7results about How to "Preserve precision" patented technology

This invention provides a flexible production line control method and system based on a large vertical domain model. The method includes: constructing a pruning index combining temporal stability and spatial similarity for the original large vertical domain model to evaluate the redundancy of the activation tensors in the intermediate layers, and performing structured pruning accordingly to obtain a sparse structure model; performing mixed-precision quantization on the pruned model: allocating quantization bits according to the sensitivity of each network layer weight to task loss, and calculating scaling factors and zeros based on the range of non-zero weights to represent sparse weights; calculating the output deviation between the quantized pruned model and the original model under the same input, and training a feedforward network with the former's output as input to fit and compensate for the output deviation; combining the quantized pruned model and the feedforward network to form a composite control model; after deployment, adding the output command of the quantized pruned model to the compensation vector during inference to generate control commands.

The invention provides a tool defect detection model construction method and a tool defect detection method, and relates to the technical field of image processing, and the tool defect detection model construction method comprises the steps: obtaining a training data set; for each first cutter image, performing feature extraction on each first cutter image by using a pre-constructed first unsupervised network to obtain a first feature map corresponding to the first cutter image; adjusting the channel dimension of the first cutter image according to a first feature map corresponding to the first cutter image to obtain a second cutter image corresponding to the first cutter image; and training the first supervised network according to the second cutter images corresponding to the plurality of first cutter images to obtain a second supervised network, and taking the second supervised network as a cutter defect detection model. Compared with the prior art, the accuracy of tool defect detection can be enhanced.

This invention provides a fast calculation method for beam transport in a de-energizer section based on the transfer matrix and Monte Carlo method, relating to the field of beam physics. The method includes: Step 1, constructing a transfer matrix model of the beam transport section before the de-energizer based on the beam optical parameters and optical element parameters at the starting point of the beam transport line, obtaining the particle phase space distribution parameters at the de-energizer inlet; Step 2, establishing a three-dimensional structural model of the de-energizer based on the particle phase space distribution parameters at the de-energizer inlet and performing Monte Carlo beam transport simulation of the de-energizer section, obtaining the position, momentum, and energy information of the beam particles at the de-energizer outlet; Step 3, performing outlier data removal processing on the position, momentum, and energy information of the beam particles at the de-energizer outlet, obtaining processed particle data. This invention achieves high-precision reconstruction of Monte Carlo particle information into optical parameters, improving the simulation efficiency of beam transport lines containing de-energizer structures.

The application relates to an intelligent content processing system based on a large model and a semantic vector, which comprises a user content preprocessing module, a user content search service module, a user content access processing module, a user content recommendation module, a user comment, opinion and dynamic classification and statistics module of a scroll bar and a user sensitive word and sentence analysis module and the like. The application also relates to a processing method. The intelligent content processing system based on the large model and the semantic vector and the method thereof adopt the application, automatically extract the abstract, the main body and the theme of the content through the large model, generate the semantic vector, combine the traditional inverted index to build a hybrid retrieval service and realize the deep fusion of keyword matching and semantic understanding. Meanwhile, the interest model is dynamically constructed based on the user behavior, various strategies are fused to realize accurate personalized recommendation, a double-layer mechanism of DFA rule and semantic vector matching is adopted for intelligent content auditing and classification, and the intelligent level, the accuracy and the system efficiency of the content processing are significantly improved.

The application provides a kind of partition flood control early warning method and system of fusion multi-source data, it is related to flood control early warning technical field, the method includes: obtaining the multi-source monitoring data set in target river basin;The multi-source monitoring data set is preprocessed, and the standardized data set is obtained;Based on the standardized data set, the flood evolution characteristic sub-model of each partition in target river basin is respectively constructed;Lightweight physical-empirical hybrid model is used to the output result of the flood evolution characteristic sub-model of each partition is coupled calculation, and the coupling calculation result is obtained;Based on machine learning algorithm analysis historical flood case data, generate the dynamic early warning threshold that is adapted to the flood evolution characteristic sub-model of each partition;Based on the coupling calculation result and the dynamic early warning threshold, generate partition early warning information.The application can fuse multi-source data, be aimed at partition characteristic, dynamically adjust threshold and efficiently calculate.

This invention discloses a method for joint synchronization of long-sequence phase of wireless signals and its ISAC action recognition method and device, belonging to the interdisciplinary field of radar signal processing and wireless communication. This invention aims to solve the problem of high-performance sensing with low-cost hardware in ISAC. Specifically, multi-carrier frequency-shift linear frequency modulation technology is used at the transmitting end to embed communication data while maintaining the characteristics of the radar detection waveform. To address the random carrier frequency offset, random sampling frequency offset, and random sampling time offset introduced by hardware non-ideals, a joint synchronization neural network is designed. A long-sequence feature extraction module captures long-sequence phase drift, achieving high-precision parameter decoupling and compensation in a single operation. Subsequently, the synchronized single-channel time-frequency features are combined with an environment-aware feature fusion network, and an adaptive gating fusion mechanism is used to eliminate static environmental clutter, achieving highly robust action gesture recognition.

The application discloses a nonlinear function quantization method in a Transformer compiler. The application adopts a quantization method to quantize and compress a transformer and a derivative model thereof, and obtains quantized weights and quantized parameters through a quantization-aware training (QAT) method. In the quantization process, a segmented linear approximation method is used to calculate a nonlinear function. In the determination of the segmented boundary, the QAT method is also used, and the linear approximation function of the nonlinear function is expressed as a differentiable function, so that it participates in the back propagation of the QAT, thereby obtaining the optimal segmented boundary combination. The application can be applied to the multi-platform deployment of a BERT model, and is especially suitable for a deployment platform such as FPGA which is suitable for rule calculation. In the quantization process, the method of the application can improve the hardware efficiency and maximize the accuracy.