Power grid bus load prediction method and system based on subtropical monsoon climate characteristics, medium and processor

Abstract

The invention provides a power grid bus load prediction method based on subtropical monsoon climate characteristics. The method comprises the following steps: S1, collecting and preprocessing historical data; s2, extracting features related to time and weather, performing coding, and performing corresponding association with the load data; s3, extracting time and weather features to perform interaction enhancement, and creating time-weather interaction features; s4, meteorological expansion features are generated after meteorological features are extracted; s5, constructing a machine learning model, and randomly dividing the features into a training set, a verification set and a test set for training, verifying and testing the machine learning model to obtain a load prediction model for the power grid bus; and S6, predicting the load of the power grid bus by using the load prediction model. The time-weather interaction characteristics can capture the complex influence of the weather factors on the power grid bus load in different time periods, so that the bus load is predicted more accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid bus load prediction, and in particular to a power grid bus load prediction method, system, medium and processor based on subtropical monsoon climate characteristics. Background Art

[0002] Subtropical monsoon climate areas have complex and distinct climate change characteristics, such as obvious seasonal changes and concentrated precipitation. Subtropical monsoon climate has high temperatures and heavy rains in summer and mild and little rains in winter. Precipitation is concentrated in summer, and extreme weather such as heavy rains may affect the operating status of power grid equipment. Therefore, the factors affecting the power grid bus are intertwined and complex.

[0003] The existing methods for predicting the load of power grid busbars all use relatively simple data usage methods, such as simply processing the collected data and using it for training the prediction model. There are problems such as incomplete consideration of meteorological factors and insufficient utilization of time period rules. For example, the existing technology may simply consider the impact of a single meteorological factor (such as only considering the temperature) on the load of power grid busbars. In subtropical monsoon climate areas, meteorological factors are complex and diverse. Not only temperature, but also humidity, precipitation, wind speed and other factors will affect the load, and there are interactions between these factors. Traditional methods may not fully consider the impact of dynamic changes in meteorological factors on the load. The subtropical monsoon climate is characterized by changeable meteorological conditions. For example, thunderstorms in summer may cause a sudden drop in temperature and change in humidity in a short period of time, thereby affecting the load. The existing technology may ignore the complex change rules of the load in different time periods (such as seasons, day and night, weekdays / weekends). In subtropical monsoon climate areas, seasonal changes are obvious, and the load characteristics of different seasons vary greatly. Therefore, the current technology uses meteorological and time characteristic data in an overly simple way, making the model inaccurate for load prediction, resulting in a lack of flexibility in power grid dispatching.

[0004] In view of this, a method, system, medium and processor for predicting power grid bus load based on subtropical monsoon climate characteristics are needed. Summary of the invention

[0005] In view of the problem that the use of meteorological and time characteristic data in the prior art is too simple, making the model's load forecasting not accurate enough, the present invention provides a power grid bus load forecasting method, system, medium and processor based on subtropical monsoon climate characteristics, which can solve the problems of incomplete consideration of meteorological factors and insufficient utilization of time cycle rules, and improve the accuracy of load forecasting. The specific technical solution is as follows:

[0006] A method for predicting power grid bus load based on subtropical monsoon climate characteristics, comprising:

[0007] S1: Collect and preprocess historical data;

[0008] S2: extract and encode time-related and meteorological features, and associate them with load data;

[0009] S3: extracting time and meteorological features for interactive reinforcement and creating time-meteorological interactive features;

[0010] S4: extract meteorological features and generate meteorological extended features;

[0011] S5: Build a machine learning model, and randomly divide the above features into a training set, a validation set, and a test set for training, validating, and testing the machine learning model to obtain a load prediction model for the power grid bus;

[0012] S6: Use the load forecasting model to forecast the load on the power grid bus.

[0013] Furthermore, the historical data includes load data, time data and meteorological data.

[0014] Furthermore, the preprocessing includes data cleaning and data standardization.

[0015] Furthermore, the time-weather interaction characteristics include delay interaction characteristics, time period-weather interaction characteristics, weather change rate-time interaction characteristics, cycle-weather interaction characteristics and cumulative weather-time interaction characteristics.

[0016] Furthermore, when generating meteorological extended features after extracting meteorological features, the meteorological extended features are generated by multiplication or division.

[0017] Furthermore, the meteorological extended features include the perceived temperature, the multiplication feature of air temperature and humidity, and the sunshine intensity divided by wind speed.

[0018] Furthermore, the calculation formula of the body temperature is as follows:

[0019] AT=T+0.348×RH-0.7×WS+0.7;

[0020] In the above formula, AT is the perceived temperature; T is the actual air temperature; RH is the relative humidity; WS is the wind speed.

[0021] A power grid bus load prediction system based on subtropical monsoon climate characteristics, applied to the power grid bus load prediction method based on subtropical monsoon climate characteristics described above, comprises:

[0022] Collection module, which is used to collect and preprocess historical data;

[0023] An encoding module is used to extract and encode time and weather-related features and associate them with load data;

[0024] An interactive module, which is used to extract time and meteorological features for interactive enhancement and create time-meteorological interactive features;

[0025] An expansion module, which is used to extract meteorological features and generate meteorological expansion features;

[0026] A training module is used to build a machine learning model, and at the same time, the above-mentioned features are randomly divided into a training set, a validation set and a test set, so as to train, validate and test the machine learning model to obtain a load prediction model for the power grid bus;

[0027] The prediction module is used to use the load prediction model to predict the load of the power grid bus.

[0028] A computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the above-mentioned method for predicting the power grid bus load with subtropical monsoon climate characteristics.

[0029] A processor is used to run a program, wherein the program executes the above-mentioned power grid bus load prediction method with subtropical monsoon climate characteristics when running.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] 1. The time-weather interaction feature can capture the complex impact of meteorological factors on the power grid bus load in different time periods. For example, in the summer afternoon, when high temperature and high humidity occur at the same time, the power grid bus load will have a specific change pattern compared with other time and meteorological condition combinations. This interactive feature allows the model to learn this complex conditional dependency, thereby more accurately predicting the bus load.

[0032] 2. Consider the impact of special meteorological events at different times. In the subtropical monsoon climate, extreme weather events such as typhoons and rainstorms occur from time to time. Through the time-meteorological interaction characteristics, the model can learn the different impacts of these extreme weather events on the power grid bus load at different times (such as day and night during typhoon landing), including load drops caused by damage to power facilities and local load increases caused by the activation of emergency equipment.

[0033] 3. Accurate prediction based on time-weather interaction characteristics can help power grid operators prepare for dispatching in advance. For example, when it is predicted that high temperature and high humidity weather will occur, the power generation capacity can be increased in advance to ensure the stable operation of the power grid. At the same time, during the low load period and when the weather conditions are suitable, maintenance work such as equipment maintenance can be reasonably arranged.

[0034] 4. It helps to formulate flexible electricity price strategies. By considering the impact of time-weather interaction characteristics on load, electricity prices can be increased during peak load periods when meteorological factors lead to tight electricity demand, guiding users to use electricity rationally, thereby achieving supply and demand balance in the power grid and efficient use of resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.

[0036] Figure 1 It is a flow chart of a method for predicting power grid bus load based on subtropical monsoon climate characteristics;

[0037] Figure 2 The schematic diagram is a structural diagram of a power grid bus load forecasting system based on subtropical monsoon climate characteristics. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are 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 are within the scope of protection of the present invention.

[0039] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0040] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0041] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0042] Embodiment 1

[0043] Climate characteristics analysis

[0044] Significant seasonal changes: The subtropical monsoon climate is hot and rainy in summer and mild and rainy in winter. In summer, due to the high temperature, residents and businesses frequently use cooling equipment (such as air conditioners), which will cause a significant increase in the load on the power grid. In winter, although the temperature is relatively mild, heating equipment (such as electric heaters) is still used, but the increase in load is usually less than in summer.

[0045] Concentrated precipitation: Precipitation is concentrated in the summer, and extreme weather such as rainstorms may affect the operating status of power grid equipment. For example, rain may cause the insulation performance of power lines to deteriorate, or even cause failures, which will indirectly affect the power grid bus load. At the same time, during precipitation, the clouds are thicker and the light is reduced, which will affect the output power of solar power generation equipment, which will also affect the supply and demand balance of the power grid bus load.

[0046] Monsoon impact: The intensity of monsoons can affect temperature and humidity. Strong monsoons may bring cooler air, lowering the temperature during the hot summer months, thereby reducing the use time and power of refrigeration equipment and correspondingly reducing the power grid bus load. Based on the above analysis of the characteristics of the subtropical monsoon climate, the following is obtained: Figure 1 A power grid bus load prediction method based on subtropical monsoon climate characteristics is shown, comprising the following steps:

[0047] S1: Collect and pre-process historical data. Further, the historical data includes load data, time data and meteorological data.

[0048] Furthermore, the load data includes collecting historical grid bus load data, including load values ​​in different seasons and under different weather conditions (sunny, rainy, cloudy, etc.). The time resolution of these data can be determined according to actual needs, such as recording every 15 minutes, every 30 minutes or every hour.

[0049] Furthermore, the meteorological data includes collecting meteorological data related to the subtropical monsoon climate, such as temperature, humidity, wind speed, precipitation, sunshine duration, etc. The temporal and spatial resolution of the meteorological data should match the load data to ensure that meteorological conditions and load changes can be accurately associated.

[0050] Furthermore, the preprocessing includes data cleaning (removing outliers and erroneous records) and data standardization (such as standardizing load data and meteorological data so that they have similar numerical ranges to facilitate subsequent model training). For example, voltage and current data that are obviously inconsistent with physical laws are eliminated. At the same time, the data is standardized so that different types of data have the same dimension to facilitate subsequent integration.

[0051] S2: Extract and encode time and weather related features, and associate them with load data.

[0052] Temporal feature encoding

[0053] Periodic encoding: For time-related features, such as hours, days, weeks, months, etc., to capture the daily, weekly, and monthly cycle changes of the load. Due to the obvious periodicity, such as 24 hours a day, 7 days a week, etc., sine and cosine transforms can be used for encoding. For example, for hourly features, you can use and To encode the hour h, this can better capture the periodicity of time, rather than simply encoding the hour as an integer.

[0054] One-hot encoding: For categorical time features such as seasons (spring, summer, autumn, and winter) and days of the week, one-hot encoding can be used. Seasonal features are particularly important for subtropical monsoon climate zones. For example, for seasonal features, if there are four seasons, four binary feature columns can be created, one for each season. When a sample belongs to a certain season, the column value corresponding to that season is 1, and the rest are 0. The season can be encoded as a one-hot vector (such as spring [1,0,0,0], summer [0,1,0,0], autumn [0,0,1,0], winter [0,0,0,1]).

[0055] Climate characteristic coding

[0056] Standardization of numerical weather features: Numerical features such as temperature, humidity, wind speed, and precipitation can be standardized, such as z-score standardization. The formula is Where x is the original eigenvalue, μ is the mean, and σ is the standard deviation. Standardized features can make deep learning models more likely to converge because it maps the eigenvalues ​​to a range with similar scales.

[0057] Furthermore, the temperature characteristics can be divided into different intervals, such as high temperature interval (greater than 30°C), comfortable interval (18-30°C), low temperature interval (less than 18°C), and encoded as classification features. Furthermore, humidity and precipitation can also be classified similarly, for example, precipitation can be divided into no rain, light rain, moderate rain, heavy rain and other categories, and encoded accordingly.

[0058] Weather condition category encoding: For categorical features such as weather conditions (such as sunny, cloudy, rainy, snowy, etc.), one-hot encoding is used. For example, if there are 5 weather conditions, 5 binary feature columns are created to represent different weather conditions.

[0059] Furthermore, fusing the encoded features includes the following steps:

[0060] Simple concatenation: concatenate different encoded feature vectors in order in the dimension. For example, after encoding the time feature, a vector of length m is obtained, after encoding the weather feature, a vector of length n is obtained, and after encoding the economic activity and special event feature, a vector of length p is obtained. Concatenate them together to obtain a new vector of length m+n+p as the fused feature representation, and all the fused new vectors constitute the input feature set. This method is simple and direct, and is suitable for situations where the features are relatively independent.

[0061] S3: extracting time and meteorological features for interactive reinforcement, and creating time-meteorological interactive features. Further, the time-meteorological interactive features include delay interactive features, time period-meteorological interactive features, meteorological change rate-time interactive features, cycle-meteorological interactive features, and cumulative meteorological-time interactive features.

[0062] Interactive reinforcement of temporal and meteorological features refers to considering the interaction between meteorological features and temporal features. For example, during the hot summer period, the load may be positively correlated with the temperature; while during the winter nights, the load may be negatively correlated with the temperature. These complex relationships can be captured by creating interactive features (such as temperature × hour, humidity × day, etc.).

[0063] Delayed interaction features

[0064] Identify delay effects: Analyze how long it takes for the bus load to respond significantly after meteorological factors change. For example, after the temperature rises, due to the thermal inertia of the building, the indoor temperature will not rise immediately, and there will be a delay in turning on refrigeration equipment such as air conditioners, which will cause a delay in the increase of bus load. By analyzing historical data, determine the time range of this delay, such as 1-3 hours.

[0065] Construct delayed interaction features: Create delayed interaction features based on the delay time range. For example, for temperature features, you can create features such as "t-1 hour temperature" and "t-2 hour temperature", and use these delayed temperature features together with the current time features (such as hours, days, etc.) as model inputs. This allows the model to learn the time delay relationship between meteorological factor changes and bus load response.

[0066] Time period-weather interaction characteristics

[0067] Divide the key time periods: Divide the key time periods according to the daily, weekly and seasonal changes of the bus load. In the subtropical monsoon climate zone, the load change patterns are different during the day and night in summer. The daytime may be mainly affected by temperature and sunshine, while the nighttime may be affected by humidity and residents' activity patterns. Divide a day into different time periods, such as daytime (08:00-18:00), evening (18:00-22:00), and night (22:00-08:00).

[0068] Create time period-weather interaction features: For each time period, create interaction features related to meteorological factors. For example, in the daytime period, you can create features such as "average daytime temperature × daytime duration", "daytime sunshine duration × daytime load peak", etc. These features can reflect the comprehensive impact of meteorological factors on bus load in different time periods, which helps the model better understand the synergy of meteorological and time factors in different time periods.

[0069] Multiply the time feature and the meteorological feature. For example, multiply the one-hot encoding vector representing the time period by the temperature feature. If a time period is encoded as [0,1,0] and the temperature is 25°C, the new feature vector obtained after multiplication is [0,25,0].

[0070] Meteorological change rate-time interaction characteristics

[0071] Calculate the meteorological change rate: In addition to the absolute value of meteorological factors, it is also important to calculate the change rate of meteorological factors. For example, for temperature, calculate the temperature change rate every hour or every half hour (such as (current temperature - temperature in the previous hour) / temperature in the previous hour). The change rate of meteorological factors can reflect the dynamic change process of the weather, and the bus load may be more sensitive to the dynamic change of the weather.

[0072] Constructing meteorological change rate-time interaction features: Combine meteorological change rate with time features to construct interaction features. For example, create a "temperature change rate × hour" feature to represent the impact of rapid temperature changes at different time points on bus load. This interaction feature can help the model capture the complex relationship between dynamic changes in meteorological conditions and bus load, especially in the case of sudden weather changes, such as a sudden drop in temperature before a rainstorm or rapid changes in humidity caused by strong winds.

[0073] Cycle-weather interaction characteristics

[0074] Consider the periodic characteristics: Consider the interaction between the periodic characteristics of bus load (daily cycle, weekly cycle, seasonal cycle) and meteorological cycle (such as daily temperature cycle, seasonal cycle of precipitation). In the subtropical monsoon climate, the impact of seasonal cycle on bus load is particularly significant.

[0075] Construct cycle-meteorological interaction features: For example, for seasonal cycles, you can create a "season × season average temperature deviation" feature, where "season average temperature deviation" refers to the difference between the current temperature and the average temperature of the season. Such features allow the model to learn the extent to which meteorological factors deviate from normal levels on bus load in different seasons. For daily cycles, you can create features such as "daily cycle stage (such as peak, trough) × daily temperature fluctuation range" to strengthen the model's understanding of daily meteorological and load change patterns.

[0076] Cumulative weather-time interaction characteristics

[0077] Calculate cumulative meteorological factors: Calculate the cumulative meteorological factors over a period of time, such as cumulative precipitation, cumulative sunshine duration, etc. These cumulative meteorological factors can reflect the potential impact of meteorological conditions on bus load over a long period of time. For example, long-term cumulative precipitation may lead to increased soil moisture, affect grounding resistance, and indirectly affect the operation of power grid equipment and bus load.

[0078] Constructing cumulative meteorological-time interaction features: Combine cumulative meteorological factors with time features to construct interactive features. For example, create the feature of "cumulative precipitation (past week) × average load within the week" to represent the impact of precipitation in the past week on the average load of this week. This interactive feature can help the model consider the relationship between the long-term cumulative effect of meteorological factors and bus load. For example, you can also multiply the cumulative precipitation of the past three days with the current hour, day of the week and other time features to obtain a new feature vector to reflect the different effects of cumulative precipitation on the target variable under different time conditions.

[0079] S4: In order to describe the influencing factors of the power grid bus load from more dimensions and more realistically reflect the environment that affects the bus load changes, after extracting meteorological related features, the meteorological extended features are generated by multiplication or division. This can not only obtain more model training resources, but also describe the impact of various meteorological features on the bus load under the interaction from more dimensions and more realistic scenarios. Furthermore, the new influencing features include body temperature;

[0080] Furthermore, the temperature characteristics can be further subdivided according to the human comfort index, considering the impact of humidity and wind speed on the perceived temperature, and the perceived temperature can be calculated and used as a feature input model.

[0081] Furthermore, the calculation formula of body temperature is as follows:

[0082] AT=T+0.348×RH-0.7×WS+0.7;

[0083] In the above formula, AT is the perceived temperature (degrees Celsius); T is the actual air temperature (degrees Celsius); RH is the relative humidity (percentage); WS is the wind speed (meters / second).

[0084] Example: When the actual air temperature is 28°C, the relative humidity is 90%, and the wind speed is 5 m / s, the perceived temperature is calculated as follows:

[0085] AT=28+0.348×90-0.7×5+0.7≈56.52℃.

[0086] Furthermore, different meteorological features can be multiplied to generate new features. For example, temperature and humidity can be multiplied because the load of refrigeration equipment may be higher in a high temperature and high humidity environment than when temperature or humidity is considered alone. This combined feature can better reflect the comprehensive impact of meteorological factors on bus load.

[0087] Furthermore, the sunshine intensity can be divided by the wind speed. When solar power generation is connected to the bus, this combination can reflect the impact of the effective utilization of solar power generation on the bus load. A high wind speed may reduce the power generation efficiency of solar panels, while the sunshine intensity represents the potential for power generation. Their ratio can be used as a new feature to consider the impact on the bus load and input into the subsequent prediction model to participate in the training and prediction of the bus load.

[0088] S5: Build a machine learning model, and randomly divide the above features into a training set, a validation set, and a test set for training, validating, and testing the machine learning model to obtain a load prediction model for predicting the power grid bus load.

[0089] Furthermore, the machine learning model adopts a decision tree model (Decision Tree) or a random forest model (Random Forest).

[0090] The decision tree model can handle nonlinear relationships well and is applicable when the relationship between the meteorological-time interaction characteristics and the target variable is complex and there are multiple interactions and conditional judgments. For example, in different seasons (time characteristics), the impact of cumulative meteorological factors (such as cumulative precipitation and cumulative wind speed) on the power grid bus load may produce different branching situations due to different seasons, and the decision tree can capture this complex relationship well.

[0091] Furthermore, the data set is divided into training set, validation set and test set. Usually, the training set is used to train the model, accounting for about 70%-80%; the validation set is used to adjust the parameters of the model and evaluate the performance of the model to prevent overfitting, accounting for about 10%-15%; the test set is used to finally evaluate the generalization ability of the model, accounting for about 10%-15%. Random sampling can be used for division, but attention should be paid to maintaining the time order of the data to avoid introducing deviations in the time series.

[0092] Furthermore, a suitable decision tree algorithm is selected, such as the CART (Classification and Regression Tree) algorithm. The main parameters of the decision tree model include the maximum depth of the tree, the minimum number of samples required for node splitting, the minimum number of samples for leaf nodes, etc. In the initial stage, some initial parameters can be set based on experience or domain knowledge. For example, the maximum depth of the tree is set to 5, the minimum number of samples required for node splitting is set to 20, and the minimum number of samples for leaf nodes is set to 10.

[0093] The time-meteorological interaction features in the training set are used as input, and the bus load data is used as the output label, and input into the decision tree model for training. The training process of the decision tree is a process of recursively dividing the feature space. For example, at the root node, the algorithm will select a time-meteorological interaction feature that can best distinguish different values ​​of the bus load as the split feature, such as selecting the interaction feature "the product of the peak hour code and the temperature". If the value of the feature is greater than a certain threshold, the sample is divided into the left subtree; otherwise, it is divided into the right subtree. Then continue to repeat this process on the subtree nodes until the stopping condition is met (such as reaching the maximum depth, the number of node samples is too small, etc.).

[0094] During the training process, record the performance indicators of the model on the validation set, such as root mean square error (RMSE) or mean absolute error (MAE). The calculation formula of RMSE is Where n is the number of samples, y i is the true load value, is the predicted load value; the calculation formula of MAE is

[0095] Adjust the parameters of the decision tree based on the performance indicators of the validation set. If the model is found to be overfitting (such as the error on the validation set starts to rise), you can reduce the depth of the tree, increase the minimum number of samples required for node splitting, or perform pruning. Pruning is an effective way to prevent overfitting. It makes the model more concise and more generalizable by removing some unimportant subtrees or branches.

[0096] You can also try different feature selection methods or interactive feature construction methods to observe the impact on model performance. For example, increase or decrease the size of the time window to construct time-weather interactive features, or select other meteorological features to combine with time features.

[0097] Use the test set to evaluate the trained decision tree model and calculate performance indicators such as RMSE or MAE on the test set to determine the generalization ability of the model. Compare the performance of the models under different parameter settings and select the model with the best performance.

[0098] In addition to error metrics, the model's feature importance can also be analyzed. The decision tree model can output the relative importance of each feature to the forecast result. By looking at the importance ranking of time-meteorological interaction features, we can understand which features are most critical to bus load forecasting, which helps to further optimize feature engineering and models.

[0099] In specific implementation, the random forest model is an integrated learning model built based on multiple decision trees. It improves the accuracy and stability of predictions by integrating the results of multiple decision trees. It performs well in dealing with high-dimensional meteorological-time interaction features and in the presence of noisy data. For example, when there are a large number of different types of meteorological-time interaction features (including cumulative meteorological features, time period features, time trend features, etc.), the random forest can automatically filter out important features and reduce the risk of overfitting.

[0100] S6: The load prediction model is used to predict the load of the power grid bus. Furthermore, the input features used for prediction are the same as the features used during training to ensure the accuracy of the model.

[0101] The beneficial effects are as follows:

[0102] 1. Capture periodic patterns:

[0103] In subtropical monsoon climate areas, seasonal and diurnal changes are obvious. The periodic changes of bus load can be well captured through time characteristics (such as hours, days, seasons, etc.). For example, in summer, due to the high temperature, the use of refrigeration equipment will cause the grid bus load to peak during the day, especially in the afternoon; while in winter, the use of heating equipment may increase the load at night. This seasonal and diurnal load change pattern can be accurately learned by the model through time characteristics.

[0104] There are also load cycles within a week, and the load patterns on weekdays and weekends are usually different. Due to the concentration of industrial production and commercial activities on weekdays, the peak hours and load amounts of the power grid bus load are significantly different from those on weekends. Using the time feature of the day of the week, the model can better predict the load changes under different date types.

[0105] 2. Reflect load change trend:

[0106] As time goes by, socio-economic development and changes in residents' lifestyles will cause the grid bus load to show a certain growth or decline trend. Time series-related features (such as the rate of change of load at adjacent time points) can help the model capture this long-term trend. For example, the upgrading of a region's industrial structure may lead to a gradual increase in industrial electricity consumption, or the growth rate of residential electricity load may slow down with the promotion of energy-saving measures. By considering time characteristics, it is possible to better predict future changes in load under this trend.

[0107] 3. Dealing with meteorological sensitivity:

[0108] The subtropical monsoon climate is characterized by high temperatures and heavy rainfall in summer and mild weather and little rainfall in winter. The temperature has a significant impact on the bus load of the power grid. During the high temperature period in summer, the rising temperature will cause a large amount of refrigeration equipment such as air conditioners to be used, thereby greatly increasing the load of the power grid; in winter, although the temperature is relatively mild, the colder weather will also increase the load of heating equipment. By introducing the meteorological feature of temperature, it is possible to directly link meteorological conditions with changes in bus load.

[0109] Humidity is also important in subtropical monsoon climates. High humidity makes people feel hotter, which further increases the frequency and duration of use of refrigeration equipment, indirectly affecting the power grid bus load. At the same time, humidity may also affect the performance of electrical equipment, which in turn has a potential impact on power grid operation. Using humidity as a feature can take these factors into account.

[0110] Precipitation is also a typical meteorological element of the subtropical monsoon climate. Heavy rain may cause some outdoor activities to stop and reduce the power consumption of commercial and public facilities, but it may also increase some loads due to the operation of drainage facilities, etc. Taking precipitation as a feature can help the model consider these complex effects.

[0111] 4. Adapt to weather variability:

[0112] The meteorological conditions of the subtropical monsoon climate are changeable, and the temperature and precipitation change frequently when the seasons change. By considering multiple meteorological characteristics and their combinations (such as the interaction between temperature and humidity), we can better adapt to this changeable meteorological environment. For example, at the turn of spring and summer when the temperature gradually rises and the humidity is high, the model can more accurately predict the change of load based on these meteorological-temporal interaction characteristics, because the use of refrigeration equipment by residents and enterprises under such weather conditions will be affected by both temperature and humidity.

[0113] 5. Accurately predict complex changes:

[0114] The time-weather interaction feature can capture the complex impact of meteorological factors on the power grid bus load in different time periods. For example, in the afternoon of summer, when high temperature and high humidity occur at the same time, the power grid bus load will have a specific change pattern compared with other combinations of time and meteorological conditions. This interactive feature allows the model to learn this complex conditional dependency, thereby more accurately predicting the bus load.

[0115] Consider the impact of special meteorological events at different times. In the subtropical monsoon climate, extreme weather events such as typhoons and rainstorms occur from time to time. Through the time-meteorological interaction characteristics, the model can learn the different impacts of these extreme weather events on the power grid bus load at different times (such as day and night during typhoon landing), including load drops caused by damage to power facilities and local load increases caused by the activation of emergency equipment.

[0116] 6. Optimize grid operation strategy:

[0117] Accurate predictions based on time-weather interaction characteristics can help power grid operators prepare for dispatching in advance. For example, when high temperature and high humidity weather are predicted, power generation capacity can be increased in advance to ensure stable operation of the power grid. At the same time, during low load periods and when weather conditions are suitable, maintenance work such as equipment overhaul can be reasonably arranged.

[0118] It helps to formulate flexible electricity price strategies. By considering the impact of time-weather interaction characteristics on load, electricity prices can be increased during peak load periods when meteorological factors lead to tight electricity demand, guiding users to use electricity rationally, thereby achieving supply and demand balance in the power grid and effective utilization of resources.

[0119] This application solution can solve the following problems:

[0120] Solve the problem of incomplete consideration of meteorological factors:

[0121] Existing technologies may simply consider the impact of a single meteorological factor (such as only considering the temperature) on the power grid bus load. In subtropical monsoon climate areas, meteorological factors are complex and diverse. Not only temperature, but also humidity, precipitation, wind speed and other factors will affect the load, and there are interactions between these factors. For example, in hot and humid weather, the duration and power of users using air conditioners may be higher than in hot weather alone. By adopting the time-meteorological interaction feature, the synergy of these factors can be comprehensively considered to more accurately predict the load. Traditional methods may not fully consider the impact of dynamic changes in meteorological factors on the load. The subtropical monsoon climate is characterized by changeable meteorological conditions. For example, thunderstorms in summer may cause a sudden drop in temperature and humidity in a short period of time, thereby affecting the load. The time-meteorological interaction feature can capture this dynamic change process. For example, by analyzing the interactive relationship between meteorological changes (such as the temperature drop and precipitation changes) and time (such as afternoon to evening) in the past few hours, the load changes caused by meteorological mutations can be better predicted.

[0122] Solve the problem of insufficient utilization of time cycle rules:

[0123] Existing technologies may ignore the complex variation of loads in different time periods (such as seasons, day and night, weekdays / weekends). In subtropical monsoon climate regions, seasonal changes are obvious, and the load characteristics of different seasons vary greatly. For example, existing methods may not be able to distinguish the load differences between summer and winter due to different climates and people's living habits. By introducing time-meteorological interaction features, such as combining seasonal characteristics with meteorological factors (such as high temperatures in summer and low temperatures in winter), these time period laws can be better used to predict loads. For the law of day and night load changes, traditional forecasts may not fully consider the different effects of meteorological factors on loads at different times of the day. For example, in the early morning, lower temperatures and humidity may result in lower electricity demand for residents and businesses, but as the sun rises, the temperature and humidity rise, and the load begins to change. The use of time-meteorological interaction features can describe in detail the relationship between meteorological conditions and loads at different time stages, thereby improving prediction accuracy.

[0124] Addressing the lack of capacity to cope with extreme weather:

[0125] In subtropical monsoon climate areas, extreme weather events such as typhoons, heavy rains, and high temperature heat waves are frequent. Existing technologies may not fully consider the comprehensive impact of these extreme weather events on the power grid when they occur at different times when predicting load. For example, when a typhoon strikes, the load on the power grid may be reduced due to the suspension of production of enterprises and the avoidance of residents, but at the same time, new load demands will be generated due to the activation of disaster relief equipment. The time-meteorological interaction feature can be combined with the time of extreme weather (such as whether the typhoon lands during the day or at night) to more accurately predict these complex load changes and help the power grid prepare for response in advance. In the recovery phase after extreme weather, meteorological conditions and load recovery processes are also closely related to time. For example, after a heavy rain, as the weather clears and time passes, residents' lives and corporate production gradually return to normal, and the load will gradually pick up. The time-meteorological interaction feature can better predict the load changes in this recovery process, which helps to reasonably arrange the repair and dispatching of the power grid.

[0126] Solve the problem of poor flexibility in grid dispatching:

[0127] Previous technologies may have led to a lack of flexibility in grid dispatching due to inaccurate load forecasting. The use of time-weather interaction features can more accurately predict load changes under different times and meteorological conditions, and the grid dispatching department can adjust the power generation plan in advance based on these predictions. For example, when it is predicted that high temperature weather is coming and the load peak will occur in the afternoon, thermal power generation or coordinated renewable energy generation (such as increasing the access of solar power generation at the right time) can be arranged in advance to meet load demand and improve the flexibility and stability of grid operation. It is also possible to reasonably arrange the inspection and maintenance of grid equipment according to the load valley period predicted by the time-weather interaction characteristics. For example, at night when the weather conditions are stable and the load is low, the inspection and maintenance of equipment such as transformers can be arranged to reduce the impact on user power supply while also ensuring the safety of maintenance personnel.

[0128] Solve the problem of unreasonable resource allocation:

[0129] Existing load forecasting technology may not take into account the combined impact of meteorological and time factors, resulting in unreasonable allocation of power generation resources. In subtropical monsoon climates, load demands vary in different seasons and weather conditions. For example, during the peak of electricity consumption in summer, if the load growth under high temperature and high humidity weather is not fully considered, insufficient power generation capacity may occur. By predicting the load through time-meteorological interaction characteristics, power generation resources can be reasonably allocated according to the load demand under different time-meteorological conditions, such as increasing the installed capacity of power generation or the investment in energy storage facilities during the peak period in summer. For the utilization of distributed energy resources (such as solar energy and wind energy), traditional methods may not combine meteorological and time factors well. In subtropical monsoon climate areas, solar power generation is affected by sunshine time and weather (such as cloud cover). The time-meteorological interaction characteristics can more accurately predict the available amount of solar power generation under different time and meteorological conditions, thereby optimizing its matching with the grid bus load, improving the absorption rate of distributed energy, and avoiding resource waste.

[0130] Embodiment 2

[0131] like Figure 2 As shown, a power grid bus load prediction system based on subtropical monsoon climate characteristics is applied to the power grid bus load prediction method based on subtropical monsoon climate characteristics described above, comprising:

[0132] Collection module, which is used to collect and preprocess historical data;

[0133] An encoding module is used to extract and encode time and weather-related features and associate them with load data;

[0134] An interactive module, which is used to extract time and meteorological features for interactive enhancement and create time-meteorological interactive features;

[0135] An expansion module, which is used to extract meteorological features and generate meteorological expansion features;

[0136] A training module is used to build a machine learning model, and at the same time, the above-mentioned features are randomly divided into a training set, a validation set and a test set, so as to train, validate and test the machine learning model to obtain a load prediction model for the power grid bus;

[0137] The prediction module is used to use the load prediction model to predict the load of the power grid bus.

[0138] Embodiment 3

[0139] A computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the above-mentioned method for predicting the power grid bus load with subtropical monsoon climate characteristics.

[0140] Embodiment 4

[0141] A processor is used to run a program, wherein the program executes the above-mentioned power grid bus load prediction method with subtropical monsoon climate characteristics when running.

[0142] The present application provides a method for predicting the load of a power grid bus based on the characteristics of a subtropical monsoon climate, including: S1: collecting and preprocessing historical data; S2: extracting and encoding time- and meteorological-related features, and correspondingly associating them with load data; S3: extracting time and meteorological features for interactive reinforcement, and creating time-meteorological interactive features; S4: generating meteorological extension features after extracting meteorological features; S5: constructing a machine learning model, and at the same time randomly dividing the above-mentioned features into a training set, a validation set, and a test set for training, validating, and testing the machine learning model to obtain a load prediction model for the power grid bus; S6: using the load prediction model to predict the load of the power grid bus. The time-meteorological interactive feature can capture the complex effects of meteorological factors on the load of the power grid bus in different time periods. For example, in the afternoon of summer, when high temperature and high humidity occur at the same time, the load of the power grid bus will have a specific change pattern compared with other combinations of time and meteorological conditions. This interactive feature allows the model to learn this complex conditional dependency, thereby more accurately predicting the bus load.

[0143] Those of ordinary skill in the art will appreciate that the units of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition of each example has been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0144] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.

[0145] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0146] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-0nlyMemory), random access memory (RAM, RandomAccessMemory), mobile hard disk, magnetic disk or optical disk, etc., which can store program code.

[0147] 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. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.

Claims

1. A method for predicting power grid bus load based on subtropical monsoon climate characteristics, characterized in that: include: S1: Collect and preprocess historical data; S2: extract and encode time-related and meteorological features, and associate them with load data; S3: extracting time and meteorological features for interactive reinforcement and creating time-meteorological interactive features; S4: extract meteorological features and generate meteorological extended features; S5: Build a machine learning model, and randomly divide the above features into a training set, a validation set, and a test set for training, validating, and testing the machine learning model to obtain a load prediction model for the power grid bus; S6: Use the load forecasting model to forecast the load on the power grid bus.

2. The method for predicting power grid bus load based on subtropical monsoon climate characteristics according to claim 1, characterized in that: The historical data includes load data, time data and meteorological data.

3. The method for predicting power grid bus load based on subtropical monsoon climate characteristics according to claim 1, characterized in that: The preprocessing includes data cleaning and data standardization.

4. The method for predicting power grid bus load based on subtropical monsoon climate characteristics according to claim 1, characterized in that: The time-weather interaction characteristics include delay interaction characteristics, time period-weather interaction characteristics, weather change rate-time interaction characteristics, cycle-weather interaction characteristics and cumulative weather-time interaction characteristics.

5. The method for predicting power grid bus load based on subtropical monsoon climate characteristics according to claim 1, characterized in that: When the meteorological extended features are generated after the meteorological features are extracted, the meteorological extended features are generated by multiplication or division.

6. The method for predicting power grid bus load based on subtropical monsoon climate characteristics according to claim 1, characterized in that: The meteorological extended features include the perceived temperature, the multiplication feature of air temperature and humidity, and the sunshine intensity divided by wind speed.

7. The method for predicting power grid bus load based on subtropical monsoon climate characteristics according to claim 6, characterized in that: The calculation formula of the body temperature is as follows: AT=T+0.348×RH-0.7×WS+0.7; In the above formula, AT is the perceived temperature; T is the actual air temperature; RH is the relative humidity; WS is the wind speed.

8. A power grid bus load forecasting system based on subtropical monsoon climate characteristics, characterized in that: The power grid bus load prediction method based on subtropical monsoon climate characteristics as described in any one of claims 1 to 7 comprises: Collection module, which is used to collect and preprocess historical data; An encoding module is used to extract and encode time and weather-related features and associate them with load data; An interactive module, which is used to extract time and meteorological features for interactive enhancement and create time-meteorological interactive features; An expansion module, which is used to extract meteorological features and generate meteorological expansion features; A training module is used to build a machine learning model, and at the same time, the above-mentioned features are randomly divided into a training set, a validation set and a test set, so as to train, validate and test the machine learning model to obtain a load prediction model for the power grid bus; The prediction module is used to use the load prediction model to predict the load of the power grid bus.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the power grid bus load prediction method with subtropical monsoon climate characteristics as described in any one of claims 1 to 7.

10. A processor, characterized in that: The processor is used to run a program, wherein the program, when running, executes the power grid bus load prediction method with subtropical monsoon climate characteristics as described in any one of claims 1 to 7.

Images