25results about How to "Guaranteed Predictability" patented technology

The invention relates to the technical field of communication interaction, in particular to a data interaction method based on an intelligent control platform. The method comprises the following steps: acquiring multi-dimensional data of production equipment, performing anomaly marking and format unification processing, generating a production line data stream, performing long and short term dependency analysis on the data stream, and constructing a time-space association network, thereby realizing prediction of an anomaly situation, generating an anomaly intelligent analysis model by using a multi-modal fusion modeling technology, and performing data analysis on the anomaly intelligent analysis model. And through the model, abnormity traceability is carried out, and abnormity reasons are clarified. According to the method, a corresponding correction strategy is constructed based on an abnormal reason, tree-shaped interaction sharing is performed through an intelligent control platform, a sharing correction instruction is formed, abnormal correction processing is performed on a production line according to the sharing instruction, feedback is collected through the intelligent control platform, an effective data interaction mechanism is formed, and the reliability and efficiency of the production line are improved. The data interaction method based on the intelligent control platform is more accurate and more comprehensive.

The invention relates to the technical field of remote water meter abnormity early warning, and discloses a remote water meter abnormity early warning method and system based on historical water volume, and the method comprises the steps: dividing a user water meter into three classes A, B and C according to the installation time of the user water meter, and employing a differentiated abnormity detection strategy. For the A-type users, establishing a water consumption prediction model based on historical consumption data of the A-type users; for the B-type users, finding similar reference water meters from the A-type users by using a K-means clustering algorithm; and for the C-type users, monitoring whether the accumulated water consumption for continuous N days is completely equal or not after the C-type users are installed. After real-time usage data is obtained, usage deviations of various users are calculated respectively, if the deviations exceed a set threshold value, abnormal early warning is triggered, and the early warning level is dynamically adjusted according to the number of abnormal duration days. According to the method, technologies such as machine learning and clustering analysis are comprehensively utilized, a set of comprehensive, flexible and efficient water consumption abnormity early warning system is constructed, and the refinement and intelligence level of urban water supply management can be remarkably improved.

The invention discloses a branch prediction unit power consumption management method and device, and the method comprises the steps: setting a multi-stage branch predictor and a low-power-consumption control logic coupled with the multi-stage branch predictor in a processor core micro-architecture, and enabling the multi-stage branch predictor to comprise three stages of predictors which are sequentially accessed in an assembly line, the hardware resource consumption and the prediction accuracy of each predictor are gradually increased from front to back; monitoring the prediction condition of each predictor in real time through low-power-consumption control logic, and closing all predictors behind the first predictor meeting the prediction requirement within a first preset time according to a front-to-back sequence; branch jump addresses are divided into high-order address fields and low-order address fields by BTBs in the multiple levels of predictors except the first-level predictor to be stored in different storage blocks, and label registers used for taking the high-order addresses as label indexes are arranged. The power consumption unit of each predictor can be finely controlled, the energy consumption is greatly reduced, system-level cooperation is not needed, and the universality is high.

The embodiment of the invention discloses a training method of an expression driving model, an expression driving method and a corresponding device. According to the main technical scheme, first training data comprising a plurality of first training samples are obtained, the first training samples at least comprise audio samples, expression control data corresponding to the audio samples and expression parameter truth values, and the expression control data at least comprise time sequence control data; the time sequence control data comprises an action type sequence corresponding to at least one time window in a time sequence of the audio sample; and training an expression driving model based on the first training data: taking the audio sample and an expression control parameter corresponding to the audio sample as input of the expression driving model, and obtaining an expression parameter sequence output by the expression driving model. According to the invention, the fine control effect of the facial expression can be improved, and the expression change is richer and more vivid.

The present application discloses a model migration method and related equipment to solve the problem in the prior art that when the deep learning model is quantized so that the GPU has resources to perform reasoning tasks, the prediction ability and accuracy of the model are reduced, resulting in the problem that it is not applicable in application scenarios with high accuracy and precision requirements. The method includes: determining the call frequency of the algorithm model in the graphics processing unit being called to perform reasoning tasks; if the call frequency is less than the preset call frequency threshold, the hierarchical structure information and model parameter information related to the reasoning task of the idle algorithm model whose call frequency is less than the preset call frequency threshold are migrated to the CPU and memory, and the thread of the idle algorithm model is put into a dormant state; when a call request of the idle algorithm model is received, the dormant thread is awakened, and the hierarchical structure information and model parameter information of the idle algorithm model are migrated from the CPU and memory to the graphics processing unit.

The invention discloses a converter FPGA (Field Programmable Gate Array) small-step real-time simulation device and method based on a per unit fixed-point reconfigurable state space solution core. The upper computer processing unit carries out per-unit processing on converter model parameters according to the per-unit reference of the introduced margin coefficient, discretization and fixed-point word length configuration are completed under the preset step length, and a discrete fixed-point matrix coefficient table is generated for different switch / topology states and is downloaded to the FPGA. A universal fixed-point matrix operation array is arranged in the FPGA simulation unit, a pre-stored matrix coefficient is called according to a switch combination index obtained through real-time analysis during operation, and per unit fixed-point state updating and output inverse per unit are completed in a fixed time sequence prediction; therefore, on-line reconstruction and small-step real-time simulation of topology and parameters of the converter are realized under the condition that the FPGA bit stream is not updated.

The invention provides a lithium battery residual life prediction method and system based on data space position analysis, and the method comprises the steps: building a battery residual life space prediction model based on a convex polytope mechanism according to the basic measurement physical quantity of a lithium battery and the health condition parameter of the battery; determining the shortest length of the online trend as a spatial trend constraint according to the battery health condition parameter of the lithium battery; extracting spatial trend characteristics based on the characteristic parameters of the online operation data of the lithium battery to be detected, and determining a spatial trend straight line by combining spatial trend constraints; and determining a residual service life prediction value corresponding to the space trend characteristics of the lithium battery to be detected by integrating the battery residual service life space prediction model. By adopting the scheme, the defects of complicated operation and insufficient prediction result precision in the prior art can be overcome, and a battery residual life prediction result with high physical significance and high interpretability can be efficiently obtained.

The invention relates to the technical field of electronic products, in particular to a multi-dimensional sensor fused electronic product all-dimensional quality detection method, and solves the problems that in the prior art, a complicated electronic product possibly needs higher-grade detection equipment and method to ensure the quality and performance of the electronic product, and the detection difficulty is gradually increased. The invention discloses a multi-dimensional sensor fused electronic product comprehensive quality detection method. The method comprises a preparation stage, a data acquisition and processing stage, a data analysis and evaluation stage, an environmental adaptability test stage, a safety and reliability evaluation stage, a comprehensive evaluation and report stage and a detection target determination stage. According to the method, clear and specific operation guidance is provided for detection personnel through definition enhancement and detailed steps, so that each step has a clear target and a clear method, fuzziness and uncertainty in operation are reduced, efficiency is improved, the detection personnel can master a detection process more quickly through detailed step guidance, learning and exploration time is shortened, and detection efficiency is improved. Therefore, the working efficiency is improved.

The invention discloses a multi-layer automatic seedbed system which comprises a rack main body and is characterized in that a motor assembly seat is fixedly mounted on the lower side of the rack main body, a servo motor is fixedly mounted on the motor assembly seat, a driving gear is arranged at the driving end of the servo motor, and the driving gear is in meshed connection with a first transmission chain; the invention relates to the technical field of automatic seedbeds, and has the beneficial effects that a multi-layer metal frame forms a framework of a system, provides stable support and realizes intensive utilization of a vertical space, and a chain and chain wheel conveying system is a key for realizing automation. The system can drive the seedling raising plates to horizontally move or vertically ascend and descend among all the layers, the space efficiency is extremely high, the most prominent advantage is that through three-dimensional stacking, the production capacity of the unit occupied area is multiplied, and the system is very suitable for intensive production in environments such as greenhouses and plants with high space cost.

The invention discloses a self-resetting anti-impact layered assembly lattice type pier, and belongs to the field of lattice type piers, the self-resetting anti-impact layered assembly lattice type pier comprises a pier body, an energy dissipation resetting assembly and a bottom energy dissipation connecting piece, the pier body comprises a plurality of pier bodies which are sequentially arranged from top to bottom, and the energy dissipation resetting assembly is installed between the two pier bodies; the bottom energy dissipation connecting piece is installed at the lower end of the pier body. Structural deformation caused by impact can be concentrated on the contact face between the top connecting pieces and the contact face between the bottom plate and the foundation, huge deformation and serious damage of a main body structure are avoided, and the structure can be used for rapid first-aid repair when disasters happen, and it is guaranteed that the traffic lifeline is smooth.

The application relates to a cloud service workload prediction method and system based on a convolution enhanced Transformer, which comprises the following steps: collecting cloud server workload data and preprocessing the cloud server workload data to decompose the cloud server workload data into a trend component and a residual component; constructing a workload prediction model, training the workload prediction model by using the trend component and the residual component, and obtaining a trained workload prediction model; predicting the cloud server workload by using the trained workload prediction model to obtain a workload prediction value; predicting the trend component by using a trend information capturing module to obtain a prediction result of the trend component; predicting the residual component by using a convolution enhanced Transformer encoder module to obtain a prediction result of the residual component; and fusing the prediction result of the trend component and the prediction result of the residual component by using a feature fusion module to obtain a final workload prediction value.

This invention relates to the field of medical data processing technology, specifically to a deep learning-based method for predicting adolescent anxiety. The method includes: quantifying gut microbiota characteristics, anxiety behavior characteristics, and environmental stress characteristics; calculating time-series weighting coefficients; integrating gut microbiota characteristics, anxiety behavior characteristics, environmental stress characteristics, and time-series weighting coefficients into a model; and outputting the prediction results. This invention fills the gap in dynamic biomarker modeling in existing anxiety disorder models by integrating research on the human gut microbiota with existing behavioral and environmental factors. Through deep learning, it seeks the long-term relationship between adolescent gut microbiota and anxiety symptoms, addressing the subjectivity, invasiveness, privacy risks, cost, and delays in existing anxiety disorder diagnosis, thus enabling early prevention. It also provides causal evidence for the "gut-brain axis" theory in adolescents, promoting the development of psychomicrobiome research.

The embodiment of the application relates to the technical field of data processing, in particular to a graph representation learning method and device for out-of-distribution generalization, equipment and a storage medium, aiming to obtain adaptive graph structure data representation of out-of-distribution environment and improve the accuracy of graph structure data related prediction. The method comprises the following steps: inputting an original graph data set into a graph structure data representation network, identifying stable subgraphs and noise subgraphs, performing representation processing on the identified graph structure data, obtaining vectorized representation of the stable subgraphs and vectorized representation of the noise subgraphs; simulating a multi-distribution environment, under the multi-distribution environment, performing prediction according to the vectorized representation of the stable subgraphs to obtain corresponding prediction results; performing loss function calculation on the prediction results and labels of the original graph structure data, optimizing parameters of the graph structure data representation network, and obtaining a graph structure data representation model; and performing a graph data related task through the model to obtain a target result of the graph data related task.

The present application discloses a fault diagnosis method based on random walk and machine learning. First, the vibration signal data of the moving parts are collected and preprocessed. For the preprocessed vibration signal, the fault feature extraction algorithm is used to extract the feature parameters, and the obtained feature parameters are subjected to principal component analysis PCA. Then, a random walk optimization algorithm is constructed, and the random walk optimization algorithm is used to optimize the hyperparameters of the machine learning model. Then, the sample data after PCA dimension reduction is divided into a training set and a test set, and the training set is put into the optimized machine learning model for training. The test set is used for fault prediction to obtain the prediction accuracy under different features. Finally, a comparative analysis is performed to select the feature with the highest accuracy as the best feature. This method solves the problem that the traditional algorithm takes too long to find the optimal solution and cannot find the global optimal solution. While ensuring the prediction performance of the model, the prediction accuracy of the model is significantly improved, thereby improving the efficiency of fault diagnosis.

The application relates to the field of power electronics, in particular to a hybrid commutation method and system. The hybrid commutation method is used for the commutation of a current source type matrix converter, and the method comprises the following steps: using a 2 / 3 PWM modulation mode to alternately control the turn-on and turn-off of the upper switch tube or the lower switch tube of any two phases of A phase, B phase and C phase of the current source type matrix converter, and meanwhile making the lower switch tube or the upper switch tube of the remaining one phase in a clamping state; obtaining the voltage difference between the upper switch tube or the lower switch tube of the any two phases; when the voltage difference is greater than a preset voltage threshold, using a voltage type commutation mode to control the switching sequence of the upper switch tube or the lower switch tube of the any two phases. Sensor-free, reduced hardware cost and complexity: the method replaces real-time sensor detection through sector prediction and current change rule calculation, and high-bandwidth current sensors and precise voltage detection circuits are omitted, so that the system hardware cost is reduced.

The application relates to the cross field of artificial intelligence and environmental health, and discloses a membrane nephropathy prognosis risk stratification method based on exposure spectrum and clinical indicators. First, demographic information, clinical detection indicators and various environmental pollutant exposure data of patients are collected to construct a multi-level exposure spectrum feature, including single pollutant concentration, same type comprehensive exposure factor and pollutant synergistic interaction term. Then, the exposure spectrum and the clinical indicators are fused, various machine learning algorithms are used to construct candidate prediction models, the optimal prognosis risk stratification model is screened through cross validation and multiple performance indicators, and the SHAP value is used to quantify the contribution degree of each feature. Further, a double machine learning method is used to control confounding variables and estimate the causal effect of key pollutants on prognosis. On this basis, individual risk stratification and main risk driving factor identification of membrane nephropathy patients are realized, and scientific basis and guidance are provided for clinical decision making.

This invention relates to the field of communication status detection technology, specifically to an intelligent method and system for detecting the communication status of 5G communication projects. It includes dividing the network transmission into segments and deploying multi-segment node detection probes; establishing a communication fault analysis model to calculate the fault status of each network element node. This invention achieves multi-channel, multi-interface, and comprehensive detection of the communication status of 5G communication projects by dividing the network transmission into segments and deploying multi-segment node detection probes. It collects network communication status data from different locations across multiple segments, and simultaneously obtains parameter data relationships between different nodes through the established communication fault analysis model. It utilizes parameter values ​​fed back from probes in key deployment nodes to classify event types and match parameter value trends, and, in conjunction with a node status calculation function, calculates the corresponding fault score in real time. This allows for real-time fault prediction of the current faulty node from multiple perspectives, avoiding the limitations of prediction based on single-node data feedback and ensuring prediction effectiveness.

The invention relates to the technical field of artificial intelligence, and discloses a network fault early warning method and device, computer equipment and a storage medium. The network fault early warning method comprises the following steps: acquiring network operation data, and preprocessing the network operation data to obtain operation characteristic data; performing prediction processing on the operation characteristic data through a preset fault prediction model to obtain a fault prediction result; the preset fault prediction model is a multi-layer perceptron network model optimized by a pelican optimization algorithm; judging whether the fault prediction result meets a preset early warning condition or not; and if the fault prediction result meets a preset early warning condition, generating fault early warning information according to the fault prediction result, and determining an early warning processing scheme corresponding to the fault early warning information. The preset fault prediction model is easy to construct and train, the prediction effect of the network fault is ensured, the accuracy of fault prediction is improved, early warning is performed in time based on the fault prediction result, the processing scheme is determined, and the network fault processing efficiency is improved.

The application relates to the technical field of performance regulation of recycled concrete, and relates to a recycled concrete performance regulation method and system based on aggregate gradation optimization, which comprises the following steps: confirming a recycled concrete performance regulation environment based on a recycled concrete performance regulation instruction, pretreating engineering slag soil by using an aggregate pretreatment unit, grading the coarse aggregate sand and the fine aggregate sand based on an aggregate gradation optimization unit, pouring recycled lightweight aggregate concrete mixture by using a curing control unit, detecting the compressive strength, apparent density and thermal conductivity of the recycled concrete test piece by using a performance evaluation and regulation unit, and comparing and analyzing the actually measured performance index set with the preset target performance index set, so that the recycled concrete performance regulation of the engineering slag soil is realized based on the regulated recycled concrete. The application can improve the accuracy, efficiency, resource utilization rate and engineering applicability of the recycled concrete performance regulation.

The embodiment of the invention provides a pancreatic cancer ferroptosis related gene prognosis model construction method and device. The method comprises the following steps: acquiring a gene expression data set and a clinical information data set of a pancreatic cancer patient; performing data preprocessing on the gene expression data set and the clinical information data set to generate a standardized gene expression matrix; performing differential expression analysis and survival analysis according to the standardized gene expression matrix in combination with a ferroptosis pathway gene set, and screening out a ferroptosis feature gene set significantly related to prognosis; constructing a deep learning prognosis prediction model based on the ferroptosis feature gene set; and evaluating the performance of the deep learning prognosis prediction model by using the independent verification set to obtain a model evaluation index. The problem that in the prior art, a pancreatic cancer prognosis prediction model is insufficient in accuracy and interpretability can be solved.

The invention provides a cognitive diagnosis method and system based on causal inference, and relates to the technical field of cognitive diagnosis, and the method comprises the steps: obtaining the multi-dimensional features of a student sample and a test question sample; performing preliminary modeling on the multi-dimensional features, inputting the multi-dimensional features to a feature fusion module, and generating a potential representation in combination with an anti-factual reasoning encoder; the potential representation comprises potential feature vectors of different knowledge points; obtaining a first internal relationship between the ability and knowledge of the student and a high-order hierarchical relationship of different knowledge hierarchies by using different attention structures in the initial prediction model, and optimizing the first internal relationship and the high-order hierarchical relationship based on the potential representation to obtain optimization parameters; adjusting the parameters of the initial prediction model based on the optimization parameters to obtain a target prediction model, and inputting the multi-dimensional features of the to-be-tested student and the to-be-tested question into the target prediction model to obtain a cognitive diagnosis result and a score prediction result of the student.

The present application relates to the field of autonomous driving technology, and in particular to a method, device, electronic device and vehicle for predicting vehicle driving intention. The method comprises: obtaining driving state information of a target vehicle at multiple historical moments; determining the historical intention result corresponding to each historical moment according to the driving state information at each historical moment in multiple historical moments; determining the forgetting factor corresponding to each historical moment; the forgetting factor is used to characterize the degree of influence of the historical intention result at the corresponding historical moment on the target prediction result; the target prediction result is used to characterize the driving intention of the target vehicle; and determining the target prediction result according to the historical intention result and the forgetting factor corresponding to each historical moment in multiple historical moments. The above method reduces the requirements for computer computing power and image clarity, and the analysis and adjustment of its related parameters are more convenient. The use of driving state data and forgetting factors at multiple historical moments can improve the accuracy of vehicle driving intention prediction.

The invention discloses a main shaft temperature measuring point optimization method and device and a storage medium, and belongs to the technical field of numerical control machine tool monitoring. The main shaft temperature measuring point optimization method comprises the following steps: starting a machine tool in a no-load state, and enabling a main shaft of the machine tool to operate at a certain rotating speed; collecting temperature data of different temperature measuring points on the main shaft and thermal error data of different thermal deformation measuring points; carrying out clustering division on the temperature measurement points of the machine tool by adopting a K-means clustering analysis algorithm improved based on a genetic algorithm to obtain a temperature measurement point grouping result; selecting typical temperature variables from each group in the temperature measurement point grouping results for combination to obtain temperature variable combination data; substituting the temperature variable combination data and the thermal error data into a multiple regression model to calculate an adjusted judgment coefficient; and determining the temperature measurement point with the maximum judgment coefficient as the optimal temperature measurement point according to the judgment coefficient. According to the method, the representative temperature measuring point of the machine tool spindle can be selected, and the robustness and the prediction effect of the thermal error model are ensured.

The invention discloses a human hand motion estimation method based on fusion of an sEMG wavelet transform image and a CNN-LSTM network, and the method comprises the following steps: 1, carrying out the preprocessing of a surface electromyogram signal, and recognizing an active segment; 2, processing and visualizing a continuous wavelet transform result of the surface electromyogram signal, and synthesizing an instantaneous image at each sampling moment; 3, constructing a CNN-LSTM neural network model; and step 4, inputting the instantaneous image into a CNN-LSTM neural network model for training so as to predict and estimate the movement of the human hand, and outputting a corresponding prediction result. According to the method, no extra information or manual design of a feature extractor is needed in the human hand motion estimation process, the method can be suitable for few-channel signals, the problems of large data processing amount and high cost are solved under the condition that the prediction result is accurate, and the training time is shortened by constructing a CNN-LSTM neural network model.

The invention relates to the technical field of artificial intelligence, and provides a multi-task prediction model establishment method and device and a prediction method and device. The method comprises the following steps: acquiring first target business data; determining a primary task and a secondary task corresponding to the first target business data according to the importance degree of the task; training a first integrated tree model by utilizing the first target business data to obtain a main task model; screening out second target business data matched with the secondary task from the first target business data; and training the main task model by using the second target business data to obtain a main task model after increment fine adjustment, and taking the main task model after increment fine adjustment as a multi-task prediction model. According to the embodiment of the invention, the performance of the multi-task prediction model can be improved, and the multi-task prediction effect is enhanced.