139 results about "Transfer model" patented technology

A dynamic optimization system for energy consumption of a refrigeration house based on digital twinning is characterized by comprising a data acquisition module used for acquiring basic structure data of the refrigeration house, technical parameters of a refrigeration system, real-time operation data and historical operation data, preprocessing the data and then outputting a standardized multi-dimensional real-time data stream; the model construction module is used for constructing a 3D geometric model, a thermodynamic transfer model and a refrigeration system mathematical model according to the multi-dimensional real-time data flow, performing machine learning calibration on model parameters through historical operation data, and performing fusion to construct a refrigeration house digital twin model; the prediction analysis module is used for predicting future energy consumption demand and load change according to the refrigeration house digital twin model and the real-time operation data, and outputting an energy consumption prediction result and a load analysis report; a strategy generation module; an execution feedback module; and a learning optimization module. Overall energy consumption of the refrigeration house is reduced, energy utilization efficiency is remarkably improved, and goods storage safety is guaranteed.

The invention discloses a personalized information accurate pushing system and method based on artificial intelligence, and relates to the technical field of personalized recommendation, and the method comprises the steps: fusing user multi-platform behavior data and external space-time environment information, and generating a situation label with confidence through employing an improved space-time density clustering algorithm; constructing a causal directed acyclic graph by adopting a causal forest algorithm, quantitatively analyzing a heterogeneity causal effect, inverting a potential intention of the user, and outputting standardized intention inversion probability distribution; in combination with a historical intention and a behavior sequence, training a Transform intention state transition model constrained by causality of a causality directed acyclic graph, and performing multi-step probability deduction to generate an intention evolution path; information is retrieved from the dynamic knowledge graph according to the prediction path, a pushing copywriting matched with the situation is generated through the NLP technology, the optimal pushing opportunity is calculated in combination with the position track of the user, and accurate reaching of personalized information is achieved.

The invention relates to the field of geotechnical engineering, and discloses a self-stress anchor rod failure evolution simulation method for jointed rock mass bolting-grouting reinforcement, which comprises the following steps: S1, constructing a three-dimensional rock mass model containing a joint structure in an FLAC3D platform, and defining joint surface space distribution and geometric parameters; s2, a bolting-grouting structure is introduced into the three-dimensional rock mass model, an anchor rod, a grouting body and the contact relation between the anchor rod and the grouting body and the rock mass are set, and an anchor rod load transfer model is established; s3, a loading path is set, displacement control is adopted as a loading mode, and a certain included angle is formed between the loading direction and the joint surface; and S4, establishing a numerical model in FLAC3D, performing grid division, and setting a rock mass material constitutive model and contact surface parameters. By accurately controlling the loading direction and the loading angle in simulation and analyzing the change influence of different loading paths on the failure mode of the bolting-grouting structure, the problem that the traditional simulation method neglects the change of the failure mode under different loading conditions is solved.

The invention discloses a load transfer method, system, equipment and medium considering new energy and load power fluctuation, and belongs to the technical field of relay protection, and the method comprises the steps: taking a switch motion moment as a first stage of load transfer, taking a set time after the switch motion as a second stage of load transfer, building a dual-stage load transfer model, and taking the set time as a second stage of load transfer; obtaining a load transfer scheme; double-stage constraint verification is executed, and operation constraint verification is carried out on the load transfer schemes of the first stage and the second stage; introducing a constraint out-of-limit penalty factor, and quantifying the degree of the load transfer scheme deviating from the constraint condition; constructing a load transfer scheme comprehensive evaluation function, and performing weighted integration; and optimizing the load transfer scheme comprehensive evaluation function by adopting an improved binary particle swarm algorithm, and determining an optimal load transfer scheme. According to the method, the limitation of a traditional static model is overcome, and the problem of operation constraint violation caused by photovoltaic output change is avoided.

The invention discloses a model reasoning method and device suitable for a question and answer scene, equipment and a medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: building a mapping relation between a new word list composed of generated target long words and an original word list through a preset word segmentation algorithm by using priori knowledge in target corpus data; when the pre-training model generates the specified content through reasoning, the final output result of the pre-training model is rapidly determined through the probability transfer model, the problems that an existing model acceleration method is low in precision and high in model training cost and calculation cost are solved, and only one probability transfer model is additionally added in the whole reasoning process, so that the calculation cost is reduced. The computing resource pressure is effectively reduced, and the generation speed of the whole system is improved. And in addition, priori knowledge in domain questions and answers is fully utilized, and the accuracy rate of domain proprietary vocabulary generation can also be improved.

The invention provides a multi-physical field compensation method and system for an exposure machine, and the method comprises the following steps: constructing a dynamic error transfer model, and obtaining a coupling parameter based on the output of a coupling function; carrying out eigenvalue decomposition on the coupling parameters, screening principal component vectors to obtain a mapping matrix, modulating the mapping matrix through a thermal weight matrix to obtain a modulation matrix, and carrying out space-time weight reconstruction based on the modulation matrix to obtain a four-dimensional compensation parameter matrix; the control parameters of the air supply pressure of the air floating guide rail, the deflection angle of the DMD micromirror and the power of the laser are adjusted. A dynamic error transfer model containing mechanical-thermal-optical coupling characteristics is constructed, a vibration spectrum, temperature field distribution and light intensity energy data are fused in real time, compensation parameters are subjected to space-time weight reconstruction to obtain a four-dimensional compensation parameter matrix, the real-time performance of matching correction is improved, and therefore the parameter compensation effect and the machining precision are improved.

The invention relates to the technical field of energy testing, in particular to a self-adaptive dynamic energy testing method based on combined heat and power generation, which comprises the following steps: constructing a thermoelectric coupling transfer model and a working condition identification model, and inputting power generation power and heat supply power into the thermoelectric coupling transfer model to correct model parameters, inputting the environmental parameters into a working condition recognition model, recognizing a current working condition type based on a preset working condition feature library, calling an initial test scheme from a test scheme library according to the current working condition type, and generating a dynamic test scheme through a machine learning algorithm in combination with the parameters corrected by the thermoelectric coupling transfer model; and controlling the test equipment to execute a test based on the dynamic test scheme, collecting test data in real time, calculating a system performance index according to the test data, comparing the system performance index with a preset threshold value, and generating feedback information for adjusting the test scheme. Therefore, the problems that in the prior art, a test scheme is fixed and cannot adapt to complex working conditions, and performance evaluation is incomplete are solved.

The present invention belongs to the field of dimension measurement technology, and specifically relates to a method, system, device, and medium for evaluating the uncertainty of a large-scale multilateral coordinate measurement system. The method inputs the acquired target point coordinates into an error transfer model to obtain three coordinate uncertainty components corresponding to the coordinate values. This uncertainty coefficient is then calculated based on the uncertainty introduced by the offset of the reflector's optical center, and the uncertainty evaluation is performed. The present invention, based on the fundamental principles of the multilateral coordinate measurement system, simplifies the system coordinate system to reduce parameters, thereby reducing the amount of calculations. The method also provides a specific mathematical formula, which facilitates rapid evaluation at the measurement site and is easy for metrologists to learn and apply.

The invention provides a ground vibration spectrum acceleration prediction method, a transfer learning method, equipment and a medium, a global model is obtained through first training by using a basic knowledge data set, the first training is evaluated by using a regional knowledge data set, and then a transfer model is obtained through second training by using the regional knowledge data set. And the second training is evaluated by adopting the regional knowledge data set, and the establishment method of the migration model retains the data features of the regional database and learns the difference between the global data set and the regional data set, so that the prediction precision of the model is ensured, and the generalization ability of the model is enhanced.

The invention discloses a micro-seismic signal identification method based on transfer learning and BiLSTM-DCNN, belongs to the technical field of mining engineering micro-seismic monitoring and signal processing, and solves the problems of data scarcity in the initial stage of mine monitoring and low identification precision under the condition of small samples. Firstly, Mel spectrum feature extraction is carried out on a micro-seismic signal; the method comprises the following steps: constructing a BiLSTM-DCNN model comprising a bidirectional long-short term memory network and a deep convolutional neural network, and carrying out pre-training by using large-scale source mine data; and the model is adapted to target mine small sample data through transfer learning, and model parameters are finely adjusted, so that high-precision classification of the signals is realized. According to the method, the recognition accuracy and the model generalization ability of the micro-seismic signals under the small sample condition are remarkably improved, the test accuracy reaches 0.9444 and is improved by 80.85% compared with an unmigrated model, and the method is suitable for an intelligent early warning and safety monitoring system for mine dynamic disasters.

A defect detection method and a model transfer method. The model transfer method comprises: acquiring a source-domain image and a corresponding annotation, and a baseline model obtained from a source domain; acquiring target-domain images, which are fully annotated, partially annotated, or unannotated; inputting the annotation of the source-domain image into a denoiser to obtain a denoised annotation; inputting the unannotated target-domain images into an initial baseline model to obtain second target-domain prediction results, performing data augmentation on the unannotated target-domain images to obtain second target-domain augmented images, and using the second target-domain prediction results as pseudo-annotations of the second target-domain augmented images; and using the source-domain image and the corresponding denoised annotation, the annotated target-domain images and the corresponding annotations, and / or the second target-domain augmented images and the corresponding pseudo-annotations to train the baseline model, so as to obtain a transfer model. The method can complete training by using a small number of annotated target-domain images, thereby solving the problem of model transfer performance being poor in the case of insufficient target-domain images.

The invention discloses a regional lake and reservoir water quality uncertainty simulation method based on Bayesian transfer learning, and the method comprises the following steps: S1, collecting water quality automatic monitoring data of a plurality of lakes and reservoirs in a region, and carrying out the preprocessing of the monitoring data; s2, numbering the plurality of lakes and reservoirs respectively, dividing the lakes and reservoirs into a target domain and a source domain, and selecting a water quality index needing to be simulated as a target index; s3, constructing a target index simulation model in the source domain by adopting a Bayesian additive regression tree, and performing model training by adopting a cross validation strategy; s4, performing parameter updating on the target index simulation model in the source domain by adopting the data of the target domain, constructing a Bayesian migration model of the target index of the target domain, and performing uncertainty simulation on the water quality target index of the target domain; and S5, according to an output result of the Bayesian migration model, identifying a key factor influencing the mean value and uncertainty of the water quality target indexes of the target domain, and analyzing a driving mechanism of the key factor to fluctuation of the water quality indexes of the target domain.

The invention relates to the technical field of data processing of stem cell culture equipment, and discloses a data processing system of stem cell culture equipment, which comprises the following modules: a data acquisition module generates a state and a control sequence by acquiring parameters in a stem cell culture process; the state modeling module constructs a multi-dimensional state matrix and a control matrix based on the collected data; the trend prediction module is used for constructing a state transition model and generating a prediction state vector at the next moment; the optimization control module solves a control strategy vector based on the prediction state vector; the control execution module is used for converting the control strategy vector into an actual control instruction; the closed-loop feedback module collects the actual state after adjustment and is used for correcting the state and the control matrix and dynamically updating the state transition model. According to the method, the state matrix and the control matrix based on the sliding time window are constructed, so that multi-dimensional time sequence information in the stem cell culture process is subjected to structured expression, and the effect of capturing the dynamic change trend of the environment in real time is achieved.

This invention discloses a lightweight BIM data transfer method based on 3DE, comprising the following steps: S1, acquiring components; S2, acquiring structure tree nodes and updating the transfer; S3, acquiring components and generating hierarchical folders; S4, acquiring lightweight geometric files and material map files, and associating the lightweight geometric files and material map files; S5, acquiring lightweight attribute files, and associating the lightweight geometric files and lightweight attribute files; S6, compressing and outputting lightweight data files. This invention solves the problem of information loss during BIM lightweight transfer caused by traditional single-feature lightweight methods, and ensures the accuracy of the association between geometric information, material maps, and attribute information of the transferred model, as well as the consistency of coordinate information and structural organization of the transferred model, thereby improving the quality and efficiency of BIM lightweight transfer.

The invention discloses a migration model training method and device, a migration method and device, electronic equipment and a computer program product. The method comprises the following steps: acquiring an image training set including style images and content images; inputting the style image and the content image into a to-be-trained migration model to obtain a migration image; inputting the style image, the content image and the migration image into a preset image feature extraction module to obtain style features, content features and migration features; calculating the model loss of a to-be-trained migration model based on the style features, the content features and the migration features; and optimizing the to-be-trained migration model according to the model loss until the model loss is converged to obtain a trained migration model which can be used for executing image color style migration processing. According to the scheme, the migration model with relatively high migration efficiency and relatively good migration effect can be obtained, so that the accurate, natural and controllable image color migration effect of various images can be realized through the migration model.

The invention provides a building interface style migration and optimization method based on multi-machine learning model coupling. According to the method, a building interface style migration model is constructed based on various machine learning technologies, so that different building style characteristics and a mapping relation thereof are obtained. Based on a de-noising diffusion probability model and a cyclic generative adversarial network, high-quality style migration among different architectural styles is realized, the stability and detail quality of an architectural interface design scheme are optimized through a neural network, and a multi-objective optimization method is combined to generate an architectural interface layout scheme meeting urban design requirements. Compared with the prior art, the method has remarkable advantages in the aspects of building interface style migration precision, generation scheme adaptability and optimization design efficiency. Meanwhile, due to the reversibility of the design process, the generation result of the scheme can be returned to any step at any time to be optimized until the generated image meets the construction design requirement.

The invention discloses an LIBS (Laser-induced Breakdown Spectroscopy) coal quality quantitative analysis method and related equipment, and the method comprises the following steps: collecting LIBS spectrums of the same coal sample through different instruments, and dividing a training set of a training migration model and a training set of a training quantitative analysis model; performing spectrum pretreatment on the spectrums respectively; carrying out wavelength offset self-correction on the spectrum to realize that the spectrum collected by the slave instrument corresponds to the spectrum collected by the master instrument in wavelength; performing feature selection on the spectrum, and taking the spectrum after feature selection as the input of a migration model; establishing and training a transfer learning model based on feature mapping and feature representation, and utilizing the trained transfer learning model to realize feature transfer from the instrument spectrum to the main instrument spectrum; and training a quantitative analysis model by utilizing the spectrum of the master instrument and the migrated spectrum of the slave instrument, and predicting the coal quality index of the slave instrument by adopting the trained quantitative analysis model. The coal quality quantitative analysis model constructed by the method can stably and efficiently run across instruments, and the modeling cost of different instruments is reduced.

The application discloses a metallogenic fluid main channel inference model based on particle filtering. The model comprises the following steps: collecting relevant data of a target deposit, establishing an inference database, and defining a state sequence of spatial discrete elements; performing statistics on ore-bearing elements known in all exploration information, taking the known ore-bearing elements as an initial particle group of particle filtering, and generating a spatial channel path of each particle; establishing a metallogenic fluid ion state transfer model coupled with probability and velocity according to the spatial position of the particle and fluid dynamics; establishing a particle weight observation model based on a particle state transfer likelihood function, and dynamically updating the weight of the fluid particle; performing spatial sampling on the particle through a resampling algorithm, updating the weight particle, and outputting a maximum posterior probability path through particle number threshold judgment. The model combines multivariate data constraint and particle filtering algorithm to dynamically estimate the main channel path of the metallogenic fluid in a three-dimensional geological space, and fundamentally solves the problems of high cost and difficult sampling sample acquisition of the traditional method.

The invention provides a disease progress probability prediction method and device based on a Monte Carlo-Markov chain, and relates to the technical field of disease progress prediction. The method comprises the following steps: acquiring multi-modal historical follow-up visit data of a patient to be tested, and constructing a state space with transfer constraint; extracting and checking time series data according to the state space; performing Monte Carlo sampling on the time sequence data, establishing a non-homogeneous transfer model and forming an individualized non-homogeneous transfer kernel set; completing path interpolation and probability filling of missing fragments based on the transfer kernel set to obtain continuous complete state sequence data; a Markov chain Monte Carlo algorithm is used for parameter updating, an individualized transfer model subjected to posteriori updating is obtained, a calibration probability result is obtained, and dynamic prediction and uncertainty evaluation of individualized disease progression are achieved. According to the method, the problems of inaccurate state determination, single transfer modeling, incomplete observation and insufficient prediction calibration in the prior art are solved.

The invention relates to the technical field of spacecraft on-orbit service, in particular to a GEO multi-spacecraft cooperative on-orbit refueling task planning method, which is characterized by comprising the following steps of: 1, establishing an orbit transfer model based on a surface and phase modulation maneuvering method, and revealing a relationship between spacecraft orbit transfer speed increment and time; 2, establishing a double-layer optimization model for collaborative on-orbit refueling task planning, and determining a task time range on the basis of minimum fuel consumption; 3, efficiently solving the optimization model by utilizing a double-layer optimization algorithm, wherein a branch and bound algorithm and a fast elite multi-objective genetic algorithm are respectively adopted in an inner layer and an outer layer; according to the method, the limitation of a traditional many-to-many on-orbit refueling strategy is broken through, multiple service spacecrafts are allowed to cooperate to complete the same refueling task, the fuel consumption and time cost of the on-orbit refueling task are reduced, the result is close to the Pareto optimal state, the constraints of multiple aspects such as dynamic scheduling and effective load limitation are fully considered, and the method is suitable for large-scale popularization and application. And actual engineering requirements are met.

The invention discloses a multi-stage rotor assembly method based on four-in-one-axis cooperative regulation and control, and belongs to the technical field of aero-engine manufacturing. The four-in-one is a combination of a rotation axis, a geometric axis, an inertia main shaft and a mass center axis. The multi-stage rotor assembling method is realized through the following steps of S1, obtaining rotor size characteristics, geometric errors, mass center coordinates and the inertia main shaft; s2, establishing a geometric error representation model of the single-stage rotor; s3, establishing an error transfer model of multi-stage rotor assembly; s4, calculating end surface centroid and centroid coordinates of each stage of rotor after assembly; s5, fitting the rotation axis, the geometric axis and the mass center axis; s6, solving an inertia tensor matrix under the assembly coordinate system; s7, calculating an inertia main shaft inclination angle under the assembly coordinate system; and S8, four-axis-in-one assembling regulation and control are carried out. The problem that vibration exceeds the standard under the ultrahigh rotating speed due to low manufacturing precision of an aero-engine rotor can be solved.

The invention provides a fault rate calculation method for a transformer cooling system considering a leakage fault, which comprises the following steps of: firstly, analyzing a dynamic transfer process among a normal operation state, a fracture fault state and a leakage fault state of a cooler pipeline, and constructing a three-state Markov transfer model of the cooler pipeline; then, sensor information such as main transformer oil temperature, winding temperature, cooler outlet oil pressure, water pressure, oil temperature and water temperature is collected, and a knowledge and data driving-based fracture fault related transfer rate calculation method is provided; thirdly, based on the frequency and the transfer rate which are easy to count or calculate, an analytical expression of the transfer rate and the frequency is established by combining a Markov model and a frequency duration method, and the transfer rate related to the leakage fault in the model is calculated; and finally, integrating the transfer rate, constructing a fault rate calculation model of the transformer cooling system, and proposing to predict the future fault rate of the cooling system based on a time convolutional network and a Transform hybrid model. According to the method provided by the invention, fault rate monitoring, analysis and early warning of the transformer cooling system can be realized.

This invention discloses a diffusion model-driven method for generating multilingual human motion, comprising: constructing a sample library; performing multilingual text translation and feature encoding using the Ali Tongyi Qianwen model; constructing a customized CondUNet1D motion denoising network with residual linear multi-head cross-attention, Dropout layer, and Rearrange layer optimization; and training using a joint strategy of exponential moving average and classifier-free guidance. During the inference phase, starting with pure noise, iterative denoising is achieved using a variant of the second-order DPMSolver++ sampler combined with training-free acceleration techniques. Simultaneously, foot slippage is identified and corrected using a vGRFs model transfer model. After secondary optimization of the posture using the diffusion model, a slippage-free motion sequence is obtained. Finally, a video is generated by a remote server and transmitted back to the local terminal. This invention achieves multilingual-driven, efficient, and highly realistic human motion generation.

This invention discloses a high-speed train bearing fault diagnosis method based on transfer learning in the field of rail transit safety monitoring technology. The method preprocesses source and target domain data separately, extracts multi-dimensional features, and uses the t-SNE algorithm for dimensionality reduction and filtering to construct source and target domain feature sets. Based on the source domain feature set, a Stacking ensemble model is constructed and trained as the source domain baseline fault diagnosis model. A multi-scale subspace alignment transfer model is constructed, projecting the source and target domain feature sets onto a shared subspace and performing geometric alignment to obtain domain-invariant features. Based on the aligned target domain features and combined with a multi-scale decision fusion strategy, fault classification is performed on the target domain data, outputting the fault diagnosis result. This method constructs a high-precision baseline model using labeled source domain data and utilizes transfer learning to align cross-domain features, achieving high-accuracy fault diagnosis in the unlabeled target domain.

The invention provides a quantitative evaluation method for the bolt connection precision retentivity of a multi-axis numerical control machine tool based on digital twinning, and the method integrates ANSYS multi-physics field simulation, Unity 3D virtual-real interaction and machine learning technologies by constructing a real-time interaction closed loop of a physical entity and a virtual model. The method comprises the following specific steps: establishing an error transfer model to quantify the influence of static offset, vibration and creep; carrying out pre-tightening force simulation and creep-vibration coupling analysis by utilizing ANSYS; by expanding Kalman filtering calibration parameters, virtual and physical entity errors are ensured; parameter adjustment and visual interaction are realized by means of Unity 3D; and finally, calculating a precision retention index, and evaluating a precision state according to a third-level judgment standard. According to the method, the limitation of single factor analysis is broken through, multi-physical field dynamic coupling capture is realized, and traditional qualitative evaluation is upgraded into a data-driven quantitative system.

The invention provides a consistency verification method and system for a state transition model and a storage medium, and the method comprises the steps: obtaining the state transition model for a protocol and an implementation code, and obtaining the description of a configuration item and a state of the state transition model; according to the description of the configuration items, determining the corresponding configuration items between the state transition models, and obtaining the position arrangement of the configuration items based on a position mapping function; screening out an effective configuration item digital value sequence set based on the configuration item option conflict condition, the self-defined digital meaning and the position arrangement of the configuration item; instantiating the state transition model by using the effective configuration item digital value sequence set to obtain a state transition sequence set; and according to the description of the state, determining and comparing corresponding state parameters in the corresponding state transition sequence, and taking a comparison result of the state transition sequence as a consistency verification result of the state transition model. According to the invention, verification resources can be reasonably utilized while the reliability and accuracy of verification results are ensured.

The application discloses a kind of photoelectric detection model transfer sharing method and internet of things monitoring evaluation system of cloud service, the transfer sharing method specifically is: by calling temperature compensation model and spectrum transfer model, the spectrum information of agricultural product sample is corrected, calling detection model is calculated to the corrected spectrum information, obtains the detection result of agricultural product sample;Internet of things monitoring evaluation system utilizes spectrum transfer sharing method to carry out spectrum information correction, and then calls detection model to calculate, and detection result is returned to detection terminal in real time, realizes remote monitoring and evaluation of agricultural product quality.The application can realize that detection model is shared in different detection terminal, and has wide application prospect in agricultural product quality sampling evaluation.

An electronic component manufacturing system is configured to identify a machine learning model trained to generate analysis or prediction data for a first substrate processing domain associated with a type of substrate processing system. The system is further configured to obtain first trajectory data relating to the first domain to train the machine learning model. The system is further configured to a migration model for a second substrate processing domain associated with the type of substrate processing system. The migration model is generated based on the first trajectory data related to the first substrate processing domain and second trajectory data related to the second substrate processing domain. At least one of the machine learning model associated with the second substrate processing domain, or current trajectory data, using the migration model, is modified to enable the machine learning model to generate analysis or prediction data associated with the second substrate processing domain.

The invention provides a data-free federated knowledge distillation and aggregation method, which comprises the following steps of: processing a knowledge distillation and aggregation process of a model through a federated learning framework consisting of a server and a plurality of clients: when the clients participate in model training, training a local model by using a data set and a lightweight transfer model locally; the knowledge of the local model is fed back to the transfer model through a training process, the server executes global model aggregation and knowledge distillation of the transfer model, and the global model and a distillation result are returned to the client; in the knowledge distillation and aggregation process, the client and the server do not depend on any public data set; the client uses the transfer model to assist in training of the local model, and performs knowledge distillation on the transfer model to obtain model knowledge of other clients; and after the local model finishes a round of training, the server generates data for distillation by calculating the average of logic values output by different transfer models, and sends the data back to the client.