Coal storage amount estimation method and device based on production dispatching data and meteorological data
By combining production and transportation data with meteorological data, and using convolutional neural networks for feature extraction and dynamic time series modeling, the problems of single data dimension and insufficient feature capture in traditional methods are solved, achieving higher accuracy and stability in coal inventory prediction.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHN ENERGY NEW ENERGY TECHNOLOGY RESEARCH INSTITUTE CO LTD
- Filing Date
- 2025-11-26
- Publication Date
- 2026-04-17
AI Technical Summary
In existing technologies, traditional methods for estimating coal reserves often rely on a single or few data sources, resulting in low accuracy in coal reserve estimation. Furthermore, traditional models struggle to effectively integrate production and transportation data with meteorological data, failing to capture multi-dimensional spatial correlations and temporal dynamics, which can lead to significant deviations in the estimation results.
A convolutional neural network-based approach is adopted to acquire multi-source historical feature data, including production and transportation data and meteorological data. Feature extraction and dynamic time-series modeling are performed using feature extraction and DeformTime modules to construct a coal inventory prediction model. Batch normalization and Dropout techniques are combined to improve the stability and adaptability of the model.
It improves the accuracy of coal inventory forecasting, can adapt to forecasting needs under different seasons and operating conditions, reduces the feature loss rate and noise impact of forecasting results, and significantly improves the comprehensiveness and accuracy of forecasting results.
Smart Images
Figure CN121882871A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of power grid technology, and in particular to a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for estimating coal reserves based on production and transportation data and meteorological data. Background Technology
[0002] In thermal power generation, heating, and large-scale industrial production processes, coal-fired boilers are core energy-consuming equipment. Accurate and reliable monitoring of their coal consumption is a key basis for enterprises to conduct cost accounting, energy efficiency management, environmental assessment, and optimize operation.
[0003] Currently, traditional methods for estimating coal reserves mostly rely on a single or a few data sources, which makes the accuracy of the estimated coal reserves low. Summary of the Invention
[0004] Therefore, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for estimating coal reserves based on production and transportation data and meteorological data, which can improve the accuracy of the estimated coal reserves, in response to the above-mentioned technical problems.
[0005] Firstly, this application provides a method for estimating coal reserves based on production and transportation data and meteorological data, including:
[0006] Acquire multi-source historical feature data within the time series window, including historical production and transportation data, historical meteorological data, and historical coal quantity data;
[0007] Based on the historical production and transportation data, the historical meteorological data, and the historical coal quantity data, determine the corresponding associated feature data under the time series window;
[0008] Based on the multi-source historical feature data and the associated feature data, a convolutional neural network model is trained to obtain a coal inventory prediction model;
[0009] The real-time feature data that matches the time series window is input into the coal inventory prediction model to obtain the coal inventory prediction result corresponding to the real-time feature data.
[0010] In one embodiment, the time window includes multiple time points, the historical production and transportation data includes the historical transportation plan quantity corresponding to each time point, the historical meteorological data includes the historical temperature, historical humidity, historical wind force and historical precipitation probability corresponding to each time point, and the historical coal quantity data includes the historical actual coal intake quantity and historical coal inventory corresponding to each time point.
[0011] The step of determining the corresponding associated feature data under the time series window based on the historical production and transportation data, the historical meteorological data, and the historical coal quantity data includes:
[0012] For each of the aforementioned time points, the difference between the historical actual coal intake and the historical planned coal transport volume at that time point is determined as the transport deviation at that time point.
[0013] The meteorological index corresponding to the given time is obtained by weighted summation of the historical temperature, historical humidity, historical wind force, and historical precipitation probability at the given time.
[0014] Based on the historical coal inventory at the given time and the historical coal inventory at the previous time, the change rate of coal inventory at the given time is obtained.
[0015] By combining the transportation deviation, meteorological index, and coal inventory change rate at the specified time, the corresponding sub-correlation feature at the specified time is obtained;
[0016] Based on the sub-association features corresponding to each time point, the association feature data corresponding to the time window is obtained.
[0017] In one embodiment, the multi-source historical feature data also includes the coal inventory estimation results corresponding to the multi-source historical feature data;
[0018] The step of training a convolutional neural network model based on the multi-source historical feature data and the associated feature data to obtain a coal inventory prediction model includes:
[0019] Determine the coal input fluctuation coefficient corresponding to the time sequence window, and the moving average of coal consumption within a preset time period under the time sequence window; the preset time period is less than the time period corresponding to the time sequence window.
[0020] By combining the coal feed fluctuation coefficient and the coal consumption moving average, the derived feature matrix corresponding to the time window is obtained.
[0021] The multi-source historical feature data, the associated feature data, and the derived feature matrix are combined to obtain fused feature data;
[0022] Based on the fused feature data and the coal inventory prediction results, a convolutional neural network model is trained to obtain a coal inventory prediction model.
[0023] In one embodiment, the time window includes multiple time points, the fused feature data includes multiple sub-feature data corresponding to each of the multiple time points, and the coal inventory prediction result includes the sub-prediction result corresponding to each of the multiple time points;
[0024] The step of training a convolutional neural network model based on the fused feature data and the coal inventory prediction results to obtain a coal inventory prediction model includes:
[0025] At each of the stated times, for each of the plurality of sub-feature data, the mutual information between the sub-feature data and the corresponding historical coal inventory at the stated time is determined;
[0026] Sort the multiple mutual information in descending order and determine the top N mutual information;
[0027] By combining the sub-feature data corresponding to each of the N mutual information, the target feature data corresponding to the time point is obtained;
[0028] Based on the target feature data and sub-prediction results corresponding to each of the multiple time points, a convolutional neural network model is trained to obtain a coal inventory prediction model.
[0029] In one embodiment, the time series window includes multiple time points; the historical coal quantity data also includes the historical actual coal intake quantity corresponding to each of the time points; determining the coal intake quantity fluctuation coefficient corresponding to the time series window includes:
[0030] Based on the historical actual coal intake at each of the multiple time points, the standard deviation and mean of the historical actual coal intake under the time window are determined.
[0031] The ratio of the standard deviation to the mean is determined as the coal feed fluctuation coefficient corresponding to the time window.
[0032] In one embodiment, the time series window includes multiple time points, and the historical coal consumption data also includes the historical coal consumption corresponding to each of the time points; determining the moving average of coal consumption within a preset time period under the time series window includes:
[0033] Based on the historical coal consumption data, a first statistical value of historical coal consumption in the last hour of the preset time period and a second statistical value of historical coal consumption in the remaining time period of the preset time period are determined.
[0034] The ratio of the sum of the first statistical value and the second statistical value to the preset duration is determined as the moving average of coal consumption within the preset duration under the time window.
[0035] Secondly, this application provides a coal inventory estimation device based on production and transportation data and meteorological data, the device comprising:
[0036] The acquisition module is used to acquire multi-source historical feature data within the time series window, including historical production and transportation data, historical meteorological data, and historical coal quantity data.
[0037] The determination module is used to determine the corresponding associated feature data under the time series window based on the historical production and transportation data, the historical meteorological data, and the historical coal quantity data.
[0038] The processing module is used to train the convolutional neural network model based on the multi-source historical feature data and the associated feature data to obtain a coal inventory prediction model.
[0039] The analysis module is used to input real-time feature data that matches the time series window into the coal inventory prediction model to obtain the coal inventory prediction result corresponding to the real-time feature data.
[0040] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:
[0041] Acquire multi-source historical feature data within the time series window, including historical production and transportation data, historical meteorological data, and historical coal quantity data;
[0042] Based on the historical production and transportation data, the historical meteorological data, and the historical coal quantity data, determine the corresponding associated feature data under the time series window;
[0043] Based on the multi-source historical feature data and the associated feature data, a convolutional neural network model is trained to obtain a coal inventory prediction model;
[0044] The real-time feature data that matches the time series window is input into the coal inventory prediction model to obtain the coal inventory prediction result corresponding to the real-time feature data.
[0045] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:
[0046] Acquire multi-source historical feature data within the time series window, including historical production and transportation data, historical meteorological data, and historical coal quantity data;
[0047] Based on the historical production and transportation data, the historical meteorological data, and the historical coal quantity data, determine the corresponding associated feature data under the time series window;
[0048] Based on the multi-source historical feature data and the associated feature data, a convolutional neural network model is trained to obtain a coal inventory prediction model;
[0049] The real-time feature data that matches the time series window is input into the coal inventory prediction model to obtain the coal inventory prediction result corresponding to the real-time feature data.
[0050] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:
[0051] Acquire multi-source historical feature data within the time series window, including historical production and transportation data, historical meteorological data, and historical coal quantity data;
[0052] Based on the historical production and transportation data, the historical meteorological data, and the historical coal quantity data, determine the corresponding associated feature data under the time series window;
[0053] Based on the multi-source historical feature data and the associated feature data, a convolutional neural network model is trained to obtain a coal inventory prediction model;
[0054] The real-time feature data that matches the time series window is input into the coal inventory prediction model to obtain the coal inventory prediction result corresponding to the real-time feature data.
[0055] The aforementioned method, apparatus, computer equipment, computer-readable storage medium, and computer program product for estimating coal inventory based on production and transportation data and meteorological data acquire multi-source historical feature data within a time-series window. This multi-source historical feature data includes historical production and transportation data, historical meteorological data, and historical coal quantity data. Based on these historical data, the corresponding associated feature data within the time-series window is determined. Then, a convolutional neural network model is trained using the multi-source historical feature data and associated feature data to obtain a coal inventory prediction model. Real-time feature data matching the time-series window is input into the coal inventory prediction model to obtain the predicted coal inventory result corresponding to the real-time feature data. Therefore, the method provided in this application, by considering production and transportation data and meteorological data during model training, avoids the situation in existing technologies where information dimensions are one-sided, leading to significant deviations in coal inventory prediction results due to missing key variables. Thus, when estimating coal inventory based on the trained model, the accuracy of the predicted coal inventory can be improved. Attached Figure Description
[0056] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0057] Figure 1 This is a flowchart illustrating a method for estimating coal inventory based on production and transportation data and meteorological data in one embodiment.
[0058] Figure 2 This is a schematic diagram of the architecture of the feature extraction module in one embodiment;
[0059] Figure 3 This is a schematic diagram of the architecture of the DeformTime module in one embodiment;
[0060] Figure 4 This is a schematic diagram of the DeformTime module in another embodiment;
[0061] Figure 5 This is a flowchart illustrating a method for estimating coal inventory based on production and transportation data and meteorological data, as described in another embodiment.
[0062] Figure 6 This is a schematic diagram of the coal inventory estimation results in one embodiment;
[0063] Figure 7 This is a structural block diagram of a coal inventory prediction device based on production and transportation data and meteorological data in one embodiment.
[0064] Figure 8 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation
[0065] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0066] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.
[0067] Currently, in the field of coal inventory estimation for thermal power plants, commonly used methods based on production and transportation data and meteorological data mainly include empirical formulas, simple statistical models such as regression analysis and Autoregressive Integrated Moving Average (ARIMA) time series models, and basic machine learning models such as traditional backpropagation (BP) neural networks. However, these existing technologies have the following three shortcomings in practical applications, making it difficult to meet the needs of refined and intelligent fuel management in thermal power plants.
[0068] On the one hand, fragmented information from multiple sources and incomplete coverage of key influencing factors limit data utilization. This is because traditional coal inventory estimation methods often rely on single or a few data sources, failing to effectively integrate production and transportation data with meteorological data, resulting in low accuracy of the estimated coal inventory. While simple statistical models can process time-series data, they only focus on production and transportation time series and cannot integrate semi-structured meteorological data, leading to a one-sided information dimension. Consequently, coal inventory estimation results are prone to significant deviations due to the lack of key variables.
[0069] On the other hand, insufficient capture of complex features and poor dynamic adaptability result in shortcomings in model capabilities. This is because changes in coal reserves at thermal power plants exhibit both spatial correlation and temporal dynamics, but traditional models lack targeted feature extraction capabilities. For example, regression analysis can only fit linear relationships and cannot characterize complex mappings such as "a nonlinear surge in coal consumption from extreme high temperatures"; the ARIMA model is only applicable to stationary time-series data, and its prediction error increases sharply when faced with sudden changes in coal consumption / transportation caused by typhoons, cold waves, etc.; although traditional BP neural networks can handle nonlinear problems, they lack a dedicated spatial feature extraction module, making it difficult to capture the correlation patterns of multi-dimensional data and resulting in insufficient adaptability to dynamic operating conditions.
[0070] In view of this, such as Figure 1 As shown, a method for estimating coal inventory based on production and transportation data and meteorological data is provided. Taking the application of this method to a server as an example, it can include the following steps:
[0071] S102, acquire multi-source historical feature data within the time series window. The multi-source historical feature data includes historical production and transportation data, historical meteorological data, and historical coal quantity data.
[0072] The time series window can be 72 hours, or 3 days, which covers the transportation cycle and the meteorological impact cycle. When there are multiple time series windows, the number of samples used for model training = total number of hours - 72 (each sample corresponds to the multi-source historical feature data of the previous 72 hours).
[0073] A time series window can include multiple points in time. Historical production and transportation data includes the actual historical transportation volume and planned historical transportation volume for each point in time. Historical meteorological data includes historical temperature, humidity, wind speed, and precipitation probability for each point in time. Historical coal volume data includes historical coal consumption, actual historical coal intake, historical coal inventory, and historical inventory baseline values for each point in time.
[0074] S104. Based on historical production and transportation data, historical meteorological data, and historical coal quantity data, determine the corresponding associated characteristic data under the time series window.
[0075] S106. Based on multi-source historical feature data and related feature data, the convolutional neural network model is trained to obtain the coal inventory prediction model.
[0076] The convolutional neural network model comprises a feature extraction module and a DeformTime module connected in series to address the problem of fixed temporal structures in existing technologies. The feature extraction module extracts spatial correlation patterns from multi-source historical feature data. The DeformTime module performs dynamic temporal modeling of features, capturing the dependencies between variables and within time.
[0077] In one embodiment, the three convolutional layers of the feature extraction module can employ 8, 16, and 32 filters respectively to extract spatial correlation patterns from multi-source historical feature data. A normalization layer accelerates training convergence and improves model stability. Furthermore, a ReLU activation function can be used to introduce nonlinear transformation capabilities; and a Dropout layer can be added to prevent overfitting and enhance the model's generalization ability.
[0078] like Figure 2 The diagram illustrates the architecture of a feature extraction module, which can be configured as follows: Convolutional Layer 1: 32 3×3 convolutional kernels, stride = 1, padding = 1, extracting local spatiotemporal features. Convolutional Layer 2: 64 3×3 convolutional kernels, stride = 1, padding = 1, enhancing the representation of higher-order features.
[0079] In one embodiment, DeformTime is a deep learning model for predicting multivariate time series (MTS). Its core innovation lies in dynamically capturing the dependencies between variables and within time through a deformable attention mechanism. It is particularly good at fusing exogenous variables (such as meteorological and production and transportation data) and endogenous variables (such as coal reserves).
[0080] The core of DeformTime consists of three parts: the Variable Deformable Attention Block (V-DAB), the Time Deformable Attention Block (T-DAB), and the input data transformation module. For example... Figure 3 As shown, a schematic diagram of the DeformTime principle is provided.
[0081] Specifically, for the input data module, learnable temporal and variable encodings can map the input data to a deformation space. By performing deformation preprocessing on multi-source historical feature data, the model's ability to perceive temporal patterns is enhanced. Learningable linear layers and ReLU activation are used to map the original input (dimensions: [batch_size, seq_len, input_dim]) to the hidden space (e.g., dimensions [batch_size, seq_len, 128]), which enhances the expressive power of the features.
[0082] In some cases, by adding specific type markers (such as transport feature marker [1,0] and meteorological feature marker [0,1]) to transport features (such as transport volume) and meteorological features (such as temperature and humidity), subsequent time-deformable attention blocks can identify feature types through these markers, thus enabling differentiated time scales. The settings provide a basis; the transformed feature matrix is directly passed to the variable deformable attention block to first complete cross-variable dependency capture, and then enters the time deformable attention block for fine modeling of the time dimension.
[0083] For the Variable Deformable Attention Block (V-DAB), an adaptive sampling offset (rather than a fixed window) is learned for the variable features at each time step, focusing on the interaction between key variables and time points (such as the fluctuation of coal inventory on the day the transportation plan is released). "Hard attention" is used to select the Top-K most relevant variable-time pairs, reducing computational complexity. Only the feature information of key variable pairs is retained, and redundant variables are eliminated, compressing the output feature dimension from "all variables" to a "subset of key variables." The "time series features of key variables" output by V-DAB are directly fed into T-DAB. At this point, T-DAB does not need to handle the time dependencies of all variables; it only needs to focus on key variables, significantly reducing the computational cost of deformable time attention and allowing subsequent GRU to more efficiently capture "high-value time series patterns."
[0084] For the Temporally Deformable Attention Block (T-DAB), by analyzing the key variable features of the V-DAB output, for each time step t, by calculating... The attention weights relative to the "critical time point t0" are normalized. The "time-focused features" (dimensions: [batch_size, seq_len, hidden_dim]) are input into the Gated Recurrent Unit (GRU). The GRU dynamically captures the long and short-term temporal dependencies of the features through the "reset gate" (determining whether historical information has been forgotten) and the "update gate" (determining whether current information has been incorporated).
[0085] As can be seen from the above, by using the three modules in the DeformTime architecture, key technologies such as deformable time attention, differentiation, and GRU can be deeply integrated through a progressive process of data augmentation, variable selection, and fine-grained time modeling, forming a complete time series prediction chain.
[0086] exist Figure 3 Based on what is shown, as Figure 4 The diagram illustrates the architecture of a DeformTime module. In embodiments related to this application, it can be based on... Figure 4 The diagram shows the coal storage capacity.
[0087] In one embodiment, when training the convolutional neural network model, the Huber loss function can be used, which balances the robustness of Mean Absolute Error (MSE) and the stability of Mean Squared Error (MSE) gradients. Specifically, the Huber loss function satisfies the following condition: when the coal inventory error is within the allowable range, it is calculated using MAE; when it exceeds the range, it is calculated using MSE.
[0088] Specifically, when the absolute value of the difference between the predicted coal inventory and the actual coal inventory is less than or equal to a preset threshold (e.g., 100 tons), the allowable error L for the coal inventory satisfies:
[0089]
[0090] Otherwise, L satisfies:
[0091]
[0092] in, This indicates the predicted coal inventory. This indicates the actual amount of coal in stock. Indicates the time scale of differentiation.
[0093] In one embodiment, when training the convolutional neural network model, the optimizer can employ AdamW, i.e., Adam with weight decay, to suppress overfitting. The initial learning rate can be set to 0.001, decaying to 0.7 times its original value every 50 epochs, for example, through torch.optim.lr_scheduler.StepLR.
[0094] S108: Input the real-time feature data that matches the time series window into the coal inventory prediction model to obtain the coal inventory prediction result corresponding to the real-time feature data.
[0095] The data format and features of the real-time feature data are the same as those used when training the convolutional neural network model. For example, when training the model, the data features include coal consumption, coal intake, transportation volume, and basic inventory values, so the real-time feature data also includes the currently collected coal consumption, coal intake, transportation volume, and basic inventory values.
[0096] The method described in the above embodiments acquires multi-source historical feature data within a time-series window. This multi-source historical feature data includes historical production and transportation data, historical meteorological data, and historical coal quantity data. Based on these historical data, the corresponding associated feature data within the time-series window is determined. Then, a convolutional neural network model is trained using the multi-source historical feature data and the associated feature data to obtain a coal inventory prediction model. Real-time feature data matching the time-series window is input into the coal inventory prediction model to obtain the predicted coal inventory corresponding to the real-time feature data. Therefore, the method provided in this application, by considering production and transportation data and meteorological data during model training, avoids the situation in existing technologies where the coal inventory prediction results are prone to significant deviations due to the lack of key variables caused by one-sided information dimensions. Thus, when determining the coal inventory based on the trained model, the accuracy of the predicted coal inventory can be improved.
[0097] In one embodiment, the method further includes: normalizing the multi-source historical feature data to obtain updated feature data; using the updated feature data as new multi-source historical feature data, and returning the step of determining the corresponding associated feature data under the time series window based on historical production and transportation data, historical meteorological data, and historical coal quantity data. Therefore, data preprocessing can improve the accuracy and efficiency of model training.
[0098] For example, the time-series window includes multiple time points. For each feature data in the multi-source historical feature data, the largest and smallest feature data are determined from the feature data corresponding to each of the multiple time points. Based on the largest and smallest feature data, the feature data at each time point is normalized to obtain the normalized result corresponding to the feature data at each time point. By combining the normalized results corresponding to the multiple feature data at each time point, the updated feature data corresponding to each time point can be obtained.
[0099] For example, taking the historical temperatures corresponding to multiple times as an example, the maximum and minimum historical temperatures are determined from the multiple historical temperatures. Then, based on the maximum and minimum historical temperatures, the historical temperatures corresponding to each time point are normalized to obtain the normalized results.
[0100] After normalizing the multi-source historical feature data, the model output is obtained by inputting the real-time feature data that matches the time series window into the coal inventory prediction model; by inversely normalizing the model output, the coal inventory prediction result corresponding to the real-time feature data can be obtained.
[0101] In one embodiment, the time series window includes multiple time points. Historical production and transportation data includes the historical planned transportation volume for each time point. Historical meteorological data includes the historical temperature, historical humidity, historical wind speed, and historical precipitation probability for each time point. Historical coal quantity data includes the historical actual coal intake and historical coal inventory for each time point. Specifically, based on the historical production and transportation data, historical meteorological data, and historical coal quantity data, the relevant characteristic data corresponding to the time series window are determined, including the following steps:
[0102] Step 1: For each time point, determine the difference between the historical actual coal intake and the historical planned coal transport volume at that time point as the corresponding transport deviation.
[0103] Step 2: Calculate the weighted sum of the historical temperature, historical humidity, historical wind force, and historical precipitation probability at the given time to obtain the corresponding meteorological index.
[0104] For example, the weather index M satisfies:
[0105] M = Temperature * k1 + Humidity * k2 + Wind Force * k3 + Precipitation Probability * k4
[0106] Where k1 to k4 are weighting coefficients, and the sum of each weighting coefficient is 1. For example, k1=0.4, k2=0.3, k3=0.2, k4=0.1.
[0107] Step 3: Based on the historical coal inventory at the current time and the historical coal inventory at the previous time, obtain the coal inventory change rate at the current time.
[0108] Specifically, the ratio of the difference between the historical coal inventory at the current time and the historical coal inventory at the previous time to the historical coal inventory at the previous time is determined as the coal inventory change rate at the current time.
[0109] Step 4: Combine the transportation deviation, meteorological index and coal inventory change rate at the given time point to obtain the sub-correlation features at that time point.
[0110] Step 5: Based on the sub-association features corresponding to each time point, obtain the association feature data corresponding to the time series window.
[0111] The associated feature data includes sub-associated features corresponding to each time point. The sub-associated features corresponding to each time point include transportation deviation, meteorological index and coal inventory change rate.
[0112] The method described in the above embodiments can determine the correlation feature matrix based on the inherent correlation between historical production and transportation data, historical meteorological data, and historical coal quantity data. By addressing the problem that existing technologies only use basic features and have insufficient fusion depth, the accuracy of the estimated coal inventory can be improved.
[0113] In one embodiment, the multi-source historical feature data also includes the coal inventory prediction results corresponding to the multi-source historical feature data. Specifically, a convolutional neural network model is trained based on the multi-source historical feature data and related feature data to obtain a coal inventory prediction model, including the following steps:
[0114] Step 1: Determine the coal feed fluctuation coefficient corresponding to the time sequence window, and the moving average of coal consumption within the preset time period under the time sequence window; the preset time period is less than the time period corresponding to the time sequence window.
[0115] The time series window includes multiple time points, and the historical coal quantity data also includes the historical coal intake corresponding to each time point. In one embodiment, determining the coal intake fluctuation coefficient corresponding to the time series window includes: determining the standard deviation and mean of the historical actual coal intake under the time series window based on the historical actual coal intake corresponding to each of the multiple time points; and determining the ratio of the standard deviation to the mean as the coal intake fluctuation coefficient corresponding to the time series window.
[0116] For example, taking a time window of 72 hours as an example, the coal intake fluctuation coefficient within 72 hours = (standard deviation of historical actual coal intake within 72 hours) / (mean of historical actual coal intake within 72 hours). Therefore, by quantifying the 3-day transportation stability and predicting risks using the coal intake fluctuation coefficient within 72 hours, and combining it with the moving average of coal consumption within a preset time window to train the model, the model's time-series prediction capability can be strengthened, further improving the accuracy of the estimated coal inventory.
[0117] The historical coal consumption data also includes the historical coal consumption at each specific moment. In one embodiment, determining the moving average of coal consumption within a preset time period under a time series window includes: based on the historical coal consumption data, determining a first statistical value of historical coal consumption in the last hour of the preset time period and a second statistical value of historical coal consumption for the remaining time period within the preset time period; and determining the ratio of the sum of the first and second statistical values to the preset time period as the moving average of coal consumption within the preset time period under the time series window.
[0118] For example, taking a preset duration of 24 hours, the moving average of coal consumption over 24 hours is calculated as (historical coal consumption over the previous 23 hours + current hourly coal consumption) / 24. Therefore, by using the moving average of coal consumption over 24 hours, daily coal consumption trends can be captured, short-term noise can be filtered out, and the accuracy of model training can be improved.
[0119] Step 2: Combine the coal feed fluctuation coefficient and the moving average of coal consumption to obtain the corresponding derived feature matrix under the time window.
[0120] Step 3: Combine multi-source historical feature data, associated feature data, and derived feature matrix to obtain fused feature data.
[0121] The fused feature data includes multiple sub-feature data corresponding to each time point. These sub-feature data at each time point can include: transportation deviation, meteorological index and coal inventory change rate, coal intake fluctuation coefficient, moving average of coal consumption, and feature data from multi-source historical feature data. Multi-source historical feature data includes historical production and transportation data, historical meteorological data, and historical coal quantity data. Therefore, the feature data in the multi-source historical feature data are the feature data from each of these three historical data points.
[0122] For example, the characteristic data in historical production and transportation data may include historical actual transportation volume; the characteristic data in historical meteorological data may include historical temperature, historical humidity, historical wind force, and historical precipitation probability; and the characteristic data in historical coal quantity data may include historical coal consumption, historical actual coal intake, and historical baseline coal inventory. Therefore, the number of characteristic data in multi-source historical characteristic data can be eight. Combining transportation deviation, meteorological indices, coal inventory change rate, coal intake fluctuation coefficient, and coal consumption moving average, the number of sub-characteristic data in the final fused characteristic data can be 13.
[0123] Step 4: Based on the fused feature data and the coal inventory prediction results, train the convolutional neural network model to obtain the coal inventory prediction model.
[0124] By adopting the method of the above embodiments, the model's time-series prediction capability can be enhanced by introducing the coal feed fluctuation coefficient and the coal consumption moving average, thereby improving the accuracy of the estimated coal inventory.
[0125] In one embodiment, the time series window includes multiple time points, the fused feature data includes multiple sub-feature data corresponding to each of the multiple time points, and the coal inventory prediction result includes the corresponding sub-prediction result for each time point. Specifically, based on the fused feature data and the coal inventory prediction result, a convolutional neural network model is trained to obtain a coal inventory prediction model, including the following steps:
[0126] Step 1: At each time point, for each sub-feature data among multiple sub-feature data, determine the mutual information between the sub-feature data and the corresponding historical coal inventory at that time point.
[0127] Step 2: Sort the multiple mutual information items in descending order and determine the top N mutual information items.
[0128] N can be set to 10 or other values.
[0129] Step 3: Combine the sub-feature data corresponding to each of the N mutual information to obtain the target feature data at time step.
[0130] Step 4: Based on the target feature data and sub-prediction results corresponding to multiple time points, train the convolutional neural network model to obtain the coal inventory prediction model.
[0131] For example, with N=10 and a time window size of 72 hours, when there are multiple time windows, the dimension of the target feature data is [number of samples, 72, 10].
[0132] By employing the method described in the above embodiments and using mutual information, redundant features can be removed to obtain target feature data. Therefore, when training a model based on the target feature data, the accuracy of model training can be improved.
[0133] In summary, such as Figure 5 As shown, a method for estimating coal inventory based on production and transportation data and meteorological data is provided. Taking the application of this method to a server as an example, it can include the following steps:
[0134] S502, obtain the multi-source historical feature data for each time point within the time series window, and the coal inventory prediction results corresponding to the multi-source historical feature data. The multi-source historical feature data includes historical production and transportation data, historical meteorological data, and historical coal quantity data.
[0135] S504: For each time point, the difference between the historical actual coal intake and the historical planned coal transport volume at that time point is determined as the corresponding transport deviation at that time point.
[0136] S506 is obtained by weighted summation of historical temperature, historical humidity, historical wind force, and historical precipitation probability at the given time, thus yielding the corresponding meteorological index at that time.
[0137] S508, based on the historical coal inventory at time t and the historical coal inventory at the previous time t, obtains the coal inventory change rate at time t.
[0138] S510 combines the transportation deviation, meteorological index, and coal inventory change rate at the given time point to obtain the corresponding sub-correlation features at that time point.
[0139] S512, based on the corresponding sub-correlation features at each time point, obtain the corresponding correlation feature data under the time series window.
[0140] S514, determine the coal feed fluctuation coefficient corresponding to the time sequence window, and the moving average of coal consumption within the preset time period under the time sequence window; the preset time period is less than the time period corresponding to the time sequence window.
[0141] S516 combines the coal feed fluctuation coefficient and the moving average of coal consumption to obtain the corresponding derived feature matrix under the time series window.
[0142] S518 combines multi-source historical feature data, associated feature data, and derived feature matrix to obtain fused feature data; the fused feature data includes multiple sub-feature data corresponding to each time point.
[0143] S520, at each time point, for each of the multiple sub-feature data, determine the mutual information between the sub-feature data and the corresponding historical coal inventory at that time point.
[0144] S522: Sort multiple mutual information in descending order and determine the top 10 mutual information.
[0145] S524, combine the sub-feature data corresponding to each of the N mutual information to obtain the target feature data at time step.
[0146] S526, based on the target feature data and sub-prediction results corresponding to multiple time points, a convolutional neural network model is trained to obtain a coal inventory prediction model.
[0147] S528: Input the real-time feature data that matches the time series window into the coal inventory prediction model to obtain the coal inventory prediction result corresponding to the real-time feature data.
[0148] The specific content of S502 to S528 can be found in the aforementioned description.
[0149] As described above, the method provided in this application is the first to systematically integrate production and transportation data with meteorological and environmental factors, overcoming the limitations of traditional methods with their single data dimension and significantly improving the comprehensiveness and accuracy of the prediction results. By capturing the spatial correlation patterns of multi-dimensional features through the feature extraction module in the convolutional neural network model, and by constructing a time window to ensure sufficient learning of temporal dynamic features, it achieves dual feature extraction of spatial correlation and temporal dynamics. During model training, regularization techniques such as batch normalization and Dropout are used, making the model robust to data noise and abnormal fluctuations, and able to adapt to the prediction needs of different seasons and operating conditions. Therefore, the method provided in this application has a clear mathematical expression and a reproducible algorithm flow. The coal inventory prediction results can directly support fuel management decisions in thermal power plants, providing technical support for intelligent operation.
[0150] To address the aforementioned method, after eliminating dimensional differences through normalization, four layers of Dropout (0.2-0.3) and L2 regularization (0.01) can be set to suppress noise. Even with coal consumption plus ±20 tons of noise and temperature fluctuations plus ±5℃, the relative prediction error remains controlled within 0.6%-0.8%, with no significant fluctuations over 7 days. Compared to traditional independent denoising methods, the feature loss rate is reduced by 40%, and model stability is significantly improved. Furthermore, to address the structural rigidity of traditional time-series models, a dynamic architecture with a three-layer CNN and temporal attention is constructed. The CNN progressively extracts multi-scale features, and global average pooling enhances temporal dimension attention, adaptively adapting to the coordination of short pulses and long-cycle meteorological features. Additionally, during training, the model training loss fluctuates significantly, but there is no uncontrolled situation of "continuously increasing loss" or "increasingly large oscillation amplitude." Verification shows an overall decreasing trend in loss, and the model gradually converges, indicating satisfactory training stability. Based on the trained model, simulations yield the following results: Figure 6The diagram shows the estimated coal inventory.
[0151] Therefore, the CNN-DeformTime coal inventory prediction method for thermal power plants, which integrates production, transportation, and meteorological information from multiple sources, provided in this application, is a core technological component of the intelligent fuel management system for thermal power plants. Its core principle is to extract spatial correlation features from the two types of data using CNN, and combine this with the DeformTime module to capture the temporal dynamic features of the data (such as transportation delay fluctuations, meteorological abrupt changes, and coal consumption trends). After fusion calculation, the method outputs the predicted coal inventory of thermal power plants for a specific future period. The core purpose of this method is to address the pain point of traditional coal inventory prediction methods being unable to handle scenarios with multiple coupled factors, providing thermal power plants with accurate coal inventory prediction basis, and assisting in the formulation of dynamic coal transportation plans and optimization of inventory strategies.
[0152] In summary, addressing the shortcomings of traditional coal inventory prediction methods for thermal power plants, such as insufficient multi-source information fusion capabilities, inadequate spatiotemporal feature extraction, and poor dynamic adaptability, this application proposes a CNN-DeformTime method for predicting coal inventory in thermal power plants that integrates multi-source information from production, transportation, and meteorology. By constructing a deep learning model with spatiotemporal feature extraction capabilities, this method achieves accurate prediction of coal inventory in thermal power plants, providing a scientific basis for coal transportation planning and inventory optimization. Specifically, it adopts a four-layer architecture consisting of a data layer, a feature layer, a model layer, and an output layer, and achieves accurate prediction of coal inventory in thermal power plants through three core processes: multi-source data fusion, dynamic time-series modeling, and intelligent prediction.
[0153] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.
[0154] Based on the same inventive concept, this application also provides a coal inventory estimation device based on production and transportation data and meteorological data for implementing the coal inventory estimation method based on production and transportation data and meteorological data mentioned above. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the coal inventory estimation device based on production and transportation data and meteorological data provided below can be found in the limitations of the coal inventory estimation method based on production and transportation data and meteorological data described above, and will not be repeated here.
[0155] In one exemplary embodiment, such as Figure 7 As shown, a coal inventory estimation device based on production and transportation data and meteorological data is provided, including: an acquisition module 702, a determination module 704, a processing module 706, and an analysis module 708, wherein:
[0156] The acquisition module 702 is used to acquire multi-source historical feature data within a time-series window, including historical production and transportation data, historical meteorological data, and historical coal quantity data; the determination module 704 is used to determine the corresponding associated feature data under the time-series window based on the historical production and transportation data, the historical meteorological data, and the historical coal quantity data; the processing module 706 is used to train a convolutional neural network model based on the multi-source historical feature data and the associated feature data to obtain a coal inventory prediction model; and the analysis module 708 is used to input real-time feature data matching the time-series window into the coal inventory prediction model to obtain the coal inventory prediction result corresponding to the real-time feature data.
[0157] In one embodiment, the time series window includes multiple time points, the historical production and transportation data includes the historical transportation plan quantity corresponding to each time point, the historical meteorological data includes the historical temperature, historical humidity, historical wind force, and historical precipitation probability corresponding to each time point, and the historical coal quantity data includes the historical actual coal intake quantity and historical coal inventory corresponding to each time point; the determination module 704 is further configured to: for each time point, determine the difference between the historical actual coal intake quantity and the historical transportation plan quantity corresponding to the time point as the transportation deviation corresponding to the time point; perform a weighted summation of the historical temperature, historical humidity, historical wind force, and historical precipitation probability corresponding to the time point to obtain the meteorological index corresponding to the time point; obtain the coal inventory change rate corresponding to the time point based on the historical coal inventory corresponding to the time point and the historical coal inventory of the previous time point; combine the transportation deviation, meteorological index, and coal inventory change rate corresponding to the time point to obtain the sub-association feature corresponding to the time point; and obtain the association feature data corresponding to the time series window based on the sub-association features corresponding to each time point.
[0158] In one embodiment, the multi-source historical feature data further includes the coal inventory prediction result corresponding to the multi-source historical feature data; the processing module 706 is further configured to: determine the coal intake fluctuation coefficient corresponding to the time series window, and the coal consumption moving average within a preset time period under the time series window; the preset time period is less than the time period corresponding to the time series window; combine the coal intake fluctuation coefficient and the coal consumption moving average to obtain the derived feature matrix corresponding to the time series window; combine the multi-source historical feature data, the associated feature data, and the derived feature matrix to obtain fused feature data; and train a convolutional neural network model based on the fused feature data and the coal inventory prediction result to obtain a coal inventory prediction model.
[0159] In one embodiment, the time-series window includes multiple time points, the fused feature data includes multiple sub-feature data corresponding to each of the multiple time points, and the coal inventory prediction result includes a sub-prediction result corresponding to each of the multiple time points; the processing module 706 is further configured to: at each of the multiple time points, for each sub-feature data in the multiple sub-feature data, determine the mutual information between the sub-feature data and the historical coal inventory corresponding to the time point; sort the multiple mutual information in descending order to determine the top N mutual information; combine the sub-feature data corresponding to each of the N mutual information to obtain the target feature data corresponding to the time point; and train a convolutional neural network model based on the target feature data and sub-prediction results corresponding to the multiple time points to obtain a coal inventory prediction model.
[0160] In one embodiment, the time series window includes multiple time periods; the historical coal quantity data also includes the historical actual coal intake quantity corresponding to each of the multiple time periods; the processing module 706 is further configured to: determine the standard deviation and mean of the historical actual coal intake quantity under the time series window based on the historical actual coal intake quantity corresponding to each of the multiple time periods; and determine the ratio of the standard deviation to the mean as the coal intake quantity fluctuation coefficient corresponding to the time series window.
[0161] In one embodiment, the time series window includes multiple time points, and the historical coal consumption data also includes the historical coal consumption corresponding to each of the time points; the processing module 706 is further configured to: based on the historical coal consumption data, determine a first statistical value of the historical coal consumption in the last hour of the preset time period and a second statistical value of the historical coal consumption in the remaining time period of the preset time period; and determine the ratio of the sum of the first statistical value and the second statistical value to the preset time period as the moving average of the coal consumption in the preset time period under the time series window.
[0162] The various modules in the aforementioned coal inventory estimation device based on production and transportation data and meteorological data can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.
[0163] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 8 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage media. The database stores data used in the coal inventory determination process. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a method for predicting coal inventory based on production and transportation data and meteorological data.
[0164] Those skilled in the art will understand that Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0165] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.
[0166] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.
[0167] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.
[0168] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.
[0169] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.
[0170] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.
[0171] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.
Claims
1. A method for estimating coal inventory based on production and transportation data and meteorological data, characterized in that, The method includes: Acquire multi-source historical feature data within the time series window, including historical production and transportation data, historical meteorological data, and historical coal quantity data; Based on the historical production and transportation data, the historical meteorological data, and the historical coal quantity data, determine the corresponding associated feature data under the time series window; Based on the multi-source historical feature data and the associated feature data, a convolutional neural network model is trained to obtain a coal inventory prediction model; The real-time feature data that matches the time series window is input into the coal inventory prediction model to obtain the coal inventory prediction result corresponding to the real-time feature data.
2. The method according to claim 1, characterized in that, The time series window includes multiple time points, the historical production and transportation data includes the historical transportation plan quantity corresponding to each time point, the historical meteorological data includes the historical temperature, historical humidity, historical wind force and historical precipitation probability corresponding to each time point, and the historical coal quantity data includes the historical actual coal intake quantity and historical coal inventory corresponding to each time point. The step of determining the corresponding associated feature data under the time series window based on the historical production and transportation data, the historical meteorological data, and the historical coal quantity data includes: For each of the aforementioned time points, the difference between the historical actual coal intake and the historical planned coal transport volume at that time point is determined as the transport deviation at that time point. The meteorological index corresponding to the given time is obtained by weighted summation of the historical temperature, historical humidity, historical wind force, and historical precipitation probability at the given time. Based on the historical coal inventory at the given time and the historical coal inventory at the previous time, the coal inventory change rate at the given time is obtained. By combining the transportation deviation, meteorological index, and coal inventory change rate at the specified time, the corresponding sub-correlation feature at the specified time is obtained; Based on the sub-association features corresponding to each of the multiple time points, the association feature data corresponding to the time window is obtained.
3. The method according to claim 1, characterized in that, The multi-source historical feature data also includes the coal inventory prediction results corresponding to the multi-source historical feature data; The step of training a convolutional neural network model based on the multi-source historical feature data and the associated feature data to obtain a coal inventory prediction model includes: Determine the coal input fluctuation coefficient corresponding to the time sequence window, and the moving average of coal consumption within a preset time period under the time sequence window; the preset time period is less than the time period corresponding to the time sequence window. By combining the coal feed fluctuation coefficient and the coal consumption moving average, the derived feature matrix corresponding to the time window is obtained. The multi-source historical feature data, the associated feature data, and the derived feature matrix are combined to obtain fused feature data; Based on the fused feature data and the coal inventory prediction results, a convolutional neural network model is trained to obtain a coal inventory prediction model.
4. The method according to claim 3, characterized in that, The time series window includes multiple time points, the fused feature data includes multiple sub-feature data corresponding to each of the multiple time points, and the coal inventory prediction result includes the sub-prediction result corresponding to each of the multiple time points. The step of training a convolutional neural network model based on the fused feature data and the coal inventory prediction results to obtain a coal inventory prediction model includes: At each of the stated times, for each of the plurality of sub-feature data, the mutual information between the sub-feature data and the corresponding historical coal inventory at the stated time is determined; Sort the multiple mutual information in descending order and determine the top N mutual information; By combining the sub-feature data corresponding to each of the N mutual information, the target feature data corresponding to the time point is obtained; Based on the target feature data and sub-prediction results corresponding to each of the multiple time points, a convolutional neural network model is trained to obtain a coal inventory prediction model.
5. The method according to claim 3, characterized in that, The time series window includes multiple time points; the historical coal quantity data also includes the historical actual coal intake quantity corresponding to each time point; determining the coal intake quantity fluctuation coefficient corresponding to the time series window includes: Based on the historical actual coal intake at each of the multiple time points, the standard deviation and mean of the historical actual coal intake under the time window are determined. The ratio of the standard deviation to the mean is determined as the coal feed fluctuation coefficient corresponding to the time window.
6. The method according to claim 3, characterized in that, The time series window includes multiple time points, and the historical coal quantity data also includes the historical coal consumption corresponding to each time point. Determining the moving average of coal consumption within a preset time period under the time window includes: Based on the historical coal consumption data, a first statistical value of historical coal consumption in the last hour of the preset time period and a second statistical value of historical coal consumption in the remaining time period of the preset time period are determined. The ratio of the sum of the first statistical value and the second statistical value to the preset duration is determined as the moving average of coal consumption within the preset duration under the time window.
7. A coal inventory estimation device based on production and transportation data and meteorological data, characterized in that, The device includes: The acquisition module is used to acquire multi-source historical feature data within the time series window, including historical production and transportation data, historical meteorological data, and historical coal quantity data. The determination module is used to determine the corresponding associated feature data under the time series window based on the historical production and transportation data, the historical meteorological data, and the historical coal quantity data. The processing module is used to train the convolutional neural network model based on the multi-source historical feature data and the associated feature data to obtain a coal inventory prediction model. The analysis module is used to input real-time feature data that matches the time series window into the coal inventory prediction model to obtain the coal inventory prediction result corresponding to the real-time feature data.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.