35217results about How to "Improve stability" patented technology

The invention provides a construction method of a prefabricated multi-ribbed cavity floor, and belongs to the technical field of prefabricated multi-ribbed plate construction. The construction method of the prefabricated multi-ribbed cavity floor comprises the steps of full-hall scaffold building, main keel laying, overall leveling and the like. The construction method comprises the steps that S1, full-hall scaffold building is conducted, specifically, when a full-hall scaffold is built, vertical rods on the upper layer and the lower layer are required to be aligned, the vertical rods are uniformly arranged from the two sides of a cast-in-place ribbed beam, the vertical rods are arranged at the intersection positions of floor elastic bidirectional lines, and scaffold boards are laid under the vertical rods; S2, main keels are laid, specifically, when the main keels are installed, the main keels are required to be arranged in order and not to be inclined, and the two vertical rods at the intersection positions must abut against the main keels; and S3, overall leveling is conducted, specifically, after the main keels are laid, overall leveling is conducted according to control lines, and full-hall scaffold formworks are hoisted on the main keels. The cavity formworks are supported by placing the modularized cavity formworks which are machined and formed in a factory-like mode, and accurate and rapid installation of an upper plate of the prefabricated multi-ribbed plate is achieved by installing limiting square pipes.

An aqueous ultraviolet-absorbing resin emulsion, and an aqueous resin emulsion composition containing the emulsion is provided, wherein the emulsion is excellent in compatibility with a synthetic resin and light resistance. Also, the emulsion improves stability in blending with an aqueous emulsion of the synthetic resin, and prevents bleed out of the ultraviolet-absorbing resin on the surface of a coating film. Further, the aqueous ultraviolet-absorbing resin emulsion imparts excellent alkali or solvent resistance to a synthetic resin poor in alkali or solvent resistance. More specifically, an aqueous resin emulsion composition comprising (I) an aqueous ultraviolet-absorbing resin emulsion and (II) an other aqueous resin emulsion is provided, wherein the emulsion (I) is obtained by urethanizing (A) a polyol component having an ultraviolet-absorbing group {e.g., 1,1-bis[3-(2H-benzotriazol-2-yl)-4-hydroxy-benzeneethanol]methane}, (B) a polyol if necessary, (C) an alkyl or aryl dialkanolamine compound, and (D) an organic polyisocyanate in (E) an organic solvent, diluting the reaction mixture with (F) an organic solvent having a boiling point lower than 100° C. to give a resin solution, neutralizing the resin solution with (G) a neutralizing agent, and dispersing the resultant in water.

The invention discloses a production method of distiller's grain fermentation liquid fertilizer specially applied to roxburgh rose. The distiller's grain fermentation liquid fertilizer is prepared bythe step of evenly mixing 50 to 80 parts of distiller's grain fermentation extracting solution, 0.5 to 1 part of microelement, 5 to 10 parts of seaweed meal, 5 to 10 parts of natural active organism,2 to 5 parts of potassium bicarbonate, 2 to 5 parts of natural active organism and 2 to 5 parts of natural surface active agent. A preparation method of the distiller's grain fermentation extracting solution comprises the steps: evenly stirring 50 to 80 parts of distiller's grain, 50 to 150 parts of rice washing water, 1 to 2 parts of EM microflora fermentation liquid and 0.5 to 1 part of sugar, performing composting fermentation at the room temperature to obtain a distiller's grain fermentation material, stirring and extracting through a sodium carbonate water solution, filtering and performing evaporation and condensation on obtained filtrate until the density is 1.1 to 1.2 to obtain distiller's grain fermentation extracting solution. The distiller's grain fermentation liquid fertilizerdisclosed by the invention has a good use effect, meanwhile can effectively prevent pest and disease damage to the roxburgh rose and is high-efficiency environment-friendly liquid fertilizer.

The invention provides an offshore wind plant current collection network topology optimization method based on an improved grey wolf optimization algorithm, and the method comprises the steps: S1, obtaining basic data of an offshore wind plant, initializing the grey wolf algorithm, and generating an initial solution set; s2, adopting polar coordinates and a minimum spanning tree to optimize and expand the initial solution set; s3, optimizing a connection mode in each group; s4, executing the optimal connection of each group of fan combination; s5, performing optimal solution set exploration on the grey wolf algorithm by adopting an operator; s6, repeatedly executing until a preset maximum number of iterations is reached, and generating a topology model of the current collection system of the offshore wind plant; the method can improve the initial solution quality and the convergence efficiency, gives consideration to the global search capability and the local optimization precision, and achieves the quick and effective processing of geometric constraints.

The invention provides a hydroxytyrosol supramolecular compound and a preparation method and application thereof, and belongs to the technical field of cosmetics and biological medicine, the supramolecular compound comprises the following chemical components by mass: 0.01-10% of hydroxytyrosol, 2-45% of an amino acid compound, 2-20% of alpha hydroxy acid, 2-30% of urea, and 15-92% of water. The supramolecular compound can promote transdermal delivery of hydroxytyrosol and maintain the biological activity of hydroxytyrosol, shows high biocompatibility, and can be applied to product development in the antibacterial field, the regeneration field and the anti-aging field.

The invention relates to a gas nano purifier. The gas nano purifier comprises a mounting seat, the gas storage bin is installed on the installation base and used for storing clean gas, the side wall of the upper portion of the gas storage bin extends upwards to form an installation plate, and a gas path through hole is formed in the middle of the installation plate; the gas-water separator is mounted on the mounting plate and is used for carrying out primary purification on to-be-cleaned gas; the adsorption tower is mounted on the mounting plate and is used for carrying out secondary purification on the to-be-cleaned gas; gas to be cleaned sequentially flows through the gas-water separator and the adsorption tower and then enters the gas storage bin; the adsorption tower is communicated with the gas storage bin through an exhaust pipe; and the exhaust pipe penetrates through the gas path through hole. Through the arrangement, the installation operation is greatly simplified, and the integrity and the space utilization rate of equipment are improved.

The invention provides an injection mold deformation compensation control method and device based on real-time monitoring, and relates to the technical field of injection molding control. The method comprises the steps that synchronous collection is conducted through a multi-source sensor according to an injection molding period; performing coupling analysis on the real-time monitoring data set, and extracting dynamic deformation features to perform attribution identification; a deformation prediction model is constructed, and prediction is conducted according to the injection molding period; reverse calculation is carried out based on the predicted deformation result, and a mold correction instruction is activated to drive an execution mechanism to carry out online compensation; and carrying out injection molding verification based on the deformation compensation effect, and carrying out iterative optimization on the deformation prediction model according to a verification result. According to the method, the technical problem that in the prior art, due to the fact that static process parameters are not matched with dynamic process disturbance, thermodynamic deformation of the mold cannot be sensed and compensated in real time, and the stability of product quality is further affected is solved, the accuracy of deformation prediction and compensation is improved, and therefore the product quality is improved.

According to the privacy enhancement type (k, d)-truss community search method for the social network, a novel KTG tree structure is constructed by fusing k-truss and G-tree indexes. According to the index structure, hierarchical community decomposition of a social graph and social distance information are fused, a refined boundary vector coding scheme is designed to support efficient distance calculation, and a double-cloud-server non-collusion architecture integrating improved homomorphic encryption and matrix encryption technologies is constructed. Through a two-stage security query process of first structure filtering and then distance verification, on the premise of protecting full-process privacy of a graph structure, a query intention, a distance matrix and an intermediate calculation result, efficient and accurate search of a close community in a large-scale social network is realized. The method is suitable for various scenes such as social recommendation, risk control, public opinion analysis and anti-fraud, and the problems of privacy disclosure and calculation efficiency in social network analysis are effectively solved.

The invention discloses an automatic causal structure generation method based on semantic representation and logical reasoning of a large language model. The method comprises the following steps: acquiring an input text; performing semantic coding and clustering on the obtained input text by utilizing a large language model, and establishing a candidate causal variable set; causal relationship detection is carried out on the established candidate causal variable set based on anti-fact intervention and do-calculation; performing causal direction judgment, and generating a directed acyclic causal graph meeting logic consistency; and on the basis of the generated directed acyclic causal graph, natural language interpretation is generated by using a large language model, and logic consistency closed-loop verification is carried out. According to the method, automatic generation from the natural language to the causal structure is realized, the causal variable set is automatically extracted and constructed from the unstructured natural language text, the defects that variables need to be manually defined and modeling depends on field experts in the existing causal modeling process are avoided, and the labor cost and professional threshold of causal structure construction are remarkably reduced.

The invention provides a preparation method of a laminated cell, the laminated cell and a photovoltaic module, and can be widely applied to the photovoltaic technical field, the preparation method of the laminated cell comprises the following steps: providing a bottom cell, and forming a composite layer on the bottom cell; a first transmission layer is formed on the composite layer, and a skeleton layer is formed on the first transmission layer through double-source evaporation; wherein a first evaporation source of the double-source evaporation comprises inorganic cesium salt, and a second evaporation source of the double-source evaporation comprises inorganic lead salt and inorganic rubidium salt; coating organic salt on the framework layer to form a perovskite layer; forming a second transmission layer and an electrode on the perovskite layer to obtain a laminated cell; wherein carriers of the first transmission layer and the second transmission layer are different. The framework layer is formed through double-source evaporation, the proportion is better controlled, through rubidium and lead, a positive effect can be generated on crystallization, stress release and the like of perovskite in the crystallization process, part of rubidium salt can be combined with PbI2 decomposed on the surface, ion migration is reduced, and the device stability is improved.

The invention relates to the technical field of deep foundation pit monitoring, and discloses a deep foundation pit micro-deformation InSAR remote sensing method and system and a storage medium. The method comprises the following steps: synchronously acquiring data through a ground radar and an InSAR satellite, and obtaining a point target set and an original scattering characteristic sequence; obtaining a net deformation phase sequence without influence through registration residual correction, phase deduction and weighted fusion processing, and carrying out phase consistency quality control and unwrapping processing to obtain an accumulated deformation quantity sequence; constructing a deformation mechanism correlation characteristic matrix based on correlation analysis of the cumulative deformation quantity sequence and radar reflection intensity amplitude attenuation characteristics; and fusing the deformation mechanism correlation feature matrix and the real-time differential interference image group, extracting deformation rate and phase gradient change, and further generating a deep foundation pit local micro-deformation risk early warning index set. According to the invention, the tiny deformation of the surrounding area of the foundation pit can be effectively detected, potential risks are warned in advance, and the safety and monitoring efficiency in the construction process are improved.

The invention discloses a dynamic obstacle avoidance method and system for a substation inspection robot. The position of the robot is positioned based on laser point cloud data, and static and dynamic obstacle features are extracted at the same time. An environment association graph is constructed through perception information in the advancing process, and feature information of each node serves as an attribute of the environment association graph; calculating the distance, the relative speed and the relative angle between the edge weights of the nodes according to the nodes; on the basis of the environment association graph, aggregating all-graph obstacle node information to generate real-time feature representation for robot nodes, and correcting a calculation mode of attention weight according to edge weight between the nodes; and finally outputting a real-time feature vector of the robot node. And defining a reinforcement learning framework and executing path planning. According to the invention, through combination of GAT and dynamic planning, efficient obstacle avoidance and path planning of the robot in a substation environment are realized, and the decision and execution capability of the robot in a dynamic complex scene can be improved.

The invention relates to the technical field of drilling parameter optimization control, and provides a petroleum drilling parameter optimization method and system and a storage medium. Comprising the steps that torque fluctuation amplitude and bit pressure response delay data are collected in real time, an original data sequence is constructed, and parameter trend data are obtained; analyzing the parameter trend data to identify a stratum change mode, extracting trend feature information, and when the trend feature information meets a preset condition, determining a corresponding abnormal mode identifier; generating a parameter adjustment direction of drilling control according to the abnormal mode identifier, and optimizing the parameter adjustment direction in combination with parameter weight analysis and a real-time feedback mechanism to obtain final optimization guidance data for guiding drilling operation; and mapping the final optimization guidance data to a control interface, and generating a dynamic adjustment instruction for drilling operation. The problems that in the existing petroleum drilling process, parameter adjustment lags behind, and stratum changes are difficult to adapt are solved, and intelligent optimization and dynamic control over drilling parameters are achieved.

The invention relates to a fiber-reinforced aerogel composite material containing a carbon and titanium dioxide biphase hybrid opacifying agent as well as a preparation method and application of the fiber-reinforced aerogel composite material. The fiber reinforced aerogel composite material containing the carbon and titanium dioxide dual-phase hybrid opacifying agent comprises an inorganic aerogel matrix, wherein a reinforced fiber skeleton and the carbon / titanium dioxide dual-phase hybrid opacifying agent are dispersed in the inorganic aerogel matrix; the carbon / titanium dioxide biphase hybrid opacifying agent comprises a titanium dioxide shell layer and carbon / titanium dioxide core-shell structure particles, wherein the surface of a reinforced fiber skeleton is coated with the titanium dioxide shell layer in situ; gaps of the reinforced fiber skeleton are filled with the carbon / titanium dioxide core-shell structure particles. The problem to be solved is to construct a two-phase hybrid shading system in which carbon / titanium dioxide core-shell particles and a titanium dioxide coating layer on the fiber surface are combined through one-step in-situ synthesis, so that the heat insulation capacity of the composite material in a high-temperature environment is synergistically improved.

The invention relates to a self-adaptive firewall rule generation method and device based on real-time threat intelligence, and the method comprises the steps: carrying out the semantic analysis of a threat intelligence text, extracting attack indexes and attack technique and tactics labels, and generating an intelligence triple set and an attack technique and tactics label set; calling a language model to generate candidate rules based on an intelligence analysis result, carrying out credibility scoring, intercepting a low-scoring rule as a high-risk illusion rule, and entering other rules into a to-be-verified set; constructing a digital twinborn shadow sub-domain to perform real-time traffic verification on the rule set to be verified, and dividing the rule set to be verified into an effective rule set and an invalid rule set according to a verification result; constructing a rule dependence graph based on the invalid rule set to perform pollution traceability, and adding nodes with excessive pollution into an isolation list; and cleaning the historical training samples based on the isolation list and the interception rule log, and performing incremental training on the language model to generate an updated model.

The invention discloses a permanent magnet synchronous motor sensorless control method based on a dynamic position error. A transition stage is provided, per-unit estimation position errors between a first position error signal obtained based on a high-frequency signal injection method and a second position error signal obtained based on a sliding mode observer are compared, an optimal switching speed point is dynamically determined, smooth fusion is carried out on the two position errors by using a nonlinear weighting function, and the optimal switching speed point is obtained. And finally, the estimated position and rotating speed are obtained through a phase-locked loop. And finally, position tracking is carried out by adopting different error signals according to the estimated rotating speed. According to the method, the problem of estimation value pulsation and jump caused by the fact that a traditional method depends on fixed experience switching points is solved, smooth and self-adaptive switching of the two estimation methods in the full-speed domain range is achieved, the control precision and operation stability of the system are remarkably improved, the structure is simple, and engineering implementation is easy.

This invention discloses a double-layer fully automatic opening and closing roller beam device for subway rail transportation, including an upper roller beam, a lower roller beam, a drive column, and side columns corresponding to the drive column. The horizontally transverse upper roller beam is rotatably mounted on the drive column at one end and rests on the side column at the other end. It also includes an automatic control mechanism and a roller beam rotation support mechanism on the same longitudinal line as the drive column. The automatic control mechanism includes a control host and several position sensors installed on the side columns and the roller beam rotation support mechanism to transmit the position information of the upper roller beam. The position sensors feed back signals to the control host to automatically control the opening and closing of the upper roller beam. It is installed on an adjustable height support beam, which includes a support beam, crossbeam support feet, and a crossbeam height adjustment mechanism. It has the advantages of fully automatic opening and closing of the roller beam, simple operation, time and labor saving, high personal safety factor, and high overall structural stability.

The invention discloses an oligopeptide from undaria pinnatifida and application thereof. The amino acid sequence of the oligopeptide is YRKD, and the molecular weight is 580.63 g / mol. The polypeptide is identified from undaria pinnatifida through a high performance liquid chromatography technology for the first time, and an unreported oligopeptide capable of inhibiting tyrosinase and cooperating with vitamin C to synergistically inhibit tyrosinase is screened out. The oligopeptide derived from undaria pinnatifida is synthesized by a solid phase method, has the characteristics of safety, no toxicity, good water solubility and good stability, and is combined with vitamin C to generate a synergistic effect and synergistically increase the inhibition effect on tyrosinase, so that the oligopeptide can achieve a better tyrosinase inhibition effect at a lower and more stable concentration; the use concentration of the oligopeptide can be greatly reduced, the stability is enhanced, and the oligopeptide can be widely applied to preparation of cosmetics, medicines, health-care products or food additives, and has important economic value and social significance.

The invention provides a large language model generation code detection method and system, which is applied to the technical field of artificial intelligence, and comprises the following steps: obtaining a to-be-detected code; a to-be-detected code is input to a trained shared encoder, a code feature vector is obtained, the shared encoder is obtained through multi-target joint training, and the multi-target joint training is used for optimizing classification loss, domain confrontation loss, comparison loss and difficult sample loss at the same time; l2 normalization is carried out on the code feature vector, and the code feature vector is mapped to a hyperspherical space to obtain spherical embedding; the sphere is embedded and input into a sphere category classifier based on sphere logistic regression for classification processing, a detection result of the to-be-detected code output by the sphere category classifier is obtained, the detection result comprises AI generation and human writing, and a decision boundary of the sphere category classifier is an intersection line of a hyperplane and a hypersphere for classification. According to the invention, the AI generation code and the human compiled code can be accurately distinguished.

The invention provides an excitation amplitude automatic closed-loop control system and method for a DLCP test. The system comprises a multi-frequency alternating current excitation and direct current bias voltage applying module which is used for automatically applying alternating current excitation signals and direct current bias voltage signals according to a preset excitation amplitude sequence under a plurality of test frequencies; the PID closed-loop control module is used for performing closed-loop correction on the applied voltage according to the response of the test port of the tested semiconductor device, so that the condition of the applied voltage at the test port meets a preset requirement; the capacitance measurement module is used for acquiring an equivalent capacitance response signal of the tested semiconductor device; the capacitance test data caching module is used for performing associated marking and classified caching on the capacitance test data according to the test frequency, the excitation level and the bias voltage set point; and the main control module is used for coordinating automatic test processes under the conditions of a plurality of test frequencies and different excitation amplitudes and outputting corresponding capacitance test data so as to improve the consistency, stability and automation degree of the DLCP test.

The invention discloses a reinforcement learning-based casting cleaning process parameter adaptive control method, which comprises the following steps of: constructing a reinforcement learning model for updating casting cleaning process parameters; the method comprises the steps that in the process that a casting cleaning system cleans a casting according to a set control period, key process signals in each control period are sequentially collected, and observation vectors are generated; inputting the obtained observation vector into a reinforcement learning model, and processing the observation vector by the reinforcement learning model to obtain a process parameter adjusting quantity, and the casting cleaning process parameters of the current control period are updated according to the process parameter adjusting quantity to obtain updated casting cleaning process parameters, and the updated casting cleaning process parameters are adopted to control a casting cleaning system of the next control period. According to the casting cleaning process parameter self-adaptive control method based on reinforcement learning, continuous and self-adaptive adjustment of the process parameters can be achieved, and the problem that fixed parameter control is insufficient in adaptability under complex working conditions is solved.

A high-order driving auxiliary lane changing trajectory planning method for an intelligent vehicle comprises the steps that laser radar and camera data are obtained, and feature vectors of obstacles are extracted after the data are processed by an ISP at the front end of a chip of the intelligent vehicle; constructing a spatio-temporal joint attention network model, inputting the feature vector of the obstacle into the spatio-temporal joint attention network model, calculating an interaction weight between the vehicle and the obstacle through a self-attention mechanism, and synchronously outputting predicted trajectory distribution of the obstacle and a candidate planning trajectory cluster of the vehicle; introducing a dynamic game in the game theory, and constructing a comprehensive cost function in combination with the interaction weight to obtain a screened track; and taking the screened trajectory as an initial value, inputting the initial value into a rear-end quadratic programming optimizer on the premise of meeting vehicle dynamics constraints, carrying out numerical optimization solution, searching an optimal programming sequence, and finally finishing lane changing trajectory planning. According to the invention, the interactive perception capability and riding comfort of the intelligent vehicle in a complex traffic scene can be improved.

The invention discloses a dynamic fusion prediction method and system for photovoltaic power generation power, and a storage medium. The method comprises the steps of collecting historical photovoltaic power generation power and corresponding multi-dimensional meteorological parameters; based on time sequence fragments in the historical data set, similarity measurement and clustering analysis are carried out on hour-level weather-power time sequence fragments in historical data by utilizing a dynamic time warping method, different weather-power mode clusters are divided, and a representative cluster center is determined for each cluster; training a long time sequence baseline prediction model and a plurality of short time sequence correction models in parallel; performing weighted fusion on the two prediction results by using the weight to obtain a final prediction value; and the prediction is updated by using the latest meteorological data by adopting an hour-by-hour rolling mechanism. According to the method, the problem that the long-period trend and the short-time fluctuation are difficult to consider at the same time is effectively solved, and the prediction precision, the stability and the adaptive capacity of the photovoltaic power generation power under the complex and changeable meteorological conditions are remarkably improved.

The invention provides a firmware online upgrading method, embedded equipment, a medium and a product, and relates to the technical field of intelligent terminals. The method comprises the following steps: after the embedded equipment is powered on, running a primary bootstrap program, and checking a first mark state: if the state is a state of updating a secondary bootstrap program, executing online updating of the program and resetting a system; and if the state is the default state, loading and running the secondary bootstrap program. Checking a second mark state; if the second mark state is an updating mirror image state, executing firmware mirror image online updating and resetting; and if the state is the default state, determining a target mirror image, copying the target mirror image from the nonvolatile memory to the random access memory, and skipping to run. And the equipment simultaneously monitors the running state of the target mirror image in real time, and executes successful confirmation or abnormal recovery operation according to the state. In this way, quick response of faults is achieved through running state feedback, the stability and reliability of firmware upgrading of the embedded device are effectively improved, and the risk of device paralysis in the upgrading process is reduced.

The invention relates to a mobile robot dynamic obstacle real-time detection method and system based on embedded vision. The method comprises the following steps: performing motion compensation on a historical frame image, generating a motion compensation image, performing frame difference analysis on the motion compensation image and a current frame image, performing adaptive threshold segmentation in combination with local texture features, and generating a motion candidate mask; and extracting feature points in the mask region, calculating an optical flow vector and generating RGB image blocks, and performing dynamic attribute verification after fusion to obtain a verified dynamic point set. And further calculating an optical flow direction angle and a fusion speed value, constructing a motion feature vector and clustering, and generating a central position coordinate of the obstacle instance. And through rotation matrix conversion and Kalman filtering processing, a smooth position estimation value and a smooth speed estimation value are obtained, and the relative distance and the dynamic threat degree of the obstacle are calculated. According to the method, through motion compensation, filtering optimization and the like, the real-time performance, precision and motion estimation accuracy of dynamic obstacle detection in an embedded scene are improved.

The utility model discloses a bed body frame of a seedling raising bed, which is used for placing hole trays for seedling raising and comprises a plurality of cross beam section bars, a plurality of supporting pipe columns and connecting rods, the cross beam section bars are arranged in parallel, the supporting pipe columns are also arranged below the cross beam section bars vertically, and the connecting rods are connected with the cross beam section bars. The adjacent supporting pipe columns are connected through the connecting rods. Each cross beam sectional material comprises a cross plate and a separation part, the separation part is vertically arranged on the upper surface of the cross plate, a sliding groove is formed by the separation part and the upper surface of the cross plate, and the hole tray is placed on the sliding groove of the adjacent cross beam sectional material and can slide in a reciprocating mode along the sliding groove. The hole tray is simple in structure, high in strength, high in precision, good in stability and capable of enabling the hole tray to slide smoothly in the sliding groove.

The invention provides a laser stripe center extraction method based on gray coefficient binarization, and aims to solve the problems of non-uniform brightness distribution of laser stripes, difficulty in extraction of weak stripes, breakage of center lines and the like. The method comprises the following steps: enhancing an image; realizing self-adaptive threshold segmentation through a local gray scale statistical model and a peak variable coefficient by using a gray scale coefficient binarization method; extracting a sub-pixel-level center point by combining a secondary positioning method of skeleton constraint; and carrying out fitting smoothing by adopting bicubic interpolation and a smooth spline method. According to the method, the problem of extraction of strong and weak stripes under complex working conditions is effectively solved, and the precision, connectivity and robustness of stripe center positioning are remarkably improved while background noise is suppressed.

The invention relates to the technical field of load prediction, and provides a power load prediction method and system based on time sequence decomposition and an attention mechanism, and the method comprises the steps: carrying out the adaptive time sequence decomposition of an obtained original load sequence, calculating the sample entropy of each decomposed component, and carrying out the clustering; constructing a group of encoder and decoder networks for each piece of clustered data, performing parallel encoding to extract features, performing serial decoding reconstruction on the features from low frequency to high frequency, and outputting prediction data from low frequency to high frequency step by step; the weight is initialized based on the sample entropy, the trained weight is obtained through optimization in the encoder and decoder network training process, and the predicted value of the power load is obtained through weighted fusion. According to the method, adaptive time sequence decomposition, a weight mechanism guided by sample entropy and an attention-enhanced encoder-decoder structure are introduced, so that multi-component collaborative modeling and cross-scale dynamic prediction are realized, and the prediction accuracy and stability in complex load data and small sample scenes are effectively improved.

The invention provides a multiple-input-multiple-output soft measurement method and device and a medium. The method comprises the steps that historical multi-sampling-rate process variable data and historical multi-quality variable data are acquired; obtaining a plurality of historical task specific features from the historical multi-sampling-rate process variable data; inputting the plurality of historical task specific features to obtain historical cross-task sharing interaction features; training according to the historical cross-task sharing interaction features, the plurality of historical task specific features and historical multi-quality variable data to obtain a trained feature interaction hybrid expert network model; and obtaining a plurality of online prediction results according to the trained feature interaction hybrid expert network model. According to the method, the device and the medium, the problems that information loss, time alignment errors and task negative migration are easily caused when different variable sampling frequencies are inconsistent in an existing multi-input multi-output soft measurement method, an irregular sampling phenomenon exists, and the prediction precision of multiple quality indexes is not high can be solved.

The invention discloses a multivariate time sequence prediction method based on dynamic complex spectrum filtering, which comprises the following steps: acquiring multivariate time sequence data of a target scene, determining a predicted target variable and a window length, and constructing a training set, a verification set and a test set based on the data; constructing a DCSFNet prediction model, wherein the model sequentially comprises a reversible instance normalization module, a frequency domain conversion module, a dynamic complex spectrum filtering module, a time domain projection module and an inverse reversible instance normalization module; and realizing multivariate time sequence prediction by using the trained DCSFNet model. According to the method, the frequency domain features are adaptively processed through the dynamic complex spectrum filtering module, and reversible normalization and time domain mapping are combined, so that the problems that an existing method is poor in adaptability to diversified time sequence data, frequency domain information is insufficient in utilization, and prediction precision and calculation efficiency are difficult to consider at the same time are effectively solved; and multivariate time sequence prediction with high precision, high efficiency and cross-scene adaptability is realized.