11results about How to "Reduce error propagation" patented technology

This invention discloses an efficient inversion method, medium, and equipment for material parameters of rockfill dams that considers target uncertainty and prior knowledge constraints. The method includes steps such as determining inversion parameters and calculating measurement points, constructing a high-precision multi-output surrogate model, constructing a multi-objective function expression, performing multi-objective parameter inversion calculations, and optimizing the solution set. By constructing a multi-objective function, the solution set space of target settlement under different spatiotemporal importance trade-offs is obtained. When generating parameters, the prior relationships between material parameters are used as constraints to avoid the generation of unreasonable results. At the same time, the uncertainty of target settlement is considered and reduced by introducing an error allowable interval and a high-precision surrogate model. Finally, a relaxed and improved Bayesian multi-objective optimization algorithm is used to achieve efficient and high-quality calculation of parameter inversion.

The invention relates to the technical field of protocol analysis, in particular to a network protocol collaborative combination analysis method based on a tree hierarchical structure, which comprises the following steps of: acquiring different network protocol analysis requirements, training a basic network protocol analysis model and a coarse granularity model, and acquiring a hierarchical structure by analyzing the coarse granularity and the fine granularity of application requirements. The method comprises the following steps: identifying and combining network protocol analysis models of the same hierarchy, constructing a tree-shaped layered structure, extracting a model identification result and an operation relationship in the tree-shaped layered structure, designing a two-dimensional mapping table, guiding input traffic through the two-dimensional mapping table, analyzing the input traffic layer by layer according to the tree-shaped structure, and positioning the next step through an operation result of each step until requirements are met. And an efficient collaborative analysis process is obtained. According to the method, the problems of high model training difficulty, high resource consumption and accumulation of errors along with layer propagation existing in a single network protocol analysis algorithm for multi-granularity and multi-layer hybrid application requirements are solved.

The invention belongs to the technical field of infrared imaging system testing, and discloses an MRTD intelligent testing method for an infrared imaging system, and the method comprises the steps: collecting the original four-rod target image data of thermal infrared imagers of the same model at different temperatures under the spatial frequency under the condition that the spatial frequency is not changed; cutting the collected original image data, and only obtaining an important area of a four-bar target image; carrying out expansion processing on the image; annotating the images, and dividing the images into a training set, a test set and a verification set; an MRTD intelligent detection model is constructed; training and accuracy testing are carried out on the MRTD intelligent detection model, label value prediction is carried out on the image data of the verification set, and the temperature difference value corresponding to the predicted label value N is the MRTD value of the thermal infrared imager under the spatial frequency. According to the method, the real-time performance of detection is ensured, and the interference of repetitive work and subjective factors in a traditional test method is reduced.

The application discloses a kind of phase correction method and system for structured light stripe profile projection, when boundary is identified, including the process of twice boundary identification, edge extraction is carried out by gray scale chart in initial boundary identification, in the second boundary identification, edge information extraction is carried out by phase calculation, boundary is identified by gray scale chart and phase diagram two directions, improve the accuracy of boundary identification, further the result of twice boundary identification is fused, so that the phase of stable area is not disturbed by jump pixel, ensure the accuracy of boundary division, based on the region after division, filtering processing is carried out to stable area, reduce the loss of feature information, provide reference information for the correction of unstable area, reduce error transmission, when the unstable area is compensated for error with stable area as reference, phase noise can be reduced, so as to reduce point cloud noise and improve precision.

The invention discloses a medical image tool enhancement processing method based on auto-reflection reinforcement learning. The method comprises the following steps: constructing a multi-modal reasoning model for vision-language joint reasoning; and performing supervision and fine tuning on the model by utilizing a cold start data set with tool calling track marks, so that the model has basic structured reasoning capability and normative tool use behaviors. An early-stage error reasoning track and a later-stage correct reasoning track which are generated by the same input at different training check points are collected, an auto-reflection training sample is constructed, and auto-reflection fine tuning is performed on the model. And based on the reinforcement learning data set containing the real tool calling scene, constructing a reward function and executing tool enhanced reinforcement learning training. After a medical image to be processed and a corresponding natural language problem are received, the trained model can generate gradual natural language thinking, a tool calling instruction and an external tool observation result, and outputs a medical image analysis conclusion, so that refined and explainable medical visual reasoning is realized.

This invention discloses a method and apparatus for detecting lane-level turning information at at-grade intersections, comprising: preprocessing the road network, cleaning trajectory data and matching maps to connect trajectory points with their respective road segments; an intersection guidance zone identification step, considering the intersection trajectory distribution characteristics and vehicle straight-ahead stopping information at stop lines, to detect the intersection guidance zone; an intersection lane identification step, based on the intersection guidance zone identification results, extracting trajectories related to the guidance zone and performing multi-angle denoising, thereby inferring the guidance zone lanes; and an intersection lane-level turning information detection step, considering the intersection lane identification results, performing lateral and longitudinal statistics on the turning trajectories of each lane, and employing a supervised learning method to detect the intersection lane-level turning information. This invention is the first attempt to identify intersection lane-level turning information based on crowdsourced trajectories, making up for the lack of systematic research in lane-level turning information mining in existing works.

The application discloses a non-invasive AC voltage measurement system based on capacitive coupling and reference signal injection, belongs to the field of electric power measurement, and comprises a detection end, a measurement unit and a shielded cable. The detection end comprises two half-circular conductive electrodes which can be opened and closed and can be clamped on an insulated wire. The measurement unit comprises an instrument amplifier, a reference signal source, an analog-to-digital converter, a processor and an isolation power supply. The shielded cable is used for connecting the detection end and the measurement unit, and the shielding layer of the shielded cable is connected with the injected reference signal. The application adopts double-frequency injection and separation calculation, utilizes frequency separation of the reference signal and the power frequency signal, avoids similar numerical interference, reduces error propagation, and the frequency of the reference signal avoids multiples of the power frequency signal, so that fast Fourier transform extraction is facilitated.

This invention belongs to the field of radar information processing technology and discloses a pre-training method for a multimodal model of radar target detection, tracking, and recognition. The invention includes collecting model pre-training data; converting the collected radar information processing data from various models into a unified data format through data preprocessing to serve as input data for the pre-trained model; achieving simultaneous input of multimodal data, including images, tracks, and category text, through a unified pre-training framework, and completing joint representation learning of the input data by the pre-training framework; and designing a multi-task joint learning optimization algorithm to achieve parallel training of radar target detection, tracking, and recognition tasks. This invention achieves integrated processing of radar target detection, tracking, and recognition tasks by establishing a multi-task learning mechanism under a unified model training framework, effectively improving the overall performance of radar information processing. Simultaneously, by enhancing versatility and generalization, the model is applicable to intelligent processing tasks of various radar models.

This invention belongs to the field of underwater acoustic communication technology. Specifically, it relates to a time-frequency joint equalization method for GMSK underwater acoustic communication. The method includes: converting and mapping the transmitted signal sequence, then modulating it according to the GMSK signal modulation method based on Laurent decomposition to obtain a GMSK complex baseband signal, and modulating it onto a carrier as the final transmitted signal, which is then transmitted into the underwater acoustic channel; the underwater acoustic channel receives the final transmitted signal, uses it as the received signal, and performs bandpass filtering, two-dimensional time-frequency synchronization, matched filtering, time-frequency equalization, demapping, and decision processing on the received signal to obtain the decided symbol, thus completing the time-frequency joint equalization.

The application discloses a USBL system installation deviation angle calibration method based on an iterative reweighted two-step method, and on the basis of a traditional two-step method, estimates a transponder position and an installation deviation angle in the same closed loop iteration process, so that position error and installation deviation angle error are mutually corrected, thereby weakening the error transmission problem in the traditional two-step method; in view of the problem that different measurement points have different qualities, a comprehensive weight is constructed for each measurement point to suppress the influence of abnormal values, gross error points and measurement points with poor geometric conditions; through the comprehensive weight, the contribution of different measurement points to position updating and installation deviation angle updating is no longer equal, thereby improving the overall estimation stability and robustness; in each iteration, the comprehensive weight is used to construct a weighted least square updating model of the transponder position and a weighted nonlinear least square updating model of the installation deviation angle, so that the alternating feedback structure and the reweighting mechanism are truly coupled, and a closed loop calibration method capable of stable convergence is formed.

The application provides a blind carrier phase recovery method and system for optical fiber communication. At the sending end, first, the QAM symbol stream is grouped, and a reference pilot with a specific fixed amplitude is inserted according to the quadrant in which the first symbol of each symbol group is located. At the receiving end, after the preliminary blind carrier phase estimation is completed, the original quadrant corresponding to the reference pilot is identified by using the minimum Euclidean distance detection mechanism, and then the actual quadrant position of the pilot in the received signal is compared, so as to analyze the phase ambiguity offset with 90 degrees as the base, and accordingly, the phase correction is implemented on the corresponding symbol group. The application effectively breaks through the four-fold rotational symmetry limitation of the traditional constellation diagram by means of amplitude modulation coding, and can accurately identify and correct the fixed-angle phase deviation occurring in the blind phase recovery process. The method can significantly suppress the error code diffusion caused by the phase ambiguity, improve the transmission performance of the system in terms of the bit error rate and the generalized mutual information, and has good algorithm compatibility and engineering implementation convenience, and the system overhead is effectively controlled.