236 results about "Interval prediction" patented technology

The invention discloses a long-tail SKU sales prediction method and system, and belongs to the technical field of sales prediction, and the method comprises the following steps: obtaining SKU sales feature data, and carrying out the preprocessing of the sales feature data; obtaining a multi-dimensional feature table based on the standardized training data set; a rule engine and a lightweight classification model are used for joint judgment to classify the multi-dimensional feature table, and four SKU category labels and probability vectors thereof are output; differentiated training models of the four SKU category labels are established, different types of SKUs are routed to the corresponding training models, and SKU sales volume point predicted values are output; probability calibration is carried out, and a plurality of quantiles are calculated to form interval prediction; establishing an SKU multi-hierarchy relationship, and performing hierarchy consistency solution on prediction results among multiple hierarchies; in the online operation process of the prediction model, input data distribution drift is monitored, and model updating is triggered when the input data distribution drift exceeds a threshold value. According to the method, the prediction precision and stability of the long-tail SKU are improved, and rapid adaptive prediction of sudden and off-peak demands is realized.

The invention discloses a photovoltaic power interval prediction method based on a GRU-LSTM combined neural network, and belongs to the technical field of photovoltaic power interval prediction. The method comprises the following steps: S1, taking historical power generation data as original wind-solar power generation power prediction data, processing the data, and screening related meteorological characteristics by adopting a Pearson correlation coefficient; s2, a GRU-LSTM combination model is constructed, and related hyper-parameters are set; s3, taking the screened related meteorological features as input for training, calculating a photovoltaic point prediction result according to a weight coefficient, and performing related error evaluation; and S4, based on the photovoltaic power point prediction result, calculating a photovoltaic power interval prediction result by using a quantile regression technology, and detecting performance evaluation through a test set. According to the method, the minimum prediction error correlation index is taken as the target, the influence of different weathers on photovoltaic power processing is considered, the Pearson's correlation coefficient analysis is utilized to select more representative meteorological characteristics, and the combined model and the quantile regression technology are utilized to finally obtain the photovoltaic power interval prediction result.

The invention relates to the technical field of ship energy consumption prediction, and discloses a ship energy consumption interval prediction method and system based on a Gaussian quantile regression model.The method comprises the steps that firstly, a ship navigation historical data set is preprocessed, a model input feature set is screened, a Gaussian process quantile regression model with a radial basis function as a kernel is constructed, and hyper-parameters are optimized; and after the prediction performance of the subset evaluation point is verified through the model, an upper quantile prediction model and a lower quantile prediction model are respectively established according to a target confidence level, and finally a prediction interval is synchronously output and an evaluation report is generated. According to the method, through combination of Gaussian process processing nonlinear relation and quantile regression estimation condition distribution, high-precision point prediction is provided, meanwhile, prediction uncertainty can be quantized, an energy consumption prediction interval corresponding to a target confidence level is output, and more comprehensive and reliable information support is provided for ship energy efficiency management and operation decision making.

The invention belongs to the technical field of artificial intelligence, and particularly relates to a wind power prediction method based on multi-objective optimization, and the method comprises the steps: wind power data collection and training data set construction, adaptive sliding window and dynamic fluctuation decomposition of wind power time series data, construction of a short-term power prediction model, and short-term power prediction. According to the method, the historical window length and the decomposition scale can be autonomously adjusted according to the inherent fluctuation characteristics of the data, and multi-scale accurate characterization of the non-stationary power sequence is realized; according to the method, a dynamic space-time diagram fusing geographic distance and instantaneous power correlation is constructed, and a diagram attention network combined with trend similarity gating is designed, so that dynamic refined modeling of a space incidence relation is realized; according to fluctuation intensity self-adaptive loss function dynamic balance point prediction precision and interval prediction reliability, synchronously outputting deterministic and probabilistic prediction results; the rated power limit and the ramp rate constraint are embedded into the model in a soft mode, and it is ensured that the prediction result conforms to the actual operation rule of the wind turbine generator.

The invention relates to the technical field of power battery life prediction, and discloses a power battery residual life prediction method based on sensor fusion. The method comprises the steps of collecting and processing multi-dimensional electrical sensor signals of historical operation of a battery, and generating a voltage platform change characteristic spectrum, a current stress characteristic spectrum and an internal resistance component evolution spectrum. And inputting the maps into a feature space alignment network for space-time dimension registration and feature cross validation to generate a unified electrochemical state panoramic feature map. Based on the atlas, multi-stage battery life state backtracking and deduction are realized by constructing a dynamic attenuation trajectory tree, and a battery aging state vector set is output. After working condition constraint correction, the set drives a residual life interval prediction module, boundary estimation and probability density propagation calculation are fused, and finally a confidence interval of the residual cycle life and probability density distribution of the confidence interval are generated. According to the method, the precision of multi-source information fusion and the reliability of life prediction are improved.

The invention belongs to the technical field of electric power transaction, and particularly relates to an electric power spot transaction clearing and bidding optimization method based on computing power prediction, which comprises the following steps: collecting and preprocessing multi-source data, and screening a core feature set through mutual information entropy, Pearson's correlation coefficients and variance expansion factors; a CNN-LSTM attention model is built, and output load, power output and electricity price fluctuation interval prediction results are calibrated through error feedback; a dynamic bidding decision mathematical model is constructed, and elastic constraints are set in combination with prediction parameters; solving by adopting an improved particle swarm algorithm, and outputting an optimal bidding strategy; a node marginal electricity price mechanism is fused for accurate clearing; and verifying a clearing result in a layered manner, and performing deviation feedback iterative optimization. According to the method, a whole-process closed-loop mechanism is constructed, the prediction precision and the bidding and clearing collaboration are improved, the system operation cost is reduced, the transaction real-time requirement is met, the power grid safety and the market subject income are guaranteed, and the method has important practical value.

The invention is suitable for the technical field of power system scheduling, and provides a new energy access provincial power grid multi-time scale intelligent scheduling method, which comprises the following steps of: obtaining related data of a provincial power grid, generating a new energy power point prediction curve of a day-ahead time scale, a new energy power fluctuation interval prediction result of an intra-day time scale, a new energy power point prediction curve of a real-time time scale and a new energy power fluctuation interval prediction result of the current moment by using the trained multi-time scale prediction model; processing by the day-ahead deep reinforcement learning agent, and outputting a day-ahead scheduling instruction; processing by the intraday deep reinforcement learning agent, and outputting an intraday adjustment instruction; and a real-time control instruction is output after the real-time deep reinforcement learning intelligent agent processes the real-time control instruction. According to the method, global optimization from day-ahead robust planning and intra-day active prevention to real-time multi-target cooperation is realized, and the cooperative scheduling control level of scheduling decision is improved.

The invention discloses an electric power spot day-ahead electricity price prediction method based on bidding space analysis, and the method comprises the steps: constructing input features, obtaining electric power market data, meteorological data and date data, and constructing a multi-dimensional input feature system; data preprocessing: carrying out missing value processing, abnormal value elimination, feature normalization and category feature coding on the data in the multi-dimensional input feature system to form a standardized feature matrix; model training: carrying out training by adopting a deep learning model, extracting a multi-scale nonlinear relationship between the bidding space and the electricity price, and fusing time sequence dependence and global feature interaction; and electricity price prediction: outputting a point prediction result and an interval prediction result of the electricity price by using the trained model. According to the method, a multi-dimensional feature system is constructed, and a model is used for training after preprocessing so as to synchronously output day-ahead electricity price point prediction and interval prediction. And high-precision and self-adaptive high-value electricity price prediction with risk assessment capability is realized.

The invention relates to the technical field of power system optimization scheduling, and discloses a source network load storage optimization scheduling method suitable for flexible supply and demand balance in extreme weather. The method comprises the following steps: firstly, acquiring new energy power interval prediction data which aims at typical extreme weather scenes and is subjected to error correction; on the basis, the flexibility requirement of the system and the supply capability of multi-type resources are quantified; and furthermore, with maximization of the comprehensive benefit of the system as a target, a flexible vacancy virtual penalty term for stabilizing fluctuation risks caused by extreme weather is particularly introduced, an optimal scheduling model is constructed and solved, and a scheduling plan is generated. According to the method, by fusing extreme weather exclusive prediction information and an operation risk hedging mechanism, efficient coordination of source network load storage resources in extreme weather is realized, the new energy power abandoning rate is effectively reduced, the dependence on traditional thermal power is reduced, and the comprehensive operation benefit and safety margin of the system are improved.

The invention discloses a regional wind-solar combined power interval prediction method and system based on multi-task learning and a hybrid neural network, and relates to the technical field of combined prediction. Meteorological data and wind-solar historical power data of a to-be-researched region are collected, and a historical data set is constructed; preprocessing the historical data set, and performing feature selection by adopting a predictive ability scoring algorithm to obtain screened meteorological data and historical power data; constructing a joint power interval prediction model; and performing combined power generation prediction based on the trained combined power interval prediction model. According to the method, the space-time correlation and the complementary relation of wind energy and photovoltaic energy are excavated, the joint power interval prediction model is constructed, and the power prediction precision is improved.

The power system inertia prediction method based on the variational Bayesian attention normalization stream comprises the following steps: constructing a system inertia data set; preprocessing the system inertia data set, and dividing the preprocessed data set into a prediction set and residual data according to a time range; respectively generating Q, K and V by adopting variational Bayes, carrying out relative position coding on the generated vectors, and carrying out weighted fusion on the Q, K and V after position coding by utilizing a multi-head attention mechanism; inputting the output of the multi-head attention mechanism into an attention normalization flow model, and capturing complex probability distribution of data through reversible transformation; carrying out model training by adopting a self-defined mixed loss function; and performing multiple Monte Carlo sampling on the trained model to obtain a sampling prediction set, calculating a prediction mean value and a standard deviation based on the sampling prediction set to obtain a significance level, and constructing an interval prediction result under a corresponding confidence interval. According to the prediction method, accurate probability prediction of the inertia of the power system is realized.

The invention discloses an assembly deviation interval rapid prediction method based on fuzzy comprehensive evaluation, and belongs to the field of assembly deviation interval prediction in the digital assembly process of aviation complex structural parts. The method comprises the following steps: constructing a simulation model of a target assembly body, executing Monte Carlo analysis, and extracting a manufacturing deviation value and an assembly deviation simulation value of a sampling sample; constructing a manufacturing deviation uncertainty quantitative model based on an assembly success rate function, wherein the assembly success rate function is obtained through probability density function integration and is defined as a fuzzy membership function; establishing an evaluation system containing three indexes of'over-small, moderate and over-large ', and calculating an assembly deviation comprehensive evaluation value; constructing a point cloud mapping model based on the assembly deviation simulation value of the sample and the assembly deviation comprehensive evaluation value, and fitting an assembly deviation prediction interval through a linear boundary function; and the accuracy of the prediction interval is verified through measured data. The method can rapidly and accurately predict the assembly deviation interval, reduces the calculation amount, and reflects the individual difference.

The invention belongs to the technical field of new energy station power prediction, and particularly relates to a new energy station power interval prediction method. In order to overcome the defect that the prediction accuracy and precision of an existing new energy station power interval prediction method have the improvement space, the invention adopts the following technical scheme: the new energy station power interval prediction method comprises the following steps: constructing a VMD-SA-TCN-BiLSTM-KDE prediction model; historical output power data of a wind power plant station are obtained, the historical output power data are preprocessed and then input into SA-TCN-BiLSTM for training, and the preprocessing comprises the step of decomposing the data through VMD; the part of historical output power data obtained after preprocessing is input into the trained SA-TCN-BiLSTM, and a prediction result is obtained; and performing KDE kernel density estimation on the prediction error, and obtaining a historical output power prediction interval of the wind power plant station according to a prediction result and the prediction error. The method has the beneficial effect of higher prediction precision.

The invention discloses a photovoltaic power prediction method based on an adaptive modified quantile regression neural network. Feature extraction is realized by constructing a double-flow hybrid neural network so as to improve prediction accuracy, a branch, combined with a multi-head attention mechanism, of the convolutional neural network is responsible for extracting long-term features, a branch of a bidirectional gating circulation unit focuses on identifying short-term fluctuation, and the double-flow hybrid neural network is combined with quantile regression. In order to solve the problems of quantile crossing and non-differentiable zero point of a loss function, a self-adaptive correction marble loss function is provided, and smooth function optimization is introduced to ensure monotone increasing of predicted quantiles and whole-domain differentiable of the loss function. According to the method, point prediction, interval prediction and probability density prediction can be realized, the prediction effect is verified through a multi-dimensional evaluation index, potential information of photovoltaic power is fully mined, and the method has practical engineering application value.

The invention relates to the technical field of chemical production, and discloses a method and device for constructing a polyethylene content prediction model in ethylene oligomerization reaction, and the method comprises the steps: obtaining a training data set; labeling each piece of training data according to the polyethylene content measured value in each piece of training data; dividing the training data set into a first subset and a second subset according to the label of each piece of training data; training a preset first machine learning model by using the labeled training data set to obtain a classification sub-model; and training a preset second machine learning model by using the second subset to obtain a prediction sub-model. A double-layer modeling strategy of a classification sub-model and a prediction sub-model is adopted, and the classification sub-model can quickly distinguish catalysts with high and low polyethylene generation risks; the prediction sub-model focuses on refined regression training of low polyethylene content samples, and interference of high content samples on low value interval prediction precision is avoided.

The invention relates to the technical field of energy consumption interval prediction, and particularly discloses a building energy consumption interval prediction method, system and equipment, and the method comprises the steps: collecting historical energy consumption data, meteorological data, building characteristic data and use mode data of a target building, and carrying out the hierarchical processing, thereby obtaining a hierarchical data set; performing wavelet transform decomposition on the historical energy consumption data, and decomposing the energy consumption time sequence into a trend term, a periodic term and a random term to obtain multi-scale decomposition features; constructing a feature evaluation model, calculating a contribution degree and dynamically allocating weights to obtain a weighted feature set; constructing an integrated prediction framework based on the weighted feature set, and generating a prediction interval through quantile regression; and performing adaptive adjustment according to the historical prediction deviation and the environmental factor change to obtain a final prediction interval. Through multi-scale feature processing, dynamic weight optimization and prediction interval adaptive calibration, the accuracy and reliability of the prediction interval are improved, and more accurate support is provided for building energy management.

The invention discloses a power grid state prediction method under high wind power permeability based on multi-source data fusion, and the method comprises the steps: firstly analyzing the correlation between variables through feature engineering and a Pearson's correlation coefficient, and expanding a data set through a sliding time window to mine time domain information; establishing a graph convolution network model of an adaptive adjacency matrix, dynamically capturing the relationship between fans, and extracting spatial features through multilayer convolution to realize information transmission; and carrying out probability density analysis on the wind power prediction error by adopting a kernel density estimation method, determining an error interval boundary value, and realizing interval prediction by combining a deterministic prediction result. The system based on the method comprises a data preprocessing module, a model building module and an uncertainty analysis module which are matched with a processor and a memory to operate. According to the scheme, the accuracy and stability of power grid state prediction in a high wind power permeability scene can be effectively improved, the certainty and the interval prediction result are considered, and a reliable basis is provided for power grid dispatching and optimized operation.

The invention discloses a power system supply and demand imbalance early warning method based on interval prediction, and belongs to the technical field of power system automation, and the method comprises the following steps: S1, data collection; s2, constructing a double-layer asymmetric prediction model; s3, scene construction; and S4, simulating an early warning model. According to the power system supply and demand imbalance early warning method based on interval prediction, in a renewable energy access power system, the potential supply and demand imbalance risk is effectively identified and predicted, accurate risk early warning is provided for power system dispatching, and safe and stable operation of the system is guaranteed.

The invention provides an energy storage battery SOC estimation method and system and an electronic device, and the method comprises the steps: obtaining the voltage, current and temperature data of an energy storage battery, and carrying out the processing of the data; performing feature data extraction on the processed data by using a two-dimensional convolutional neural network to obtain first feature data; introducing an attention mechanism to enable a neural network to pay attention to important information in the first feature data to obtain second feature data; and inputting the second feature data into the QRLSTM for regression prediction, thereby realizing SOC point and interval prediction of the energy storage battery. The system is used for realizing the SOC estimation method. The electronic device includes a memory, a processor, and a computing program stored in the memory and executable on the processor. According to the method, the prediction precision is improved, a more comprehensive and deep data analysis tool is provided for an energy storage battery management system, and the safety and efficiency of energy storage battery use are effectively improved.

The invention relates to an interval value carbon price prediction method based on multi-source data fusion and deep learning integration, and the method comprises the following steps: obtaining the actual historical transaction data of a carbon market, collecting the maximum price and the minimum price of the daily actual transaction of the carbon market, and determining an interval value carbon price sequence; key external driving factors influencing the carbon transaction price are screened through feature importance analysis, and data alignment is conducted on the key external driving factors of different market transaction days through a cubic spline interpolation method; cEEMDAN-VMD dual decomposition is adopted, and higher-precision multi-scale characterization of an original interval value carbon price sequence is achieved; and by taking the decomposition result and the determined external key driving factor as input and the carbon price prediction interval as output, constructing a carbon price interval prediction model, and finally outputting to obtain the carbon price prediction interval containing upper and lower boundary values. The method can effectively cope with the non-stationary and non-linear characteristics of the carbon valence sequence, and is good in expansibility and adaptability.

The invention discloses a power load probability prediction method and system based on a neural network quantile regression model and multiple linear regression, and belongs to the technical field of power system load prediction. The method comprises the following steps: obtaining standardized data by using a longitudinal data analysis method; identifying key influence factors of the power load in the standardized data based on a Pearson correlation analysis method, and constructing a factor analysis model to quantify influence weights of the key influence factors on the power load; constructing a neural network quantile regression model based on seasonal trend decomposition to fit the key influence factors with different influence weights to obtain a quantile prediction result; based on the quantile prediction result, estimating a continuous probability distribution curve of a load common factor by adopting a non-parametric kernel density technology to obtain an interval prediction result; and constructing a multiple linear regression model to predict the load scale change, and adjusting the interval prediction result. According to the invention, the precision and calculation efficiency of load prediction are effectively improved.

The invention relates to the technical field of energy-saving optimization, in particular to an energy-saving quantity prediction and energy efficiency optimization method in a simulation environment, and the method achieves the real-time correction through defining a dynamic simulation-field mismatch factor and comparing a simulation predicted cooling capacity with an actually measured cooling capacity, and guarantees that the energy-saving prediction is closer to a real field operation state. And a thermal inertia-exogenous driving response index is designed, and the dynamic response of passenger flow disturbance to the cooling capacity demand is quantified, so that the heat storage effect and control lag of the building are explicitly described, and the physical reasonability of prediction is improved. A simulation consistency-hydraulic matching index and a thermal inertia response index are fused into a core feature, a coupling penalty factor is introduced, a measurable credibility function is formed, and a risk boundary is provided for energy-saving prediction. Based on the corrected simulation energy consumption difference, energy-saving quantity point estimation is calculated in combination with core characteristics, an instant energy-saving predicted value is output, interval prediction can be expanded, and the requirements of operation and maintenance decision making and energy-saving auditing are met.

The application belongs to the field of micro-grid optimal scheduling, and discloses a micro-grid dynamic interval optimal scheduling method considering source-load uncertainty, comprising the following steps: step 1, photovoltaic power prediction; step 2, photovoltaic power dynamic interval prediction; step 3, power load prediction; step 4, load dynamic interval prediction; and step 5, dynamic interval optimal scheduling of the micro-grid considering source-load uncertainty, wherein a dynamic interval optimal scheduling model of the micro-grid considering source-load uncertainty is first established, a target function and a constraint condition of the dynamic interval optimal scheduling model of the micro-grid are further established, and an optimal scheduling result of the micro-grid is obtained by solving. The application uses a ConvLSTM-Attention-LSTM fusion prediction model to perform photovoltaic power prediction, and the model can more accurately capture the spatial and temporal dependencies in time series data, thereby improving the prediction accuracy.

The invention provides a clearing price interval prediction method based on an LSTM network and a similar point matching algorithm, and the method comprises the following steps: carrying out the optimization of an independent variable feature set, taking a real-time clearing price as a target variable based on the historical data of an electricity market at an interval of 15 minutes, constructing a point prediction model based on the LSTM network, and carrying out the calculation of the point prediction model. On the basis of optimization of an independent variable feature set, a long-short-term memory neural network is adopted to construct a real-time clearing price point prediction model, high-precision point prediction, similar point matching and statistical feature extraction are realized, feature vectors of to-be-predicted time points are constructed to be used for matching similar time points in a historical data set, and a real-time clearing price point prediction model is obtained. And constructing upper and lower limits of the prediction interval. A real-time clearing price point prediction model is constructed based on an LSTM network, and an upper limit and a lower limit of a real-time clearing price prediction interval are formed in combination with a similar point matching algorithm and an interval feature statistical analysis method, so that uncertainty in an electricity market is quantified better, and more decision bases are provided for market participants.

The invention discloses a cross-scene self-adaptive flexible load dynamic interval prediction method. The method comprises the following steps of multi-modal scene identification, dynamic interval prediction and cross-scene self-adaptive migration. According to the cross-scene adaptive flexible load dynamic interval prediction method provided by the invention, the cross-scene adaptive capability is realized, automatic identification of industrial, commercial and resident scenes is realized, cross-scene prediction MAPE errors are reduced to be within 3%, which is better than that of a traditional model, the interval reliability is improved, quantile non-cross constraint enables PICP to be stabilized to be more than 95%, PINAW is reduced to be 5%-8%, and the interval reliability is improved. According to the method, the requirement of power grid dispatching for a narrow and reliable interval is met, knowledge migration is efficient, through field adaptive migration learning, the new scene model training time is shortened by 60%, the data demand quantity is reduced by 50%, and the engineering application cost is reduced.

This invention relates to the field of subway equipment fault prediction technology. It provides a method for adaptive threshold early warning of subway equipment fault intervals. The method includes: organizing original fault events, calculating the time interval between two adjacent faults to obtain a fault interval sequence, and cleaning the fault interval sequence; converting the cleaned fault interval sequence into a dimensionless health index sequence, and generating two adaptive threshold sequences on the health index sequence using a fixed-length sliding window; performing ordinary least squares regression on the health index sequence to obtain the slope and intercept, simultaneously calculating the Theil-Sen slope, and then performing a Mann-Kendall trend test to calculate the statistic, variance, standardized value, and two-tailed values ​​to obtain a trend conclusion; based on fault interval prediction, combined with hysteresis window determination, event proportion threshold, and cooling step size mechanism, generating a final alarm, and outputting corresponding maintenance suggestions based on the trend conclusion. This invention achieves adaptive early warning for non-stationary drift.

The invention relates to the technical field of capacitor voltage transformers, in particular to a CVT state online monitoring method and device, electronic equipment and a storage medium, and the method comprises the steps: synchronously collecting secondary voltage signals of a three-phase CVT during normal operation; based on a Pisareenko harmonic decomposition method, main components of each phase of secondary voltage signal are extracted; acquiring secondary voltage characteristic data of each phase as a training sample according to the secondary voltage signal of each phase and the main component of the secondary voltage signal, and training a single-class support vector machine model according to the training sample; obtaining secondary voltage characteristic data of each phase of CVT in the current time period as a plurality of test samples of a primary test, and identifying an abnormal ratio of abnormal samples in the plurality of test samples based on a trained single-class support vector machine model; performing interval prediction on the abnormal ratio of multiple tests by adopting a linear regression algorithm to obtain a prediction interval; and giving an alarm for the abnormal ratio which is not in the prediction interval. The technical scheme is used for on-line monitoring of the CVT state, and monitoring can be carried out accurately and effectively.

The invention belongs to the technical field of data processing, and provides a power market price prediction method and system based on two-stage optimization, and the method comprises the steps: determining an initial generating capacity node value and an initial unit price node value of each generator set, and carrying out the interpolation refining operation of the initial generating capacity node value and the initial unit price node value, obtaining a refined generating capacity node value and a refined unit price node value; establishing an output interval prediction 0-1 planning model of the generator set; performing discrete optimization solution on the output interval prediction 0-1 planning model to obtain a predicted generating capacity interval of each generator set; according to the predicted generating capacity interval, the refined generating capacity node value and the refined unit price node value, establishing an average electricity price prediction quadratic programming model; and performing continuous optimization solution on the average electricity price prediction quadratic programming model to obtain an average electricity price prediction value. The scheme can adapt to market dynamic changes, and prediction precision, flexibility and reliability are effectively improved.

This invention relates to a method for predicting robot position errors and their intervals. The steps include: acquiring the joint angle vectors of an industrial robot during operation and calculating the difference between the theoretical and actual positions to construct a dataset containing the joint angle vectors and position errors; inputting the dataset containing the joint angle vectors and position errors into a neural network-based prediction model; predicting the position error interval by calculating the predicted interval coverage probability and the average predicted interval width, while ensuring the predicted interval coverage probability meets the standard and minimizing the average predicted interval width; generating relative weights for the predicted interval boundaries by defining learnable parameters and using mean squared error loss to regress the predicted values ​​to predict the position error; and outputting the position error value and its interval prediction result from the prediction model. Compared with existing technologies, this invention can not only predict position errors but also predict the error range, improving the applicability and effectiveness of position error prediction.

To provide methods and devices for decoding video signals using inter prediction.SOLUTION: A method comprises constructing a reference picture list of a current picture in the video signal, and performing a prediction for a current picture by using the reference picture list. The step of constructing the reference picture list comprises, if a first entry of the reference picture list corresponds to an STRP entry, obtaining a POC difference value between a picture related to the first entry and another picture, and if a second entry of the reference picture list corresponds to an LTRP entry, obtaining a POC modulo value of a picture related to the second entry. A reference picture list for identifying a picture may be generated in a simple and effective manner, and thus, the compression performance may be increased and the computation time may be decreased.SELECTED DRAWING: Figure 2