5 results about "Adaptive bias" patented technology

The invention discloses a self-adaptive control method and system for a condenser for a nuclear power station, relates to the field of condenser control, and aims to predict a future load change trend in a prospective manner by analyzing historical data of a unit power instruction and steam flow, and calculate feed-forward circulating water flow based on a prediction result. On the basis, a feedback control loop based on actually measured pressure is reserved. More importantly, a self-adaptive bias correction mechanism is introduced, and continuous learning and correction are carried out on output of feedback control, so that steady-state errors of the system are eliminated, and uncertainty of the model is compensated. Finally, feed-forward, feedback and self-adaptive correction components are intelligently fused to form a final control instruction which can quickly respond to load changes and can accurately maintain pressure stability, and the problems of severe pressure fluctuation and energy waste under the dynamic working condition are effectively solved.

The application provides a visual language large model adaptive bias evaluation method for evaluating bias performance of a target model on a specific task, and the method comprises the following steps: S1, obtaining an initial evaluation dataset, wherein the initial evaluation dataset comprises a plurality of evaluation data; S2, fine-tuning a preset large language model using the initial evaluation dataset in a preset fine-tuning manner to obtain an initial data generator; and S3, generating evaluation data based on the initial data generator to perform multiple rounds of bias evaluation on the target model, wherein in each round of bias evaluation, based on all evaluation data corresponding to the last round of bias evaluation, a data generator corresponding to the last round of bias evaluation is iteratively updated according to a preset objective function to obtain a current round data generator, and the current round data generator is used to generate a current round evaluation dataset to perform bias evaluation on the target model.

The invention discloses a large model semantic alignment method and system based on multi-level ontology. Generating an ontology semantic state shoot matrix by linearly mapping a large model hidden state to a five-level ontology semantic space and applying hierarchical attenuation; performing multiplicative semantic modulation on the original logits based on the matrix to form an implicit value function of ontology perception; fusing a future state regular item and an exponential hierarchical attenuation structure reward, and reconstructing an endogenous reward signal; generating controlled strategy distribution in combination with recursively accumulated semantic deviation penalty; and finally, semantic consistency evaluation of probability weighting and time decay is carried out on the complete sequence, an echo correction value is calculated, and system parameters are reversely optimized. According to the method, under the condition that gradient updating or manual labeling is not needed, real-time structured alignment and self-adaptive correction in the large model generation process are achieved, the interpretability, compliance and multi-round consistency of the generated content are remarkably improved, and the method can be widely applied to intelligent question answering and decision support systems in the fields of finance, government affairs, medical treatment and the like.

This invention discloses a dynamic control method for adaptive bias voltage in the energy region of a CZT photon counting detector, belonging to the field of radiation detection and photon counting control technology. The method acquires energy events of incident photons within a fast sliding time window and forms an energy spectrum histogram, extracting energy region composition information and flux status information for low, medium, and high energy regions. Within a slow sliding time window, it acquires count rate change information, energy peak drift information, pulse tail change information, leakage current status information, temperature status information, and current operating bias voltage information. Based on the energy region composition information, flux status information, flux change trend, and current operating bias voltage information, it matches them with a preset energy region bias voltage mapping relationship to generate candidate operating bias voltages. A polarization risk index is constructed based on count rate change, energy peak drift, pulse tail change, leakage current status, and temperature status, and a bias voltage compensation amount is generated based on the level and trend of the polarization risk index.

The embodiment of the invention provides a CSI (Channel State Information) noise reduction visible light positioning method based on adaptive bias and a related device. The method comprises the following steps: processing a received optical signal to obtain a channel impulse response data vector; inputting the channel impulse response data vector into a recurrent neural network, and adaptively selecting an optimal offset value; adding the optimal offset value to a channel impulse response data vector according to a preset proportion, and cutting to obtain a target channel impulse response data vector; estimating the line-of-sight path according to the target channel impulse response data vector, and calculating the receiving power of the line-of-sight path; calculating a distance value from the LED light source to the receiver according to the receiving power; and obtaining a final position coordinate of the receiver by using a least square method based on the distance value and the position information of the LED light source. Based on this, the embodiment of the invention can reduce the interference of the estimation value error on the receiving power, and realize a more accurate positioning result.