5results about How to "Accurate call" patented technology

This invention relates to the field of exercise prescription generation and recommendation technology, and discloses an LLM-based generative recommendation method for exercise prescriptions. The method involves acquiring and preprocessing data, constructing a knowledge graph based on the preprocessed data (S1), pre-training the LLM model, inputting the constructed knowledge graph into the LLM model, and optimizing it through reinforcement learning. The PPO algorithm is used to optimize the LLM model's output strategy. Based on the optimized LLM model, the exercise prescription is derived and generated in natural language using FITT parameters. This method combines user characteristics and scenario information to generate high-quality, personalized exercise prescriptions. The invention utilizes a multi-dimensional reward function framework to quantify prescription quality and optimize the generation strategy, ensuring that the generated exercise prescriptions are not only scientifically sound but also conform to the user's actual situation and preferences. Synthetic data is generated through a user simulator, achieving an upgrade from static recommendation to dynamic adaptation, improving user experience and the level of intelligence in exercise and health management.

The application provides a large model fine-tuning method and device for a railway design intelligent agent tool call, the method comprising: constructing a supervised fine-tuning data set in the field of railway design; constructing a direct preference optimization data set based on the success and failure feedback of the large model intelligent agent call; using the constructed supervised fine-tuning data set and direct preference optimization data set to pre-train a large language model; using the supervised fine-tuning data set to perform single supervised fine-tuning on the pre-trained large language model to obtain a large language model after supervised fine-tuning; using the direct preference optimization data set to perform multi-round preference optimization fine-tuning with feedback on the large language model after supervised fine-tuning to obtain a large language model after optimization fine-tuning. By constructing high-quality data and fine-tuning the large language model, the application significantly reduces the risk of the large language model fabricating tools or misadjusting interfaces in complex design tasks, so that the intelligent agent has stable and controllable tool collaboration capability in the real railway design software environment.

The invention relates to the technical field of environmental assessment text processing, and discloses an intelligent auditing system for an environmental assessment report based on dynamic classification and knowledge mapping, which enables a classification system to automatically identify a novel pollution type and dynamically adjust a type boundary through multi-source data fusion and an incremental learning algorithm. The method ensures that the classification result always meets the newest environmental policy requirement, and ensures that the knowledge base can synchronize the newest standard specification content in real time by establishing a classification system and a trigger updating mechanism of the standard specification, thereby achieving the effect of comprehensively improving the professionalism and timeliness of the environmental assessment system. The system can automatically adapt to environmental policy changes and technical standard updating, can maintain scientificity of a classification system and integrity of a knowledge base without manual intervention, ensures that newest and most relevant standard specifications can be called for professional judgment when various environmental assessment reports are audited, and remarkably reduces audit deviation caused by knowledge lag.

The invention discloses an electric drive system rotary transformer zero position self-learning and curing system, which comprises a mechanical alignment module for driving a motor rotor to rotate to a preset mechanical zero position and generating a zero position offset value; after the motor reaches the rated rotating speed, the control module sends a rotary transformer zero-position self-learning enable signal, so that a motor controller executes current closed-loop control based on a zero-position offset value, a d-axis current instruction in a rotor synchronous rotating coordinate system is set to be a negative value and a q-axis current instruction is set to be zero, and an actual torque value is obtained; the zero calibration module obtains an average torque value through the actual torque value, calculates the deviation between the average torque value and a zero reference torque value, and when the deviation is larger than a positive deviation threshold value or smaller than a negative deviation threshold value, adjusts a zero offset value and triggers the control module to recollect the actual torque value; and obtaining a final zero offset value until the deviation is smaller than or equal to the positive deviation threshold value and larger than or equal to the negative deviation threshold value. The torque control precision and the operation reliability of the electric drive system are improved.

This invention relates to a closed-loop coin detection module based on a novel eddy current magnetic core. The main control unit retrieves temperature drift curves for n temperatures from a storage unit based on temperatures obtained from a temperature sensor. It obtains sampled values ​​based on the sampling points corresponding to the temperatures in the drift curves, and outputs the coin detection result by comparing the sampled values ​​with a set range, thus achieving closed-loop coin detection. The storage unit stores temperature drift curves obtained from temperature data sampling within a set temperature range for different coin types. The voltage stabilizing unit stabilizes the voltage of the central control unit's operating units. This invention eliminates deviations caused by coin identification speed, calculation time, and data communication between modules, enabling precise curve retrieval within a limited time, thereby achieving closed-loop control for coin detection.