New energy data acquisition and control method based on scheduling master station

Abstract

The invention discloses a new energy data acquisition and control method based on a scheduling master station, relates to the technical field of new energy, and aims to calculate each type of to-be-acquired data, namely each type of data used for training a machine learning model in a data driving manner to obtain a reasonable acquisition proportion. Specifically, a comprehensive score of each type of data is obtained through an influence index, a correlation index and a fluctuation index, and then the comprehensive score is compared with a total score to obtain a collection proportion of each type of data; therefore, under the premise that the balance of each type of data used for model training is effectively ensured, the situation that a certain type of data is insufficient in representativeness can be avoided, namely under the premise that the representativeness of each type of data is met, the precision of the model during training can be effectively ensured, meanwhile, the accuracy of the model during subsequent prediction can be effectively improved, and the prediction efficiency of the model is improved. And thus, a more scientific and accurate control strategy can be formulated on the premise of a more accurate prediction result.

Description

Technical Field

[0001] The present invention relates to the field of new energy technology, and in particular to a new energy data collection and control method based on a dispatching master station. Background Art

[0002] With the growth of global demand for clean energy and the development of power systems towards intelligence and automation, how to effectively manage and optimize the performance of renewable energy power generation systems has become a key issue.

[0003] In current practice, in order to achieve effective management and optimization of renewable energy power generation, a series of advanced monitoring and control systems are usually used to complete data collection tasks. These systems rely on remote terminal units, smart meters and other types of sensor equipment installed at various renewable energy power generation sites to monitor and report key parameters such as voltage, current, power output, etc. in real time. In addition, the environmental condition information provided by the meteorological station is also crucial because they have a direct impact on the prediction of renewable energy power generation. Subsequently, big data technology or machine learning algorithms can be used to predict the future use trend of renewable energy power generation from massive historical and real-time data, and control decisions can be made accordingly. At the same time, the dispatching master station sends instructions to each power generation site based on the control decision and the automatic control system to realize power regulation, frequency control and other functions. This system can greatly improve the efficiency and reliability of renewable energy power generation.

[0004] Although the existing data collection and control methods based on the dispatching master station have made significant progress, there are still some defects in practical applications:

[0005] In the prior art, data collection for machine learning model training often fails to fully consider the balance between various types of data. Certain types of data are easily over-emphasized or neglected, resulting in an unbalanced proportion of training data. For example, for wind farms, too much attention is paid to wind speed and direction data, while relatively ignoring factors such as temperature and humidity that are equally important to power generation efficiency; for solar power stations, too much emphasis is placed on light intensity, while insufficient attention is paid to influencing factors such as changes in cloud cover. This unbalanced data collection will cause some of the data used for subsequent model training to lack representativeness. Once this happens, it is very easy to cause the model prediction results to be biased towards performance under specific conditions, while ignoring other potentially important variables, which will not only easily lead to a decrease in the accuracy of model training, but also easily affect the accuracy of subsequent model predictions, and thus easily lead to the lack of scientificity and accuracy in the final control strategy.

[0006] Therefore, the prior art urgently needs a technical solution for the new energy data collection and control method based on the dispatching master station. Summary of the invention

[0007] In order to solve the above technical problems, the present invention provides a new energy data collection and control method based on a dispatching master station, which specifically includes the following steps:

[0008] Step S1, classify the data to be collected according to their nature, perform analysis on each type of data, and obtain the collection ratio of each type of data based on the analysis results;

[0009] Step S1a, by analyzing the historical data, determining the influence of each type of data on the prediction results of the prediction model, and obtaining the influence index of each type of data;

[0010] Step S1a1, collecting usage of each type of data in the historical data, wherein the usage includes the usage frequency, application scope and coverage of each type of data;

[0011] Step S1a2: based on the usage frequency, application scope and coverage of each type of data, calculate the influence of each type of data on the prediction results of the prediction model;

[0012] Step S1a3, performing a conversion operation on the influence degree value of each type of data on the prediction result of the prediction model to obtain the influence index of each type of data;

[0013] The calculation formula for the impact index of each type of data is:

[0014] I i =f(α·U i +β·A i +γ·C i );

[0015] In the formula, I i represents the influence index of the i-th type of data; f represents a monotonically increasing function; U i represents the usage frequency of the i-th type of data; A i Represents the application scope of the i-th category of data; C i represents the coverage rate of the i-th category of data; α, β, and γ represent the weight coefficients of usage frequency, application scope, and coverage rate, respectively.

[0016] Step S1b, analyzing the correlation between each type of data and the prediction results of the prediction model by statistical methods to obtain a correlation index for each type of data;

[0017] Step S1b1, obtaining the time series value of each type of data and the time series value of the prediction result of the prediction model;

[0018] Step S1b2, using a sliding window method to segment the time series values ​​of each type of data and the prediction results of the prediction model, to obtain a data set of several sub-time periods of each type of data and the prediction results of the prediction model, wherein the data set includes several data points;

[0019] Step S1b3, using each data point in the data set of each sub-time period of each type of data, and each data point in the data set of each sub-time period of the prediction model prediction result, apply the Pearson correlation coefficient formula to analyze the sub-correlation between each type of data and the prediction model prediction result, and obtain the sub-correlation coefficient of each type of data in each sub-time period;

[0020] Among them, the calculation formula for obtaining the sub-correlation coefficient of each type of data in each sub-time period is:

[0021]

[0022] In the formula, r i,j represents the sub-correlation coefficient of the i-th category of data in the j-th sub-time period; X i,j,k represents the kth data point in the data set of the i-th category of data in the j-th sub-time period; represents the mean of all data points in the data set of the i-th category of data in the j-th sub-time period; Y j,k The kth data point in the data set representing the prediction result of the prediction model in the jth sub-time period; Represents the mean of all data points in the data set in the jth sub-time period in the prediction model’s prediction result; n j represents the number of data points in the dataset for the jth sub-period;

[0023] Step S1b4, combining the sub-correlation coefficients of each type of data in each sub-time period and taking the average, to obtain the correlation index of each type of data;

[0024] Among them, the calculation formula for the correlation index of each type of data is:

[0025]

[0026] In the formula, R i Represents the correlation index of the i-th category of data; r i,j represents the sub-correlation coefficient of the i-th category of data in the j-th sub-time period; m represents the number of sub-time periods.

[0027] Step S1c, analyzing the volatility of each type of data to obtain a volatility index for each type of data;

[0028] Step S1c1, calculating the standard deviation of each type of data;

[0029] Step S1c11, calling the time series value of each type of data and the data set of one of the sub-time periods after the time series value is divided through step S1b1 and step S1b2, including a number of data points contained in the data set;

[0030] Step S1c12, calculating the mean of several data points included in the data set, and obtaining the standard deviation of each type of data in a sub-time period based on the mean;

[0031] Step S1c2, calculating the standard deviation change rate of each type of data;

[0032] Step S1c21, calling the time series value of each type of data and the data sets of several sub-time periods after the time series value is divided through step S1b1 and step S1b2, including several data points contained in the data sets;

[0033] Step S1c22, repeat step S1c12 to calculate the standard deviation of each type of data in each sub-time period;

[0034] Step S1c23, calculating the standard deviation normalization rate of each type of data in two adjacent sub-time periods through the standard deviation of each type of data in two adjacent sub-time periods;

[0035] The calculation formula for the standard deviation standardization rate of each type of data in two adjacent sub-time periods is:

[0036]

[0037] In the formula, Δs i represents the standard deviation of the i-th type of data in two adjacent sub-time periods; s (t+1)，i represents the standard deviation of the i-th type of data in the t+1 sub-time period; s t，i Represents the standard deviation of the i-th type of data in t sub-time periods; Δ t represents the time interval between t+1 sub-periods and t sub-periods;

[0038] Step S1c3, combining the standard deviation and standard deviation change rate of each type of data to obtain the volatility index of each type of data;

[0039] Among them, the calculation formula for the volatility index of each type of data is:

[0040]

[0041] Where V i Represents the volatility index of the i-th type of data; s i Represents the standard deviation of the i-th type of data in a sub-time period; Δs i Represents the standard deviation of the i-th type of data in two adjacent sub-time periods;

[0042] Step S1d, obtaining a comprehensive score for each type of data based on the influence index, correlation index and volatility index of each type of data;

[0043] The calculation formula for obtaining the comprehensive score of each type of data is:

[0044]

[0045] In the formula, S i Represents the comprehensive score of the i-th category data; I i Represents the impact index of the i-th category of data; R i Represents the correlation index of the i-th category data; V i Represents the volatility index of the i-th category of data.

[0046] Step S1e, obtaining the collection ratio of each type of data according to the comprehensive score of each type of data;

[0047] Step S1e1, summing up the comprehensive scores of each category of data to obtain the total scores of all categories of data;

[0048] Step S1e2, obtaining the collection ratio of each type of data according to the ratio of the comprehensive score of each type of data to the total score;

[0049] Among them, the calculation formula for obtaining the collection ratio of each type of data is:

[0050]

[0051] Where P i represents the collection ratio of the i-th type of data; S i represents the comprehensive score of the i-th category data; q represents the number of data categories.

[0052] Step S2, performing data collection according to the collection ratio of each type of data, and performing preprocessing on each type of collected data;

[0053] Step S3: training a pre-built machine learning model using each type of data collected in the collection ratio and pre-processed to obtain a trained machine learning model, and inputting the data to be predicted into the trained machine learning model to obtain a prediction result;

[0054] Step S4: Make a control decision based on the prediction result, and the dispatching master station sends a control instruction to each power station through the automatic control system according to the control decision.

[0055] The embodiments of the present invention have the following technical effects:

[0056] The present invention aims to calculate a reasonable collection ratio for each type of data to be collected, that is, each type of data used for machine learning model training, in a data-driven manner. Specifically, a comprehensive score for each type of data is obtained through the influence index, correlation index and volatility index, and then the comprehensive score is compared with the total score to obtain the collection ratio of each type of data. This can avoid the situation where a certain type of data is insufficiently representative while effectively ensuring the balance of each type of data used in model training. That is, it can effectively ensure the accuracy of the model during training while satisfying the representativeness of each type of data, and at the same time can effectively improve the accuracy of the model in subsequent predictions, and thus can formulate more scientific and precise control strategies based on more accurate prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0058] Figure 1 It is a flow chart of a new energy data collection and control method based on a dispatching master station provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work belong to the scope of protection of the present invention.

[0060] Embodiment 1: Figure 1 As shown, the present invention provides a new energy data collection and control method based on a dispatching master station, comprising the following steps:

[0061] Step S1, classify the data to be collected according to their nature, perform analysis on each type of data, and obtain the collection ratio of each type of data based on the analysis results;

[0062] In order to ensure the scientificity and representativeness of data collection, in this embodiment, the data to be collected are classified according to the physical properties, time characteristics, spatial characteristics, etc. of the data; wherein the physical properties refer to the specific physical phenomena or attributes reflected by the data. According to this property, the data to be collected are divided into the following categories: temperature and humidity data, wind speed and wind direction data, light intensity data, air pressure and precipitation data, and power parameter data, such as voltage, current, power, etc. The time characteristic refers to the time distribution characteristics of the data, including real-time, periodic and seasonal changes. According to this characteristic, the data is divided into the following categories: real-time data, such as power generation equipment status data and power grid frequency and voltage data; periodic data, such as daily power consumption mode data and weekly weather forecast data; seasonal data, such as quarterly user behavior data and annual external influencing factor data. Spatial characteristics refer to the spatial distribution characteristics of data, including geographical location and coverage. According to this characteristic, the data is divided into the following categories: geographical area data, such as regional meteorological data and distributed power generation site data. Local environmental data, such as environmental data around a specific power station. National data, national power load data and national weather forecast data.

[0063] Step S1a, by analyzing the historical data, determining the influence of each type of data on the prediction results of the prediction model, and obtaining the influence index of each type of data;

[0064] Step S1a1, collecting usage of each type of data in the historical data, wherein the usage includes the usage frequency, application scope and coverage of each type of data;

[0065] Frequency of use refers to the number of times a certain type of data is accessed, processed or used within a certain period of time. The specific steps are as follows: first select a suitable time window, such as hourly, daily or weekly, and then for each time window, count the number of times the data is accessed or used, such as obtaining wind speed data from a weather station once a day, obtaining electricity consumption data from a smart meter every 15 minutes, etc.; then sum up the number of times used in all time windows, and then divide it by the total number of time windows to get the frequency of use of the data.

[0066] The application scope refers to the application field or scenario of a certain type of data in the system. The specific steps are as follows: first list all the application scenarios involved in the actual operation of this type of data, such as power parameter data is used for grid stability monitoring, equipment status monitoring, etc.; then, according to the importance of the application scenario, use a method based on historical data analysis to determine the weight of this type of data in each application scenario; finally, sum up the weights of each application scenario of this type of data to obtain the application scope of this type of data.

[0067] Coverage refers to the degree of coverage of a certain type of data in the entire system or geographic area. The specific steps are as follows: first, clarify the geographical scope or system boundary of data collection, such as distributed power generation site data covering multiple wind farms or solar power stations; then for each geographical or system boundary, count the number of collection points for this type of data, such as 5,000 smart meters installed throughout the city for power load data collection; finally, divide the number of collection points obtained by the statistics by the total number of collection points to obtain the coverage of this type of data.

[0068] Step S1a2: based on the usage frequency, application scope and coverage of each type of data, calculate the influence of each type of data on the prediction results of the prediction model;

[0069] It is worth noting that when calculating the influence of each type of data on the prediction results of the prediction model, the frequency of use, application scope and coverage of each type of data are combined. This is because the frequency of use refers to the number of times a certain type of data is accessed, processed or used within a certain period of time. It reflects the activity and importance of this type of data in actual operations. Therefore, data with high frequency of use often means that this type of data has high immediacy and criticality for the operation and decision-making of the system. For example, in the power system, the frequency of use of real-time power load data is very high because it is directly related to the balance of supply and demand and the stability of the power grid. This can further explain that data with high frequency of use contains more up-to-date information and can more accurately reflect the current status, so it has a greater impact on the prediction results of the prediction model. Similarly, data with low frequency of use contains less up-to-date information, but it may still be of great significance to certain specific scenarios. Therefore, it also has a greater impact on the prediction results of the prediction model. This is enough to show that frequency of use is an important dimension for evaluating the prediction results of the prediction model.

[0070] Similarly, regarding the scope of application, widely used data can provide valuable information in different scenarios, thereby effectively improving the comprehensiveness and accuracy of the prediction results of the prediction model. Although narrowly applied data plays an important role in specific scenarios, its influence on global predictions is relatively small. Therefore, the scope of application is also another important dimension for evaluating the impact of the prediction results of the prediction model. And the coverage rate is also another important dimension for evaluating the impact of the prediction results of the prediction model, which will not be elaborated here.

[0071] Step S1a3, performing a conversion operation on the influence degree value of each type of data on the prediction result of the prediction model to obtain the influence index of each type of data;

[0072] The calculation formula for the impact index of each type of data is:

[0073] I i =f(α·Ui +β·A i +γ·C i );

[0074] In the formula, I i represents the influence index of the i-th type of data; f represents a monotonically increasing function; U i represents the usage frequency of the i-th type of data; A i Represents the application scope of the i-th category of data; C i represents the coverage rate of the i-th category of data; α, β, and γ represent the weight coefficients of usage frequency, application scope, and coverage rate, respectively.

[0075] It is worth further explaining that α·U i +β·A i +γ·C i The calculation result of represents the influence of the i-th type of data on the prediction results of the prediction model.

[0076] Step S1b, analyzing the correlation between each type of data and the prediction results of the prediction model by statistical methods to obtain a correlation index for each type of data;

[0077] Step S1b1, obtaining the time series value of each type of data and the time series value of the prediction result of the prediction model;

[0078] Step S1b2, using a sliding window method to segment the time series values ​​of each type of data and the prediction results of the prediction model, to obtain a data set of several sub-time periods of each type of data and the prediction results of the prediction model, wherein the data set includes several data points;

[0079] Step S1b3, using each data point in the data set of each sub-time period of each type of data, and each data point in the data set of each sub-time period of the prediction model prediction result, apply the Pearson correlation coefficient formula to analyze the sub-correlation between each type of data and the prediction model prediction result, and obtain the sub-correlation coefficient of each type of data in each sub-time period;

[0080] Among them, the calculation formula for obtaining the sub-correlation coefficient of each type of data in each sub-time period is:

[0081]

[0082] In the formula, r i,j represents the sub-correlation coefficient of the i-th category of data in the j-th sub-time period; X i,j,k represents the kth data point in the data set of the i-th category of data in the j-th sub-time period; represents the mean of all data points in the data set of the i-th category of data in the j-th sub-time period; Y j,kThe kth data point in the data set representing the prediction result of the prediction model in the jth sub-time period; Represents the mean of all data points in the data set in the jth sub-time period in the prediction model’s prediction result; n j represents the number of data points in the dataset for the jth sub-period;

[0083] Step S1b4, combining the sub-correlation coefficients of each type of data in each sub-time period and taking the average, to obtain the correlation index of each type of data;

[0084] Among them, the calculation formula for the correlation index of each type of data is:

[0085]

[0086] In the formula, R u Represents the correlation index of the i-th category of data; r u,j represents the sub-correlation coefficient of the i-th category of data in the j-th sub-time period; m represents the number of sub-time periods.

[0087] Step S1c, analyzing the volatility of each type of data to obtain a volatility index for each type of data;

[0088] Step S1c1, calculating the standard deviation of each type of data;

[0089] It is worth noting that when calculating the standard deviation of each type of data, we mainly analyze the standard deviation of each type of data within a fixed period of time;

[0090] Step S1c11, calling the time series value of each type of data and the data set of one of the sub-time periods after the time series value is divided through step S1b1 and step S1b2, including a number of data points contained in the data set;

[0091] Step S1c12, calculating the mean of several data points included in the data set, and obtaining the standard deviation of each type of data in a sub-time period based on the mean;

[0092] Step S1c2, calculating the standard deviation change rate of each type of data;

[0093] It is worth noting that when calculating the standard deviation change rate of each type of data, the main thing is to analyze the standard deviation standardization rate of each type of data in different time intervals;

[0094] Step S1c21, calling the time series value of each type of data and the data sets of several sub-time periods after the time series value is divided through step S1b1 and step S1b2, including several data points contained in the data sets;

[0095] Step S1c22, repeat step S1c12 to calculate the standard deviation of each type of data in each sub-time period;

[0096] Step S1c23, calculating the standard deviation normalization rate of each type of data in two adjacent sub-time periods through the standard deviation of each type of data in two adjacent sub-time periods;

[0097] The calculation formula for the standard deviation standardization rate of each type of data in two adjacent sub-time periods is:

[0098]

[0099] In the formula, Δs i represents the standard deviation of the i-th type of data in two adjacent sub-time periods; s (t+1)，i represents the standard deviation of the i-th type of data in the t+1 sub-time period; s t，i Represents the standard deviation of the i-th type of data in t sub-time periods; Δ t represents the time interval between t+1 sub-periods and t sub-periods;

[0100] Step S1c3, combining the standard deviation and standard deviation change rate of each type of data to obtain the volatility index of each type of data;

[0101] Among them, the calculation formula for the volatility index of each type of data is:

[0102]

[0103] Where V i Represents the volatility index of the i-th type of data; s i Represents the standard deviation of the i-th type of data in a sub-time period; Δs i Represents the standard deviation of the i-th type of data in two adjacent sub-time periods;

[0104] Step S1d, obtaining a comprehensive score for each type of data based on the influence index, correlation index and volatility index of each type of data;

[0105] The calculation formula for obtaining the comprehensive score of each type of data is:

[0106]

[0107] In the formula, S i Represents the comprehensive score of the i-th category data; I i Represents the impact index of the i-th category of data; R i Represents the correlation index of the i-th category data; V i Represents the volatility index of the i-th category of data.

[0108] It is worth noting that when calculating the comprehensive score of each type of data, the influence index can be used to quantify the relative importance of each type of data in the prediction model, ensuring that key data receive sufficient attention, and helping to reasonably allocate data collection resources, giving priority to collecting data that have a significant impact on the prediction results; the correlation index can be used to identify which data are strongly correlated with the prediction results, thereby improving the prediction accuracy of the model, and helping to screen out the key features that best reflect changes in the prediction results, avoiding noise caused by redundant data; the volatility index can be used to evaluate the stability and reliability of the data, ensuring that the data used to train the model has a certain stability, and the volatility index can also help the model better adapt to dynamic changes in data and improve its ability to cope with complex environments; therefore, the comprehensive score takes into account the importance of the data, namely the influence index, the relevance, namely the correlation index, and the stability, namely the volatility index, and can provide a more comprehensive data evaluation, thereby helping to scientifically and reasonably set the data collection ratio, ensuring that key data are fully collected, while avoiding unnecessary data redundancy.

[0109] Step S1e, obtaining the collection ratio of each type of data according to the comprehensive score of each type of data;

[0110] Step S1e1, summing up the comprehensive scores of each category of data to obtain the total scores of all categories of data;

[0111] Step S1e2, obtaining the collection ratio of each type of data according to the ratio of the comprehensive score of each type of data to the total score;

[0112] Among them, the calculation formula for obtaining the collection ratio of each type of data is:

[0113]

[0114] Where P i represents the collection ratio of the i-th type of data; S i represents the comprehensive score of the i-th category data; q represents the number of data categories.

[0115] Step S2, performing data collection according to the collection ratio of each type of data, and performing preprocessing on each type of collected data;

[0116] Preprocessing usually includes data cleaning and missing value processing, and the detailed processing process will not be described in detail here.

[0117] Step S3: training a pre-built machine learning model using each type of data collected in the collection ratio and pre-processed to obtain a trained machine learning model, and inputting the data to be predicted into the trained machine learning model to obtain a prediction result;

[0118] The process of building a machine learning model, including model training and prediction, is an existing conventional technical means and will not be elaborated on here.

[0119] Step S4: making a control decision based on the prediction result, and the dispatching master station sends a control instruction to each power station through the automatic control system according to the control decision;

[0120] Based on the prediction results, formulate scientific and reasonable control strategies to optimize the operating efficiency of power plants and grid stability, including control strategy formulation, control command generation, command sending and execution, etc.;

[0121] The control strategy formulation includes goal setting: such as maximizing power generation efficiency to ensure that each power station can improve power generation efficiency as much as possible while meeting demand; minimizing costs by optimizing power generation plans to reduce fuel consumption and other operating costs; ensuring grid stability, that is, maintaining grid frequency and voltage within a reasonable range to ensure the quality of power supply. And strategy selection: such as adjusting the output power of each power station in real time according to predicted short-term load changes; planning the maintenance and overhaul arrangements of power generation equipment in advance based on long-term forecast results; and formulating emergency plans for emergencies such as natural disasters and equipment failures to ensure rapid response and recovery.

[0122] The control command generation includes command types: such as specifying the specific power generation of each power station, such as increasing or decreasing the output of a generator; controlling the operating status of specific equipment, such as starting, stopping, switching working modes, etc.; managing the charging and discharging status of the energy storage system, and optimizing energy storage and release. And the command format: such as using a unified communication protocol such as IEC61850 to ensure the transmission and parsing consistency of the command; setting different command priorities according to the urgency and importance to ensure that key commands are executed first.

[0123] The instruction sending and execution include instruction sending: such as using a secure and reliable communication network such as optical fiber and wireless communication to send instructions from the dispatching master station to each power station, and ensuring identity authentication and encryption during the instruction sending process to prevent unauthorized operations. And instruction execution: such as each power station is equipped with a local control system such as PLC and SCADA, receiving and parsing instructions from the dispatching master station, and establishing a feedback mechanism to ensure that the execution status and results can be fed back in time after the instruction is executed, which is convenient for monitoring and adjustment.

[0124] It should be noted that the terms used in the present invention are only for describing specific embodiments, rather than limiting the scope of the present application. As shown in the present specification, unless the context clearly indicates an exception, the words "one", "a", "a kind of" and / or "the" do not specifically refer to the singular, but may also include the plural. The terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method or device. In the absence of more restrictions, the elements defined by the sentence "include one..." do not exclude the presence of other identical elements in the process, method or device including the elements.

[0125] It should also be noted that the terms "center", "up", "down", "left", "right", "vertical", "horizontal", "inside", "outside", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. Unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", etc. should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims

1. A new energy data collection and control method based on a dispatching master station, characterized in that: The following steps are involved: Step S1, classify the data to be collected according to their nature, perform analysis on each type of data, and obtain the collection ratio of each type of data based on the analysis results; Step S2, performing data collection according to the collection ratio of each type of data, and performing preprocessing on each type of collected data; Step S3: training a pre-built machine learning model using each type of data collected in the collection ratio and pre-processed to obtain a trained machine learning model, and inputting the data to be predicted into the trained machine learning model to obtain a prediction result; Step S4: Make a control decision based on the prediction result, and the dispatching master station sends a control instruction to each power station through the automatic control system according to the control decision.

2. The new energy data collection and control method based on the dispatching master station according to claim 1 is characterized in that: The collected data is classified according to its nature, and each type of data is analyzed. Based on the analysis results, the collection ratio of each type of data is obtained, including: Step S1a, by analyzing the historical data, determining the influence of each type of data on the prediction results of the prediction model, and obtaining the influence index of each type of data; Step S1b, analyzing the correlation between each type of data and the prediction results of the prediction model by statistical methods to obtain a correlation index for each type of data; Step S1c, analyzing the volatility of each type of data to obtain a volatility index for each type of data; Step S1d, obtaining a comprehensive score for each type of data based on the influence index, correlation index and volatility index of each type of data; Step S1e: According to the comprehensive score of each type of data, the collection ratio of each type of data is obtained.

3. The new energy data collection and control method based on the dispatching master station according to claim 2 is characterized in that: By analyzing historical data, we can determine the degree of influence of each type of data on the prediction results of the prediction model and obtain the influence index of each type of data, including: Step S1a1, collecting usage of each type of data in the historical data, wherein the usage includes the usage frequency, application scope and coverage of each type of data; Step S1a2: based on the usage frequency, application scope and coverage of each type of data, calculate the influence of each type of data on the prediction results of the prediction model; Step S1a3: Perform a conversion operation on the influence degree value of each type of data on the prediction result of the prediction model to obtain the influence index of each type of data: I i =f(α·U i +β·A i +γ·C i ); In the formula, I i represents the influence index of the i-th type of data; f represents a monotonically increasing function; U i represents the usage frequency of the i-th type of data; A i Represents the application scope of the i-th category of data; C i represents the coverage rate of the i-th category of data; α, β, and γ represent the weight coefficients of usage frequency, application scope, and coverage rate, respectively.

4. The new energy data collection and control method based on the dispatching master station according to claim 2 is characterized in that: The correlation between each type of data and the prediction results of the prediction model is analyzed by statistical methods to obtain the correlation index of each type of data, including: Step S1b1, obtaining the time series value of each type of data and the time series value of the prediction result of the prediction model; Step S1b2, using a sliding window method to segment the time series values ​​of each type of data and the prediction results of the prediction model, to obtain a data set of several sub-time periods of each type of data and the prediction results of the prediction model, wherein the data set includes several data points; Step S1b3, using each data point in the data set of each sub-time period of each type of data, and each data point in the data set of each sub-time period of the prediction model prediction result, apply the Pearson correlation coefficient formula to analyze the sub-correlation between each type of data and the prediction model prediction result, and obtain the sub-correlation coefficient of each type of data in each sub-time period: In the formula, r i，j represents the sub-correlation coefficient of the i-th category of data in the j-th sub-time period; X i，j，k represents the kth data point in the data set of the i-th category of data in the j-th sub-time period; represents the mean of all data points in the data set of the i-th category of data in the j-th sub-time period; Y j，k The kth data point in the data set representing the prediction result of the prediction model in the jth sub-time period; Represents the mean of all data points in the data set in the jth sub-time period in the prediction model’s prediction result; n j represents the number of data points in the dataset for the jth sub-period; Step S1b4: synthesize the sub-correlation coefficients of each type of data in each sub-time period and take the average to obtain the correlation index of each type of data: In the formula, R i Represents the correlation index of the i-th category of data; r i，j represents the sub-correlation coefficient of the i-th category of data in the j-th sub-time period; m represents the number of sub-time periods.

5. The new energy data collection and control method based on the dispatching master station according to claim 2 is characterized in that: Analyze the volatility of each type of data and obtain the volatility index of each type of data, including: Step S1c1, calculating the standard deviation of each type of data; Step S1c11, calling the time series value of each type of data and the data set of one of the sub-time periods after the time series value is divided through step S1b1 and step S1b2, including a number of data points contained in the data set; Step S1c12, calculating the mean of several data points included in the data set, and obtaining the standard deviation of each type of data in a sub-time period based on the mean; Step S1c2, calculating the standard deviation change rate of each type of data; Step S1c21, calling the time series value of each type of data and the data sets of several sub-time periods after the time series value is divided through step S1b1 and step S1b2, including several data points contained in the data sets; Step S1c22, repeat step S1c12 to calculate the standard deviation of each type of data in each sub-time period; Step S1c23: Calculate the standard deviation normalization rate of each type of data in two adjacent sub-time periods by using the standard deviation of each type of data in two adjacent sub-time periods: In the formula, Δs i represents the standard deviation of the i-th type of data in two adjacent sub-time periods; s (t+1)，i represents the standard deviation of the i-th type of data in the t+1 sub-time period; s t，i Represents the standard deviation of the i-th type of data in t sub-time periods; Δ t represents the time interval between t+1 sub-periods and t sub-periods; Step S1c3, combining the standard deviation and standard deviation change rate of each type of data to obtain the volatility index of each type of data: Where V i Represents the volatility index of the i-th type of data; s i Represents the standard deviation of the i-th type of data in a sub-time period; Δs i Represents the standard deviation standardized rate of the i-th type of data in two adjacent sub-time periods.

6. The new energy data collection and control method based on the dispatching master station according to claim 2 is characterized in that: The calculation formula for obtaining the comprehensive score of each type of data is: In the formula, S i Represents the comprehensive score of the i-th category data; I i Represents the impact index of the i-th category of data; R i Represents the correlation index of the i-th category data; V i Represents the volatility index of the i-th category of data.

7. The new energy data collection and control method based on the dispatching master station according to claim 2 is characterized in that: According to the comprehensive score of each type of data, the collection ratio of each type of data is obtained, including: Step S1e1, summing up the comprehensive scores of each category of data to obtain the total scores of all categories of data; Step S1e2: According to the ratio of the comprehensive score of each type of data to the total score, the collection ratio of each type of data is obtained: Where P i represents the collection ratio of the i-th type of data; S i represents the comprehensive score of the i-th category data; q represents the number of data categories.

Images