calibration transformation matrix is calculated; and S5, performing calibration error verification on the hand-eye calibration transformation matrix, judging whether a precision requirement is met or not, if so, returning to the step S2 to perform calibration again until the precision requirement is met, and if not, not performing any operation. Compared with the prior art, the method has the advantages of high calibration precision and the like.

The invention provides a water quality parameter inversion method and system based on spectrum correction and electronic equipment, and relates to the technical field of water environment remote sensing. According to the method, the water quality parameter and the water body reflectivity value of the surface feature spectrometer are measured synchronously when the remote sensing image passes through the scene, spectrum correction is carried out on the remote sensing image by using actually measured data as a reference, and the accuracy of water quality parameter inversion is improved. Firstly, water quality parameter and water body reflectivity value measurement of a surface feature spectrometer are synchronously carried out innovatively when a remote sensing image passes through a scene, so that the consistency of measured data and the remote sensing image in time and space is ensured, and a reliable basis is provided for subsequent spectrum correction and water quality parameter inversion modeling. Secondly, a spectrum correction model is established based on the synchronously acquired measured data, the spectrum of the remote sensing image can be adjusted according to the measured data, and the reflectivity deviation caused by atmospheric correction errors and sensor calibration errors is reduced, so that the accuracy of the spectrum data is improved.

The invention discloses a line structured light sensor light plane calibration error evaluation method, and the calibration process of a light plane comprises the steps: obtaining a target image of which the surface is projected with a laser light bar, and extracting the pixel coordinates of the central point of the laser light bar; respectively converting the central point of the light bar and the target plane to a camera coordinate system by using calibration parameters of the camera; under a camera coordinate system, connecting the optical center with the central point of each light strip, and recording the intersection point of the optical center and the target plane as a reference point; fitting a light plane by using the reference point to complete calibration; the error evaluation method comprises the following steps: S1, marking intersection points of connecting lines between the optical center of the camera and each datum point and a light plane as theoretical points; and S2, calculating a deviation value between each theoretical point and the corresponding reference point, or calculating a deviation value between a back projection point of each theoretical point on the target image and a light strip center point in the target image, and taking the deviation value as an error of light plane calibration. The method is simple and reliable in evaluation process, and can improve the precision of light plane calibration.

The invention provides a calibration piece and a calibration method for a multi-port waveguide antenna test system, and a test method and a test system for a multi-port waveguide antenna, in the calibration piece, a positioning structure can ensure the repeatability of each connection, which is the premise of high-precision calibration, and the calibration accuracy is greatly improved. The load unit, the offset short circuit unit and the offset load unit are provided with a plurality of waveguide ports which are the same as those of a tested piece and can be directly butted, the internal structure of the waveguide port of the load unit is an antenna, electromagnetic wave energy can be radiated out instead of being absorbed internally, and high precision and stability of the load standard are guaranteed; according to the method, the reference surface can be determined at the interface of the tested piece, the calibration error model can be directly tested on the reference surface, the accuracy is good, and the solved calibration error model contains all time-varying characteristics of the whole test link during each calibration, so that the method can be used for frequently calibrating according to requirements, and the calibration accuracy is high. Measurement errors caused by temperature drift and link loss can be handled, and the efficiency is high.

According to the embodiment of the invention, leakage characteristic data of a plurality of sensors are collected in real time; performing filtering processing on the leakage characteristic data and then determining an area where a leakage source of the vacuum equipment is located; determining the confidence coefficient of the area where the leakage source is located, determining the confidence coefficient of the position of the leakage source based on the confidence coefficient of the area where the leakage source is located, and giving an alarm when the confidence coefficient of the position of the leakage source reaches a confidence threshold value; and injecting helium to the leakage source position of the vacuum equipment in a preset time period to perform calibration test, calculating a calibration error of the leakage source position of the vacuum equipment, and when the calibration error exceeds an error threshold value, recalculating the confidence coefficient of the leakage source position and resetting the confidence coefficient threshold value. The problems of high cost, low detection precision, slow response speed, dependence on artificial experience and incapability of real-time monitoring due to long-time use of a helium mass spectrum detection method when leakage positioning detection is performed on a leakage source of vacuum equipment in the prior art are solved.

A traffic scene external-field-free continuous calibration method comprises the steps that a traffic scene image is collected, geometric prior features in the image are extracted, and the geometric prior features are used for representing points existing on a distortion curve on the traffic scene image; an error function is defined, distortion parameters are optimized based on the error function, and the distortion parameters are used for representing geometric offset generated by lens distortion in the image and comprise radial distortion parameters and tangential distortion parameters; based on the optimal distortion parameter, performing orthodontic processing on the acquired image by adopting a nonlinear optimization algorithm; ground coplanar feature points are extracted from the orthodontic image, and camera external parameters including a rotation matrix and a translation vector are solved by combining corresponding world coordinate system positions of the ground coplanar feature points; calibration error distribution is evaluated based on simulation or actual scene data, distortion model parameters or optimization strategies are dynamically adjusted, and online parameter updating is triggered when it is detected that in-place posture deviation exceeds a threshold value. The method can simplify the calibration process and improve the calibration efficiency.

The invention discloses a multi-modal task model training method and device, a multi-modal task processing method and device, equipment and a computer program product, and the method comprises the steps: obtaining multi-modal sensor data, including laser point cloud data and image data; performing data enhancement processing on the multi-modal sensor data by using a data enhancement strategy; calibration data of the multi-modal sensors are obtained, training data of a multi-modal task model are constructed according to the enhanced multi-modal sensor data and the calibration data between the multi-modal sensors, and the calibration data comprise good calibration data and weak calibration data; and training the multi-modal task model by using the training data and a preset knowledge distillation strategy. According to the multi-modal model training method based on online knowledge distillation, on the premise of not depending on accurate calibration, the model can dynamically adapt to different calibration errors, the model training precision is maintained, and it is ensured that an automatic driving system can maintain the high-precision sensing ability in different environments.