169results about How to "Reduce error accumulation" patented technology

The invention belongs to the technical field of text processing, and relates to a text factual proofreading method and system based on a large language model. The method comprises the following steps: constructing a task-adaptive large language model cue word framework; decomposing the text into a plurality of independent simplified fact expression texts; generating a fact query statement; establishing a local own knowledge base and a trusted website library; performing fine tuning training to obtain a trained large language model; performing three-way parallel judgment verification by using the trained large language model; and outputting a judgment verification result and an information source. According to the method, the powerful text analysis processing capacity of a large language model is fully utilized, error accumulation is reduced, and higher-precision proofreading is achieved; according to the method, multiple knowledge bases are supported, knowledge bases dynamically changing along with time can be obtained, a specially-assigned person does not need to be assigned to carry out real-time maintenance and custom fact texts, high precision and high real-time performance of facts are ensured, factuality verification is carried out, and the situation that correction is delayed due to delay caused by updating waiting of the local knowledge bases is avoided.

The invention relates to the technical field of mixed reality, in particular to a mixed reality multi-user interaction large-space positioning, aligning and collaborating system and method, and aims to solve the problems that in the prior art, the user interaction experience is improved only by collecting multi-aspect data for processing, efficient collaboration work of multiple users in a large space cannot be achieved, and the user experience is poor. Information sharing, task distribution and real-time interaction among users cannot be realized; information sharing, task distribution and real-time interaction among users are realized through the multi-user cooperation module, cross-device operation consistency is ensured through a multi-device sharing coordinate system, configuration is simplified, cooperation efficiency is improved, real-time data transmission is realized through multi-device network synchronization, delay is reduced, cooperation fluency is enhanced, and user experience is improved. The interaction reality sense is enhanced through physical interaction logic synchronization, the working process is optimized, and the cooperation experience is improved.

The invention provides a satellite video super-resolution reconstruction method and system oriented to multiple degradation processes, and the whole process of the method comprises the steps: generating corresponding low-resolution video data through a satellite video random degradation module, and carrying out the feature extraction of the low-resolution data through a feature extraction module, the method comprises the following steps of: extracting features, transmitting each frame feature of a whole video sequence by adopting a high-order grid bidirectional propagation scheme according to the extracted features, realizing feature alignment of video data by using a feature alignment module, and finally performing pixel recombination on the aligned features, and adding the aligned features with low-resolution features after up-sampling to obtain a reconstructed high-resolution image. Three satellite video data experiments show that the high fidelity of the satellite video can be effectively recovered, the reconstruction quality of the remote sensing image is remarkably improved, and meanwhile, the method has a certain denoising capability.

The invention discloses a physically-guided critical region fluid nonlinear wave motion prediction method, system and device and a medium. The method comprises the steps that a critical region fluid computational domain and a nonlinear time-varying equation are established; constructing a sample data set; constructing a physical information neural network model; constructing a loss function including nonlinear time-varying equation residual error loss, initial point error loss and boundary point error loss; training the physical information neural network model to obtain a critical region fluid free water surface displacement wave envelope prediction model; weighting internal matching point samples in each iteration round by adopting a gradient-free nonlinear attention mask method based on residual accumulation in the training process; and inputting the time-space coordinates of the to-be-predicted point of the critical region fluid into the prediction model to obtain the free water surface displacement wave envelope of the to-be-predicted point fluid. According to the method, error accumulation is effectively reduced, the model convergence speed is increased, and the prediction precision of the nonlinear wave motion of the fluid in the critical region is improved.

The invention provides an earth-rock volume calculation method, system and device and a storage medium, and the method comprises the steps: obtaining the three-dimensional coordinate information of the earth-rock surface in advance, and correcting the three-dimensional coordinate information to obtain the corrected point cloud coordinate information; performing curved surface fitting through a radial basis function according to the point cloud coordinate information to obtain a fitted curved surface equation; determining the ranges of the coordinate X, the coordinate Y and the coordinate Z according to the fitting curved surface equation; according to the ranges of the coordinate X, the coordinate Y and the coordinate Z, the earth-rock volume is calculated through an integration method. The problems that in earthwork volume calculation, a high-resolution image is obtained through aerial photography of an unmanned aerial vehicle, the image quality is possibly influenced by weather conditions and illumination changes, so that the modeling precision is influenced, and a modeling method is limited by software functions are solved.

The invention relates to the technical field of machine tool intelligent machining, and discloses a machining method based on deep learning and multi-sensor fusion, which comprises the following steps: collecting data in the machining process of a numerical control machine tool, and preprocessing the data in the machining process of the numerical control machine tool to obtain a data set; inputting the obtained data set corresponding to the data acquired by the plurality of sensors into a Kalman filtering algorithm for Kalman recursive filtering processing to obtain fused actual observation data acquired by the sensors; inputting a data set corresponding to the fused actual observation data acquired by the sensor and the data acquired by the numerical control system into a depth Q network algorithm, and predicting to obtain an optimal tool path; a magnetic field is applied to the machined workpiece through the excitation module; and machining a machined workpiece model along the optimal tool path by adopting an ultrasonic tool high-frequency machining method. According to the invention, the surface quality of the to-be-processed workpiece can be obviously improved.

The invention discloses a nanometer length metering system and method based on chromium atom transition frequency, and belongs to the field of nanometer length metering, the nanometer length metering system comprises a physical wavelength reference and a working standard substance, the physical wavelength reference is a 425.55 nm laser wavelength locked to a chromium atom (7S3-> 7P4) transition spectral line, and the working standard substance is a laser wavelength locked to a chromium atom (7S3-> 7P4) transition spectral line. The working standard substance is chromium self-tracing series gratings prepared by using an atomic photoetching method, grating periods are traced to a physical wavelength reference, and the chromium self-tracing series gratings comprise a one-dimensional chromium self-tracing grating and a two-dimensional chromium self-tracing grating. According to the nanometer length metering system based on the chromium atom transition frequency, the chromium atom transition frequency, which is a natural constant, is materialized to the chromium self-traceability grating in a laser convergence atom deposition manner, so that the grating has independent traceability and ultrahigh accuracy and consistency at the same time.

The invention discloses a standard value prediction method, an error monitoring method and a device thereof for a current transformer, and relates to a power system monitoring technology. The prediction method comprises the following steps: acquiring time sequence data of secondary sides of multiple current transformers; processing the time sequence data through a transposition embedding layer to obtain high-dimensional features; performing feature extraction on the high-dimensional features through a gating network attention network to obtain time sequence trend features; encoding the time sequence trend characteristics through a feedforward neural network to obtain encoding characteristics; and processing the coding characteristics through an output layer to obtain a predicted standard current value of the current transformer. According to the method, the accurate standard current value can be predicted and obtained to serve as the reference value of the current transformer, and therefore error monitoring of the current transformer in a circuit is achieved under the condition that no standard current transformer exists.

The invention discloses a seven-degree-of-freedom series-parallel structure mechanical arm for a biped robot, and relates to the field of robots, the seven-degree-of-freedom series-parallel structure mechanical arm is installed on a supporting body, and the supporting body is located on an installation fixing base. Comprising a five-degree-of-freedom series arm and a two-degree-of-freedom parallel wrist structure which is located at the front end of the series arm and connected with the series arm. According to the mechanical arm with the series-parallel connection structure, the rigidity and the bearing capacity of the mechanical arm are improved, the working space is expanded to adapt to more complex operation tasks, the stability and the reliability of operation are enhanced, error accumulation is reduced, and the accuracy of overall operation is improved.

The invention provides an online domain change continuous learning method and system based on pseudo label filtering, and relates to the technical field of computer vision, and the method comprises the steps: obtaining a pre-training model, predicting changed target domain data through the pre-training model, and generating a pseudo label for online self-adaption; based on binary classification, deriving a lemma based on a threshold filtering false label in the line domain change continuous learning, and designing a threshold setting principle in the line domain change continuous learning according to the lemma; a designed threshold setting principle is utilized to filter a pseudo label with low confidence coefficient of model prediction, and meanwhile, a class prior alignment method is introduced to encourage the model to fairly predict an unknown domain sample; and updating and optimizing the model by using the filtered pseudo labels to obtain a classification prediction result in online domain change continuous learning. According to the method, the self-adaptive threshold value capable of adapting to the CTTA process is established, so that the quality of the false label is ensured.

The invention discloses a cable sensor network time service compensation method, device and equipment and a medium, and belongs to the technical field of network time service, and the method comprises the following steps: collecting process historical data of a cable network actual error compensation system, and taking error data of a satellite second clock and a crystal oscillator second clock as an initial sample; aggregating the initial samples to obtain aggregated samples under each step size scale; constructing an error feature vector containing an aggregation sample of the current scale and an accumulated error of the current time point information; performing modeling on the error feature vectors by using a Gaussian-like process regression model to obtain a Gaussian-like process regression model of each step size scale; error estimation values of all step size scales are generated in sequence, and a time service error estimation value of the current moment is obtained by combining the difference error estimation values of the lowest step size scale and the highest step size scale; and compensating the accumulated error of the crystal oscillator clock based on the time service error estimation value at the current moment. According to the invention, the synchronization precision and practicability are improved.

The embodiment of the invention provides a seal analysis method and system, electronic equipment and a storage medium, and belongs to the field of target detection and text recognition. The method comprises the following steps: acquiring a to-be-analyzed image, inputting the to-be-analyzed image into a feature extraction module in a pre-constructed analysis model, extracting image feature information and feature position information, and outputting a fused feature matrix through matrix fusion; inputting the fusion feature matrix into a feature coding module to obtain global features of the fusion feature matrix; and inputting the global feature and the target feature learning matrix into a prediction module to obtain a seal type, a seal frame and seal text content. Image end-to-end seal analysis is provided, a seal detection task and a seal identification task are constructed into an analysis model, the use efficiency of feature coding can be improved, the capability and training speed of a coding module can be further improved through constraint of two prediction heads on the coding module, and the method is more suitable for popularization and application. And error accumulation caused by a plurality of steps is reduced.

The invention relates to the field of electronic manufacturing and intelligent control, and discloses a patch bearing positioning calibration method and system, a storage medium and a robot, environmental parameters are monitored in real time through a distributed sensor network, and data preprocessing and integration are performed by using edge computing equipment to generate comprehensive environmental state data; a nanoscale micro-motor driving system is utilized to realize high-precision motion control and real-time positioning calibration of the suction nozzle; path optimization is carried out through the quantum computing technology, the path is dynamically adjusted based on real-time positioning calibration data and environment state data, the response speed and precision of the chip mounting process are improved, AI continuous learning and anomaly detection are further included, and self-adaptive optimization and long-term stability of the system are achieved through real-time data collection and model updating and optimization; according to the invention, the performance of the existing patch calibration technology is obviously improved, efficient and accurate patch bearing, positioning and calibration are realized through advanced technical means and optimized data processing flow, and the production quality and efficiency are improved.

The invention provides a BIM-based electromechanical installation position analysis method, and relates to the field of equipment position detection, and the method comprises the steps: obtaining a BIM-based space structure model of a to-be-analyzed building through point cloud data; performing enhanced display on the obtained fuzzy region in the BIM-based building space model by fusing IFC semantic information and geometric similarity; and analyzing and judging the actual mechanical and electrical installation position by combining the processed BIM building space model for the building and the set mechanical and electrical installation information. The invention further discloses an electromechanical installation position analysis device based on the BIM. The electromechanical installation position analysis device comprises an electromechanical equipment position acquisition module, a central processing module, a position coordinate extraction module and a comparative analysis module. Spatial comparison is carried out on the basis of the BIM building space model and the set installation position, whether the installation error between the electromechanical installation position and the set position exceeds the error allowable value or not is judged, error accumulation in the data transmission process can be reduced through cleaning use of the space model, and the accuracy of the electromechanical installation position is improved.

The invention discloses a short-term traffic flow prediction method based on a dynamic multi-dimensional global perception graph convolutional network, and belongs to the field of traffic flow prediction. The method comprises the following steps: acquiring an input dynamic graph feature sequence, and generating space-time element parameters; carrying out feature enhancement on the dynamic graph feature sequence, fusing the dynamic graph feature sequence with the space-time element parameters, and then respectively carrying out space-time feature coding; fusing the spatio-temporal characteristics obtained by coding the spatio-temporal characteristics; and carrying out space-time trend alignment decoding on the fused space-time features. According to the method, the defects of limited modeling precision, insufficient dynamic adaptability, incomplete spatial-temporal feature capture and the like in short-term traffic flow prediction are overcome, and the accuracy and the applicability of short-term traffic flow prediction are improved.

The invention relates to the technical field of numerical simulation, and discloses a seismic wave forward numerical simulation method, device and system and a storage medium, and the method comprises the steps: determining a first objective function based on a traditional regular grid finite difference format three-dimensional scalar wave equation; based on the first objective function, establishing a second objective function of a three-dimensional effective vector wave number range; a regularization parameter is introduced into the second objective function, an optimal solution is determined by solving a regularization problem, and an optimized finite difference coefficient is obtained; and substituting the optimized finite difference coefficient into the second objective function to obtain the maximum value of the effective vector wave number range. According to the method, the obtained three-dimensional space is established on the basis of the minimum norm, the domain second-order partial derivative rule grid finite difference coefficient is obtained, and error accumulation in numerical simulation of the deep target can be better reduced.

The invention belongs to the technical field of speech synthesis, and discloses a ByT5-based speech synthesis method, device and equipment and a storage medium, and the method comprises the steps: obtaining a bit sequence of an input text, and converting the bit sequence into a text feature vector through a ByT5 model; obtaining a part-of-speech embedding vector and a syntactic structure embedding vector; summing the text feature vector, the part-of-speech embedded vector and the syntactic structure embedded vector, then sequentially inputting the summed text feature vector, the part-of-speech embedded vector and the syntactic structure embedded vector into a full connection layer of a generative adversarial network speech model, and activating a function for feature fusion to obtain text fusion features; and a decoder and a vocoder of the generative adversarial network voice model are used to successively acquire a target language Me < l > spectrum and a target language waveform. According to the method, the problem that a token-based model is limited by a dictionary is solved, an original text can be directly converted to multilingual unified bit representation, syntax and structure information is fused on the basis of the strong general processing capacity of a ByT5 model framework, and the quality of synthesized speech can be effectively improved.

The invention discloses a forming method of a high-speed wheel spoke plate, which further comprises a shaping die used in a shaping step, and the method comprises the following steps: 1) plate shearing and blanking are carried out, and a product blank is prepared from raw materials; (2) stretching forming is conducted, and a radial plate contour piece is manufactured through the product blank; 3) punching: punching a mounting hole and a middle hole in the radial plate contour piece; (4) flanging is conducted, specifically, the edge of the radial plate contour piece is flanged; 5) shaping; according to the method, the position precision of the middle hole and the chamfer of the product is remarkably improved. Cooperative shaping machining of chamfers, mounting holes, middle holes and turned-over edges is completed in the same working procedure through the shaping die. And through the synchronous shaping process, the runout control precision and the chamfering position degree of the radial plate are remarkably improved. According to the method, the production efficiency is greatly improved, accumulative errors are reduced, and high-quality and high-efficiency production of the high-speed wheel spoke plate is achieved.

The present invention relates to the field of gas metering technology, and discloses a civil gas metering method based on ultrasonic metering, including: using an ultrasonic gas meter to record the user's gas consumption, recording the gas-using equipment, gas consumption duration and time period in each gas usage cycle, and ensuring that the equipment heat load is a constant value. By screening the independent gas consumption time period of each device, its unit time gas consumption is calculated as historical data. At the same time, the noise of the gas pipeline is monitored. If the noise exceeds the limit, it is recorded as an abnormal period, and the abnormal time period and marking equipment are obtained. According to the marking equipment and the average unit gas consumption, a first data calibration strategy is performed, and the actual consumption and the estimated consumption are compared after the cycle ends. If there is no error, it is recorded. If there is an error, the second data calibration strategy is executed, and statistical calibration is performed according to the error value; the accuracy of gas metering is guaranteed by the first data calibration strategy, the second data calibration strategy and statistical calibration.

The invention discloses a wind speed short-term multi-step prediction method fusing direct prediction and recursive prediction, and the method comprises the steps: obtaining an original data set of a wind speed short-term multi-step prediction task, and carrying out the normalization processing, and obtaining a training data set; constructing a multi-channel LSTM coding layer based on a global encoder and a single-step encoder; inputting the training data set into a multi-channel LSTM coding layer for coding to obtain coding state features; constructing a two-way decoder, and inputting the coding state features into the two-way decoder for global recursive decoding and single-step direct decoding to obtain global decoding features and single-step decoding features; splicing and fusing the global decoding features and the single-step decoding features to obtain fused features; and performing training prediction on the fusion features based on a residual network and a filling strategy to obtain a multi-step prediction result. According to the invention, the prediction precision can be improved, and the error accumulation during the prediction period can be reduced at the same time. The method can be widely applied to the technical field of wind speed short-term multi-step prediction.

The invention discloses a text-driven action generation method, a text-driven action generation device, text-driven action generation equipment, a storage medium and a computer program product, and relates to the technical field of artificial intelligence. In the application, firstly, an input text and a null motion sequence are input into an autoregressive motion converter; obtaining a basic motion sequence; then, sequentially processing the basic motion sequence through an encoder and a quantizer in the vector quantization variational auto-encoder to obtain a residual index; thirdly, inputting the input text, the basic motion sequence and the residual index into a non-autoregressive motion converter to obtain a residual motion sequence; and finally, processing the residual motion sequence through a decoder in the vector quantization variational auto-encoder to obtain a target motion sequence.

The present invention discloses a system and method for collecting and processing data for national land space planning, and specifically relates to the technical field of data processing. The system improves the detail accuracy of low-resolution satellite images in national land space planning through a data processing and analysis method for multi-resolution remote sensing images, obtains image data of different resolutions from different satellites according to needs, and evaluates the detail fidelity and captures the fluctuation state of low-resolution images through image segmentation and feature ratio analysis. After comprehensive analysis, the images are divided into high, medium and low accuracy levels, and abnormal prediction, data correction and optimization decision-making are performed for medium-accuracy images. This can not only effectively reduce the cumulative errors caused by low-resolution images, but also enhance the accuracy and long-term feasibility of planning, thereby ensuring the scientificity and practicality of national land space planning.

This invention proposes a multi-field-of-view coupled drilling robot rig positioning method, applicable to complex underground roadway environments in coal mines. Current drilling rig positioning methods primarily rely on vision sensors, which suffer from difficulties in selecting borehole points, potential damage to the anchor mesh structure during drilling, and the presence of dust and water mist during drilling, which obstructs vision sensors and causes image loss. This application proposes a multi-field-of-view coupled drilling rig positioning scheme based on the spatial positions of the drilling robot's components, and formulates a borehole point positioning strategy to avoid interference between the borehole and the anchor mesh during drilling. By identifying and locating the boreholes in the image, precise anchoring and positioning of the drilling rig are achieved. This not only has direct significance for improving support efficiency and safety but also provides important reference for positioning technologies of other underground engineering equipment.

A locking mechanism for a diagonal tie rod in an outward-opening, top-hung hardware system is disclosed. The diagonal tie rod includes a movable tie rod slidably mounted on a base for positioning a first connecting rod. The base has a first connecting rod with a limiting groove at one end near the base. The positioning device includes a limiting block with a locking part. When the window sash needs to be opened to its maximum position, the locking part extends out of the base and engages with the limiting groove to form a limiting connection, thus locking the window sash. The movable tie rod has an unlocking part. When the window sash needs to be closed, the locking part slides within the base driven by the handle. The unlocking part abuts against the limiting block, creating a lever effect that disengages the locking part from the limiting groove, thereby releasing the locked state of the window sash.

The invention relates to the technical field of vehicle bodies, in particular to a B-column-free vehicle body structure, a vehicle and an assembling method of the B-column-free vehicle body structure. The B-column-free vehicle body structure comprises a front windshield assembly, a top cover longitudinal beam, a top cover rear assembly, a back door assembly and a lower vehicle body assembly; the front end of the top cover longitudinal beam is fixedly connected with the middle of the upper end of the top cover rear assembly. Gull-wing door installation parts are arranged on the left side and the right side of the middle of the top cover longitudinal beam. The lower end of the front windshield assembly is fixedly lapped with the front part of the lower vehicle body assembly, the lower end of the top cover rear assembly is fixedly lapped with the rear part of the lower vehicle body assembly, and the back door assembly is fixedly connected with the rear sides of the top cover rear assembly and the lower vehicle body assembly; the front windshield assembly, the top cover longitudinal beam, the top cover rear assembly, the back door assembly and the lower automobile body assembly are all aluminum alloy integrally-cast forming parts. The European wing type opening mode of the whole side vehicle door can be met, and on the premise that the performance is met, light-weight and lean design and production of a vehicle body are achieved.

The invention discloses an unmanned aerial vehicle remote control signal-oriented sweep interference suppression method, which relates to the technical field of communication, and realizes interference signal reconstruction in a digital domain based on sweep interference frequency change estimation of an instantaneous frequency estimation algorithm and in combination with estimated instantaneous power information so as to achieve an interference signal suppression function. Therefore, interference is effectively reduced while useful signals are received. The invention also discloses a sweep interference suppression device for the remote control signal of the unmanned aerial vehicle, the normalized least mean square algorithm (NLMS) is selected, the method is more suitable for interference suppression in a low signal-to-noise ratio scene, error accumulation can be effectively reduced, and the filtering precision is improved. Through the improvement, the anti-interference capability of the system is remarkably improved, so that remote control signals of the unmanned aerial vehicle can still be normally received under the condition of relatively strong interference.

The invention provides a target tracking method based on an inertial sensor and a visual camera, which comprises inertial navigation, visual detection and multi-sensor fusion, and the inertial navigation comprises the following steps: S11: data acquisition: an accelerometer, a gyroscope and other IMU sensors built in a mobile phone acquire acceleration and angular velocity data of a target in three directions in real time, basic data is provided for subsequent processing; s12, data processing: carrying out filtering and de-noising preprocessing on the collected data; s13, position calculation: carrying out two times of integration on the acceleration data to obtain displacements of the target in three directions so as to calculate the position of the target, and meanwhile, calculating the attitude and course of the target by utilizing gyroscope data; and S14, error correction: carrying out auxiliary correction on the positioning technology by adopting a PDR algorithm so as to improve the positioning precision. The advantages of the inertial sensor and the visual camera can be fully utilized, more accurate tracking of the target object is realized, and error accumulation is reduced.

The invention provides an auxiliary welding device and a welding method. The auxiliary welding device comprises a support. The bracket includes: a first rod extending in a first direction; the second rod extends in the second direction, is arranged on the first rod and is provided with a first clamping part; the first clamping part is used for being clamped with a first position of a supporting rod of the rotating brake; the first direction is perpendicular to the second direction; the rotating brake comprises a main rod and a plurality of supporting rods connected with the main rod; the first guide piece is provided with a second clamping part; the second clamping part is used for being clamped with a second position of the supporting rod of the rotating brake, so that the first clamping part and the second clamping part are arranged at intervals in the first direction, and the supporting rod is fixed in the first direction; and the second guide piece is used for being clamped with the main rod of the rotating brake, so that the length direction of the connecting hole of the main rod is the same as the first direction, and one end of the supporting rod can be located in the connecting hole of the main rod in the length direction. According to the welding auxiliary device, the welding machining convenience of the welding auxiliary device on the main rod and the supporting rod of the full-height rotating brake can be improved.

One or more embodiments of the invention provide a training method of a voice unit prediction model, a voice synthesis method and electronic equipment. The training method comprises the following steps: acquiring a first training sample and a second training sample; wherein the first training sample comprises a plurality of texts, and the second training sample comprises a plurality of texts and voice units thereof; the number of texts in the first training sample is greater than the number of texts in the second training sample; performing self-supervised pre-training on a preset neural network model by using the first training sample to obtain a language model for learning language expression information from the first training sample; performing supervised training on the language model by using the second training sample to obtain a voice unit prediction model; the voice unit prediction model is used for processing an input text to obtain a voice unit sequence. According to the embodiment of the invention, pre-training is performed to obtain the language model learning the language expression information, and then supervised training is performed on the language model, so that the prediction accuracy of the voice unit prediction model is improved.