634results about How to "Verify validity" patented technology

The invention discloses an image-text matching method based on progressive semantic aggregation and structured cognitive enhancement, and relates to the technical field of image-text matching, comprising the following steps: S1, multi-modal feature extraction; s2, introducing a hierarchical self-enhanced semantic perception pooling mechanism; s3, introducing a structured knowledge auxiliary learning module; s4, introducing a cross-modal semantic alignment mechanism; s5, introducing a multi-scale semantic association matrix; according to the method, through multi-level semantic enhancement and fine-grained level self-adaptive feature aggregation, a significant word field is enhanced, and consistency and unification between local representation and global representation are achieved; extracting structured semantics between entities from an image title corpus, mining high-order relationships in long and short sentences, and expanding concepts missed in text global embedding by utilizing the priori knowledge; a cross-modal relation interaction mechanism is introduced, semantic embedding is enhanced, and more consistent semantic understanding is promoted; a multi-scale semantic association matrix is introduced, and a more comprehensive and accurate similarity measurement matrix is generated.

The invention discloses an integrated energy system economic dispatching model method based on deep reinforcement learning, and the method comprises the steps: constructing a park equipment environment model, and carrying out the mathematical description of a park integrated energy system low-carbon economic dispatching problem through taking economy and carbon emission as targets; secondly, the scheduling problem is expressed as a Markov decision process framework, and a state space, an action space and a reward mechanism of the system are reasonably designed; then, a near-end strategy optimization algorithm is adopted to carry out low-carbon economic dispatching decision making, and an action noise dynamic updating strategy is provided to gradually improve the exploration capability and learning capability of a network model; according to the method, the action space, the state space, the reward mechanism, the neural network structure and the like of the intelligent agent are designed in detail, the minimization of the system operation cost is considered on the basis, and a deep reinforcement learning method is adopted for scheduling simulation. The method provided by the invention can perform optimal scheduling according to the load demand of the system, and has good scheduling performance and self-learning ability.

The invention provides a multi-modal sentiment analysis method based on unwrapping representation learning, which comprises the following steps: acquiring a multi-modal sentiment analysis data set, and extracting features of texts, audios and videos; constructing a representation learning network, extracting private features and shared features, and processing through adversarial learning optimization and time smoothness constraint; under the guidance of different language scales, the private features of the audio and the video are fused; further fusing text, audio and video features through cross-modal attention and a gating mechanism; and finally, inputting the fusion features into an emotion analysis module, and generating an emotion prediction result in combination with an emotion classification or regression task. According to the method, the consistency of the audio and video features in continuous time is considered, redundant information is effectively reduced, the time dimension consistency is ensured, deep fusion of the audio and video features is realized under text guidance, and the cross-modal fusion effect is remarkably improved.

The invention relates to the technical field of network security, and discloses an LLM-based network threat flow detection rule automatic generation method and system. According to the system, an automatic rule generation framework based on a large language model is constructed for network threat traffic, and the system comprises a cue word module, a large language model module, a rule engine module and a rule base management module; the four modules are executed in sequence to form a complete link from traffic information input to rule automatic generation and application verification; according to the method, domain knowledge is fused into a pre-training model through LoRA fine tuning, small sample cue word engineering and a correction mechanism based on engine feedback, verification is carried out through an actual detection engine, and the effectiveness of a rule automatically generated by the model is verified. According to the method, the network threat traffic sample can be fully and effectively analyzed, meanwhile, the requirement for training data is reduced, the rule set capable of being practically applied is generated, and the network defense capability is improved.

The invention relates to the technical field of computer deep learning, and relates to an equipment fault diagnosis network model training method based on multi-source domain knowledge joint migration, and the method comprises the steps: constructing a training sample set, a verification set and a test set, and training a pre-constructed equipment fault diagnosis network model through the training sample set, optimizing parameters of the equipment fault diagnosis network model to obtain a trained equipment fault diagnosis network model; verifying the trained equipment fault diagnosis network model by adopting a verification set, evaluating the diagnosis performance of the trained equipment fault diagnosis network model, and selecting an optimal equipment fault diagnosis network model; and testing by adopting the test set optimal equipment fault diagnosis network model, and evaluating the performance of the optimal equipment fault diagnosis network model. According to the method, multi-source-domain cooperative training is carried out on the model by utilizing a collected multi-source-domain data set which is composed of a plurality of labeled source-domain data sets and unlabeled source-domain data sets and contains a small number of labels, domain invariant features in fault samples are learned, and shared diagnosis knowledge is migrated to a target domain from labeled data and unlabeled data together; therefore, the trained equipment fault diagnosis network model can complete a cross-domain fault diagnosis task without obtaining a target sample in advance.

The invention discloses a conflict detection processing method and system based on a BIM model, a medium and a product, and relates to the technical field of constructional engineering, and the method comprises the steps: obtaining a construction project design scheme; based on a preset model framework, determining a BIM model according to the construction project design scheme; predicting potential conflict points in the BIM model based on a preset conflict mode library; matching the potential conflict points with the building structure, the electromechanical pipeline and the equipment layout by using a space mapping algorithm, and determining possible conflict areas; on the basis of a natural language processing technology, operable solution suggestions are generated according to the possible conflict areas; after the operable solution suggestions are sent to the user, the solution suggestions determined by the user are received; and adjusting the BIM model based on the solution suggestion. By implementing the method, the accuracy and intelligence of conflict detection can be improved.

The invention belongs to the technical field of oil and gas production and development, and discloses a sand production pressure difference and sand production quantity quantitative prediction method suitable for a high-temperature and high-pressure sandstone reservoir, which comprises the following steps: calculating static rock mechanical parameters of a target well; predicting initial crustal stress; establishing a porosity and permeability dynamic evolution model; establishing a three-dimensional finite element model; establishing a seepage stress coupling transient nonlinear finite element equation; applying well bottom pressure drop to simulate production, and determining sand production pressure difference; and establishing a sand production rate prediction model, and quantitatively predicting the sand production rate. According to the method, factors such as pore permeability stress, temperature change and water content influencing rock strength are comprehensively considered, a three-dimensional finite element model is established, a seepage stress coupling finite element equation and solving steps are established based on a reservoir elastic-plastic consolidation theory, critical plastic strain is introduced to quantitatively characterize the sand production risk, a sand production prediction model is established, and the rock strength is predicted. And finally, quantitative and accurate calculation and scientific pressure control of the critical sand production pressure difference and the sand production amount are realized.

The invention discloses a lithium battery SOC (State of Charge) estimation method based on constrained optimization Kalman filtering. The method comprises the following steps: building a battery RC (Resistance-Capacitance) model; establishing a Kalman filtering model and initializing parameters; processing the collected battery data; iteratively updating Kalman filtering parameters continuously through constraint optimization to obtain more accurate values; mSE and RMSE are used for a Kalman filter to evaluate the performance of the filter. In order to overcome the limitation of the Kalman filter on parameter setting and the sensitivity of the Kalman filter on model change, the parameter adjustment process of the Kalman filter is optimized by introducing constraint conditions, and the parameters of the filter are dynamically adjusted in combination with an advanced optimization algorithm, so that the filter can keep the optimal estimation performance under different working conditions. Meanwhile, through limitation of constraint conditions, it is ensured that the estimation result conforms to the physical characteristics and actual working conditions of the battery, and the accuracy and reliability of SOC estimation are further improved.

The invention discloses a wind power prediction method based on improved LSTM and FA-KELM, and the method comprises the steps: obtaining the wind power data of a wind power plant, carrying out the normalization of the wind power data, and carrying out the decomposition of the wind power data through employing CEEMDAN; calculating the fuzzy entropy of the obtained subsequences, performing secondary decomposition on the subsequence with the maximum fuzzy entropy by using a PSO-VMD algorithm, and dividing other sequences into a high-entropy sequence and a low-entropy sequence according to the fuzzy entropy; the method comprises the following steps: improving an LSTM network, and replacing an input gate, a forgetting gate and an output gate in the LSTM network with a gate control structure with a cross coupling structure to obtain ConvsLSTM; mIC is adopted to screen out meteorological factors with large correlation with wind power in the meteorological factors of the wind power plant; and in combination with the obtained meteorological factor data, predicting the obtained high-entropy sequence by using ConvsLSTM, and predicting the low-entropy sequence by using FA-KELM. And superposing the predicted values of all the components, and carrying out reverse normalization to obtain a final ultra-short-term wind power prediction result. According to the method, the appropriate prediction model can be selected for the power subsequences with different complexities, and the overall wind power prediction precision is improved.

The invention discloses an unsupervised domain adaptive medical image segmentation method and device based on multi-scale features, and relates to the technical field of computers, and the method comprises the steps: constructing a multi-scale feature fusion segmentation network based on double attention; taking the multi-scale feature fusion segmentation network as a generator, and constructing a first discriminator and a second discriminator; training based on the generator, the first discriminator and the second discriminator to obtain a trained multi-scale feature fusion segmentation network; and segmenting a medical image to be processed by using the trained multi-scale feature fusion segmentation network. According to the method, multi-scale features are introduced, feature representation of different levels is considered, and alignment of source domain and target domain features is realized through an adversarial feature alignment method and adversarial learning, so that the performance and generalization ability of a segmentation model on a target domain are improved; moreover, the method does not need to use a target domain label, and can well solve a problem of unsupervised domain adaptation of a medical image segmentation task.

The invention discloses a hyperspectral and multispectral image fusion method and system based on an improved Transform, and the method comprises the steps: carrying out the up-sampling of a hyperspectral image, and extracting the potential features of the hyperspectral image; multi-scale features of the hyperspectral image and the multi-spectral image are respectively extracted by using a double-branch structure, and global feature extraction is carried out on feature maps through a multi-layer deformable convolution crossover Transform; and adding the extracted features according to the same direction, splicing the features in a channel dimension, and gradually reconstructing a multispectral image through cascade up-sampling. The invention provides a new method based on an up-sampling recursive sub-network and a multilayer deformable convolutional cross Transformer (URSMDCT) for hyperspectral and multispectral image fusion, potential features and multiscale features of the image can be fully extracted, and spectral information and spatial information are effectively fused.

The invention relates to the technical field of building design, in particular to a low-carbon building design method based on a full life cycle. The method comprises the following steps: acquiring building material data; performing data preprocessing on the building material data to generate building material processing data; obtaining a building three-dimensional diagram and construction process data; performing material carbon emission coefficient calculation on the building material processing data to generate material carbon emission data; building a building dynamic process three-dimensional diagram by using the building three-dimensional diagram and the construction flow data, endowing material attributes by using building material data, endowing carbon emission by using material carbon emission data, and generating a three-dimensional dynamic carbon emission model; therefore, by constructing the full-life-cycle carbon emission three-dimensional dynamic model, the problem that in traditional building carbon emission evaluation, a static model is difficult to adapt to dynamic environment changes is solved, and the accuracy and real-time performance of building full-life-cycle carbon emission management are improved.

The invention provides a soil nutrient detection method based on a gamma-ray energy spectrum and an XGboost algorithm, and the method comprises the steps: constructing a training sample for soil nutrient detection, the training sample comprises radioactive element data, and the label of the training sample is soil nutrient data with the same soil point location as the radioactive element data; the training sample is input into an XGboost model, the XGboost model extracts features of the training sample and predicts soil nutrients of the training sample, the soil nutrients serve as nutrient predicted values, a loss function is constructed according to the nutrient predicted values and labels of the training sample, the XGboost model is trained, and a soil nutrient detection model is obtained; acquiring soil radioactive elements of to-be-detected soil through a gamma-ray energy spectrometer, and inputting the soil radioactive elements into the soil nutrient detection model to obtain a nutrient detection result of the to-be-detected soil.

The invention discloses a structural deformation measurement method based on geometric correction of a multi-view panoramic camera. A method for measuring bridge displacement by using a single multi-view panorama camera is provided, multi-view segmentation is performed on a panorama image according to a panorama camera equivalent model and a cube projection method, and independent calibration and correction are performed to obtain a distorted image capable of covering a 360-degree full-field range; therefore, a single panoramic camera is equivalent to a vertical four-camera system without overlapped view fields, so that the reference target in any direction can be used for displacement correction. The invention further provides a camera disturbance elimination method based on a multi-view circle center array target and pose transformation matrix purification optimization, and a coordinate system transformation matrix after purification optimization is obtained through a high-precision multi-feature point tracking algorithm from camera single-view moving target pose measurement. And obtaining a multi-view geometric correction theoretical model of any view of the measurement target and the reference target in a 360-degree range by utilizing a multi-view equivalent camera fixed connection relationship.

The invention discloses an electric vehicle charging path planning method based on a dynamic energy consumption model and road network information. According to the method, the influence of different temperatures on the battery capacity is analyzed according to the charging and discharging experimental data of the lithium battery, the power consumption generated in the driving process of the electric vehicle is analyzed, and a refined air conditioner energy consumption model and a mileage energy consumption model are established. Based on an improved A * algorithm, a method for searching a reachable path with the minimum energy consumption according to the residual electric quantity of a vehicle-mounted battery, the position of a charging station and the real-time road condition is established. The A * algorithm of the model solving algorithm serves as an optimal priority search algorithm in graph theory search methods, and the optimality requirement can be met by reasonably designing a function; and meanwhile, aiming at an urban road network, a heuristic A * algorithm is adopted to be matched with the path optimization solution of a large-scale traffic network in a road-electricity coupling scene. The effectiveness of the provided model in the charging path decision is verified by respectively comparing three targets of time optimization, path optimization and comprehensive energy consumption optimization provided by the invention.

The invention relates to a method for converting Lao characters into phonemes based on transfer learning, and aims to realize efficient phoneme conversion by utilizing a model pre-trained on other languages and finely adjusting limited Lao data. According to the method, the problem of data scarcity can be relieved, the training process of the model can be accelerated, and the generalization ability of the model on new tasks is improved. A more robust Lao character transphoneme model can be constructed through transfer learning, and a foundation is laid for subsequent speech synthesis. As Thai with high language similarity with Lao is also a low-resource language and cannot be used as a pre-training language, Chinese is selected as the pre-training language. A Chinese open source data set is firstly used for training a Chinese G2P model based on a Transform architecture, decoding end parameters of the model are initialized to a student model, fine tuning training is performed by using a Lao data set with the scale of ten thousand, and the accuracy of a Lao character-to-phoneme model is improved.

The invention relates to a radar PRI modulation type identification method based on a neural network and a multi-head attention mechanism, and belongs to the field of radar signal processing. First-order difference is carried out on a radar pulse TOA sequence to obtain sample data of a PRI sequence, a median filter is adopted to carry out filtering processing on the sample data to reduce the influence of missing or false pulses, and a model integrating a multi-scale convolution extrusion excitation neural network, a bidirectional gating circulation unit neural network and a multi-head attention mechanism is constructed. The model can extract data local refinement features and capture data long-term time sequence correlation features, a multi-head attention mechanism is used for integrating global information, more weights are applied to important time steps in a gating circulation unit, the model is trained and tested, and the result shows that compared with other neural network methods, the model has the advantages that the algorithm is simple, and the algorithm is easy to implement. According to the method, the recognition rate of six kinds of radar PRI modulation signals is higher under the condition that the ratio of missing pulses to stray pulses is high.

The invention provides an agent development platform and method, a medium and equipment, and relates to the technical field of intelligent algorithm development. The platform comprises a hybrid cloud heterogeneous computing resource management module used for monitoring a plurality of integrated computing resources and scheduling and distributing the plurality of computing resources; the algorithm model automatic development module is used for providing a target development environment and a target algorithm template for a user and generating an initial intelligent algorithm model in response to an editing instruction of the user for the target algorithm template in the target development environment; the intelligent game distributed training module is used for performing distributed training on the initial intelligent algorithm model based on the target data set to obtain a target intelligent algorithm model; and the intelligent algorithm model evaluation module is used for evaluating the target intelligent algorithm model based on the evaluation index and verifying the effectiveness of the target intelligent algorithm model. Therefore, a user is comprehensively assisted in developing a new algorithm model, and the research and development threshold and time cost of intelligent task planning are reduced.

The invention discloses a breakable crop stalk discrete element simulation model construction method, which comprises the following steps: firstly, measuring the actual feature size of a crop stalk, and layering the internal structure of the crop stalk based on a microstructure; then performing an experiment to obtain intrinsic parameters and contact parameters of the stalks, constructing a multi-layer geometry three-dimensional model according to measurement characteristics, filling each layer of the geometry three-dimensional model with different types of particles, establishing an initial discrete element simulation model, introducing a contact model, and constructing an initial breakable discrete element simulation model; and finally, adjusting the mesoscopic parameters of the simulation model according to the actually measured data until the mesoscopic parameters are kept consistent. According to the method, the contact model is introduced, different structural layers are filled with particles with different material attributes, the biological characteristics and mechanical characteristics of multiple layers of real stalks from outside to inside are restored, and errors are smaller when mechanical behaviors are simulated.

The invention provides a point cloud semantic segmentation method for improving a RandLA-Net network. The method comprises the following steps: step 1, carrying out data preprocessing grid downsampling; step 2, calculating a point cloud normal vector by a local curved surface fitting method; thirdly, local multi-dimensional feature fusion is carried out, and local geometric information is captured; fourthly, important features are automatically learned and aggregated through a self-attention module and attention pooling; 5, cascading an up-sampling result with the intermediate feature of the coding layer through jump connection; and 6, predicting the semantic tag by using a full connection layer. According to the invention, high-precision and high-efficiency automatic point cloud semantic segmentation in a large-scale outdoor scene is realized.

The invention relates to a lockage ship reservation scheduling optimization method. The method comprises the following steps: S1, obtaining ship information of a water transportation junction, attribute information of a ship lock and ship remote lockage declaration information; s2, taking the minimization of the ship lock waiting time and the ship reservation adjustment rate as optimization objectives, constructing a ship reservation scheduling double-objective optimization model, and establishing constraint conditions satisfied by lockage ship reservation scheduling; and S3, performing iterative solution on the ship reservation scheduling dual-objective optimization model based on a fast non-dominated sorting genetic algorithm to obtain an optimized lockage ship scheduling scheme. The technical problems that existing ship lock resources are not enough in utilization and the ship dam passing efficiency is not high are solved, the waterway traffic jam problem can be effectively relieved, the overall ship lock passing efficiency can be improved, the ship waiting time is shortened by optimizing the matching relation between the ship dam passing number and the ship lock service capacity, and the ship dam passing efficiency is improved. And the ship reservation adjustment rate is reduced, so that the ship lockage efficiency is improved.

The embodiment of the invention discloses an intelligent prediction method and device for adverse drug reaction based on multi-source data fusion, relates to the technical field of drug-gene-adverse drug reaction neural networks, and aims to solve the problem of low ADR prediction performance due to lack of a more effective feature analysis method and design of a more efficient model in the prior art. The method comprises the following steps: acquiring multi-source data; performing standardization processing on the multi-source data, constructing drug descriptor features, and obtaining a drug similarity vector set; constructing adverse drug reaction descriptor features, and obtaining an adverse drug reaction similarity vector set; and training the target adverse drug reaction intelligent prediction model based on the drug similarity vector set and the adverse drug reaction similarity vector set, and after training is completed, predicting an adverse reaction generated when a target drug is applied to a target individual based on the target adverse drug reaction intelligent prediction model.

The invention provides an electric vehicle charging prediction method and model fused with time-space node information. The prediction method comprises the following steps: constructing an adjacent matrix according to traffic position characteristics, and obtaining a historical sub-number set according to historical number set slicing so as to obtain a node physical graph; obtaining a PageRank value according to the node physical graph so as to obtain a time sequence matrix and a spatio-temporal information graph; based on a multi-head attention mechanism, respectively constructing a space attention matrix and a time attention matrix according to the time sequence matrix; obtaining a spatial dynamic correlation matrix according to the spatial attention matrix and the adjacency matrix, and combining the spatial-temporal information graph to obtain a fusion graph; and obtaining a spatial convolutional network based on graph convolution according to the fusion graph, and obtaining a time convolutional network based on causal convolution according to the time attention matrix, so as to obtain a prediction model for electric vehicle charging prediction. According to the electric vehicle charging prediction method fusing the space-time node information, the space-time coupling effect can be comprehensively considered, and the prediction precision is high.

The invention discloses a self-adaptive voiceprint recognition system based on a Mama model. The self-adaptive voiceprint recognition system comprises a voice acquisition module, a voice enhancement module, a voice feature extraction module, a spectrogram enhancement module, a model training module and a model decision module, according to the invention, through well-designed network architecture and algorithm optimization, high identification accuracy is ensured, and good robustness is shown; the system can keep a low equal error rate under various noise conditions and in different speaking styles, which indicates high robustness and accuracy of the system in a voiceprint recognition task; a bidirectional state space model of the Mama module can effectively capture local details and global context of voiceprints, and the recognition capability of the system on complex voice features is enhanced; the TDNN module strengthens the sensitivity of the system to time sequence features, the DenseNet module improves the transmission and fusion of the features through the dense connection of the DenseNet module, and the combination of the TDNN module and the DenseNet module effectively improves the feature extraction fineness of the Mamba module and improves the robustness of the model.

The invention relates to the field of pole and tower landslide disasters, in particular to a system and method for modeling, simulating and monitoring pole and tower landslide disasters based on an unmanned aerial vehicle. The complex terrain around the pole tower is monitored by means of unmanned aerial vehicle operation modeling, ground disaster hidden danger identification, three-dimensional monitoring and early warning, hidden danger dynamic prevention and control and the like. The reliability and effectiveness of the method are verified by carrying out a large number of tests on towers under different topographic conditions, the dependence on hardware in the embedded software testing process is reduced, and comprehensive real-time monitoring of topographic changes and accurate identification of landslide signs are realized. According to the invention, the monitoring and early warning capability of tower landslide disasters is obviously improved, and safe and stable operation of a power system is powerfully guaranteed.

The invention discloses a combustion chamber wall surface pit array structure design method and a combustion chamber. The method comprises the following steps: extracting a three-dimensional fluid model and carrying out bench test; establishing and verifying an in-cylinder combustion and emission simulation model; performing temperature field bench test on the cylinder cover, the piston and the cylinder sleeve; establishing and verifying a heat transfer simulation model; establishing a heat engine coupling simulation model; parameterization design is conducted on the pit array structures on the piston, the cylinder sleeve and the cylinder cover, a collaborative optimization mathematical model of pit array structure design is established, an optimal solution set is obtained, and typical solutions are selected; key parts in the typical solution are machined, an installation test is carried out, and the optimal pit array structure design scheme is obtained through the test. According to the pit array structure obtained through the collaborative optimization design method, the weight of components can be reduced, a turbulent layer can be formed on the near wall face of the combustion chamber to promote oil-gas mixing, then nitric oxide and particulate matter are reduced, and the combustion performance of an engine is improved.

The invention discloses a method for constructing a hyperbolic discrete diffusion model with a three-dimensional RNA (Ribonucleic Acid) structure inversely folded, which comprises the following steps of: embedding rich geometrical characteristics and other topological properties of the RNA three-dimensional structure into a hyperbolic space; a forward diffusion process is utilized to carry out step-by-step noise addition on a real one-dimensional sequence of an RNA three-dimensional structure and carry out step-by-step noise removal on a hyperbolic isovariant graph neural network noise removal model, and on the basis of a limited training sample, nucleotide distribution of a target RNA three-dimensional structure, namely the one-dimensional sequence, is effectively recovered, so that the target RNA three-dimensional structure is obtained. According to the method, the hierarchical structure of RNA is represented through a hyperbolic isovariant graph neural network denoising model, and the structural similarity measurement of three-dimensional reverse folding is enhanced. The invention provides a powerful artificial intelligence method for the generative design of a functional RNA structure, and is beneficial to the design of RNA drugs and RNA biomolecules and the application of the RNA drugs and RNA biomolecules in biological medicines.

The invention discloses a scientific research partner recommendation method based on an attention mechanism and a graph neural network, and provides an academic partner recommendation model AGNNRec based on the graph neural network and the attention mechanism. According to the method, a recommendation model is constructed based on a graph neural network, a context path is introduced to capture a high-order relationship between nodes, a multi-head relationship attention mechanism is introduced to learn embedded vectors of node high-order neighbors in parallel, and finally, recommendation of partners is completed by using a cosine similarity algorithm. Experiments are carried out on the model on a self-built data set and a public data set DBLP respectively, in the experiment using the DBLP data set, the effects of the model on Precision (at) 5 and NDGG (at) 5 are improved by 19.07% and 1.17% respectively, and the effectiveness of AGNNRec in science and technology partner recommendation tasks is shown.

The invention discloses a Yungang grotto environmental parameter prediction method and system based on an artificial neural network, and relates to the field of neural network prediction algorithms, and the method comprises the steps: collecting and preprocessing external and internal environmental parameters of a grotto through a sensor, and constructing a data sample set through the preprocessed data; constructing a CNN-LSTM network prediction model according to the sample set, establishing hyper-parameters of the prediction model, setting a combination interval of the hyper-parameters to be optimized, and executing a BO algorithm on the hyper-parameters; and selecting a next group of hyper-parameters to carry out a new round of training according to a probability model acquisition function until the number of iterations is reached, and outputting a predicted value. According to the method, the BO-CNN-LSTM model is combined with the Bayesian optimization algorithm, hyper-parameter optimization is remarkably improved, more excellent performance indexes are achieved, the BO-CNN-LSTM model has remarkable advantages in prediction of the internal environment parameters of the Yungang grotto, microclimate changes in the grotto are effectively captured and predicted, an actual environment parameter sample is accurately predicted, and the prediction efficiency is improved. And a comprehensive and accurate prediction result is provided for the internal environment parameters of the Yungang grotto.

The invention discloses a position deployment and resource allocation method for multi-unmanned aerial vehicle communication. The method comprises the following operation steps: S1, initializing parameters and a model; s2, modeling: modeling for unmanned aerial vehicle position deployment and resource allocation problems; s3, splitting the problem by using a block coordinate descent method, and splitting the mixed integer non-convex problem by using the block coordinate descent method; s4, user association and the three-dimensional position of the unmanned aerial vehicle, wherein the user association and the three-dimensional position of the unmanned aerial vehicle are obtained through an ant lion algorithm; s5, transmitting power of the unmanned aerial vehicle, and optimizing the transmitting power of the unmanned aerial vehicle by using a successive convex approximation technology; and S6, an iterative optimization algorithm is carried out. Under the condition of multi-unmanned aerial vehicle deployment, the transmission rate requirement of the ground users can be met at the same time, the position deployment and resource allocation of the multiple unmanned aerial vehicles are optimized, the transmission rate is maximized, the system throughput can be improved, and the lowest rate requirement of all the ground users can be met.