28 results about "Penalty factor" patented technology

The invention relates to the technical field of power load prediction, and discloses a power load prediction method and system based on a VMD and MoE system, and the method comprises the steps: inputting original power load data into a trained VMD decomposer, dynamically adjusting a modal number K value and a penalty factor # imgabs0 # value according to a power load fluctuation level, and decomposing the data into K modal components; thirdly, inputting the modal components into the trained MoE system, distributing M groups of expert models and weights for the modal components through a gating network, and adjusting the number of the expert models of the high-frequency and low-frequency modal components according to the fluctuation level; and finally, at least one group of expert models is selected from the M groups of expert models, prediction results are fused through weighted summation based on the selected model and the weight thereof, and a final power load prediction value is obtained. Through the method, the precision and robustness of power load prediction are improved.

The method comprises the following steps: obtaining a power quality signal containing noise; selecting permutation entropy as an adaptive function of genetic algorithm, calling variational mode decomposition through genetic algorithm, and iteratively optimizing a penalty factor α and a decomposition mode number k of the variational mode decomposition to determine optimal parameters; decomposing signal data into k mode components through the variational mode decomposition, and determining effective mode components and noise mode components through a correlation coefficient; for improved wavelet threshold, a parameter-adjustable threshold function is proposed, and the concept of wavelet energy entropy is introduced into the threshold function; the noise mode components are denoised through the improved wavelet threshold, and the effective mode components and the denoised noise mode components are reconstructed to obtain a denoised power quality disturbance signal. The method can effectively remove noise interference while retaining singular information of mutation points of the collected signal, and provides help for subsequent analysis and treatment of the power quality disturbance signal.

The invention is suitable for the technical field of data processing, and provides a news credibility evaluation method based on a multivariate fusion weighting and nonlinear punishment mechanism, and the method comprises the following steps: carrying out the multi-dimensional scoring of a news article through a large language model; determining a subjective weight matrix of the news article through the AHP expert matrix; objective weights are calculated based on an entropy method and a plurality of pieces of historical news of the same type, and weight fusion is carried out; penalty factors are applied to the key indexes, and weighted scores are adjusted; and summing the weighted scores and carrying out nonlinear normalization to output a total score. According to the method, the news credibility analysis process is divided into a plurality of detailed evaluation dimensions, and the score of each dimension can be independently output, so that a solid technical basis is provided for manual rechecking, intelligent auditing and transparent responsibility asking. And a weight distribution mechanism gives consideration to subjective expert experience and historical data statistics, and through fusion of AHP and an entropy weight method, the problem that single subjective configuration is easily influenced by personal experience and prejudice is overcome, and the accuracy is improved.

The present invention belongs to the technical field of edge computing, and specifically relates to an adaptive risk-resistant Kubernetes cluster scheduling method based on a penalty factor. The present invention resists sudden high-load computing tasks by reserving a certain amount of computing resources and adaptively changing the reserved computing resource pool according to the state of the cluster; the present invention improves the preference algorithm in the default scheduling algorithm of Kubernetes, introduces the mean and standard deviation to evaluate the cluster resource state, and adds corresponding penalty factors to balance the resource states in different dimensions and optimize the scoring rules; at the same time, an influence weight is introduced as a supplement to the scoring rules. When the resource states and scoring scores of two Worker nodes are all the same, the historical scores obtained by this node are incorporated into the scoring rules to preferentially select a more suitable Worker node for this load task, effectively ensuring the load balance of the cluster.

The invention discloses a variational mode decomposition method based on a multi-objective crayfish algorithm, and the method comprises the following steps: S1, constructing an original variational mode decomposition model, determining a parameter design space, and optimizing an objective function; s2, improving a multi-target crayfish algorithm, wherein the multi-target crayfish algorithm comprises the steps of introducing non-dominated sorting to carry out comprehensive evaluation on a plurality of optimization targets; introducing a Pareto optimal frontier based on crowding degree calculation to determine a population updating strategy; s3, establishing a multi-target crayfish algorithm mathematical model, and performing iterative calculation to obtain an optimal parameter combination; and S4, inputting the optimal parameter combination into the variational mode decomposition model to obtain an optimal variational mode decomposition result, and verifying the superiority of variational mode decomposition based on the multi-target crayfish algorithm. The invention discloses a variational mode decomposition method based on a multi-target crayfish algorithm, which overcomes the defects of the existing algorithm, and in variational mode decomposition, an optimal mode number K and a penalty factor alpha are selected to realize optimal variational mode decomposition of signals.

The invention discloses a priori calculation method and device for a penalty multiplication factor for determining constraint strength in a projection type reduced-order model, and solves the problems that an existing value taking mode of the penalty multiplication factor seriously depends on manual empirical adjustment, the efficiency is low, a large amount of redundant calculation is generated due to the fact that the reduced-order model is solved for multiple times, and the calculation cost is high. The calculation cost and the calculation time consumption are obviously increased. The method comprises the following steps: solving a control equation for a parameterization problem under a plurality of representative parameters by adopting a discrete method to obtain a full-order solution sample, thereby obtaining a characteristic mode through intrinsic orthogonal decomposition. A reduced-order model is constructed through a Galerkin projection method, boundary constraints are introduced in combination with penalty multiplication, and then an optimal penalty multiplication factor is obtained through an optimization algorithm. And taking the result as an initial value, searching a minimum value of an error cost function through an optimization algorithm, and adjusting the value of a penalty multiplication factor to realize the balance of calculation cost control and precision improvement.

The embodiments of the present disclosure disclose a method and device for generating information. A specific implementation of the method includes: receiving a target adjacency vector of two target commodities, the target adjacency vector represents the number of times the target commodity and other commodities are purchased at the same time; inputting the target adjacency vector into a pre-trained target coding model to obtain a target embedding vector of the target commodity, the loss function of the target coding model includes a penalty factor, and the penalty factor represents the probability that the two commodities are purchased at the same time; inputting the target embedding vectors of the two target commodities into a pre-trained target information generation model to obtain target information, the target information generation model represents the correspondence between the similarity of the two target embedding vectors and the probability that the two target commodities are purchased at the same time, and the target information represents the probability that the two target commodities are purchased at the same time. The target embedding vector can reflect the deep characteristics of the target commodities being purchased at the same time, and improve the accuracy of predicting the probability of commodities being purchased at the same time.

The invention provides a training method of a decision tree, and relates to the technical field of computers, in particular to the technical field of the training method of the decision tree, a method for determining a risk user and machine learning. According to the specific implementation scheme, a sample set containing user behavior characteristics is constructed; respectively calculating the value of the Gini index of each user behavior feature in the sample set; training an initial decision tree based on a penalty factor function and the value of the Gini index of each user behavior feature to obtain a trained decision tree; wherein the trained decision tree is used for determining risk users according to user behavior characteristics. The recall effectiveness can be improved, and then missing detection of risk users can be avoided.

The invention discloses a reasonable penalty factor setting method and system based on unit quotation and sensitivity. The method comprises the following steps: reading unit quotation, a power flow transfer factor and section data related to a section and a line; calculating a single-period section shadow price based on the section cost; different types of section punishment shadow prices are calculated; performing penalty factor limitation setting of the day-ahead market or real-time market power generation dispatching optimization model based on a duality theory; calculating and setting penalty factors under multi-stage requirements; a penalty factor is selected and added into a scheduling and pricing model for calculation, an optimization model of the spot market is solved according to input data under the condition of following corresponding standby constraints, and reasonable unit clearing amount and market bid winning price are obtained. Based on the quotation and the sensitivity factor between the devices, the power dispatching optimization calculation efficiency and the spot market quantity-price matching degree can be improved according to actual demand setting.

The invention discloses an electric energy quality signal noise reduction and optimization detection method and system and application thereof, and the method specifically comprises the following steps: S1, dynamically optimizing a penalty factor delta and a mode number beta of variational mode decomposition (VMD) through a particle swarm optimization (PSO) algorithm, constructing a fitness function with a composite index (permutation entropy and mutual information) as a core, and carrying out the dynamic optimization of the penalty factor delta and the mode number beta of the variational mode decomposition (VMD) through the particle swarm optimization (PSO) algorithm; and screening an optimal decomposition parameter. S2, performing VMD decomposition on the power quality signal according to the optimized decomposition parameters delta and beta to obtain a plurality of intrinsic mode function components, namely IMF components; s3, screening an optimal signal component from the IMF components based on the composite index to realize signal noise reduction; according to the method, the parameters of VMD decomposition can be optimized for the electric energy quality signal, so that noise interference in the signal is overcome, the problem of low signal-to-noise ratio of the signal is solved, finally, the detection precision of the electric energy quality is effectively improved, and the measurement error is below 2%.

The application discloses a pipeline leakage signal denoising method based on improved WOA-VMD combined with SVD, relates to the technical field of signal processing, and acquires a pipeline leakage signal of a pipeline defect position; a penalty factor alpha and a mode number K in a VMD algorithm are used as optimization targets of a whale optimization algorithm (WOA) algorithm; the VMD algorithm is set according to the optimal penalty factor alpha and the optimal mode number K obtained by the WOA algorithm, and VDM decomposition is carried out on the pipeline leakage signal to obtain K IMF components; the K IMF components are identified by using a correlation coefficient method, and effective IMF components are identified; SVD decomposition and denoising processing are respectively carried out on each effective IMF component by using an SVD algorithm, and noise signals in each effective IMF component are removed; and each effective IMF component after the denoising processing is synthesized to obtain a pipeline leakage signal after denoising. The application combines the WOA, VMD and SVD algorithms, forms an integrated signal processing framework, and has a significant improvement in performance compared with a single method.

The invention discloses a multi-energy load prediction method, system and device for an integrated energy system and a medium, and the method comprises the following steps: obtaining meteorological factor data and historical load data of the integrated energy system, and enabling meteorological factors corresponding to Pearson's correlation coefficients meeting conditions to form an input feature set; searching an optimal decomposition layer number and a penalty factor of variational mode decomposition in the parameter space, and performing variational mode decomposition on each type of load sequence by using the optimal decomposition layer number and the penalty factor to obtain an intrinsic mode component; combining the input feature set with an intrinsic mode component to construct a prediction model; and verifying the prediction model until the prediction model meets a set standard. According to the method, the meteorological factors are screened through the Pearson's correlation coefficient, redundant factors with relatively low load correlation are eliminated, the input feature dimension is further reduced, and the calculation complexity of model training is reduced.

The invention provides a train bogie axle box bearing composite fault intelligent diagnosis method based on sound signals in combination with deep learning. The method comprises the following steps: firstly, aiming at the problem of more noise of a sound signal of a composite fault, carrying out global optimization on a parameter combination in a VMD decomposition process through an FGO optimization algorithm, determining two parameter combinations of a modal component and a penalty factor, and decomposing an original sound signal into K intrinsic mode function components (IMF) through an optimized variational mode (VMD) to carry out noise reduction processing; and finally, inputting the decomposed sound signals into a Transform-BiGRU architecture neural network model, and carrying out classification and identification on various composite faults of the axle box bearing.

The invention belongs to the technical field of aircraft design, and particularly relates to a layout optimization design method for airfoil reinforcing ribs. The traditional airfoil reinforcing rib layout design seriously depends on the engineering experience of a designer, a reference prototype and a series of'design-check-modification 'trial and error processes. According to the method, through the SIMP model and the penalty factor, the mathematical solution problem of the '0-1' discrete optimization problem is ingeniously solved, and the final result is clear. Through combined control of the volume constraint and the filtering technology, the sparse rib layout with the minimum material consumption and the minimum number can be naturally generated, and the layout is continuous and smooth and can be easily converted into an actual engineering component. The efficiency of the gradient-based optimization algorithm in processing large-scale variables is far higher than that of a genetic algorithm and other gradient-free algorithms, so that the method is suitable for engineering practice.

This invention relates to the field of aircraft propulsion system technology, and discloses a comprehensive evaluation and decision-making method for multiple engine concept design schemes. By introducing game theory concepts, it achieves deep coupling of subjective and objective weights, and combines the maximum entropy criterion and deviation penalty factor to seek the optimal compromise between expert experience judgment and the inherent laws of data, thereby obtaining the optimal combination of weights. Based on this, a quantifiable comprehensive evaluation index is constructed to achieve a comprehensive and objective assessment and ranking of engine concept design schemes. Simultaneously, the Euclidean distance considering the standard deviation of technical indicators is introduced into the ranking of alternative schemes, enhancing the discriminative power and robustness of the evaluation results, effectively solving the problem of the difficulty in comprehensively balancing multiple technical indicators such as performance, structure, and cost. This invention can identify and screen the most promising innovative schemes, providing decision support for subsequent component design, thereby improving the overall engine performance and reasonably controlling R&D costs.

The application discloses a text generation method and device, equipment and storage medium, and belongs to the technical field of artificial intelligence. The method can avoid the generation of unnecessary repetitive content by setting optimization strategies in the training stage and the inference stage. In detail, in the training stage, a penalty term is added to the loss function, wherein the penalty term is related to the first penalty factor and the content repetition degree of the predicted text output by the large model. That is, if the large model attempts to generate unnecessary repetitive content, the large model will be greatly punished due to the existence of the penalty term, thereby encouraging the large model to generate different content. In the inference stage, the large model remembers the generated content during the output generation word process, and uses the content repetition degree as the judgment standard to prevent the large model from generating unnecessary repetitive content by adjusting the generation probability.

The invention discloses an integrated energy system multi-load prediction method and device, equipment, a medium and a product, and relates to the field of integrated energy system load prediction, and the method comprises the steps: obtaining historical time series data of multi-element loads, and carrying out the preprocessing; carrying out adaptive optimization on the penalty factor; the optimized penalty factor is used for executing MVMD decomposition on the multi-element load historical time sequence data; calculating the sample entropy SE of each IMF component and the residual component, setting a threshold value according to the SE value, and reconstructing all the components into a high-frequency sub-sequence, a low-frequency sub-sequence and a residual sub-sequence; an MVMD-SF-DMS multi-element load prediction model is constructed and trained; utilizing a DMS mechanism to dynamically select an optimal base model for prediction to obtain a prediction result of each subsequence; and carrying out weighted summation on the prediction result according to the original reconstruction weight to obtain a final power, heat, cold and gas multi-element load short-term prediction value. According to the method, the multi-element load prediction precision and robustness are effectively improved, and reliable data support is provided for optimal scheduling of the integrated energy system.

The application discloses a railway track circuit frequency identification and fault detection method, utilizes a whale optimization algorithm to adaptively optimize the mode decomposition number and a penalty factor of variational mode decomposition, takes the minimum local envelope entropy as an adaptability evaluation index, and obtains an optimal parameter combination; based on the combination, variational mode decomposition is performed on a track circuit signal, valid mode components are screened from obtained intrinsic mode functions according to a preset screening criterion, each valid mode component is combined as an independent channel in a channel dimension, a multi-channel one-dimensional feature matrix retaining time sequence information and frequency component information is constructed, the matrix is input into a one-dimensional convolutional neural network feature extraction model, convolution feature extraction is performed along a time axis, a fixed-length one-dimensional high-order feature vector is output through a pooling layer or a full connection layer, the vector is input into an integrated classification model based on gradient boosting for classification, and a frequency category or a fault type is output.

The invention relates to a resource allocation method and device for public test tasks, equipment, a storage medium and a program product. The method comprises the steps of obtaining a to-be-tested public test task, wherein the to-be-tested public test task comprises task description data; inputting the task description data into a preset index scoring model for processing, and calculating to obtain a multi-dimensional evaluation result, a target skill label and a vulnerability category of the to-be-tested public test task; determining a dynamic regulation factor according to the multi-dimensional evaluation result, the target skill label and the vulnerability category; the dynamic regulation factor comprises a market supply and demand coefficient and a risk penalty factor; and allocating corresponding resources to the to-be-tested public test task according to the multi-dimensional evaluation result and the dynamic adjustment factor. By performing multi-dimensional evaluation on the crowd test task, effective identification of the self state of the crowd test task is realized, the dynamic regulation factor is determined based on the self information of the crowd test task, accurate identification of the external environment is realized, the incentive limit can be accurately generated, and the accuracy of the allocation result is improved.

The application provides a random noise suppression method, device, equipment and storage medium, wherein the method comprises the following steps: configuring the number of components and the penalty factor parameters of variational mode decomposition by using a genetic algorithm; and suppressing random noise by using the variational mode decomposition with the configured parameters, so as to improve the quality of seismic data. The method provided by the application suppresses the random noise in the seismic data by GA-VMD, and improves the quality of the seismic data. The GA-VMD realizes intelligent optimal configuration of the VMD parameters by using GA, reduces the workload of people, avoids the decomposition error caused by the subjective factors of people, and thus improves the noise suppression effect, and creates a good data basis for subsequent inversion and interpretation.

The invention relates to a time-varying reliability analysis method for optimizing SVR based on a dragonfly algorithm. The invention discloses a time-varying reliability analysis method for optimizing SVR based on a dragonfly algorithm, and the method comprises the steps: carrying out the normalization processing of a selected initial sample, inputting the preprocessed sample into a dragonfly algorithm in which a structural risk-oriented objective function is embedded, and carrying out the global optimization of a penalty factor and a kernel coefficient of SVR, and constructing a DA-SVR model by using an optimal penalty factor and an optimal kernel coefficient output by a dragonfly algorithm, carrying out double convergence criterion discrimination on the DA-SVR model, screening the processed samples through a sample adaptive screening strategy, and carrying out adaptive updating on the DA-SVR model until a converged DA-SVR model is obtained. And outputting a predicted value of the time-varying reliability failure probability. According to the time-varying reliability analysis method for optimizing the SVR based on the dragonfly algorithm, the sample demand quantity of the time-varying reliability agent model can be remarkably reduced, the model training efficiency and precision are effectively improved, and the reliability analysis efficiency and analysis precision of the time-varying event are improved.

The invention relates to the technical field of Internet information processing, and discloses a news optimization method, a terminal and a storage medium. The method comprises the following steps: acquiring a news set to be processed, and extracting entities and events contained in each piece of news to form an information unit set; constructing a binary coverage matrix between the news set and the information unit set; based on the binary coverage matrix, defining a coverage function and a redundancy function, and introducing a redundancy penalty factor to construct a news optimization objective function; converting the news optimization objective function into a linear programming model; and constructing a heuristic algorithm based on a Lagrange multiplier to solve the linear programming model, and iteratively selecting news by calculating marginal comprehensive increment of candidate news until a preset stop condition is met, thereby obtaining an optimal news set. According to the method, under the constraint of limited news quantity, the coverage of key enterprises and event information units is maximized, and meanwhile, the redundant influence caused by repeatedly reporting the same information by multiple pieces of news is reduced.

The invention relates to a power resource data scheduling method, and the method comprises the steps: obtaining the operation data of a target object, and the operation data comprises the power resource data generated by the target object and the power resource data consumed by the target object; constructing a power resource data scheduling objective function based on the operation data; obtaining a constraint condition of the electric power resource data objective function; constructing a Lagrange function based on the target function, the constraint condition, a preset Lagrange multiplier and a preset penalty factor, wherein the Lagrange function is used for decomposing the target function into a plurality of sub-target functions; and carrying out iterative calculation on the plurality of sub-objective functions, obtaining a target solution of the objective function under the condition that an iterative result meets a preset convergence condition, and obtaining a scheme of generating the power resource data by the target object and a scheme of consuming the power resource data by the target object based on the target solution.

The application discloses a kind of self-adapting optimization methods for the collaborative operation of multi-community integrated energy system, comprising: establishing the feasible region of safe operation of power transformer, for evaluating the load capacity of oil-immersed transformer;Establish a rolling optimization scheduling model of multi-community integrated energy system, with the minimum total operation cost of the system as the objective function, and impose community operation constraints;Using an improved adaptive ADMM algorithm based on spectral penalty factor to distribute the rolling optimization scheduling model for iterative optimization, to obtain the optimal scheduling strategy.The method of the present application provides a fast distributed collaborative optimization method with higher safety and new energy consumption rate for the operation of multi-community integrated energy system, and can protect the privacy and security of each subject.

This paper proposes a PSO-VMD-LSTM-based oil well production forecasting method. This method builds on the traditional VMD-LSTM forecasting method by applying the PSO algorithm to both the VMD model and the LSTM network. Specifically, the PSO algorithm optimizes the VMD model to obtain the optimal variational mode decomposition (VMD) and penalty factor, avoiding the errors caused by observational VMD and penalty factor calculations. Simultaneously, the PSO algorithm optimizes the LSTM network, enabling optimal parameterization, further improving the accuracy of the forecasting method.

The invention relates to a test case generation method and system, and the method comprises the steps: carrying out the change weighting of all change line numbers, change timestamps, change depths and change operators, which are extracted from a Git change record of a current version, through employing a time exponential decay model; meanwhile, a penalty factor of author diversity entropy generated by a change operator is introduced in the weight change process to obtain row-level change popularity, and a code calling graph, a code inheritance graph and a data dependency graph are extracted from a complete code library of the current version through a static analysis tool to construct a code heterogeneous graph; the method comprises the following steps: establishing a large language model, introducing line-level change popularity and carrying out weight aggregation to generate an influence map, extracting and executing a hotspot method from the influence map to generate a to-be-covered path, and inputting the to-be-covered path into the large language model to output a recall test case and a newly-added test case. Therefore, the method not only has the capability of generating the new test case, but also ensures the recall rate and the accuracy of the test case.

The present invention relates to a parameter optimization bearing fault diagnosis method and system based on FOA-VMD, which includes: using the fruit fly algorithm to perform a global optimal combination search on the VMD decomposition parameters of the penalty factor and the number of decompositions, selecting the information entropy increment combined with the kurtosis index as the objective function to obtain the optimal parameter combination; decomposing the original fault vibration signal by VMD after obtaining the optimal parameter combination to obtain several intrinsic mode components, and selecting the best component; performing envelope analysis processing on the best component, and further processing the best component by using the 1.5D spectrum to eliminate the interference of other components and highlight the fault characteristic frequency and its multiple frequencies, so as to obtain the fault type of the bearing. The present invention can effectively obtain the fault characteristics of the rolling bearing and improve the recognition accuracy of the fault characteristics. The present invention can be applied in the technical field of mechanical equipment fault diagnosis.

The invention relates to the technical field of harmonic detection, in particular to a power quality disturbance detection algorithm based on IGWO-VMD-DWT, and the algorithm comprises the steps: decomposing an intrinsic mode function in a variational mode decomposition algorithm; optimizing the decomposition number k and the penalty factor a of the intrinsic mode function by adopting an improved grey wolf optimization algorithm; and decomposing the harmonic-containing characteristic signal into k intrinsic mode function components, and carrying out time-frequency analysis on the k intrinsic mode function components by adopting a discrete wavelet algorithm.