12 results about "Elimination method" patented technology

The invention discloses a large model illusion elimination method and device, equipment and a storage medium, and relates to the technical field of artificial intelligence natural language processing. Carrying out atomization disassembly on the initial generation content to obtain an atomic fact set; performing internal knowledge base semantic matching and networking retrieval cross validation on each atomic fact in the atomic fact set to obtain an evidence coverage rate; when the evidence coverage rate is lower than a preset coverage rate threshold value, generating an evidence vacancy report; performing differentiation rewriting on the initial generation content based on an evidence vacancy report to obtain target generation content, performing evidence coverage rate quantitative checking on the generation content through atomization disassembly and an internal and external double-source cross validation mechanism, blocking illusion cascade in the early stage of reasoning, and feeding back a verification result for iteration rewriting, so as to obtain the target generation content. And high-confidence content subjected to dominant verification is obtained.

The application claims a time-oriented RDF quad model for MongoDB storage and a redundant attribute elimination method, which comprises the following steps: step 1, constructing a time-oriented RDF quad data model, taking time information as a new tuple of RDF triple extension, expanding the triple into a quad model containing time information, and defining the expanded time-oriented RDF quad data model; step 2, storing time-oriented RDF quad instance data in a document-oriented non-relational MongoDB database; step 3, using a weighted-based design algorithm to find out the instance attributes with higher redundancy from the time-oriented RDF quad data in the form of text stored in the MongoDB database, and eliminating these redundant instance attributes in the ontology library that has been constructed. The RDF quad-based query model designed in the application has a more significant effect on resources with time characteristics, and at the same time solves the problem of ontology ambiguity caused by repetitive instance attributes.

The invention discloses a self-adaptive backlash elimination method, device and equipment based on dual-motor coaxial driving and a storage medium, and relates to the technical field of electromechanical control, and the method comprises the steps: controlling a first motor and a second motor to carry out calibration when an output shaft of a control gearbox is fixed, and obtaining a backlash elimination value; based on the backlash elimination value and the user target position, motion track planning is obtained; and when the first motor and the second motor are controlled to move according to the motion track plan, adjusting the backlash elimination value according to the current value in the backlash elimination stage and the current value in the target motion stage, so that the adjusted backlash elimination value gradually reaches the actual backlash value. The technical problems that due to the fact that the tooth space value changes along with operation abrasion, the preset compensation value loses efficacy, the double motors move asynchronously, and the transmission shaft generates torsion stress are solved, the tooth space value is eliminated, the tooth space value can automatically change along with the actual tooth space value, and therefore position synchronization and torque balance are maintained for a long time in a double-motor coaxial driving system.

The application provides a kind of active elimination method of DC ripple of wind power cluster sending-out system through diode dispersion rectification. In view of the strong time-varying characteristics of the harmonic amplitude and phase of the DC side of the diode rectification system changing with the output of the wind turbine and the reference phase of the AC side, which leads to the problem that the ripple suppression effect of the traditional fixed uniform interleaved angle control is limited, the method proposes an active elimination mechanism of DC ripple based on the negative gradient algorithm of the central controller. The central controller extracts the 12th harmonic amplitude and phase information of each DC branch current in real time, uses the negative gradient algorithm to minimize the total amplitude after the superposition of the 12th harmonic, dynamically adjusts the reference phase of the sinusoidal AC voltage of each wind turbine grid side, compensates the amplitude difference through active phase regulation, and realizes the closed cancellation of unequal amplitude harmonics under the condition of output imbalance. The application does not need to increase hardware, effectively suppresses the DC sending-out current ripple, significantly reduces the size and cost of the filter device, and is suitable for space-limited scenarios such as offshore wind power.

The invention discloses a DC component digital elimination system and elimination method.The elimination system comprises a photoelectric detection module, an elimination module, a digital comparison output module and a control module, and the control module is used for closing an emitting element after elimination is triggered, controlling a first elimination digital-to-analog converter to traverse from the highest weight elimination gear and controlling a second elimination digital-to-analog converter to traverse from the highest weight elimination gear; determining a first elimination gear combination according to a comparison result of the current elimination gear, starting a transmitting element in a subsequent sampling period, controlling a first elimination digital-to-analog converter to counteract an ambient light signal, and controlling a second elimination digital-to-analog converter to traverse from the highest weight elimination gear when executing to a preset sampling gap, and determining the second elimination gear combination according to the comparison result of the current elimination gear to offset the direct current component in the reflected light signal, thereby dynamically determining the elimination gear combination in a gear-by-gear manner, effectively eliminating the direct current component in the ambient light signal and the reflected light signal, and ensuring the accuracy and reliability of the detection result.

The invention provides a large model refusal response elimination method and device, a storage medium and electronic equipment, and relates to the field of large language models. The electronic equipment obtains a mistaken rejection sample set and a conventional sample set; obtaining a first activation vector set when the to-be-processed model processes the false rejection sample set and a second activation vector set when the to-be-processed model processes the conventional sample set; determining an answer rejection network layer from the to-be-processed model according to the difference between the first activation vector set and the second activation vector set; and performing intervention on the answer rejection network layer, so that the to-be-processed model can normally process the harmless problem which is mistakenly rejected before intervention. Thus, by comparing the internal activation modes of the model for harmless but rejected questions and normally answerable questions, the rejection network layers dominating rejection behaviors are identified, and directional intervention is performed for the rejection network layers, so that the excessive defense response of the model can be weakened on the premise of minimizing the influence on other functions.

The objective is to obtain a biological elimination device that can prevent the target organism from becoming accustomed to a specific elimination method, depending on the number of times the target organism has invaded. [Solution] The system includes: an acquisition unit 7 that acquires detection information indicating that an organism to be eliminated 3 has entered the monitoring area 12 from a detection unit 5 that detects when an organism to be eliminated 3 has entered the monitoring area 12; and a control unit 11a that controls the elimination execution unit 6, which performs elimination to prevent the organism to be eliminated 3 from entering an intrusion-prohibited area, to change the elimination means executed by the elimination execution unit 6 to a different elimination means than the one executed when the cumulative number of eliminations was less than the predetermined number, if the cumulative number of eliminations, which is the number of times the organism to be eliminated 3 has entered the monitoring area 12, is greater than or equal to a predetermined number, based on the detection information.

The application provides a kind of self-interference continuous elimination method for the cooperative exploration of multiple unmanned ships in unknown areas, belonging to the technical field of autonomous exploration, autonomous positioning and map construction of multiple unmanned ships. To solve the self-interference problem that occurs when multiple unmanned ships cooperatively explore unknown areas, a coordinate relationship established by SLAM and multiple unmanned ship map fusion is used in combination with the iterative closest point registration algorithm to propose a laser radar point cloud method for continuously processing dynamic obstacles, which eliminates the self-interference of multiple unmanned ships in the cooperative exploration of unknown areas. In addition, for the laser radar point cloud data generated by dynamic obstacles, random points are formed in the map, which leads to poor subsequent multi-map fusion results and ultimately fails the exploration task. Through the coordinate system conversion relationship between unmanned ships, the application calculates the occurrence time of the self-interference problem and the initial position of the dynamic obstacle, and uses the iterative closest point to continuously track the position of the dynamic obstacle, achieving the purpose of continuously eliminating the influence of self-interference.

The application discloses a multi-modal large model structured pruning and parameter redundancy elimination method, system, device and medium, comprising: initializing the structure of the multi-modal large model; dividing the pruning process into a far-end stage and a near-end stage; constructing a cross-modal perception structured parameter group based on the importance of low-rank matrix gradient estimation parameters, and calculating the group importance of each structured parameter group according to the importance of the parameters; arranging the structured parameter groups in ascending order of group importance, pruning a preset number of parameter groups with the lowest importance, and performing non-uniform pruning on the remaining parameter groups; performing redundancy matching on the multi-modal large model, using a greedy algorithm to solve the optimal batch scheduling sequence, and performing cache and reuse of embedded representation of redundant nodes between adjacent batches according to the redundancy matching result; and performing efficient parameter adjustment after pruning. The application provides a feasible solution for efficient deployment of multi-modal large models in resource-constrained environments such as edge computing devices.

An elimination method for executing instructions that calculate constant values ​​based on load value prediction is disclosed. One aspect of the method involves predicting the value that a load instruction in the first iteration of a loop will load, and executing multiple instructions that occur after the load instruction in the first iteration of the loop to generate multiple results. Each of the multiple results depends only on the value, one or more constant values, values ​​derived from the value and / or one or more constant values, or any combination thereof. The method also includes generating multiple results for the multiple instructions during each iteration in one or more iterations following the first iteration of the loop without re-executing the multiple instructions. Other methods, apparatuses, and systems are also disclosed.

The invention discloses a data elimination method and device, equipment and a storage medium, is applied to a cache system, and relates to the technical field of cache management, and the data elimination method comprises the steps of determining a preset number of candidate data items when the usage amount of a cache space of the cache system reaches a preset threshold value; performing weighted calculation on the feature vector of each candidate data item through a preset weighted scoring model to obtain an elimination priority score; the feature vectors are vectors representing access frequency, timeliness, data size, access delay and relevance; the relevance is the ratio of the member number of the data cluster corresponding to the data item to the target number; the target number is the member number of the historical maximum data cluster; and executing a data elimination operation on the candidate data items corresponding to the elimination priority scores meeting a preset score condition. Therefore, the influence of the data cluster to which the data item belongs on the data item is considered, and data value evaluation is performed in combination with the multi-dimensional feature vector, so that the data elimination accuracy can be improved to match actual access requirements.

The invention discloses a DC component simulation elimination system and elimination method.The elimination system comprises a photoelectric detection module, an elimination module, a simulation comparison output module and a control module, controlling the first elimination digital-to-analog converter to sequentially traverse from the highest weight elimination gear to determine a first elimination gear combination; after a first elimination gear combination is determined, a transmitting element is turned on, a first elimination digital-to-analog converter is controlled to counteract an ambient light signal, and when a sleep period of a current sampling period is executed, a second elimination digital-to-analog converter is controlled to sequentially traverse from the highest weight elimination gear, determining a second elimination gear combination according to the comparison result of the current elimination gear to offset the direct current component in the reflected light signal; according to the application, the elimination gear combination can be dynamically determined step by step, the direct current component in the ambient light signal and the reflected light signal is effectively eliminated, and the accuracy and reliability of the detection result are ensured.