414 results about "Performance enhancement" patented technology

The invention relates to the technical field of industrial equipment control, and discloses an industrial equipment intelligent control system and method based on deep reinforcement learning. The system comprises an equipment state monitoring module, a reinforcement learning decision-making module, a control parameter adjustment module, an abnormity intervention module and an equipment performance optimization module. The equipment state monitoring module collects real-time operation data of equipment, analyzes state change by means of a deep reinforcement learning algorithm, and generates a state feature data set; the reinforcement learning decision-making module calculates a control strategy by using a deep reinforcement learning model and generates a control strategy parameter set; the control parameter adjusting module optimizes the operation efficiency according to the adjusted parameters and generates an optimized control parameter set; the abnormity intervention module monitors abnormity, recognizes intervention through deep reinforcement learning, and generates an abnormity intervention adjustment data set; and the equipment performance optimization module optimizes the performance indexes according to the parameters and generates a performance optimization parameter table. The system realizes accurate monitoring of equipment, intelligent optimization of strategies, timely intervention of abnormities and performance improvement, and meets intelligent management and control requirements of the equipment.

The invention provides an arch dam shape efficient intelligent optimization method and system based on an agent model. The method comprises the steps that an arch dam physical-numerical model is constructed, a double optimization target with structural safety and economical efficiency as the core is determined, design parameters are selected, a constraint function is set, and an evaluation index system is established; samples are generated through Latin hypercube sampling, a data set is constructed in combination with finite element calculation, and a multi-task learning architecture is adopted to train a high-precision agent model. And then, coupling the proxy model with a multi-objective optimization algorithm, quickly searching a Pareto optimal solution set, and screening out a comprehensive optimal figure by using a multi-attribute decision-making method. And finally, through a finite element simulation verification result, prediction precision and performance improvement are ensured. The method has the advantages of lightweight modeling, efficient prediction and accurate search, and provides an effective tool for intelligent optimization and rapid decision making of hydraulic structures such as arch dams and the like.

The invention belongs to the technical field of artificial intelligence, and relates to a fine-grained instruction following evaluation method and system based on a large language model. The method comprises the steps that (1) an instruction is collected from an actual application scene, a corresponding response is generated for the instruction, constraints in the instruction are decomposed, a constraint check list is generated, a plurality of comments are generated for each constraint, and each comment comprises explanation and judgment; 2) selecting an optimal comment for each constraint through a comment filtering mechanism, and forming a training data set based on the instruction, the corresponding response, the constraint check list and the optimal comment; 3) supervising and fine-tuning the large language model to obtain a supervised and fine-tuned large language model; 4) performing preference optimization on the supervised and fine-tuned large language model to obtain a preference-optimized large language model, and 5) performing fine-grained instruction following evaluation by using the preference-optimized large language model, so that greater performance improvement can be brought under lower calculation overhead, and the reliability of instruction following evaluation is improved.

The invention discloses an interpretable recommendation optimization system based on user feedback driving, and the system is characterized in that the system comprises an offline strategy reinforcement learning framework, a personalized reward prompt of target perception, multi-view interpretation optimization based on Pareto optimum, an overall target function, and a complete learning algorithm; by introducing user subjective evaluation and a multi-objective optimization strategy, the problems that recommendation interpretation quality is low and user requirements cannot be accurately reflected are solved. According to the method, the performance of the original basic model is remarkably improved in the aspects of recommendation quality and generation of the recommendation explanation text.

The invention discloses a formula optimization method and device of an insulating material, equipment and a storage medium, and relates to the technical field of chemical materials.The method comprises the steps that initial formula information corresponding to the current insulating material is obtained, and a physical constraint model is constructed according to the initial formula information; the physical constraint model is composed of a component compatibility constraint condition, a process feasibility constraint condition and a material performance threshold condition; generating an initial candidate formula according to a physical constraint model based on the initial formula information; performing iterative optimization on the initial candidate formula by adopting a gradient response optimization algorithm to obtain a current candidate formula; and performing physical property prediction on the current candidate formula, and determining a target optimization formula according to a physical property prediction result. The optimization process is based on a scientific model and algorithm, the limitation of traditional experience can be broken through, the relation between different performance indexes is balanced, the formula combination which may be ignored is explored, and a new possibility is provided for performance improvement of the insulating material.

The invention discloses a method and system for assisting weak supervision full slide image classification performance improvement in the technical field of computer vision and computational pathology, and the method comprises the steps: obtaining an image block instance set of WSI through a data preprocessing module, and obtaining an instance feature matrix through a pre-training feature extractor; learning features which are related to tasks but possibly contain deviation by using observation branches, and adaptively fusing observation representation and potential causal representation by intervention branches through a causal gate unit so as to correct false correlation; and combining the two-branch classification loss and cross-branch adversarial supervision loss joint optimization model, and finally outputting WSI classification prediction fusing double-branch results. According to the method and the system for assisting weak supervision full slide image classification performance improvement, shortcut learning can be effectively inhibited without explicitly annotating confounding factors, the generalization performance of a model on data outside distribution is remarkably improved, and the method and the system can be flexibly embedded into various mainstream MIL frames as a universal module.

The invention provides a GPU acceleration system and method based on medium sparseness large language model sparse matrix multiplication, and belongs to the technical field of computers. The system cuts a large model weight matrix through an unstructured cutting module to obtain a sparse weight matrix; a sparse matrix is converted into an MLSF format through a sparse tensor kernel layer, a slot filling layer and a residual element layer in the sparse matrix preprocessing module so as to minimize storage and decoding overhead; a sparse matrix multiplication GPU operation module is used, an MLSF format and a dense matrix are used as input, a fragmentation strategy and asynchronous loading are combined, a pipeline mechanism of data loading, decoding and calculation is constructed, and high overlapping of calculation and memory access operation is achieved. According to the method, dense matrix calculation with sparse matrix multiplication performance exceeding highly optimized performance is realized under medium sparseness, performance improvement of 5.29 times at most is obtained, and technical support is provided for efficient deployment of a large language model.

The invention discloses an imitated radio frequency division multiplexing signal detection method based on low-complexity optimization expected propagation, and belongs to the signal detection technology of multi-carrier communication. According to the method, information is updated through iteration to approach an ML result, and the complexity is reduced from an exponential level to a cubic level; in order to reduce the complexity of the EP, sparse and quasi-banded channel structures in the DAFT domain are utilized. Through matrix block decomposition, upper and lower triangular decomposition and upper triangular matrix forward substitution, low-complexity matrix inversion is successfully realized, and the complexity of the EP algorithm is reduced to a linear level. In addition, the EP algorithm performance is further improved through a deep learning optimization moment matching strategy, so that the process becomes more adaptive through trainable positive and negative component parameters. In addition, DAFT domain iteration detection and decoding can be achieved based on the method, external information of a decoder is fed back to a detector, and the reliability of the whole system is remarkably improved. According to the invention, effective balance between performance improvement and calculation complexity is realized.

The invention relates to the technical field of communication, particularly discloses a phased array transmitting and receiving circuit and a transmitting control method, and brings double advantages of system structure simplification and performance improvement through highly integrated design of three core links of baseband, frequency conversion and beam control. In the aspect of the structure, the physical size of the circuit is obviously reduced, and the hardware complexity and the manufacturing cost are reduced. In addition, in terms of performance, signal loss and delay are effectively suppressed by shortening an internal signal transmission path. Furthermore, the same signal processing unit uniformly outputs the baseband signal and the beam control signal, so that accurate and low-delay synchronization between data processing and beam pointing adjustment is realized, and the inherent time sequence error of a discrete scheme is avoided.

The embodiment of the invention discloses a thread priority adjustment method and related equipment which are applied to electronic equipment, and the electronic equipment runs a first application. A first process of the first application comprises a plurality of first threads; identifying an associated thread of the first thread from the threads of the second process; the second process is a process which is operated by the electronic equipment, except the first process and the resident process and provides service for the first application; when it is detected that the running state of the first application is changed, the priority of the first thread is adjusted, the priority of the associated thread is adjusted, and the priority adjustment trend of the associated thread is consistent with the priority adjustment trend of the first thread. By adopting the embodiment of the invention, the response speed and performance of the system can be improved, and the user experience is improved.

The invention provides an OTFS performance enhancement method based on adaptive fractional delay and Doppler index. The OTFS performance enhancement method comprises the following steps: establishing an OTFS-ISAC system model comprising M sub-carriers and N sub-symbols; the method comprises the following steps: based on modeling of an OTFS waveform in receiving and transmitting signals of a communication system and a sensing system, obtaining a symbol domain matrix; estimation of system delay and Doppler parameters is realized through an amplitude gravity center calibration algorithm: firstly, calculating an amplitude matrix of a symbol domain matrix of a receiving end; secondly, searching a peak value of the amplitude matrix as integer delay and Doppler index input; finally, further fractional order retrieval is carried out on the delay peak value and the Doppler peak value, and the delay parameter and the Doppler parameter are calibrated through comparison; under fractional order delay and Doppler parameter estimation, a system model for simulating system communication and sensing performance is established, and OTFS performance is enhanced. According to the method, the overall performance of the OTFS-ISAC system is improved, and the target detection precision is improved through more accurate parameter estimation; and the communication error rate performance is optimized.

The invention discloses a blast furnace smelting safety reinforcement learning decision optimization method based on expert strategy constraint. According to the method, an optimization strategy can be learned from an off-line expert track on the premise that operation safety is ensured. Specifically, a conditional generative adversarial mechanism is introduced to realize the alignment of strategy distribution and expert decision, and the exploration ability in a security domain is retained while the expert experience is inherited. And an independent state-action safety evaluation network is designed and a discount factor is introduced, so that the long-term accumulation risk caused by the large hysteresis characteristic of the blast furnace is effectively dealt with. In addition, memory is adopted to enhance network coding historical information, incomplete state observation is supplemented, and information deviation in decision is reduced. According to the method, effective integration of triple guidance mechanisms is realized, knowledge inheritance is realized through distributed alignment, performance improvement is driven through reward optimization, and risk prevention and control are ensured through explicit security constraints.

The invention belongs to the technical field of software engineering, particularly relates to an automatic driving system performance enhancement method based on an expert hybrid architecture, and aims to solve the core problems that an end-to-end automatic driving system is confronted with semantic fuzziness to cause unreliable decision, multi-task interference hinders optimization planning, too long reasoning delay increases driving risks and the like. According to the method, an ExpertAD framework is provided, task key features are amplified through a perception adapter (PA), and the relevance of scene context understanding is guaranteed; related driving tasks are dynamically activated through a sparse expert mixture (MoSE), and task interference is minimized; and in combination with a customized training loss function, collaborative optimization of planning effectiveness and reasoning efficiency is realized. Experiments show that compared with an existing method, the method has the advantages that the average collision rate is reduced by 20%, the reasoning delay is reduced by 25%, higher multi-skill planning capacity is achieved in rare scenes (such as accident handling and first-aid vehicle avoiding), and good generalization is achieved for unseen urban environments.

The invention relates to a microprism reflective film with three kinds of triangular pyramid combinations and a manufacturing method of a mold of the microprism reflective film, the surface of the microprism reflective film is formed by an array of identical parallelogram unit bodies (1), each parallelogram unit body (1) is composed of four identical triangular pyramids A, one triangular pyramid B and three identical triangular pyramids C, the directions of the four triangular pyramids A and the directions of the three triangular pyramids C and the triangular pyramids B form 180 degrees, and the three triangular pyramids C and the triangular pyramids B are arranged among the four triangular pyramids A in a staggered manner; the bottom surfaces of the triangular pyramids A, B and C are the same equilateral triangles, the inclination angles of the three pyramids of each triangular pyramid A are alpha / 2, the inclination angles of the three pyramids of each triangular pyramid B are beta-alpha / 2, and the inclination angles of the three pyramids of each triangular pyramid C are alpha / 2, alpha / 2 and beta-alpha / 2 respectively. According to the method, the performance under different application conditions can be reasonably optimized, so that wide-angle performance improvement and optimization and balance of overall comprehensive performance can be realized.

The invention provides a remote sensing large model performance improvement method and device based on composite visual coding, and relates to the technical field of remote sensing image large language models. According to the method, a network structure is rewritten on the basis of an open-source visual universal language model, a composite visual coding ConvToMe layer used for extracting local details and capturing global semantics is added, the ConvToMe layer extracts the local details through a ConvNeXt layer, the global semantics are captured in combination with a ToMe ViT layer, a remote sensing large model with improved performance is obtained, and the performance of the remote sensing large model is improved. Through composite visual coding, the global context and local details are reserved at the same time, the robustness is improved, and the model precision is improved at the same time.

The invention provides an antenna control method, electronic equipment and a computer readable storage medium. In the communication process of the electronic equipment and other equipment, the electronic equipment can continuously monitor the posture of the electronic equipment and control the antenna based on the posture of the electronic equipment, so that the situation that the natural performance is reduced after the posture of holding the electronic equipment by a user is changed is avoided. According to the method, the communication performance of the antenna is improved, and the call efficiency is improved.

The invention discloses a method for improving the performance of double-independent quantum key distribution of actual reference system measurement equipment, belongs to the technical field of quantum key distribution (QKD), and aims to improve the performance of the double-independent quantum key distribution when non-ideal factors such as a post-pulse effect of a detector and finite sample statistical fluctuation are considered. The invention relates to a method and a system for improving the performance of a reference system measurement device by introducing an advantage extraction technology into a double independent quantum key distribution (RFI-MDI-QKD) protocol of the reference system measurement device. Simulation results show that when the post-pulse effect and the statistical fluctuation effect are considered, the advantage extraction method can effectively improve the security key rate and the security transmission distance of the RFI-MDI-QKD protocol on the premise of not changing optical hardware. According to the invention, a valuable reference technology can be provided for practical research of the RFI-MDI-QKD system.

The invention relates to the technical field of buck-boost switching control, and discloses a buck-boost seamless switching MPPT control method which has remarkable advantages. The single-side boost-buck function can stably control the output voltage in a full-voltage range, and the problem that the efficiency and stability of the system are affected due to the fact that the traditional topology cannot accurately regulate and control the power under the photovoltaic high-voltage working condition is effectively solved. A buck-boost mode is seamlessly switched, the response speed and reliability are improved, and faults caused by improper mode switching are avoided; when the method is applied to an energy storage product, the PV high voltage condition does not exist, the additional control logic can be reduced, the system complexity is reduced, the reliability is improved, meanwhile, the development cost and the maintenance difficulty are reduced, and powerful support is provided for photovoltaic system performance improvement and energy storage product optimization.

The invention provides a SiC controller electromagnetic-thermal-force full-coupling simulation design method and system, and belongs to the technical field of power electronic device simulation design, and the method comprises the steps: constructing a bidirectional real-time coupling model of an electromagnetic field, a thermal field and a mechanical stress field; the irradiation dose is determined based on the expected working environment of the SiC controller, irradiation damage factors are introduced to correct SiC material parameters, and a bidirectional real-time coupling model is input; different temperature conditions and stress conditions are input into the bidirectional real-time coupling model for multi-working-condition simulation, performance parameters of the SiC material are output, and a quantitative relation model of the performance parameters, the temperature and the stress is established to reveal the performance degradation law of the SiC material; and based on the bidirectional real-time coupling model and the performance degradation law of the SiC material, taking the switching loss and the temperature rise as optimization targets, and carrying out collaborative optimization on parameters of the SiC controller. The simulation precision is improved, the performance of the SiC controller is optimized, the reliability is improved, and the service life is prolonged.

The invention discloses a composite cutter bar design parameter optimization method. The method comprises the following steps: constructing a low-fidelity data set, a high-fidelity data set and a dual-stage deep learning model; in the first stage, low-fidelity features are learned based on an LSTM network, and a cutter bar performance prediction value is output; in the second stage, high-fidelity data and a low-fidelity prediction result are input for residual learning, and a comprehensive prediction model is obtained. Performing global optimization in the cutter bar design parameter space through a particle swarm optimization algorithm to obtain optimized design parameters; performing experimental verification to obtain actually-measured dynamic performance parameters and errors between the calculated predicted values and the actually-measured dynamic performance parameters; when the error does not meet the preset threshold value, according to the error distribution extension parameter value interval, new high-fidelity data is generated, then the comprehensive prediction model is updated, optimization verification is repeated, and final cutter bar design parameters are output. According to the method, efficient optimization and dynamic performance improvement of cutter bar design parameters are achieved.

The invention provides an ultra-thin wing anti-bending performance quantitative evaluation method which comprises the following steps: acquiring geometric parameters and material characteristic data of an ultra-thin wing, constructing a digital twin model by adopting a finite element analysis method, and simulating stress distribution under boundary conditions to obtain initial mechanical response characteristics; the obtained risk assessment matrix is adopted to calibrate the limitation of a static test, and boundary condition constraints are fused to obtain a calibrated anti-bending energy index set; constructing an evaluation framework of a quantitative system through the calibrated anti-bending energy index set, classifying performance levels in different load scenes by adopting a support vector machine algorithm, and determining an overall structure reliability score; and iteratively updating the parameter setting of the digital twin model according to the obtained anti-bending performance improvement scheme, simulating the optimized flight performance, and obtaining the final quantitative evaluation system output.

The invention discloses a transient performance enhancement-oriented multi-spacecraft attitude cooperative tracking control method, which provides a fixed time controller through an inversion control strategy, realizes quick response of a multi-spacecraft attitude system, and ensures that a consistency error is converged to a sufficiently small domain within fixed time. A fuzzy weight updating law is designed, dependence on system model information is reduced, and adaptability and high robustness are achieved. According to the asymmetric fractional obstacle Lyapunov function, adherence of multiple spacecrafts to time-varying constraints is guaranteed, meanwhile, the problem that the control amplitude related to a large constraint boundary in a traditional logarithmic obstacle Lyapunov function is too large is solved, and therefore the proposed method can process constrained and unconstrained scenes. A saturation threshold event triggering strategy is designed, a multi-spacecraft system is allowed to select three different triggering strategies by adjusting parameters, the transmission frequency of control signals is flexibly changed, and control bus resources are reasonably distributed.

The invention relates to the technical field of artificial intelligence chips, and provides a data processing method, electronic equipment, a storage medium and a program product, and the method comprises the steps: obtaining input data stored in a first data format; in the process of loading the input data to the main cache, the input data is converted into a second data format from the first data format, and the second data format is a hardware dedicated data format supported by the tensor core; based on the tensor core, using the data stored in the main cache in the second data format to execute matrix multiplication to generate result data; and converting the result data from the second data format to the first data format, and outputting the converted result data. According to the method, an automatic format conversion mechanism is introduced in a key link of data circulation, a channel between an upper-layer application and bottom-layer tensor core hardware is successfully opened, on the premise that usability is not sacrificed, effective utilization of a tensor core is achieved, and remarkable performance improvement is brought.

The invention relates to the technical field of extremely-small curved surface structures, in particular to an optimal design method, system and equipment for an extremely-small curved surface structure and a storage medium. The minimum curved surface structure optimization design method comprises the steps that a multi-objective optimization problem is determined according to design variables and optimization objectives of a three-period minimum curved surface lattice structure; constructing an approximate function relationship between the optimization target and the design variable by utilizing a self-adaptive multi-target engineering optimization method of a Kriging agent model; latin hypercube sampling and a sequential quadratic programming optimization algorithm are combined to carry out optimization design on the three-period minimal curved surface lattice structure. According to the design method, the light / bearing / wave-absorbing minimum curved surface lattice sandwich structure with good structure-performance dual characteristics can be designed, the wave-absorbing structure is filled with the minimum curved surface lattice structure in a shape follow-up mode, design and manufacturing of a wave-absorbing and bearing integrated structure are achieved, and rapid iteration and performance improvement of design of the wave-absorbing and bearing structure are supported.

The invention relates to the technical field of Internet of Things communication, in particular to a LoRa network dynamic link self-adaptive control method and system based on environment awareness, and the method comprises the steps that the system firstly presets initial communication parameters for a terminal and collects the data and link information of the terminal; based on the building and traffic data on the transmission path, an environmental perception model is constructed to quantify environmental impact factors, and a dynamic evaluation matrix is established in combination with performance indexes. Calculating the comprehensive performance index of the current link based on the matrix, and predicting the change of the signal-to-noise ratio to obtain a relative difference value; the system generates optimization suggestions of communication parameters by analyzing a performance index change trend and applying an adaptive algorithm, and predicts brought performance improvement. By balancing energy consumption and time delay cost required by performance improvement and adjustment, parameter adjustment is executed only when the income is greater than the overhead. And finally, completing the optimization of all terminals within the fixed network distribution time, and evaluating the overall performance of the network. According to the invention, the accuracy of dynamic link adaptive control of the LoRa network is improved.

The embodiment of the invention discloses a floating-point number processing method and device, equipment, a storage medium and a program product. The floating-point number processing method comprises the steps that a first floating-point number and a second floating-point number which need to be processed currently are obtained; the data formats of the first floating-point number and the second floating-point number are first data formats; performing format conversion on the first floating-point number and the second floating-point number based on the loop indication, and determining a first operation result after the first floating-point number and the second floating-point number are multiplied based on the converted first floating-point number and the converted second floating-point number; the loop indication is used for indicating the tensor calculation engine to execute a processing mode of format conversion and the number of loops; the converted first floating-point number and the converted second floating-point number are in a second data format, and the second data format is the simplification of the first data format. Therefore, the structure of the tensor calculation engine does not need to be changed, and 2.66 times of performance improvement can be obtained only by fewer hardware resources.

An X-ray detector module of a multi-module X-ray detector array having a local processing unit configured to provide real-time control and computational capabilities at the x-ray detector module. In various embodiments, by providing these capabilities in the detector module itself, as opposed to a component located downstream of the detector module, there may be an immediate cost reduction, opportunities to include additional features, and performance improvement opportunities not otherwise practical to implement in conventional X-ray imaging systems.

The invention relates to a dynamic power control method and system for a storage and charging all-in-one machine, and the method comprises the following steps: data collection and monitoring: collecting and analyzing key parameters in real time through multi-source data fusion; dynamic safety threshold calculation: dynamically calculating a safety operation threshold of each subsystem based on real-time data, and ensuring that the system operates in an optimal state; multi-objective optimization control is carried out, an intelligent algorithm is adopted, and an optimal balance point is found among battery safety, power grid stability and charging efficiency; according to a fault processing and redundancy mechanism, a multi-layer fault response system is established, and the safety and continuity of the system under the abnormal working condition are ensured; according to the energy efficiency management and strategy optimization, long-term performance improvement and economical efficiency optimization are achieved based on a data closed loop, and the problem that potential safety hazards are caused due to the fact that a traditional storage and charging all-in-one machine system neglects the common influence of multiple factors such as SOC, temperature and power grid loads is solved.

The invention belongs to the field of artificial intelligence and machine learning, aims to solve the problems of high manual generation cost, poor authenticity, low negative sample quality and insufficient scene coverage of multi-modal tool calling data, and comprises four core modules: a tool knowledge base module stores API metadata through PostgreSQL, and generates various user intentions and coherent calling sequences in combination with pre-training LLM; the multi-agent interaction module generates conversation and calling data containing multi-modal input through cooperation of a user, an assistant and an environment agent; the adversarial environment simulation module generates high-quality negative samples in batches by means of three adversarial strategies and automatically marks the high-quality negative samples; and the human feedback iteration module determines convergence by a performance improvement rate according to a model failure case optimization strategy. The system supports local / cloud deployment, adapts to LLM and Agent model training, and is suitable for training and robustness evaluation of dialogue type AI assistants and vertical field intelligent Agents.

A mining method performed by node devices connected to a blockchain network, comprises: a step in which a specific node device acquires a transaction including a first block having a first block number and a first trie node related to specific state data of the blockchain network, and state change data for changing the state of a specific account; a step in which the specific node device generates a second block including a second trie node and a second block number, the second trie node being generated based on the transaction and the first trie node; and a step in which, if the hash value of a second node corresponding to a state path-being a path from a specific leaf node corresponding to the specific account to a specific root node-among one or more second nodes included in the second trie node satisfies a predetermined criterion.