Gas consumption demand prediction method, system and equipment and storage medium
By constructing multiple prediction models and combining the weight factors of each model, the problem that a single model is difficult to capture complex gas usage laws is solved, and a more accurate prediction of urban gas usage needs is achieved.
Patent Information
- Application Number
- CN202310807606.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-07-03
- Publication Date
- 2025-01-03
AI Technical Summary
In the prior art, in the forecast of urban gas demand, a single prediction model is difficult to reflect the superposition effects of trend, periodicity and sudden factors, resulting in poor prediction results.
Build multiple gas demand prediction models, including recurrent neural network model, multi-layer perceptron neural network model and time series prediction network model. Through historical data training and combining the weight factors of each model, the advantages of each model are integrated to improve Prediction accuracy.
Through multi-model fusion, the prediction accuracy of urban gas demand is improved, and complex gas usage laws can be more accurately reflected.
Smart Images
Figure CN119250244A_ABST
Abstract
Description
Technical Field
[0001] The embodiments of the present disclosure relate to the field of big data technology, and specifically to a gas demand prediction method, system, device and storage medium. Background Art
[0002] Natural gas has many application areas, and customer characteristics are complex and diverse. There are many factors that affect customer gas demand, and the correlation coupling is high. City gas is the industry with the highest proportion of gas consumption in my country. With the continuous increase in the length of upstream transmission and distribution pipelines and the continuous increase in downstream gas demand, the scale of the industry continues to expand. The fluctuation of gas consumption of city gas companies will have a huge impact on resource allocation and pipeline network operation. Therefore, it is very important to predict the gas demand of city gas. City gas companies are usually regional gas supply companies. There are many factors that affect gas consumption, covering macroeconomics, meteorology, prices, industry, real estate, electricity, policies and other dimensions. The gas demand law of city gas companies is trend, cyclical, seasonal and impacted by sudden factors. When using single prediction models such as time series forecasting network model (NBeats, Neural basis expansion analysis for interpretable time series forecasting), multilayer perceptron neural network model (MLP, Multilayer Perceptron), recurrent neural network model (RNN, Recurrent Neural Network), the effect is not good.
[0003] Figure 1 The schematic diagram of the prior art for predicting gas demand through a model is shown schematically; Figure 1As shown in the figure (Output represents output data, x1, x2, x3 represent input data, and forecast represents predicted value), the current technology for demand forecasting is mainly divided into the following steps: First, find the data factors related to the forecast target, determine the correlation between the selected factors and the historical values of the target to be forecasted by calculating the Pearson correlation coefficient and other methods, and select the factors with high correlation as the input conditions of the forecast model. Combine the historical values of the target to be forecasted with the corresponding related factor data to form an analysis matrix. Split the historical data into a training set and a validation set, input the training set data into ridge regression, lasso algorithm (Lasso, Least absolute shrinkage and selection operator), MLP, Nbeats, supervised algorithm (Xgboost, EXtreme Gradient Boosting) and other models for fitting, calculate the forecast target through the real factor data and the fitting relationship in the training set, and then compare the actual target data with the forecast target data to determine the fitting effect. Correlation coefficients such as R2 (coefficient of determination), MAE (mean absolute error), RMSE (root mean square error), and MAPE (mean absolute percentage error) are usually used for judgment, and the best fitting model is selected as the model for future predictions.
[0004] In summary, the gas demand law of urban gas enterprises is a superposition of trends, cycles, seasonality, and the impact of sudden factors. Commonly used prediction models usually only have a good fitting effect for certain scenarios. For example, the MLP model is good at fitting long-term trend states, the RNN model has a better fitting effect in short-term shocks, and NBeats is suitable for data prediction with strong periodicity. Therefore, if only a single prediction model is used to predict the gas demand of urban gas enterprises, it is difficult to reflect the full picture of the data law, resulting in poor prediction results. Summary of the invention
[0005] The embodiments of the present disclosure provide a gas demand prediction method, system, device and storage medium to solve or alleviate one or more of the above technical problems in the prior art.
[0006] According to one aspect of the present disclosure, a method for predicting gas demand is provided, comprising:
[0007] Build multiple gas demand forecasting models;
[0008] Obtaining fitting results of gas demand forecast values and actual gas demand values of a plurality of gas demand forecast models;
[0009] Determine the weight factor of each gas demand forecasting model at each time node according to the fitting result;
[0010] Multiple gas demand forecasting models are combined with the weight factor of each gas demand forecasting model at the current time node to forecast gas demand.
[0011] In a possible implementation, the building of multiple gas demand prediction models includes:
[0012] Acquire historical data, wherein the historical data includes actual gas consumption and data of factors affecting gas consumption demand;
[0013] Dividing the actual gas consumption value and influencing factor data into a training set and a validation set;
[0014] Establish multiple initial prediction models;
[0015] The training sets are respectively input into a plurality of initial prediction models, and the plurality of initial prediction models are trained to form a plurality of gas demand prediction models.
[0016] In a possible implementation, the obtaining of fitting results of the gas demand forecast values and the actual gas demand values of the gas demand forecast models includes:
[0017] Inputting the verification set into multiple gas demand prediction models respectively;
[0018] Calculate the fitting results of the gas demand forecast value and the actual gas demand value of each trained forecast model.
[0019] In a possible implementation, determining the weight factor of each gas demand forecasting model at each time node according to the fitting result includes:
[0020] Inputting the verification set into multiple gas demand forecasting models respectively to generate a fitting curve corresponding to each gas demand forecasting model;
[0021] According to the fitting curve, a fitting value with the smallest difference with the actual gas consumption value at each time node is calculated;
[0022] When there is a fitted value with the smallest difference from the actual gas consumption value at each time node, the weight factor of the gas demand forecasting model corresponding to the fitted value is set to 1, and the weight factors of other gas demand forecasting models are set to 0;
[0023] When there are multiple fitted values with the smallest difference from the actual gas consumption value at each time node, the weight factors of the gas demand forecasting models are evenly divided by the multiple fitted values, and the weight factors of other gas demand forecasting models are set to 0.
[0024] In a possible implementation, the combining of multiple gas demand forecasting models with the weight factor of each gas demand forecasting model at the current time node to forecast the gas demand includes:
[0025] Inputting the influencing factors corresponding to the gas demand to be predicted into multiple gas demand prediction models respectively, and calculating the gas demand prediction value of each gas demand prediction model at the time node to be predicted;
[0026] Based on the corresponding relationship between weight factors and time nodes, the weight factors of each gas demand forecasting model at the time node to be predicted are expanded through the time series algorithm;
[0027] The product of each gas demand forecast value and its corresponding weight factor is added together to obtain a target gas demand forecast value.
[0028] In a possible implementation, the influencing factor data includes daily high temperature data, daily low temperature data, wind data, holiday data, city night light data, second-hand housing transaction price data and natural gas price data.
[0029] In a possible implementation, the gas demand prediction model includes a recurrent neural network model, a multilayer perceptron neural network model, and a time series prediction network model.
[0030] According to one aspect of the present disclosure, a gas demand forecasting system is provided, comprising:
[0031] A construction unit, used for constructing multiple gas demand prediction models;
[0032] An acquisition unit, used to acquire fitting results of gas demand forecast values and actual gas demand values of a plurality of gas demand forecast models;
[0033] A determination unit, used to determine a weight factor of each gas demand forecasting model at each time node according to the fitting result;
[0034] The prediction unit is used to combine multiple gas demand prediction models with the weight factor of each gas demand prediction model at the current time node to predict the gas demand.
[0035] According to one aspect of the present disclosure, there is provided a gas demand prediction device, comprising:
[0036] Processor and memory;
[0037] The memory is used to store a computer program, and the processor calls the computer program stored in the memory to execute any of the gas demand forecasting methods described above.
[0038] According to one aspect of the present disclosure, a computer-readable storage medium is provided, in which a computer program is stored. When the computer program is executed by a processor, the processor is enabled to execute any one of the gas demand forecasting methods described above.
[0039] The exemplary embodiments of the present disclosure have the following beneficial effects: the exemplary embodiments of the present disclosure can determine the weight factor of each gas demand prediction value through the fitting results between the gas demand prediction values and the actual gas consumption values of multiple gas demand prediction models, and predict a prediction result that is closer to the actual gas consumption value based on each gas demand prediction value and its weight factor, thereby achieving the effect of improving the prediction accuracy of urban gas demand.
[0040] The details of one or more embodiments of the present application are presented in the following drawings and descriptions. Other features and advantages of the present application will become apparent from the accompanying drawings. It should be understood that the above general description and the detailed descriptions below are only exemplary and explanatory and cannot limit the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] The accompanying drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure, and together with the specification are used to explain the principles of the present disclosure. Obviously, the accompanying drawings described below are only some embodiments of the present disclosure, and for ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without creative work.
[0042] Figure 1 The schematic diagram of predicting gas demand through a model in the prior art is schematically shown;
[0043] Figure 2 A flow chart of a method for predicting gas demand of this exemplary embodiment is schematically shown;
[0044] Figure 3 The following diagram schematically shows the prediction effect of running the time series prediction network model alone:
[0045] Figure 4 The following diagram schematically shows the prediction effect of running the recurrent neural network model alone:
[0046] Figure 5 The following diagram schematically shows the prediction effect of running the multilayer perceptron neural network model alone:
[0047] Figure 6 A schematic diagram of a prediction effect obtained by using a gas demand prediction method of this exemplary embodiment is schematically shown;
[0048] Figure 7 A block diagram of a gas demand forecasting system of the exemplary embodiment is schematically shown;
[0049] Figure 8 The structure block diagram of the electronic device of this exemplary embodiment is schematically shown. DETAILED DESCRIPTION
[0050] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as being limited to the examples set forth herein; on the contrary, these embodiments are provided so that the present disclosure will be more comprehensive and complete, and the concepts of the example embodiments are fully conveyed to those skilled in the art. The described features, structures, or characteristics may be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure may be practiced while omitting one or more of the specific details, or other methods, components, devices, steps, etc. may be adopted. In other cases, known technical solutions are not shown or described in detail to avoid obscuring various aspects of the present disclosure.
[0051] In addition, the accompanying drawings are only schematic illustrations of the present disclosure and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and thus their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware units or integrated circuits, or implemented in different networks and / or processor 1 devices and / or microcontroller devices.
[0052] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the steps. For example, some steps may be decomposed, while some steps may be combined or partially combined, so the actual execution order may change according to the actual situation.
[0053] The terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein, for example.
[0054] In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or sub-modules is not necessarily limited to those steps or sub-modules explicitly listed, but may include other steps or sub-modules not explicitly listed or inherent to these processes, methods, products, or apparatuses.
[0055] Figure 2 A flow chart of a method for predicting gas demand in this exemplary embodiment is schematically shown. Figure 2 As shown, an exemplary embodiment of the present disclosure provides a gas demand forecasting method, comprising:
[0056] S1. Build multiple gas demand prediction models;
[0057] S2. Obtain fitting results of gas demand forecast values and actual gas demand values of a plurality of gas demand forecast models;
[0058] S3. Determine the weight factor of each gas demand forecasting model at each time node according to the fitting result;
[0059] S4. Combine multiple gas demand forecasting models with the weight factor of each gas demand forecasting model at the current time node to forecast the gas demand.
[0060] Exemplarily, the multiple gas demand forecasting models in this embodiment include NBeats, RNN and MLP models. This embodiment integrates the NBeats, RNN and MLP forecasting model algorithms, automatically allocates model weights based on the optimal model fitting effect, and reconstructs the fitting function. This allows each model to take advantage of its respective areas of expertise, thereby improving the forecasting effect of urban gas demand.
[0061] This embodiment can determine the weight factor of each gas demand prediction value through the fitting results between the gas demand prediction values of multiple gas demand prediction models and the actual gas consumption values, and predict a prediction result that is closer to the actual gas consumption value based on each gas demand prediction value and its weight factor, thereby achieving the purpose of improving the prediction accuracy of urban gas demand.
[0062] Specifically, the construction of multiple gas demand prediction models includes:
[0063] Acquire historical data, wherein the historical data includes actual gas consumption and data of factors affecting gas consumption demand;
[0064] Dividing the actual gas consumption value and influencing factor data into a training set and a validation set;
[0065] Establish multiple initial prediction models;
[0066] The training sets are respectively input into a plurality of initial prediction models, and the plurality of initial prediction models are trained to form a plurality of gas demand prediction models.
[0067] Specifically, the obtaining of fitting results of the gas demand forecast values and the actual gas demand values of the gas demand forecast models includes:
[0068] Inputting the verification set into multiple gas demand prediction models respectively;
[0069] Calculate the fitting results of the gas demand forecast value and the actual gas demand value of each trained forecast model.
[0070] In this embodiment, three target prediction models (NBeats, RNN and MLP models) are determined based on the fitting relationship between the predicted gas demand value of the city gas and the actual gas consumption value in multiple trained gas demand prediction models for city gas demand prediction, where different target prediction models have different corresponding fitting relationships.
[0071] For example, the actual gas demand of urban gas enterprises is used as the prediction target of the prediction model, and the daily high temperature, daily low temperature, wind power, holidays, urban night light data, second-hand housing transaction prices, and LNG price data are used as relevant influencing factors as the input data matrix of the prediction model;
[0072] The real gas demand of urban gas enterprises, daily high temperature, daily low temperature, wind power, holidays, urban night light data, second-hand housing transaction price, and LNG price data in the historical data are divided into training set and validation set. The observed data features are used to configure the operating parameters of the NBeats, RNN, and MLP models. The real gas demand of urban gas enterprises, daily high temperature, daily low temperature, wind power, holidays, urban night light data, second-hand housing transaction price, and LNG price data in the training set are used to train the NBeats, RNN, and MLP models respectively;
[0073] The actual gas demand of urban gas companies in the verification set, daily high temperature, daily low temperature, wind power, holidays, urban night light data, second-hand housing transaction prices, and LNG price data were input into the NBeats, RNN and MLP models respectively, and the fitting results of the actual gas consumption forecast values and actual values of urban gas companies corresponding to the NBeats, RNN and MLP models were obtained.
[0074] Specifically, determining the weight factor of each gas demand forecasting model at each time node according to the fitting result includes:
[0075] Inputting the verification set into multiple gas demand forecasting models respectively to generate a fitting curve corresponding to each gas demand forecasting model;
[0076] According to the fitting curve, calculate the fitting value with the smallest difference from the actual gas consumption value at each time node;
[0077] When there is a fitting value with the smallest difference from the actual gas consumption value at each time node, set the weight factor of the gas demand prediction model corresponding to this fitting value to 1, and set the weight factors of other gas demand prediction models to 0;
[0078] When there are multiple fitting values with the smallest difference from the actual gas consumption value at each time node, evenly divide the weight factor among the gas demand prediction models corresponding to the multiple fitting values, and set the weight factors of other gas demand prediction models to 0.
[0079] In this embodiment, according to the distance between the fitting value and the actual value of the model at each validation set time node in the three prediction models, determine the weight factors of the three prediction models at the current time node.
[0080] Exemplarily, input the same input data matrix into the NBeats, RNN, and MLP models respectively to obtain three corresponding fitting curves;
[0081] Calculate the fitting value with the smallest difference from the actual value among the three fitting values at each time node;
[0082] If there is an optimal fitting value, set the weight factor of the model corresponding to the target prediction output value to 1, and set the weight factors of the other 2 models to 0; if there are multiple optimal fitting values at the same time, evenly divide the weight among the corresponding models.
[0083] For example, each prediction model will output a fitting point estimate (yhat) at any historical time node. Assume that at time node i, the yhats of NBeats, RNN, and MLP are Ni, Ri, and Mi respectively, and the weight settings of the three prediction models at this time point are ai, bi, and ci. Select the item with the best fitting effect among Ni, Ri, and Mi as the new yhat value after fusion, and mark the weight of this model at time point i as 1. If the Ni value is the best, then (ai, bi, ci) = (1, 0, 0); if Ni = Ri < Mi, then (ai, bi, ci) = (0.5, 0.5, 0).
[0084] Specifically, the prediction of gas demand by combining multiple gas demand prediction models with the weight factors of each gas demand prediction model at the current time node includes:
[0085] Input the influencing factors corresponding to the gas demand to be predicted into multiple gas demand prediction models respectively, and calculate the gas demand prediction values of each gas demand prediction model at the time node to be predicted;
[0086] Based on the corresponding relationship between weight factors and time nodes, the weight factors of each gas demand forecasting model at the time node to be predicted are expanded through the time series algorithm;
[0087] The product of each gas demand forecast value and its corresponding weight factor is added together to obtain a target gas demand forecast value.
[0088] Specifically, the influencing factor data include daily high temperature data, daily low temperature data, wind data, holiday data, city night light data, second-hand housing transaction price data and natural gas price data.
[0089] Specifically, the gas demand prediction model includes a recurrent neural network model, a multi-layer perceptron neural network model and a time series prediction network model.
[0090] In this embodiment, three target prediction models are determined based on the fitting relationship between the predicted value of urban gas demand and the actual value in multiple trained urban gas demand prediction models, wherein the fitting relationships corresponding to different target prediction models are different; based on the distance between the fitting value and the actual value of the model at each validation set time node in the three prediction models, the weight factors of the three prediction models at the current time node are determined; the weight factors of the three models are input as a new prediction model; and the urban gas demand is predicted according to the new prediction model.
[0091] In this embodiment, the weight factors of the three models are combined with the corresponding time nodes to form a data matrix as the input of the final prediction model. Figure 3 The following diagram schematically shows the prediction effect of running the time series prediction network model alone: Figure 4 The following diagram schematically shows the prediction effect of running the recurrent neural network model alone: Figure 5 The following diagram schematically shows the prediction effect of running the multilayer perceptron neural network model alone: Figure 6 A schematic diagram of the prediction effect obtained by using a gas demand prediction method of this exemplary embodiment is shown schematically; Figure 6 and Figure 3-5 By comparison, it can be seen that after reconstructing the estimated value (yhat) of the fitting point at each time node in the validation set, the new fitting curve shows the fitting effect (see Figure 6 ) is better than any single model fitting effect (see Figure 3-5 ). It is worth noting that Figure 3-6 The horizontal axis is time, and the vertical axis is gas demand.
[0092] In this embodiment, the city gas demand is predicted according to the new prediction model.
[0093] For example, the daily high temperature, daily low temperature, wind speed, holidays, city night light data, second-hand housing transaction price, and LNG price data of the city gas demand to be predicted are input into the NBeats, RNN, and MLP models respectively to obtain the predicted output values of the three models;
[0094] Based on the correspondence between historical weight factors and dates, the time series algorithm can automatically expand the weights of the three models at the time node to be predicted;
[0095] The product of each predicted output value and its weight factor is added together to obtain the final predicted value of city gas demand.
[0096] In summary, the present invention can determine the weight factor of each predicted output value through the fitting results between the predicted output values of multiple prediction models and the actual values, and predict a prediction result that is closer to the actual value based on each predicted output value and its weight factor, thereby improving the prediction accuracy of urban gas demand.
[0097] Figure 7 A block diagram of a gas demand forecasting system of this exemplary embodiment is schematically shown; Figure 7 As shown, an exemplary embodiment of the present disclosure provides a gas demand forecasting system, comprising:
[0098] A construction unit 10 is used to construct multiple gas demand prediction models;
[0099] An acquisition unit 20, used to acquire fitting results of gas demand prediction values and actual gas demand values of a plurality of gas demand prediction models;
[0100] A determination unit 30, configured to determine a weight factor of each gas demand forecasting model at each time node according to the fitting result;
[0101] The prediction unit 40 is used to predict the gas demand by combining the multiple gas demand prediction models with the weight factor of each gas demand prediction model at the current time node.
[0102] Figure 8 The structure block diagram of the electronic device of this exemplary embodiment is schematically shown. Figure 8As shown, corresponding to the gas demand prediction method provided above, the present invention also provides a gas demand prediction device. Since the embodiment of the device is similar to the above method embodiment, the description is relatively simple. For relevant parts, please refer to the description of the above method embodiment. The device described below is only schematic. The device may include: a processor (processor) 1, a memory (memory) 2 and a communication bus (i.e., the above device bus) and a search engine, wherein the processor 1 and the memory 2 communicate with each other through the communication bus and communicate with the outside through the communication interface. The processor 1 can call the logic instructions in the memory 2 to execute the gas demand prediction method.
[0103] In addition, the logic instructions in the above-mentioned memory 2 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: a storage chip, a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a disk or an optical disk, and other media that can store program codes.
[0104] On the other hand, an embodiment of the present invention further provides a processor-readable storage medium, on which a computer program 3 is stored. When the computer program 3 is executed by the processor 1, the gas demand forecasting method provided in the above embodiments is implemented.
[0105] The processor-readable storage medium can be any available medium or data storage device that can be accessed by the processor 1, including but not limited to magnetic storage (such as floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO), etc.), optical storage (such as CD, DVD, BD, HVD, etc.), and semiconductor storage (such as ROM, EPROM, EEPROM, non-volatile memory (NANDFLASH), solid-state drive (SSD)), etc.
[0106] The above are only preferred implementations of the present disclosure, and the protection scope of the present disclosure is not limited to the above embodiments. All technical solutions under the concept of the present disclosure belong to the protection scope of the present disclosure. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present disclosure should be regarded as the protection scope of the present disclosure.
Claims
1. A method for predicting gas demand, characterized in that: include: Build multiple gas demand forecasting models; Obtaining fitting results of gas demand forecast values and actual gas demand values of a plurality of gas demand forecast models; Determine the weight factor of each gas demand forecasting model at each time node according to the fitting result; Multiple gas demand forecasting models are combined with the weight factor of each gas demand forecasting model at the current time node to forecast gas demand.
2. The method for predicting gas demand according to claim 1, characterized in that: The construction of multiple gas demand prediction models includes: Acquire historical data, wherein the historical data includes actual gas consumption and data of factors affecting gas consumption demand; Dividing the actual gas consumption value and influencing factor data into a training set and a validation set; Establish multiple initial prediction models; The training sets are respectively input into a plurality of initial prediction models, and the plurality of initial prediction models are trained to form a plurality of gas demand prediction models.
3. The method for predicting gas demand according to claim 2, characterized in that: The obtaining of fitting results of the gas demand forecast values and the actual gas demand values of the gas demand forecasting models includes: Inputting the verification set into multiple gas demand prediction models respectively; Calculate the fitting results of the gas demand forecast value and the actual gas demand value of each trained forecast model.
4. The method for predicting gas demand according to claim 3, characterized in that: Determining the weight factor of each gas demand forecasting model at each time node according to the fitting result includes: Inputting the verification set into multiple gas demand forecasting models respectively to generate a fitting curve corresponding to each gas demand forecasting model; According to the fitting curve, a fitting value with the smallest difference with the actual gas consumption value at each time node is calculated; When there is a fitted value with the smallest difference from the actual gas consumption value at each time node, the weight factor of the gas demand forecasting model corresponding to the fitted value is set to 1, and the weight factors of other gas demand forecasting models are set to 0; When there are multiple fitted values with the smallest difference from the actual gas consumption value at each time node, the weight factors of the gas demand forecasting models are evenly divided by the multiple fitted values, and the weight factors of other gas demand forecasting models are set to 0.
5. The method for predicting gas demand according to claim 1, characterized in that: The method of combining multiple gas demand forecasting models with the weight factor of each gas demand forecasting model at the current time node to forecast the gas demand includes: Inputting the influencing factors corresponding to the gas demand to be predicted into multiple gas demand prediction models respectively, and calculating the gas demand prediction value of each gas demand prediction model at the time node to be predicted; Based on the corresponding relationship between weight factors and time nodes, the weight factors of each gas demand forecasting model at the time node to be predicted are expanded through the time series algorithm; The product of each gas demand forecast value and its corresponding weight factor is added together to obtain a target gas demand forecast value.
6. The method for predicting gas demand according to any one of claims 2 to 5, characterized in that: The influencing factor data include daily high temperature data, daily low temperature data, wind data, holiday data, city night light data, second-hand housing transaction price data and natural gas price data.
7. The method for predicting gas demand according to any one of claims 1 to 5, characterized in that: The gas demand prediction model includes a recurrent neural network model, a multi-layer perceptron neural network model and a time series prediction network model.
8. A gas demand forecasting system, characterized in that: include: A construction unit, used for constructing multiple gas demand prediction models; An acquisition unit, used to acquire fitting results of gas demand forecast values and actual gas demand values of a plurality of gas demand forecast models; A determination unit, used to determine a weight factor of each gas demand forecasting model at each time node according to the fitting result; The prediction unit is used to combine multiple gas demand prediction models with the weight factor of each gas demand prediction model at the current time node to predict the gas demand.
9. A gas demand forecasting device, characterized in that: include: Processor and memory; The memory is used to store a computer program, and the processor calls the computer program stored in the memory to execute the gas demand forecasting method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the processor is enabled to execute the gas demand forecasting method according to any one of claims 1 to 7.
Citation Information
Cited By
Automatic control method and system for air compressor
CN121932370A
Port liquefied natural gas loading and unloading demand structured prediction method
CN121981769A