Big data-based short-term change cost prediction method and system, and medium

Abstract

The invention provides a short-term change cost prediction method and system based on big data and a medium, and relates to the technical field of cost prediction, and the method comprises the steps that an interactive meteorological platform collects and obtains meteorological prediction data of a preset time window, and the meteorological prediction data comprise power utilization influence meteorological data and power generation influence meteorological data; performing electricity demand analysis to generate predicted electricity demand data; performing new energy generating capacity prediction to generate predicted new energy generating capacity data; performing deviation calculation to obtain predicted thermal power compensation demand data; pre-constructing a short-term change cost prediction model; performing cost prediction analysis to obtain a new energy power generation cost prediction value and a thermal power compensation cost prediction value; and performing addition calculation, and outputting a short-term variable power generation cost prediction value. The technical problem that a traditional power cost prediction method mainly depends on historical data and lacks dynamic response ability to real-time meteorological conditions and new energy fluctuation, resulting in insufficient prediction precision is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of cost prediction, and particularly to a short-term variable cost prediction method, system and medium based on big data. Background Art

[0002] With the large-scale access of renewable energy sources (such as wind power, photovoltaic power, etc.), the operational complexity of the power system has been continuously increasing. Traditional power generation methods (such as thermal power) dominate in power dispatching, but their high carbon emission characteristics have promoted the gradual development of the power system towards a low-carbon direction. Although the popularization of new energy power generation has made remarkable progress in terms of environmental protection and sustainable development, its randomness and intermittency have also brought new challenges to the cost control and dispatching of the power system. Therefore, how to accurately predict the new energy power generation cost and the thermal power compensation cost has become the core issue of short-term power generation cost control. However, existing technologies usually predict short-term power generation costs through historical data analysis, power load prediction models, and power generation cost accounting models. This makes the prediction results mainly rely on historical data and lack the dynamic response ability to real-time meteorological conditions and the volatility of new energy sources, resulting in insufficient prediction accuracy. Summary of the Invention

[0003] This application provides a short-term variable cost prediction method, system and medium based on big data, aiming to solve the technical problem that traditional power cost prediction methods mainly rely on historical data and lack the dynamic response ability to real-time meteorological conditions and the volatility of new energy sources, resulting in insufficient prediction accuracy.

[0004] In the first aspect disclosed in this application, a short-term variable cost prediction method based on big data is provided. The method includes: an interactive meteorological platform collects meteorological prediction data for a preset time window, where the meteorological prediction data includes meteorological data affecting electricity consumption and meteorological data affecting power generation; with the preset time window as a constraint, based on the meteorological data affecting electricity consumption, electricity demand analysis is performed to generate predicted electricity demand data; based on the meteorological data affecting power generation, new energy power generation prediction is performed to generate predicted new energy power generation data; deviation calculation is performed on the predicted electricity demand data and the predicted new energy power generation data, and predicted thermal power compensation demand data is obtained according to the calculation result; a short-term variable cost prediction model is pre-constructed, where the short-term variable cost prediction model includes a new energy power generation cost prediction branch and a thermal power compensation cost prediction branch; the predicted new energy power generation data and the predicted thermal power compensation demand data are respectively synchronized to the new energy power generation cost prediction branch and the thermal power compensation cost prediction branch of the short-term variable cost prediction model for cost prediction analysis to obtain a predicted new energy power generation cost value and a predicted thermal power compensation cost value; the predicted new energy power generation cost value and the predicted thermal power compensation cost value are added and calculated to output a predicted short-term variable power generation cost value.

[0005] The second aspect disclosed in this application provides a short-term variable cost prediction system based on big data. The system is used for the above-mentioned short-term variable cost prediction method based on big data, and the system includes: a meteorological prediction data acquisition module, which is used to interact with a meteorological platform to collect meteorological prediction data within a preset time window. Among them, the meteorological prediction data includes meteorological data affecting electricity consumption and meteorological data affecting power generation; an electricity demand analysis module, which is used to perform electricity demand analysis based on the meteorological data affecting electricity consumption with the preset time window as a constraint, and generate predicted electricity demand data; a power generation prediction module, which is used to perform new energy power generation prediction based on the meteorological data affecting power generation, and generate predicted new energy power generation data; a deviation calculation module, which is used to calculate the deviation between the predicted electricity demand data and the predicted new energy power generation data, and obtain predicted thermal power compensation demand data according to the calculation result; a prediction model construction module, which is used to pre-construct a short-term variable cost prediction model. Among them, the short-term variable cost prediction model includes a new energy power generation cost prediction branch and a thermal power compensation cost prediction branch; a cost prediction analysis module, which is used to synchronize the predicted new energy power generation data and the predicted thermal power compensation demand data to the new energy power generation cost prediction branch and the thermal power compensation cost prediction branch of the short-term variable cost prediction model respectively, perform cost prediction analysis, and obtain a predicted new energy power generation cost value and a predicted thermal power compensation cost value; an addition calculation module, which is used to perform addition calculation on the predicted new energy power generation cost value and the predicted thermal power compensation cost value, and output a predicted short-term variable power generation cost value.

[0006] The third aspect disclosed in this application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned short-term variable cost prediction method based on big data are implemented.

[0007] One or more technical solutions provided in this application have at least the following beneficial effects:

[0008] The interactive meteorological platform can obtain meteorological data affecting power consumption and meteorological data affecting power generation in real time, and dynamically adjust the prediction in combination with the actual meteorological conditions. This data-driven method improves the input accuracy of the model, providing a high-quality data basis for subsequent power consumption demand and power generation prediction; analyzing power consumption demand and new energy power generation based on different meteorological variables, refining the prediction of the supply-demand relationship, reducing uncertainty, and providing an accurate input basis for the formulation of power generation plans, thereby avoiding waste of dispatching resources caused by prediction deviations in demand or power generation; calculating the deviation between the predicted power consumption demand data and the predicted new energy power generation data, accurately generating the predicted thermal power compensation demand data by quantifying the gap between the two, effectively addressing the intermittency and volatility problems of new energy power generation, providing a scientific basis for the regulation of thermal power, and avoiding waste of power generation resources caused by over-regulation or under-regulation; decomposing the short-term variable cost into two parts: new energy power generation cost and thermal power compensation cost, and independently modeling them through prediction branches respectively, so that the model has higher sensitivity to the cost influencing factors of different power generation types; synchronizing the predicted new energy power generation and thermal power compensation demand to the two branches of the model respectively, and independently analyzing the cost contributions of the two power generation types, improving the accuracy of power generation cost prediction; adding the predicted new energy power generation cost value and the predicted thermal power compensation cost value to output the short-term variable power generation cost, forming a comprehensive cost prediction result, which directly serves the decision-making of power dispatching, cost control, and power generation resource allocation, helping the power system achieve higher economic efficiency.

[0009] The above description is only an overview of the technical solution of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features, and advantages of this application more obvious and understandable, the specific embodiments of this application are given below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 It is a schematic flowchart of a method for predicting short-term variable costs based on big data provided by an embodiment of this application.

[0011] Figure 2 It is a schematic structural diagram of a system for predicting short-term variable costs based on big data provided by an embodiment of this application.

[0012] Description of the reference numerals: Meteorological prediction data acquisition module 10, power consumption demand analysis module 20, power generation prediction module 30, deviation calculation module 40, prediction model construction module 50, cost prediction analysis module 60, summation calculation module 70. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0013] Embodiments of the present application provide a short-term variable cost prediction method, system, and medium based on big data, which solve the technical problem that traditional power cost prediction methods mainly rely on historical data and lack the dynamic response ability to real-time meteorological conditions and new energy volatility, resulting in insufficient prediction accuracy.

[0014] After introducing the basic principle of the present application, various non-limiting embodiments of the present application will be specifically introduced below in conjunction with the accompanying drawings of the specification. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0015] Embodiment 1, as Figure 1 shown, embodiments of the present application provide a short-term variable cost prediction method based on big data, and the method includes:

[0016] The interactive meteorological platform collects and obtains meteorological prediction data within a preset time window, where the meteorological prediction data includes meteorological data affecting electricity consumption and meteorological data affecting power generation.

[0017] The meteorological platform is a system integrating meteorological data collection, prediction, and analysis. Its main function is to provide data on meteorological conditions according to a preset time window to support subsequent predictions of electricity demand and power generation; the preset time window refers to the time range for meteorological data collection, usually a time period in hours, days, or weeks. In short-term variable cost prediction, the preset time window is generally set to a relatively short period, such as the next 24 hours or 48 hours.

[0018] The collected meteorological prediction data includes meteorological data affecting electricity consumption and meteorological data affecting power generation. Among them, meteorological data affecting electricity consumption refers to meteorological factors that affect electricity demand, including temperature, humidity, wind speed, precipitation, etc. These meteorological factors are closely related to consumers' electricity consumption behavior; meteorological data affecting power generation refers to meteorological factors that affect new energy power generation, including wind speed, solar radiation, cloud cover, etc. Wind speed is a key factor affecting wind power generation, and solar radiation is the main factor affecting photovoltaic power generation.

[0019] Taking the preset time window as a constraint, based on the meteorological data affecting electricity consumption, electricity demand analysis is performed to generate predicted electricity demand data.

[0020] By analyzing the meteorological data affecting electricity consumption and combining the limitations of the time window, the electricity demand within the future preset time window is predicted. Specifically, the time characteristics in historical electricity consumption data are analyzed, and time laws related to electricity consumption are extracted, including time-of-day characteristics, seasonal characteristics, and holiday characteristics. Using the time characteristics and the meteorological data affecting electricity consumption, a prediction model is constructed, and the predicted electricity demand data within the future time window is output, including electricity load values for different time periods.

[0021] Based on the power generation-influencing meteorological data, new energy power generation is predicted to generate predicted new energy power generation data.

[0022] By analyzing the power generation-influencing meteorological data and combining with the limitation of the time window, the new energy power generation such as wind power and photovoltaic power in the future preset time window is predicted to generate predicted new energy power generation data. Specifically, according to variables such as wind speed, wind direction, and air pressure, combined with the performance curve of the wind turbine, the wind power generation is predicted. According to variables such as solar radiation intensity and cloud cover, combined with the performance parameters of the photovoltaic module, the photovoltaic power generation is predicted. The wind power prediction result and the photovoltaic prediction result are combined to generate predicted new energy power generation data.

[0023] The deviation between the predicted electricity demand data and the predicted new energy power generation data is calculated, and the predicted thermal power compensation demand data is obtained according to the calculation result.

[0024] Calculate the difference between the predicted electricity demand data and the predicted new energy power generation data. If the deviation value is positive, it means that the new energy power generation is insufficient and thermal power needs to be compensated. The positive part of the deviation value is used as the thermal power compensation demand to generate the predicted thermal power compensation demand data, that is, the electricity generation that needs to be compensated by thermal power in the future time window. If the deviation value is negative, it means that the new energy power generation is excessive, and strategies such as energy storage or power transmission can be adopted without subsequent thermal power compensation.

[0025] A short-term variable cost prediction model is pre-constructed, where the short-term variable cost prediction model includes a new energy power generation cost prediction branch and a thermal power compensation cost prediction branch.

[0026] A short-term variable cost prediction model is pre-constructed, including a new energy power generation cost prediction branch and a thermal power compensation cost prediction branch. Among them, the new energy power generation cost prediction branch is used to predict the cost of new energy power generation, and the thermal power compensation cost prediction branch is used to predict the cost of thermal power compensation power generation. Specifically, taking the wind power-related variables as the input and the historical wind power cost data as the output, a machine learning model is used for training to construct a wind power cost prediction model; taking the photovoltaic-related variables as the input and the historical photovoltaic cost data as the output, a similar method is used to construct a photovoltaic cost prediction model, and the new energy power generation cost prediction branch is obtained by combining the wind power cost prediction model. The thermal power compensation cost prediction branch needs to consider factors such as fuel price fluctuations, unit power generation cost, and start-up cost, and the output is the predicted value of the thermal power compensation cost in the future time window. The new energy power generation cost prediction branch and the thermal power compensation cost prediction branch are connected in parallel to form a complete short-term variable cost prediction model.

[0027] Synchronize the predicted new energy power generation data and the predicted thermal power compensation demand data to the new energy power generation cost prediction branch and the thermal power compensation cost prediction branch of the short-term variable cost prediction model respectively, conduct cost prediction analysis, and obtain the new energy power generation cost prediction value and the thermal power compensation cost prediction value.

[0028] Input the predicted wind power and photovoltaic power generation data into the wind power cost prediction unit and the photovoltaic cost prediction unit of the new energy power generation cost prediction branch respectively to predict the wind power operation cost and the photovoltaic operation cost. Sum up the wind power cost prediction value and the photovoltaic cost prediction value to obtain the new energy power generation cost prediction value. Input the thermal power compensation demand volume into the thermal power compensation cost prediction branch, and conduct the total cost prediction of thermal power compensation by combining relevant data such as historical fuel costs and operation and maintenance costs, and output the thermal power compensation cost prediction value.

[0029] Perform addition calculation on the new energy power generation cost prediction value and the thermal power compensation cost prediction value, and output the short-term variable power generation cost prediction value.

[0030] Add the new energy power generation cost prediction value and the thermal power compensation cost prediction value to obtain the short-term variable power generation cost prediction value within the future time window, that is, within the future time window, after integrating new energy power generation and thermal power compensation, the total variable power generation cost. This cost prediction value provides an important basis for power dispatching, price setting, and market optimization.

[0031] Furthermore, the interactive meteorological platform collects and obtains meteorological prediction data for a preset time window. Among them, the meteorological prediction data includes meteorological data affecting electricity consumption and meteorological data affecting power generation. The method includes:

[0032] Pre-define power generation correlation variables, where the power generation correlation variables include wind power correlation variables and photovoltaic correlation variables; within the preset time window, with the wind power correlation variables as constraints, the interactive meteorological platform collects and obtains wind power meteorological data; within the preset time window, with the photovoltaic correlation variables as constraints, the interactive meteorological platform collects and obtains photovoltaic meteorological data; integrate the wind power meteorological data and the photovoltaic meteorological data as the meteorological data affecting power generation; pre-define electricity consumption correlation variables, and within the preset time window, with the electricity consumption correlation variables as constraints, the interactive meteorological platform collects and obtains the meteorological data affecting electricity consumption.

[0033] The power generation correlation variables include wind power correlation variables and photovoltaic correlation variables. Among them, the wind power correlation variables are the main influencing factors for the output power of wind power generation, including wind speed, wind direction, air density, etc., and the photovoltaic correlation variables are the main influencing factors for the output power of photovoltaic power generation, including solar radiation intensity, sunshine duration, cloud cover, etc.

[0034] Within the specified preset time window, using predefined wind power correlation variables, the interactive meteorological platform collects relevant meteorological data. That is, with the wind power correlation variables as the constraint conditions, relevant data is collected through API or database query to obtain wind power meteorological data, providing input for wind power generation prediction.

[0035] Within the specified preset time window, using predefined photovoltaic correlation variables, the interactive meteorological platform collects relevant meteorological data. That is, with the photovoltaic correlation variables as the constraint conditions, relevant data is collected through API or database query to obtain photovoltaic meteorological data, providing input for photovoltaic power generation prediction.

[0036] Integrate the separately collected wind power meteorological data and photovoltaic meteorological data to form unified power generation impact meteorological data, providing input for subsequent new energy power generation prediction.

[0037] Identify the key variables affecting electricity demand to obtain electricity demand correlation variables. Electricity demand is affected by multiple factors such as meteorology, time, and season. The meteorological factors of electricity demand correlation variables usually include temperature, humidity, rainfall, etc. According to the electricity demand correlation variables, the interactive meteorological platform obtains electricity demand impact meteorological data, providing input for electricity demand prediction.

[0038] Furthermore, based on the electricity demand impact meteorological data, conduct electricity demand analysis to generate predicted electricity demand data. The method includes:

[0039] Extract time features from the preset time window to obtain electricity demand time features; based on the electricity demand time features and the electricity demand impact meteorological data, conduct electricity demand analysis to generate predicted electricity demand data.

[0040] Extract time features related to electricity demand from the preset time window to provide input data for subsequent electricity demand prediction. Specifically, electricity demand has obvious time regularity, and the daily time period features, seasonal features, and holiday features can be extracted as electricity demand time features to characterize the electricity demand time characteristics.

[0041] Establish an electricity demand analysis model. For example, model the periodicity and trend in time based on a time series model, such as seasonal fluctuations and short-term electricity demand changes. Use the electricity demand time features and the electricity demand impact meteorological data as model inputs, combine historical electricity data to construct a feature set, train the above model with historical data so that it can predict the electricity demand in future time periods, evaluate the prediction accuracy of the model through cross-validation or test sets, adjust the model parameters, and the output predicted electricity demand data is the electricity demand in units of hours, minutes, or days, providing a basis for power grid scheduling to ensure power supply safety during peak electricity consumption periods.

[0042] Furthermore, the electricity consumption time characteristics include electricity consumption period characteristics, electricity consumption seasonal characteristics, and electricity consumption holiday characteristics.

[0043] The electricity consumption time characteristics include electricity consumption period characteristics, electricity consumption seasonal characteristics, and electricity consumption holiday characteristics. Among them, the electricity consumption period characteristics reflect the electricity consumption patterns at different times of the day. For example, during peak electricity consumption periods, the electricity demand is usually high due to increased household and commercial activities, while during off-peak electricity consumption periods, most users are at rest and the electricity demand is low. During the extraction process, the period categories of each time point are marked at hourly and minute granularities to obtain early peaks, late peaks, valleys, etc. The electricity consumption seasonal characteristics reflect the impact of different seasons on electricity demand. For example, in summer, due to the increase in air conditioning load, the electricity demand increases significantly, and in winter, the heating demand may lead to an increase in electricity consumption. During the extraction process, the seasons are divided according to the months to which the time window belongs, and the seasonal trends are analyzed by combining historical electricity consumption data. The electricity consumption holiday characteristics are used to mark whether each time point is a holiday and the type of holiday, such as weekends, legal holidays, etc., through calendar information.

[0044] Furthermore, the pre-constructed short-term variable cost prediction model, in which the short-term variable cost prediction model includes a new energy power generation cost prediction branch and a thermal power compensation cost prediction branch. The method includes:

[0045] Constrained by the power generation correlation variables, based on historical power generation records, obtain the historical meteorological data set and historical new energy power generation cost data set of the historical time window; use the historical meteorological data set as the input and the historical new energy power generation cost data set as the output to construct the new energy power generation cost prediction branch; and so on, construct the thermal power compensation cost prediction branch, and parallel the thermal power compensation cost prediction branch and the new energy power generation cost prediction branch to obtain the short-term variable cost prediction model.

[0046] The power generation correlation variables include wind power correlation variables and photovoltaic correlation variables, which directly affect the new energy power generation efficiency and power generation volume. The historical power generation records include the historical power generation data of the power grid operation, including new energy power generation volume, thermal power generation volume, etc., and also include the historical cost records of new energy power generation, including the costs of wind power and photovoltaic power generation, including equipment operation and maintenance costs, power conversion costs, etc. According to the power generation correlation variables, the historical meteorological conditions are associated with the actual power generation volume and power generation cost to obtain the historical meteorological data set and historical new energy power generation cost data set of the historical time window.

[0047] Use the historical meteorological data and power generation cost data to train the prediction model, and adjust the model parameters through cross-validation or test sets to ensure the prediction accuracy and generalization ability. The obtained new energy power generation cost prediction branch can predict the wind power and photovoltaic power generation costs according to the input future meteorological data.

[0048] The variables related to the thermal power compensation cost include fuel cost, load demand, operating efficiency, etc. The thermal power compensation cost is usually determined by the load demand and fuel cost. A modeling method similar to that of new energy can be used to construct a prediction branch for the thermal power compensation cost, which can predict the unit power generation cost of the thermal power compensation demand according to the input variables.

[0049] Parallelize the new energy power generation cost prediction branch and the thermal power compensation cost prediction branch to form a complete short-term variable cost prediction model. In the same time window, input meteorological data, load demand prediction and other data into the two branch models, and the new energy power generation cost and the thermal power compensation cost can be output respectively.

[0050] Furthermore, the method for constructing the new energy power generation cost prediction branch includes:

[0051] Constrained by the wind power correlation variables, based on the historical power generation records, obtain the historical wind power meteorological data set and the historical wind power generation cost data set in the historical time window; constrained by the photovoltaic correlation variables, based on the historical power generation records, obtain the historical photovoltaic meteorological data set and the historical photovoltaic power generation cost data set in the historical time window; construct the wind power generation cost prediction unit and the photovoltaic power generation cost prediction unit of the new energy power generation cost prediction branch based on the feedback neural network; use the historical wind power meteorological data set and the historical wind power generation cost data set as training data to train the wind power generation cost prediction unit until the preset convergence condition is reached; use the historical photovoltaic meteorological data set and the historical photovoltaic power generation cost data set as training data to train the photovoltaic power generation cost prediction unit until the preset convergence condition is reached.

[0052] The wind power correlation variables mainly include meteorological factors closely related to the wind power generation efficiency, such as wind speed, wind direction, wind shear, etc. The historical time window corresponds to the preset time window, which is the past period. Interact with the meteorological platform to obtain the historical wind power meteorological data set in the historical time window, and obtain the corresponding historical wind power generation cost data set. The wind power cost includes equipment maintenance cost, depreciation cost, land lease cost, etc. Align the historical cost data in time to ensure matching with the meteorological data set.

[0053] The photovoltaic correlation variables mainly include meteorological factors closely related to the photovoltaic power generation efficiency, such as solar radiation intensity, sunshine duration, cloud cover, etc. Similar to wind power, set a fixed time window, interact with the meteorological platform to obtain the historical photovoltaic meteorological data set and the historical photovoltaic power generation cost data set in the historical time window, and synchronize the historical cost records in time.

[0054] Feedback neural networks are suitable for processing time series data. Both meteorological data and power generation costs have time dependencies. For example, the wind speed and radiation intensity at a certain moment will affect the subsequent power generation costs. Through the feedback mechanism, the feedback neural network can capture the influence of historical data on the current moment. Based on the feedback neural network, a prediction unit for wind power and photovoltaic power generation costs is constructed, and the time series characteristics are utilized to achieve accurate prediction of future power generation costs.

[0055] Using historical wind power meteorological data sets and historical wind power generation cost data sets as training data, they are divided into a training set, a validation set, and a test set. For example, 70% is used for the training set, 15% for the validation set, and 15% for the test set. The feedback neural network is used as the model framework to capture the dynamic dependencies of time series data. During the training process, the training set is input into the model in the form of a time series, and the network weights are gradually adjusted. After each round of iteration, the loss is calculated on the validation set to evaluate the model performance until the loss value on the validation set decreases and stabilizes, or the set maximum number of training rounds is reached. The test set is used to evaluate the prediction ability of the model. If the performance is not good, the model parameters need to be adjusted or the feature input needs to be redesigned.

[0056] In the same way, the prediction unit for photovoltaic power generation costs is trained. For the sake of simplicity of the specification, it will not be elaborated here.

[0057] Furthermore, the method further includes:

[0058] Within a preset time window, power generation is monitored to obtain actual power generation data; based on the predicted power consumption demand data and the actual power generation data, supply-demand matching is performed to obtain a power supply-demand deviation value; when the power supply-demand deviation value is greater than or equal to a preset deviation threshold, a scheduling decision for the energy storage system is generated based on the power supply-demand deviation value, where the scheduling decision for the energy storage system includes a predicted value of the energy storage system scheduling cost; the predicted value of the energy storage system scheduling cost is added to the predicted value of the short-term variable power generation cost.

[0059] Within a preset time window, the power generation monitoring system real-time collects the power generation data of each power source, including wind power, photovoltaic power, thermal power, etc., summarizes the actual power generation of all power generation equipment, and generates actual power generation data.

[0060] Calculate the deviation value between the actual power generation data and the predicted power consumption demand data to obtain the power supply-demand deviation value. If the deviation value is greater than zero, it means that the power supply is excessive. If the deviation value is less than zero, it means that the power supply is insufficient. The deviation value can be used as the core basis for guiding subsequent scheduling to help optimize the system operation.

[0061] According to the operating requirements of the power system, a deviation threshold is set, which can be dynamically adjusted according to the operating characteristics of the power grid. Compare the power supply-demand deviation value with the deviation threshold. If the power supply-demand deviation value is greater than or equal to the preset deviation threshold, it is necessary to start the energy storage system scheduling program and charge the energy storage system to absorb the excess power; otherwise, if the power supply-demand deviation value is less than the preset deviation threshold, it is necessary to discharge the energy storage system to supplement the power supply. Comprehensively analyze the operating costs of the energy storage system, including charging costs, discharging costs, and cycle life consumption costs, predict the scheduling costs of the energy storage system, and obtain the predicted value of the energy storage system scheduling costs.

[0062] Add the predicted value of the energy storage system scheduling costs to the predicted value of the short-term variable generation costs, and output the predicted value of the short-term variable generation costs including the energy storage scheduling costs, providing a basis for the optimal operation of the power system and market quotation.

[0063] Furthermore, the method further includes:

[0064] In a power market environment, monitor the market power price in real time; according to the market power price and the predicted value of the short-term variable generation costs, perform dynamic optimization of the short-term variable cost prediction model.

[0065] Use a power market trading platform or a power dispatching system to obtain real-time market power price data, including the day-ahead market price, which reflects the power trading price of the next day; the real-time market price, which reflects the real-time power supply-demand balance price of the current period; and the peak-valley electricity price, which characterizes the characteristics of the electricity price fluctuating over time.

[0066] Calculate the deviation between the market power price and the predicted value of the short-term variable generation costs. The magnitude of the deviation reflects the degree of coincidence between the prediction model and the actual market price. Calibrate the model according to the deviation value. The calibration goal is to minimize the deviation between the predicted value and the actual value. For example, feedback the deviation value into the prediction model and dynamically adjust the parameters of the model, such as weights and biases, to optimize the prediction results and achieve the dynamic optimization of the short-term variable cost prediction model.

[0067] In summary, the short-term variable cost prediction method based on big data provided by the embodiments of the present application has the following technical effects:

[0068] The interactive meteorological platform can obtain meteorological data affecting power consumption and meteorological data affecting power generation in real time, and dynamically adjust predictions in combination with actual meteorological conditions. This data-driven method improves the input accuracy of the model and provides a high-quality data basis for subsequent predictions of power consumption demand and power generation. By analyzing power consumption demand and new energy power generation based on different meteorological variables, the prediction of the supply-demand relationship is refined, reducing uncertainty and providing an accurate input basis for the formulation of power generation plans, thereby avoiding waste of dispatching resources caused by prediction deviations in demand or power generation. Calculate the deviation between the predicted power consumption demand data and the predicted new energy power generation data, and accurately generate the predicted thermal power compensation demand data by quantifying the gap between the two, effectively addressing the intermittency and volatility issues of new energy power generation, providing a scientific basis for the regulation of thermal power, and avoiding waste of power generation resources caused by over-regulation or under-regulation. Decompose the short-term variable cost into two parts: new energy power generation cost and thermal power compensation cost, and independently model them through prediction branches respectively, so that the model has higher sensitivity to the cost influencing factors of different power generation types. Synchronize the predicted new energy power generation and thermal power compensation demand to the two branches of the model respectively, and independently analyze the cost contributions of the two power generation types, improving the accuracy of power generation cost prediction. Add the predicted new energy power generation cost value and the predicted thermal power compensation cost value to output the short-term variable power generation cost, forming a comprehensive cost prediction result, which directly serves the decision-making of power dispatching, cost control, and power generation resource allocation, helping the power system achieve higher economic efficiency.

[0069] Embodiment 2, based on the same inventive concept as the short-term variable cost prediction method based on big data in the foregoing embodiment, as Figure 2 shown, the embodiment of the present application provides a short-term variable cost prediction system based on big data, and the system includes:

[0070] A meteorological prediction data acquisition module 10 is used to interact with a meteorological platform to collect meteorological prediction data for a preset time window. Among them, the meteorological prediction data includes meteorological data affecting power consumption and meteorological data affecting power generation; a power consumption demand analysis module 20 is used to perform power consumption demand analysis based on the meteorological data affecting power consumption with the preset time window as a constraint, and generate predicted power consumption demand data; a power generation prediction module 30 is used to predict the new energy power generation based on the meteorological data affecting power generation, and generate predicted new energy power generation data; a deviation calculation module 40 is used to calculate the deviation between the predicted power consumption demand data and the predicted new energy power generation data, and obtain predicted thermal power compensation demand data according to the calculation result; a prediction model construction module 50 is used to pre-construct a short-term variable cost prediction model. Among them, the short-term variable cost prediction model includes a new energy power generation cost prediction branch and a thermal power compensation cost prediction branch; a cost prediction and analysis module 60 is used to synchronize the predicted new energy power generation data and the predicted thermal power compensation demand data to the new energy power generation cost prediction branch and the thermal power compensation cost prediction branch of the short-term variable cost prediction model respectively, perform cost prediction and analysis, and obtain a predicted new energy power generation cost value and a predicted thermal power compensation cost value; an addition calculation module 70 is used to perform an addition calculation on the predicted new energy power generation cost value and the predicted thermal power compensation cost value, and output a predicted short-term variable power generation cost value.

[0071] Furthermore, the meteorological prediction data acquisition module 10 includes:

[0072] A power generation-related variable definition unit is used to pre-define power generation-related variables. Among them, the power generation-related variables include wind power-related variables and photovoltaic-related variables; a wind power meteorological data acquisition unit is used to interact with a meteorological platform to collect wind power meteorological data within the preset time window with the wind power-related variables as a constraint; a photovoltaic meteorological data acquisition unit is used to interact with a meteorological platform to collect photovoltaic meteorological data within the preset time window with the photovoltaic-related variables as a constraint; a meteorological data integration unit is used to integrate the wind power meteorological data and the photovoltaic meteorological data as the meteorological data affecting power generation; a meteorological data affecting power consumption acquisition unit is used to pre-define power consumption-related variables, and interact with a meteorological platform to collect the meteorological data affecting power consumption within the preset time window with the power consumption-related variables as a constraint.

[0073] Furthermore, the power consumption demand analysis module 20 includes:

[0074] A time feature extraction unit is used to extract time features from the preset time window to obtain power consumption time features; a power consumption demand analysis unit is used to perform power consumption demand analysis based on the power consumption time features and the meteorological data affecting power consumption, and generate predicted power consumption demand data.

[0075] Furthermore, the power consumption time characteristics include power consumption period characteristics, power consumption seasonal characteristics, and power consumption holiday characteristics.

[0076] Furthermore, the prediction model construction module 50 includes:

[0077] A historical data acquisition unit, configured to obtain a historical meteorological data set and a historical new energy power generation cost data set of a historical time window based on historical power generation records with the power generation correlation variable as a constraint; a prediction branch construction unit, configured to construct the new energy power generation cost prediction branch with the historical meteorological data set as an input and the historical new energy power generation cost data set as an output; a prediction model construction unit, configured to construct the thermal power compensation cost prediction branch in the same way, and parallel the thermal power compensation cost prediction branch and the new energy power generation cost prediction branch to obtain the short-term variable cost prediction model.

[0078] Furthermore, the prediction branch construction unit includes:

[0079] A first historical data set acquisition channel, configured to obtain a historical wind power meteorological data set and a historical wind power generation cost data set of a historical time window based on historical power generation records with the wind power correlation variable as a constraint; a second historical data set acquisition channel, configured to obtain a historical photovoltaic meteorological data set and a historical photovoltaic power generation cost data set of a historical time window based on historical power generation records with the photovoltaic correlation variable as a constraint; a cost prediction unit construction channel, configured to construct a wind power generation cost prediction unit and a photovoltaic power generation cost prediction unit of the new energy power generation cost prediction branch based on a feedback neural network; a first training channel, configured to use the historical wind power meteorological data set and the historical wind power generation cost data set as training data to train the wind power generation cost prediction unit until a preset convergence condition is reached; a second training channel, configured to use the historical photovoltaic meteorological data set and the historical photovoltaic power generation cost data set as training data to train the photovoltaic power generation cost prediction unit until a preset convergence condition is reached.

[0080] Furthermore, the system further includes:

[0081] The power generation monitoring module is used to monitor power generation within a preset time window to obtain actual power generation data; the supply-demand matching module is used to perform supply-demand matching based on the predicted power consumption demand data and the actual power generation data to obtain a power supply-demand deviation value; the scheduling decision generation module is used to generate an energy storage system scheduling decision based on the power supply-demand deviation value when the power supply-demand deviation value is greater than or equal to a preset deviation threshold, where the energy storage system scheduling decision includes a predicted value of the energy storage system scheduling cost; the predicted value adding module is used to add the predicted value of the energy storage system scheduling cost to the predicted value of the short-term variable power generation cost.

[0082] Furthermore, the system further includes:

[0083] The electricity price monitoring module is used to monitor the market electricity price in real time in the electricity market environment; the model dynamic optimization module is used to dynamically optimize the short-term variable cost prediction model according to the market electricity price and the predicted value of the short-term variable power generation cost.

[0084] Through the foregoing detailed description of the short-term variable cost prediction method based on big data in this specification, those skilled in the art can clearly know the short-term variable cost prediction system based on big data in this embodiment. Since it corresponds to the method disclosed in Embodiment 1, the description is relatively simple, and the relevant parts can be referred to the description in the method section.

[0085] Embodiment 3, based on the same inventive concept as the short-term variable cost prediction method based on big data in Embodiment 1 above, the present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements each step of the above-mentioned short-term variable cost prediction method embodiment based on big data and can achieve the same technical effect. To avoid repetition, it will not be elaborated here.

[0086] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A short-term variable cost forecasting method based on big data, characterized in that: The method comprises: The interactive meteorological platform collects and obtains meteorological forecast data for a preset time window, wherein the meteorological forecast data includes meteorological data on the impact of electricity consumption and meteorological data on the impact of power generation; Taking the preset time window as a constraint, based on the electricity consumption-influencing meteorological data, an electricity demand analysis is performed to generate predicted electricity demand data; Based on the power generation-influencing meteorological data, forecast the power generation of new energy sources and generate forecast power generation data of new energy sources; Performing deviation calculation on the predicted electricity demand data and the predicted new energy power generation data, and obtaining predicted thermal power compensation demand data according to the calculation result; Pre-constructing a short-term variable cost prediction model, wherein the short-term variable cost prediction model includes a new energy power generation cost prediction branch and a thermal power compensation cost prediction branch; The predicted new energy power generation data and the predicted thermal power compensation demand data are synchronized to the new energy power generation cost prediction branch and the thermal power compensation cost prediction branch of the short-term variable cost prediction model, respectively, and cost prediction analysis is performed to obtain a new energy power generation cost prediction value and a thermal power compensation cost prediction value; The predicted value of the new energy power generation cost and the predicted value of the thermal power compensation cost are added and calculated to output a predicted value of the short-term variable power generation cost.

2. The short-term variable cost forecasting method based on big data according to claim 1, characterized in that: The interactive meteorological platform collects and obtains meteorological forecast data for a preset time window, wherein the meteorological forecast data includes meteorological data on electricity consumption and meteorological data on power generation, and the method includes: Predefine power generation related variables, wherein the power generation related variables include wind power related variables and photovoltaic related variables; Within the preset time window, the interactive meteorological platform collects and obtains wind power meteorological data with the wind power related variables as constraints; Within the preset time window, the interactive meteorological platform collects and obtains photovoltaic meteorological data with the photovoltaic-related variables as constraints; Integrating the wind power meteorological data and the photovoltaic meteorological data as the power generation impact meteorological data; Predefine electricity consumption related variables, and within the preset time window, use the electricity consumption related variables as constraints, and the interactive meteorological platform collects and obtains the electricity consumption-affecting meteorological data.

3. The short-term variable cost forecasting method based on big data as claimed in claim 2 is characterized in that: The method of performing power demand analysis based on the power consumption-influencing meteorological data to generate predicted power demand data includes: Extracting time features from the preset time window to obtain power consumption time features; Based on the electricity consumption time characteristics and the electricity consumption-influencing meteorological data, an electricity demand analysis is performed to generate predicted electricity demand data.

4. The short-term variable cost forecasting method based on big data as claimed in claim 3 is characterized in that: The electricity consumption time characteristics include electricity consumption time period characteristics, electricity consumption seasonal characteristics, and electricity consumption holiday characteristics.

5. The short-term variable cost forecasting method based on big data as claimed in claim 2, characterized in that: The pre-constructed short-term variable cost prediction model includes a new energy power generation cost prediction branch and a thermal power compensation cost prediction branch, and the method includes: Obtaining a historical meteorological data set and a historical new energy power generation cost data set for a historical time window based on the power generation associated variable constraints and historical power generation records; Taking the historical meteorological data set as input and the historical new energy power generation cost data set as output, constructing the new energy power generation cost prediction branch; By analogy, the thermal power compensation cost prediction branch is constructed, and the thermal power compensation cost prediction branch and the new energy power generation cost prediction branch are connected in parallel to obtain the short-term variable cost prediction model.

6. The short-term variable cost forecasting method based on big data as claimed in claim 2, characterized in that: The method for constructing the new energy power generation cost prediction branch includes: Taking the wind power associated variables as constraints and based on historical power generation records, a historical wind power meteorological data set and a historical wind power generation cost data set of a historical time window are obtained; Taking the photovoltaic-related variables as constraints and based on historical power generation records, a historical photovoltaic meteorological data set and a historical photovoltaic power generation cost data set of a historical time window are obtained; Constructing a wind power generation cost prediction unit and a photovoltaic power generation cost prediction unit of the new energy power generation cost prediction branch based on a feedback neural network; Using the historical wind power meteorological data set and the historical wind power generation cost data set as training data, training the wind power generation cost prediction unit until a preset convergence condition is reached; The photovoltaic power generation cost prediction unit is trained using the historical photovoltaic meteorological data set and the historical photovoltaic power generation cost data set as training data until a preset convergence condition is reached.

7. The short-term variable cost forecasting method based on big data according to claim 1, characterized in that: The method further comprises: Monitor power generation within the preset time window to obtain actual power generation data; Matching supply and demand based on the predicted power demand data and the actual power generation data to obtain a power supply and demand deviation value; When the power supply and demand deviation value is greater than or equal to a preset deviation threshold, an energy storage system scheduling decision is generated based on the power supply and demand deviation value, wherein the energy storage system scheduling decision includes a predicted value of the energy storage system scheduling cost; The energy storage system dispatch cost forecast value is added to the short-term variable power generation cost forecast value.

8. The short-term variable cost forecasting method based on big data as claimed in claim 1, characterized in that: The method further comprises: In the power market environment, real-time monitoring of market power prices; The short-term variable cost prediction model is dynamically optimized based on the market electricity price and the short-term variable power generation cost prediction value.

9. The short-term variable cost forecasting system based on big data is characterized by: The system for implementing the short-term variable cost forecasting method based on big data according to any one of claims 1 to 8 comprises: A meteorological forecast data acquisition module is used for the interactive meteorological platform to collect and obtain meteorological forecast data for a preset time window, wherein the meteorological forecast data includes meteorological data on electricity consumption and meteorological data on power generation; An electricity demand analysis module is used to perform electricity demand analysis based on the electricity consumption-influencing meteorological data and generate predicted electricity demand data with the preset time window as a constraint; A power generation prediction module, used to predict the power generation of new energy based on the power generation-influencing meteorological data, and generate predicted power generation data of new energy; A deviation calculation module, used to perform deviation calculation on the predicted electricity demand data and the predicted new energy power generation data, and obtain predicted thermal power compensation demand data according to the calculation result; A prediction model building module, used to pre-build a short-term variable cost prediction model, wherein the short-term variable cost prediction model includes a new energy power generation cost prediction branch and a thermal power compensation cost prediction branch; A cost forecasting and analysis module is used to synchronize the predicted renewable energy power generation data and the predicted thermal power compensation demand data to the renewable energy power generation cost forecasting branch and the thermal power compensation cost forecasting branch of the short-term variable cost forecasting model, respectively, to perform cost forecasting and analysis, and obtain a renewable energy power generation cost forecast value and a thermal power compensation cost forecast value; The sum calculation module is used to sum the predicted value of the new energy power generation cost and the predicted value of the thermal power compensation cost, and output the predicted value of the short-term variable power generation cost.

10. The short-term variable cost forecasting system based on big data according to claim 9, characterized in that: The weather forecast data acquisition module comprises: A power generation associated variable definition unit, used to predefine power generation associated variables, wherein the power generation associated variables include wind power associated variables and photovoltaic associated variables; A wind power meteorological data acquisition unit, configured to acquire wind power meteorological data through an interactive meteorological platform within the preset time window and with the wind power associated variables as constraints; A photovoltaic meteorological data collection unit is used to collect and obtain photovoltaic meteorological data through an interactive meteorological platform within the preset time window and with the photovoltaic-related variables as constraints; A meteorological data integration unit, used for integrating the wind power meteorological data and the photovoltaic meteorological data as the power generation impact meteorological data; The electricity consumption-affected meteorological data collection unit is used to predefine electricity consumption-related variables. Within the preset time window, the interactive meteorological platform collects and obtains the electricity consumption-affected meteorological data with the electricity consumption-related variables as constraints.

11. The short-term variable cost forecasting system based on big data according to claim 10, characterized in that: The electricity demand analysis module includes: A time feature extraction unit, used to extract the time feature of the preset time window to obtain the power consumption time feature; The power demand analysis unit is used to perform power demand analysis based on the power consumption time characteristics and the power consumption-influencing meteorological data to generate predicted power demand data.

12. The short-term variable cost forecasting system based on big data according to claim 11, characterized in that: The electricity consumption time characteristics include electricity consumption time period characteristics, electricity consumption seasonal characteristics, and electricity consumption holiday characteristics.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the short-term variable cost forecasting method based on big data as described in any one of claims 1 to 8 are implemented.

Images