Electricity charge income prediction method, computer device and storage medium

Abstract

This invention provides a method, computer device, and storage medium for predicting electricity revenue. The method includes: collecting multi-dimensional model input data through a power grid company's marketing system, including historical electricity sales data and historical electricity sales price data corresponding to transaction methods, electricity consumption categories, voltage levels, peak and off-peak periods; establishing a general prediction model corresponding to multiple prediction indicators; using the general prediction model corresponding to each prediction indicator to predict the model input data, obtaining the predicted electricity sales volume and predicted electricity sales price for each prediction indicator in the current month to be predicted; summarizing the predicted electricity sales volume and predicted electricity sales price in the current month to be predicted; and obtaining the predicted electricity revenue amount for the current month to be predicted based on the predicted electricity sales volume and predicted electricity sales price. Applying this invention enables refined prediction of electricity revenue, improving prediction accuracy and stability.

Description

Technical Field

[0001] This invention relates to the field of electricity revenue forecasting technology, specifically to an electricity revenue forecasting method, a computer device using the electricity revenue forecasting method, and a computer-readable storage medium using the electricity revenue forecasting method. Background Technology

[0002] Power companies need more accurate electricity revenue forecasts to formulate reasonable market strategies, financial plans, and risk management measures. However, in reality, the accuracy of electricity revenue forecasting faces various challenges and difficulties. Existing electricity revenue forecasting technologies mainly focus on monthly aggregated electricity revenue (a single indicator). This method prevents users from gaining in-depth understanding of detailed error situations, limiting their grasp of more granular indicator data and making it difficult to identify potential problems in operations or production. It also limits the potential for accuracy optimization.

[0003] To address these issues, some researchers have proposed solutions. For example, one researcher proposed a method for predicting electricity revenue based on Markov state transition matrices. This method primarily involves calculating monthly cash flow forecasts and visualizing the predicted monthly cash flow data based on historical indicator data. The predicted monthly cash flow data is then input into a Markov state transition model to output the daily electricity revenue cash inflow. However, this method has some limitations. First, it relies solely on the Markov state transition matrix for model prediction. The diversity of historical data prevents multi-dimensional prediction; it can only filter historical data that conforms to the Markov state transition matrix type and output the results. This model cannot consider the actual business data for multi-dimensional predictive analysis, failing to meet the actual needs and challenges of enterprises.

[0004] Therefore, a more optimized method for predicting electricity revenue needs to be considered. Summary of the Invention

[0005] The primary objective of this invention is to provide a method for predicting electricity revenue that enables more refined forecasting of electricity revenue, thereby improving forecast accuracy and stability.

[0006] The second objective of this invention is to provide a computer device that can achieve refined prediction of electricity revenue, thereby improving prediction accuracy and stability.

[0007] A third objective of this invention is to provide a computer-readable storage medium that enables refined prediction of electricity revenue, thereby improving prediction accuracy and stability.

[0008] To achieve the first objective of this invention, the electricity revenue forecasting method provided by this invention includes: collecting model input data from multiple dimensions through the power grid enterprise's marketing system, including transaction methods, electricity consumption categories, voltage levels, and historical electricity sales data and historical electricity sales unit price data corresponding to peak, flat, and valley periods; establishing a general forecasting model corresponding to multiple forecasting indicators; using the general forecasting model corresponding to each forecasting indicator to forecast the model input data, obtaining the sub-model predicted electricity sales volume and sub-model predicted electricity sales unit price corresponding to each forecasting indicator for the current forecast month; summarizing the sub-model predicted electricity sales volume and sub-model predicted electricity sales unit price to obtain the predicted electricity sales volume and predicted electricity sales unit price for the current forecast month; and obtaining the predicted electricity revenue amount for the current forecast month based on the predicted electricity sales volume and predicted electricity sales unit price.

[0009] As can be seen from the above scheme, the electricity revenue prediction method of the present invention collects model input data from multiple dimensions, uses a general prediction model corresponding to multiple prediction indicators to predict the model input data, and summarizes the predicted electricity sales volume and predicted electricity sales unit price obtained from the sub-model to obtain the predicted electricity sales volume and predicted electricity sales unit price for the current month to be predicted, thereby obtaining the electricity revenue amount due for the current month to be predicted. It can realize multi-dimensional and multi-category analysis combined with actual business needs, and the application range of the model calculation results is relatively wide. It can perform more precise control from the dimensions of electricity consumption category, peak and valley, and voltage level, thereby improving the accuracy and stability of prediction.

[0010] In a further proposed solution, after collecting model input data from multiple dimensions through the power grid enterprise marketing system, the solution also includes: distinguishing the model input data to obtain first characteristic data of time series with trends or stability and second characteristic data of time series without trends or stability.

[0011] Therefore, it is evident that the time series data of every indicator cannot be guaranteed to exhibit good trend or stability. Consequently, it is necessary to differentiate the data characteristics of different indicators in the model input data so that more appropriate prediction methods can be adopted for indicators with different characteristics in subsequent prediction stages.

[0012] In a further proposed solution, the steps of distinguishing the model input data to obtain first-characteristic data of time series with trends or stability and second-characteristic data of time series without trends or stability include: performing mean normalization on the model input data corresponding to each predictive indicator and calculating the coefficient of variation; if the coefficient of variation is less than or equal to the coefficient of variation fluctuation threshold, the model input data corresponding to the predictive indicator is the first-characteristic data; if the coefficient of variation is greater than the coefficient of variation fluctuation threshold, the model input data corresponding to the predictive indicator is the second-characteristic data.

[0013] Therefore, it can be seen that by using the coefficient of variation to distinguish between the first characteristic data of time series with trends or stability and the second characteristic data of time series without trends or stability, the accuracy of classification can be improved.

[0014] In a further scheme, the steps of using the general prediction model corresponding to each prediction indicator to predict the model input data include: using at least two preset time series algorithms to combine and predict the first characteristic data to obtain the sub-model predicted electricity sales volume and sub-model predicted electricity sales unit price corresponding to the first characteristic data; and / or weighting and summing the indicator value of the same period last year and the indicator value of the previous month to obtain the sub-model predicted electricity sales volume and sub-model predicted electricity sales unit price corresponding to the second characteristic data.

[0015] Therefore, using at least two pre-defined time series algorithms to predict the first characteristic data can fully leverage the advantages of each algorithm while compensating for their shortcomings, thereby achieving accurate prediction of trending or stable indicator data. By weighting and summing the indicator values ​​from the same period last year and the previous month to obtain the predicted data for the second characteristic data, the impact of strong time series fluctuations on the predicted values ​​can be minimized.

[0016] In a further proposed solution, after collecting model input data from multiple dimensions through the power grid enterprise marketing system, the solution also includes: organizing and cleaning the model input data.

[0017] As can be seen, due to the numerous and complex dimensions involved in the model input data, the original data may contain various irregularities, inaccuracies, and anomalies. Therefore, before conducting further in-depth analysis, it is necessary to carefully organize and clean the data to ensure its quality and usability.

[0018] In a further proposed solution, the steps for cleaning and tidying up the model input data include: using an outlier filtering method with a preset IQR multiple to filter the model input data.

[0019] Therefore, by filtering the model input data using the outlier filtering method with a preset IQR multiple, the influence of dirty and outlier data can be avoided, ensuring that the data used for analysis and prediction are relatively "clean" data within a reasonable range.

[0020] In a further proposed solution, the model input data also includes external variable data that affect the actual amount of electricity bills collected; after obtaining the predicted amount of electricity bills due for the current month based on the predicted electricity sales volume and the predicted electricity sales price, the solution further includes: establishing an electricity bill collection amount prediction model, using the electricity bill collection amount prediction model to predict the external variable data and the predicted amount of electricity bills due, and obtaining the predicted amount of electricity bills collected.

[0021] Therefore, considering that the electricity bills receivable and the electricity bills actually received in the same city should have a strong correlation, and that the electricity bills actually received are also affected by external variables such as weather and holidays, it is necessary to combine external variable data and the predicted amount of electricity bills receivable to make predictions, thereby improving the accuracy of the predicted amount of electricity bills actually received.

[0022] A further proposed solution, after obtaining the predicted electricity bill amount, also includes: performing a model evaluation operation.

[0023] Therefore, performing model evaluation can ensure the accuracy and effectiveness of the model.

[0024] To achieve the second objective of the present invention, the present invention provides a computer device including a processor and a memory, the memory storing a computer program, which, when executed by the processor, implements the steps of the above-described electricity revenue prediction method.

[0025] To achieve the third objective of the present invention, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a controller, implements the steps of the above-described electricity revenue prediction method. Attached Figure Description

[0026] Figure 1 This is a flowchart of an embodiment of the electricity revenue prediction method of the present invention.

[0027] The present invention will be further described below with reference to the accompanying drawings and embodiments. Detailed Implementation

[0028] Example of an electricity revenue forecasting method:

[0029] The electricity revenue forecasting method of the present invention is a computer program applied in a computer device for power grid companies to forecast electricity revenue.

[0030] In this embodiment, the electricity revenue prediction method first executes step S1, collecting model input data from multiple dimensions through the power grid company's marketing system. This model input data includes transaction methods, electricity consumption categories, voltage levels, and historical electricity sales data and historical electricity sales unit price data corresponding to peak, off-peak, and valley periods. The model input data also includes external variable data affecting the actual amount of electricity collected. Through the power grid company's marketing system, data information involving multiple key dimensions is collected extensively. For example, in terms of transaction methods, it includes three different types of transaction methods: guaranteed minimum users, agent-purchased electricity users, and market-based transaction users. The electricity sales situation under each method will be the focus of the analysis. Regarding electricity consumption categories, it meticulously divides them into large industrial electricity consumption, non-industrial general electricity consumption, agricultural electricity consumption, commercial electricity consumption, residential electricity consumption, and rice paddy irrigation electricity consumption, among others. Different electricity consumption categories have different demand characteristics and consumption patterns; comprehensively collecting this data helps to gain a deeper understanding of the electricity consumption behavior of different user groups. Voltage levels also cover multiple levels, including 1-10 kV, 35-110 kV, and 220 kV and above. Different voltage levels correspond to different scales and types of electricity consumption scenarios, and their electricity sales data is of great value for analyzing the load distribution and power supply efficiency of the power grid. Furthermore, peak and off-peak periods are also considered, including flat periods, off-peak periods, peak periods, and peak periods. The differences in electricity sales volume and unit price during these periods often reflect the temporal distribution patterns of user electricity consumption and the supply and demand situation of the power grid at different times. In addition to the data of the above core dimensions, the potential impact of external variable data on electricity sales is also fully considered, specifically collecting data on holidays and information such as monthly minimum or maximum temperatures. During holidays, changes in people's life and production activities often lead to significant fluctuations in electricity demand. Temperature factors, including both monthly minimum and maximum temperatures, are closely related to the frequency of electrical equipment use by residents and businesses. For example, in cold winters (when monthly minimum temperatures are low), the extensive use of heating equipment significantly increases electricity consumption, while in hot summers (when monthly maximum temperatures are high), the frequent operation of cooling equipment similarly raises electricity consumption. These external variable data will serve as important supplementary information, combining with the core dimension data to provide a more comprehensive basis for subsequent model analysis and prediction.

[0031] After acquiring model input data from multiple dimensions, step S2 is executed to organize and clean the model input data. Because the model input data involves numerous and complex dimensions, the raw data may contain various irregularities, inaccuracies, and anomalies. Therefore, before conducting further in-depth analysis, meticulous data organization and cleaning are essential to ensure data quality and usability. The multi-dimensional data sources can lead to significant differences in data format, data volume, and data recording methods. For example, the units used to record electricity sales may not be entirely consistent under different transaction methods, or different electricity consumption categories may use different precisions when statistically analyzing voltage level-related data. If these differences are not addressed, they will cause considerable inconvenience to subsequent data analysis and model calculations, and may even lead to erroneous results. By organizing the data, the data format can be unified, and the data recording method standardized, enabling data from different dimensions to be processed and analyzed within a relatively consistent framework.

[0032] In this implementation, the steps for cleaning and tidying the model input data include: filtering the model input data using an outlier filtering method with a preset IQR (interquartile range). The preset IQR can be pre-set based on experimental data; preferably, it is 1.5 times. Outlier filtering methods are well-known to those skilled in the art and will not be elaborated upon here. Filtering the model input data using the preset IQR outlier filtering method avoids the influence of dirty and outlier data, ensuring that the data ultimately used for analysis and prediction is relatively "clean" data within a reasonable range.

[0033] After processing and cleaning the model input data, step S3 is executed to differentiate the data, obtaining the first characteristic data of time series with trends or stability and the second characteristic data of time series without trends or stability. Since it cannot be guaranteed that every indicator's time series will have good trends or stability, it is necessary to differentiate the data characteristics of different indicators in the model input data so that more appropriate prediction methods can be adopted for indicators with different characteristics in subsequent prediction stages.

[0034] In this embodiment, the step of distinguishing the model input data to obtain first characteristic data of time series with trends or stability and second characteristic data of time series without trends or stability includes: performing mean normalization on the model input data corresponding to each prediction indicator and calculating the coefficient of variation; if the coefficient of variation is less than or equal to the coefficient of variation fluctuation threshold, the model input data corresponding to the prediction indicator is the first characteristic data; if the coefficient of variation is greater than the coefficient of variation fluctuation threshold, the model input data corresponding to the prediction indicator is the second characteristic data. The coefficient of variation fluctuation threshold can be preset based on experimental data; preferably, the coefficient of variation fluctuation threshold is 0.5.

[0035] Specifically, for each predictive indicator, the first step is to normalize the model input data. The purpose of mean normalization is to map the data to a relatively uniform interval, making data of different magnitudes comparable. This is done by subtracting the mean of the indicator's data sequence from each data point, and then dividing by the mean again, resulting in a new, mean-normalized data sequence. Next, the coefficient of variation (CV) is calculated, which is the standard deviation divided by the mean. By calculating the standard deviation of the mean-normalized data sequence and then dividing it by the mean of the sequence, the coefficient of variation for that indicator is obtained. Finally, the calculated coefficient of variation is compared with a set fluctuation threshold. If the coefficient of variation is less than or equal to the fluctuation threshold, the predictive indicator is considered to have good trend or stability and is identified as first-characteristic data; conversely, if the coefficient of variation is greater than the fluctuation threshold, the predictive indicator is considered to lack good trend or stability and is identified as second-characteristic data.

[0036] After differentiating the model input data, step S4 is executed to establish a general prediction model corresponding to multiple prediction indicators. After clarifying whether each prediction indicator has a trend or stability, appropriate prediction models need to be established for these two different characteristics to achieve accurate prediction of each sub-indicator. For the first characteristic data, the established general prediction model is: using at least two preset time series algorithms to combine and predict the first characteristic data, obtaining the sub-model predicted electricity sales volume and sub-model predicted electricity sales price corresponding to the first characteristic data. For the second characteristic data, the established general prediction model is: based on the indicator value of the same period last year and the indicator value of the previous month, a weighted sum is performed to obtain the sub-model predicted electricity sales volume and sub-model predicted electricity sales price corresponding to the second characteristic data. The weighting values ​​of the indicator values ​​of the same period last year and the previous month can be preset based on experimental data; preferably, the weighting values ​​of the indicator values ​​of the same period last year and the previous month are both 0.5. The preset time series algorithms can be set as needed. In this embodiment, the preset time series algorithms include ARIMA algorithm, Prophet algorithm, exponential smoothing method, grey model algorithm, autoregressive conditional heteroscedasticity model algorithm, and threshold autoregressive model algorithm. When using at least two preset time series algorithms to predict the first characteristic data in combination, each algorithm can be used to predict separately, and the optimal predicted value can be selected as the final predicted value from the prediction results. Alternatively, after each algorithm predicts separately, the prediction results of all algorithms can be weighted and superimposed to obtain the final predicted value.

[0037] For example, for primary characteristic data such as sub-indicators exhibiting trends or stability, a combination of ARIMA and Prophet time series algorithms is used for forecasting. ARIMA excels at capturing autocorrelation and moving averages, and handles linear relationships well, but struggles with non-linear relationships and the incorporation of external variables. Prophet, on the other hand, excels at handling data with non-linear relationships and holiday effects, but is less sensitive to seasonality and trends. Combining these two algorithms leverages their respective strengths and compensates for their weaknesses, thus achieving accurate forecasting of trending or stable indicator data. For secondary characteristic data such as sub-indicators lacking trends or stability, a relatively simple yet effective forecasting method is employed. This involves using the indicator values ​​from the same period last year and the previous month, calculating 0.5 * the indicator value from the same period last year + 0.5 * the indicator value from the previous month as the predicted indicator value for the current month. The advantage of this method is that it minimizes the impact of strong time series fluctuations on the predicted value, as it comprehensively considers both recent and relatively distant indicator values, providing a relatively reasonable forecast even when data fluctuates significantly and lacks a clear trend.

[0038] After establishing a general prediction model corresponding to multiple prediction indicators, step S5 is executed to predict the model input data using the general prediction model corresponding to each prediction indicator, thereby obtaining the sub-model predicted electricity sales volume and sub-model predicted electricity sales price for each prediction indicator in the current month to be predicted. After establishing the general prediction model for the prediction indicators, the general prediction model can be used to predict the electricity sales volume and electricity sales price data. In this embodiment, the step of using the general prediction model corresponding to each prediction indicator to predict the model input data includes: using at least two preset time series algorithms to combine and predict the first characteristic data to obtain the sub-model predicted electricity sales volume and sub-model predicted electricity sales price corresponding to the first characteristic data; and weighting and summing the indicator values ​​of the same period last year and the previous month to obtain the sub-model predicted electricity sales volume and sub-model predicted electricity sales price corresponding to the second characteristic data.

[0039] The key to this step is to accurately input the most granular data from each dimension into the model to obtain accurate prediction results. Data collected from the power grid company's marketing system, covering various transaction methods, electricity consumption categories, voltage levels, peak and off-peak periods, and other granular dimensions, will be accurately input into the established general prediction model according to the model's requirements. During the input process, the completeness and accuracy of the data must be ensured; any missing or incorrect data may lead to inaccurate predictions. After receiving the input data, the model will simultaneously predict both electricity sales volume and electricity price. Since there is a close relationship between electricity sales volume and electricity price, and their changes often influence each other, simultaneous prediction provides a more comprehensive reflection of the actual situation of the power grid company's electricity sales business. Through the model's calculations, the predicted values ​​for the detailed indicators for the current month can be obtained. For example, the predicted electricity sales volume and predicted electricity price for the current month under the guaranteed user transaction method can be obtained from the sub-model.

[0040] After obtaining the predicted electricity sales volume and predicted electricity price per unit for each predicted indicator of the current month to be predicted, step S6 is executed to summarize the predicted electricity sales volume and predicted electricity price per unit for the current month to be predicted. After obtaining the predicted electricity sales volume and predicted electricity price per unit for each sub-indicator, these predicted values ​​need to be summarized. By summarizing the predicted values ​​of sub-indicators under different transaction methods, electricity consumption categories, voltage levels, peak and off-peak periods, etc., the predicted electricity sales volume and predicted electricity price per unit for the current month to be predicted in a certain city can be obtained. This summarization method can reflect the overall electricity sales situation of the city from a macro perspective, providing a basis for subsequent operations such as calculating the amount of electricity receivable.

[0041] After obtaining the predicted electricity sales volume and predicted electricity price for the current month to be predicted, proceed to step S7 to obtain the predicted electricity bill receivable amount for the current month to be predicted based on the predicted electricity sales volume and predicted electricity price. The predicted electricity bill receivable amount for the current month to be predicted is obtained by the following formula: Predicted electricity bill receivable amount for the current month to be predicted = Predicted electricity sales volume for the current month to be predicted × Predicted electricity price for the current month to be predicted.

[0042] After obtaining the predicted electricity bill receivable for the current month, proceed to step S8 to establish a prediction model for the actual electricity bill received. This model is then used to predict the external variable data and the predicted electricity bill receivable to obtain the predicted actual electricity bill received. Considering that the electricity bill receivable and the actual electricity bill received in the same city should have a strong correlation, and that the actual electricity bill received is also affected by external variables such as weather and holidays, it is necessary to combine the external variable data and the predicted electricity bill receivable for prediction to improve the accuracy of the predicted actual electricity bill received.

[0043] To construct a model for predicting actual electricity bill revenue, the following features were selected: the number of holidays, the number of workdays, the number of weekends, the lowest (or highest) temperature, and the amount of electricity bill due in the month. These features were chosen based on an in-depth analysis of the factors influencing actual electricity bill revenue. During holidays, changes in people's daily routines and production activities may affect actual electricity bill revenue; the distribution of workdays and weekends also impacts revenue, as people's electricity consumption habits differ at different times; weather factors, including both the lowest and highest temperatures, are closely related to the frequency of use of electrical equipment by residents and businesses, thus affecting actual electricity bill revenue; and the amount of electricity bill due in the month itself serves as an important reference value and is closely related to the actual revenue. When selecting an algorithm to predict the actual electricity bill revenue for the month, the Gradient Boosting Decision Tree (XGBoost) algorithm was chosen. This is because XGBoost can effectively capture linear or complex nonlinear relationships between features, and as a intensive learning algorithm, it can improve model accuracy by training weak classifiers through multiple iterations. Furthermore, according to experimental verification, southern regions are more sensitive to minimum temperatures, while northern regions are more sensitive to maximum temperatures. Therefore, when selecting weather features, choosing the minimum or maximum temperature of the month based on the geographical location of the city can better reflect the actual local conditions.

[0044] After obtaining the predicted electricity bill amount, step S9 is executed to perform a model evaluation. This evaluation ensures the model's accuracy and effectiveness. In this embodiment, the Mean Absolute Percentage Error (MAPE) is used to evaluate the model's performance. The evaluation scope of this embodiment includes all prediction indicators, including the prediction performance of sub-indicators. By evaluating all prediction indicators, sub-indicators with poor prediction performance can be identified, allowing users to check for abnormal events or human error that may have caused unusual data performance. This evaluation method helps users identify problems promptly, adjust the model, or handle abnormal data, thereby improving the model's accuracy and effectiveness.

[0045] In a specific example, the model consists of five sub-models: monthly electricity sales forecast model, monthly electricity receivable forecast model, monthly actual electricity receivable forecast model, daily actual electricity receivable forecast model, and daily actual electricity receivable rolling model.

[0046] The main task of the monthly electricity sales forecasting model is to predict the monthly electricity sales for each transaction type. For example, the monthly total electricity sales forecast is based on historical data from the electricity sales details table (excluding fuel oil prices), and is predicted using a combination of models (ARIMA, Prophet, exponential smoothing, grey model, autoregressive conditional heteroscedasticity model, and threshold autoregressive model). Separate forecasts are calculated for different user categories, such as agent-purchased electricity, guaranteed customers, and market-based transactions. Taking agent-purchased electricity as an example, the monthly total electricity sales forecast is combined with the agent-purchased electricity's monthly electricity sales percentage, and then decomposed to the corresponding month to obtain the future monthly forecast for agent-purchased electricity sales. The monthly electricity sales forecast for agent-purchased electricity = the agent-purchased electricity's percentage of the total electricity sales in the same period of the previous year * the monthly total electricity sales forecast.

[0047] The monthly electricity receivable forecasting model first predicts the monthly electricity receivable based on historical data from the electricity sales details and the correlation between various fees. Then, it predicts future monthly prepaid electricity charges and prepaid-to-actual-charge charges based on historical values ​​of these amounts. Finally, these three factors are used to calculate the final predicted monthly electricity receivable.

[0048] For example, take the monthly agent purchasing electronic model as an example:

[0049] (1) Monthly transmission and distribution fee forecast for electricity purchase agents:

[0050] The first step is to obtain the unit price of transmission and distribution based on the electricity consumption category in the provincial transmission and distribution price catalog published by the state; the second step is to first calculate the transmission and distribution fee for the peak and valley level of electricity consumption by using the unit price of transmission and distribution output in the first step and the monthly electricity sales forecast; then the transmission and distribution fees are gradually summarized to obtain the transmission and distribution fee for the electricity consumption category level.

[0051] (2) Monthly forecast of electricity purchase fees for agents: The first step is to obtain the unit price of agent purchase based on the electricity category in the provincial electricity price catalog published by the state; the second step is to calculate the agent purchase fee for the electricity category of peak and valley levels by first using the unit price of agent purchase output in the first step and the monthly electricity sales forecast; then the agent purchase fee is gradually summarized to obtain the agent purchase fee for the electricity category level.

[0052] (3) Monthly forecast of electricity charges for power factor adjustment for customers purchasing electricity through agents: Based on historical data, the historical value of electricity charges for power factor adjustment is predicted by combining models (ARIMA, exponential smoothing, grey model, autoregressive conditional heteroscedasticity model, threshold autoregressive model).

[0053] (4) Monthly basic electricity cost forecast for agency electricity purchase users: The first step is to directly obtain the unit price of the corresponding transformer capacity (kVA) and the unit price of the maximum demand (kW); the second step is to predict the transformer capacity (kVA) and the maximum demand (kW) based on historical data through a combination model (ARIMA, exponential smoothing method, grey model, autoregressive conditional heteroscedasticity model, threshold autoregressive model); the third step is to obtain the maximum capacity cost and the maximum demand cost according to the following calculation formula: maximum capacity cost = transformer capacity (kVA) * transformer capacity (kVA) unit price, maximum demand cost = maximum demand (kW) * maximum demand (kW) unit price; the fourth step is to obtain the basic electricity cost forecast value according to the following calculation formula: basic electricity cost forecast value = maximum capacity cost + maximum demand cost;

[0054] (5) Monthly forecast of fund and surcharge for agent electricity purchase users: The first step is to forecast the unit price of fund and surcharge based on the electricity sales details table - historical data of agent electricity purchase users by using a combination model (ARIMA, exponential smoothing method, grey model, autoregressive conditional heteroscedasticity model, threshold autoregressive model); the second step is to calculate the fund and surcharge by using the forecasted unit price of fund and surcharge with the forecasted monthly electricity sales.

[0055] (6) Monthly electricity receivable forecast for agents purchasing electricity: The monthly electricity receivable is calculated based on the reconciliation between transmission and distribution fees, agent purchase electricity fees, power factor adjustment fees, basic electricity fees, funds and surcharges.

[0056] The projected electricity bill receivable for this month for customers who purchase electricity through agents = transmission and distribution fees + electricity fees for purchasing electricity through agents + power factor adjustment fees + basic electricity fees + funds and surcharges.

[0057] The monthly electricity bill prediction model analyzes monthly electricity bills and other historical data that may affect monthly electricity bills (such as user payment habits and holiday factors) to construct a monthly electricity bill calculation equation and predict monthly electricity bills.

[0058] The daily actual electricity charge prediction model decomposes the daily actual electricity charge based on the results of the monthly actual electricity charge prediction model. The predicted daily actual electricity charge can be obtained by multiplying the proportion of the historical data of the same period of the previous year with the results of the monthly actual electricity charge prediction model.

[0059] The daily electricity bill rolling model goes a step further by using the known actual electricity bills for the current month to update the forecasts for future dates.

[0060] It should be noted that the monthly electricity sales forecast model, the monthly electricity receivable forecast model, and the monthly actual electricity receivable forecast model are the main models, while the daily actual electricity receivable forecast model and the daily actual electricity receivable rolling model are used as auxiliary supplements.

[0061] As described above, the electricity revenue forecasting method of the present invention collects model input data from multiple dimensions, uses a general forecasting model corresponding to multiple forecasting indicators to forecast the model input data, and summarizes the predicted electricity sales volume and predicted electricity sales unit price obtained from the sub-models to obtain the predicted electricity sales volume and predicted electricity sales unit price for the current month to be forecasted, thereby obtaining the electricity revenue amount due for the current month to be forecasted. It can realize multi-dimensional and multi-category analysis combined with actual business needs, and the application range of the model calculation results is relatively wide. It can perform more precise control from the dimensions of electricity consumption category, peak and off-peak periods, and voltage level, thereby improving the accuracy and stability of forecasting.

[0062] Computer device embodiment:

[0063] The computer device in this embodiment includes a controller, which executes the steps in the above-described electricity revenue prediction method embodiment when executing a computer program.

[0064] For example, a computer program can be divided into one or more modules, one or more of which are stored in memory and executed by a controller to perform the present invention. One or more modules can be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in a computer device.

[0065] A computer device may include, but is not limited to, a controller and memory. Those skilled in the art will understand that a computer device may include more or fewer components, or a combination of certain components, or different components; for example, a computer device may also include input / output devices, network access devices, buses, etc.

[0066] For example, a controller can be a Central Processing Unit (CPU), or other general-purpose controllers, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose controller can be a microcontroller or any conventional controller. The controller is the control center of a computer device, connecting all parts of the computer device through various interfaces and lines.

[0067] The memory can be used to store computer programs and / or modules. The controller implements various functions of the computer device by running or executing the computer programs and / or modules stored in the memory, and by accessing data stored in the memory. For example, the memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound receiving function, sound-to-text function, etc.), etc.; the data storage area may store data created based on the use of the mobile phone (such as audio data, text data, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, RAM, plug-in hard disk, SmartMediaCard (SMC), Secure Digital (SD) card, FlashCard, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0068] Examples of computer-readable storage media:

[0069] If the modules integrated into the computer device in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above-described electricity revenue prediction method embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a controller, it can implement the steps of the above-described electricity revenue prediction method embodiments. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The storage medium can include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content contained in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0070] It should be noted that the above are only preferred embodiments of the present invention, but the design concept of the invention is not limited thereto. Any non-substantial modifications made to the present invention using this concept also fall within the protection scope of the present invention.

Claims

1. A method for predicting electricity revenue, characterized in that: include: The model input data is collected from the power grid enterprise marketing system in multiple dimensions, including transaction method, electricity consumption category, voltage level, and historical electricity sales data and historical electricity sales unit price data corresponding to peak, flat and valley periods. The model input data is distinguished to obtain first characteristic data of time series with trends or stability and second characteristic data of time series without trends or stability; Establish a general forecasting model corresponding to multiple forecasting indicators; The general prediction model corresponding to each of the prediction indicators is used to predict the model input data to obtain the sub-model predicted electricity sales volume and sub-model predicted electricity sales unit price corresponding to each of the prediction indicators for the current month to be predicted. The predicted electricity sales volume and predicted electricity price per unit of the sub-model are summarized to obtain the predicted electricity sales volume and predicted electricity price per unit of the current month to be predicted. The predicted electricity bill amount for the current month to be predicted is obtained based on the predicted electricity sales volume and the predicted electricity sales price. The step of distinguishing the model input data to obtain first characteristic data of time series with trends or stability and second characteristic data of time series without trends or stability includes: For each of the predicted indicators, the model input data is normalized to the mean and the coefficient of variation is calculated. If the coefficient of variation is less than or equal to the coefficient of variation fluctuation threshold, then the model input data corresponding to the prediction index is the first characteristic data; If the coefficient of variation is greater than the coefficient of variation fluctuation threshold, then the model input data corresponding to the prediction index is the second characteristic data; The step of using the general prediction model corresponding to each prediction indicator to predict the model input data includes: At least two preset time series algorithms are used to combine and predict the first characteristic data to obtain the sub-model predicted electricity sales volume and the sub-model predicted electricity sales price corresponding to the first characteristic data; and / or By weighting and summing the indicator values ​​from the same period last year and the previous month, we obtain the sub-model predicted electricity sales volume and the sub-model predicted electricity sales unit price corresponding to the second characteristic data.

2. The electricity revenue forecasting method according to claim 1, characterized in that: After collecting multi-dimensional model input data through the power grid enterprise marketing system, the process also includes: The input data of the model is organized and cleaned.

3. The electricity revenue forecasting method according to claim 2, characterized in that: The steps for organizing and cleaning the model input data include: An outlier filtering method with a preset multiple of IQR was used to filter the model input data.

4. The electricity revenue forecasting method according to any one of claims 1 to 3, characterized in that: The model input data also includes external variable data that affects the actual amount of electricity bills collected; After obtaining the predicted electricity bill amount for the current month to be predicted based on the predicted electricity sales volume and the predicted electricity sales price, the method further includes: Establish a model for predicting actual electricity bill revenue; The predicted electricity bill amount is obtained by using the external variable data and the predicted electricity bill amount to predict the actual electricity bill amount.

5. The electricity revenue forecasting method according to claim 4, characterized in that: After obtaining the projected electricity bill revenue, the following is also included: Perform model evaluation.

6. A computer device comprising a processor and a memory, characterized in that: The memory stores a computer program that, when executed by the processor, implements the steps of the electricity revenue forecasting method as described in any one of claims 1 to 5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the controller, it implements the steps of the electricity revenue forecasting method as described in any one of claims 1 to 5.

Images