Intelligent power distribution network load regulation and control method based on data collaboration

Abstract

The invention provides a power distribution network load intelligent regulation and control method based on data collaboration. The method comprises the following steps: calculating a quality aging score; identifying the historical data duration of the current power distribution network, and if the power distribution network is in a source domain, calculating a load prediction value of a next time period according to the load related characteristics and historical load information; if the power distribution network is a target domain, matching a source domain, extracting historical load information of the source domain, and calculating a load prediction value of the target domain in a next time period; and dynamically regulating and controlling the load of the power distribution network according to the load predicted value. By adopting the method, it can be ensured that the subsequent load prediction value can be calculated based on the high-precision and high-timeliness load related characteristics. And the target domain can borrow historical load information of the source domain to realize a cross-power-distribution-network data collaboration effect.

Description

Technical Field

[0001] This invention relates to distribution network load, specifically to a data-driven intelligent control method for distribution network load. Background Technology

[0002] In the intelligent operation of distribution networks, mature load regulation strategies have been widely applied to distribution network automation platforms, such as adjusting the operation of energy storage devices and charging piles. These strategies themselves have the ability to effectively and dynamically adjust the load.

[0003] However, a key prerequisite for achieving precise regulation is obtaining highly accurate load forecasts for future periods. The current core problem lies in the insufficient accuracy of load forecasts, primarily due to two key factors. First, the quality and timeliness of the data sources supporting the forecasts are problematic. Data collected by load monitoring equipment often suffers from reduced completeness, accuracy, or timeliness due to transmission channel noise, equipment failure, or communication delays. Second, the historical operating time varies significantly across different distribution networks. Mature lines possess sufficient historical data to support complex, high-precision calculation models, but newly added or upgraded lines lack effective historical data, forcing the use of simple algorithms with limited accuracy, such as proportionally adjusting the current load value based on time to obtain the load forecast. When data is scarce or lacks historical support, even highly optimized regulation algorithms can easily lead to deviations from actual needs, resulting in localized overloads or resource waste. Summary of the Invention

[0004] To overcome existing technical problems, this invention provides a data-coordinated intelligent load control method across distribution networks.

[0005] The present invention adopts the following technical solution.

[0006] A data-driven intelligent load control method for distribution networks includes the following steps: By collecting load-related characteristics of the current power distribution network through multiple load monitoring devices, the transmission quality coefficient of each load monitoring device is calculated. According to the transmission quality coefficient Calculate the quality timeliness score based on data update time. If the quality timeliness score If the data falls below the preset timeliness threshold, the corresponding load monitoring equipment will re-collect data. The calculated quality timeliness score will be determined based on the re-collected data. If the timeliness threshold is lower than the preset timeliness threshold, alternative data is obtained through data substitution. The system identifies the duration of historical data for the current distribution network. If this duration exceeds a preset reference duration, the distribution network is considered a source region. Based on load-related characteristics and historical load information, the load forecast for the next time period is calculated. And upload load-related characteristics and historical load information to the cloud database; If the load is less than the preset reference duration, the distribution network is considered the target domain. Load-related characteristics are uploaded to the cloud database to match the source domain, historical load information of the source domain is extracted, and the load forecast value for the next time period of the target domain is calculated. ; Based on load forecast values Dynamically regulate the load of the distribution network.

[0007] As a further improvement to the present invention, the transmission quality coefficient of each load monitoring device is calculated. The specific steps include: , in, It is the fault sensitivity factor of the i-th load monitoring device. Is the i-th load monitoring device in the most recent time? Total number of failures occurring within the period It is the packet loss penalty coefficient for the i-th load monitoring device. It is the most recent time The number of packets lost within, It is the most recent time The number of packets sent within, and It is a transmission quality weighting factor. It is a packet loss penalty amplification factor.

[0008] As a further improvement of the present invention, based on the transmission quality coefficient Calculate the quality timeliness score based on data update time. The specific steps include: , in, It is the timeliness impact factor of the i-th load monitoring device. It is the current time point. This is the time of the most recent data update. It is a timeliness adjustment factor. It is a dynamic quality weighting coefficient. It is the base weight value of the dynamic quality weight coefficient. It is a transmission quality reward / penalty adjustment factor. It is the basic value for transmission quality rewards and penalties; The specific steps of the data substitution method include: calculating the quality timeliness score of the nearest similar load monitoring device to the load monitoring device. If the time limit is higher than the preset time threshold, the data collected by the same type of load monitoring equipment will be extracted.

[0009] As a further improvement of the present invention, the distribution network is equipped with a historical database storing historical load information and historical temperature information, and the load-related features include real-time load values. Real-time temperature value and weather forecast temperature values ; Calculate the load forecast for the next period based on load-related characteristics and historical load information. The specific steps include: ， ， ， ， ， in, It is a dynamic weighting coefficient for temperature changes. It is the deviation between historical predicted temperature values ​​and historical actual temperature values. It refers to the number of time periods for which historical temperature data was collected. This is the maximum allowable predicted temperature error value. It is the minimum temperature load change weighting factor. It is the temperature load variation coefficient, used to convert temperature changes into load changes. It predicts changes in temperature values. The next forecast period +1 is the weather forecast temperature value. The current time period The real-time temperature value, It is the sign correction factor. It is a data quality and security amplification factor. It is the transmission quality factor of the load monitoring equipment that collects temperature data. It is the smallest quality and safety amplification factor. It is a dynamic weighting coefficient based on historical load data trust and temperature changes. It is a load reference value for the same historical period. It is the total number of weeks collected. It is the time decay weighting coefficient. It was before Historical load changes for the same time period of the week compared to the previous period. It is the time decay factor. It is a dynamic base threshold. It is the load temperature influencing factor. It is the first The temperature difference between the same time period of the week and the current time period. It is an optimized control deviation compensation item, calculated based on historical load re-prediction values ​​and historical actual load values.

[0010] As a further improvement to the present invention, the temperature load variation coefficient The calculation expression is as follows: , in, This refers to the number of days containing temperature load variation data of the same type within the same temperature range. The data type for temperature load variation includes weekdays and rest days. It is the current time period on day i. The actual load value, It is the time period preceding the current time period on day i. The actual load value, Current time period on day i Historical actual temperature values, The previous time period on day i The historical actual temperature value.

[0011] As a further improvement to the present invention, the control deviation compensation term is optimized. The calculation expression is as follows: ， , in, It is the number of recent historical load re-prediction values ​​and historical actual load values ​​collected. It is the most recent A historical load true value, It is the most recent Historical load re-prediction value Used to reserve margin settings; The calculation steps for historical load reprediction are as follows: based on the real-time temperature values ​​for the current period... ,renew and The value, based on the updated and The value, and the previous Calculation of load-related characteristics and historical load information for each time period The load forecast value for the +1 time period is obtained. Historical load re-prediction values.

[0012] As a further improvement of the present invention, the load-related features include the shape of the daily load curve and the main load types; The specific steps for matching the source domain include: the cloud database divides the daily load curves of the source and target domains into multiple time periods, calculates the average historical actual load value for each time period, calculates the peak-to-valley difference, morning peak percentage, evening peak percentage, and load factor based on the average historical actual load value for each time period, selects the corresponding source domain as a candidate reference object based on the main load type of the target domain, and calculates the daily load characteristic difference between the target domain and each candidate reference object. , , in, It represents the proportion of the early peak in the target domain. It represents the early peak percentage of the source domain. It represents the proportion of late-peak hours in the target domain. It is the proportion of the late peak in the source domain. It is the load rate of the target domain. It is the load rate of the source domain. and It is a weighting adjustment factor; Matching daily load characteristic gap The smallest source domain.

[0013] As a further improvement of the present invention, the historical load information includes the value of the temperature load change coefficient, the value of the optimization control deviation compensation term coefficient, the historical load change value, and time data. Calculate the load forecast for the target domain in the next time period. The specific expression is as follows: ， ， ， ， ， ， ， in ， It is a dynamic weighting coefficient for temperature changes. This represents the deviation between historical predicted temperature values ​​and historical actual temperature values. If no historical predicted or actual temperature values ​​exist, this value will be 0. This represents the number of time periods for which historical temperature data was collected. If no historical temperature data exists, this value is 1. This is the maximum allowable predicted temperature error value. It is the minimum temperature load change weighting factor. It is the temperature load variation coefficient of the source domain, used to convert temperature changes into load changes. It predicts changes in temperature values. The next forecast period +1 is the weather forecast temperature value. The current time period The real-time temperature value, It is the sign correction factor. It is a data quality and security amplification factor. It is the transmission quality factor of the load monitoring equipment that collects temperature data. It is the smallest quality and safety amplification factor. It is a dynamic weighting coefficient based on historical load data trust and temperature changes. It is a reference value for mixed loads from the source and target domains during the same historical period. This refers to the total number of weeks the target domain was collected. It is the time decay weighting coefficient. Is the target domain before Historical load changes for the same time period of the week compared to the previous period. It is the source domain trust weight coefficient. The daily load characteristic difference conversion factor, This is the total number of weeks of data collection from the source domain. It is the time decay weighting coefficient. It is the source domain before Historical load changes for the same time period of the week compared to the previous period. It is the time decay factor. It is a dynamic base threshold. It is the load temperature influencing factor. It is the first The temperature difference between the same time period of the week and the current time period, if there is no first time period. If the temperature value is the same during the same period of the week, then this item is 1. It is an optimized control deviation compensation item, calculated based on historical load forecast values ​​and historical actual load values.

[0014] The beneficial effects of this invention are as follows: when the quality and timeliness score fails to meet the standard, the corresponding load monitoring equipment is triggered to re-collect data, ensuring that subsequent load forecasts can be calculated based on high-precision and high-timeliness load-related characteristics. Based on the duration of historical data from the distribution network, it can intelligently divide the data into source domains with sufficient historical data and target domains with insufficient data, and implement targeted forecasting strategies. For the target domain, it can compensate for its own historical data deficiencies by borrowing historical load information from similar source domains, achieving cross-distribution network data collaboration. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating the present invention. Detailed Implementation

[0017] The accompanying drawings are for illustrative purposes only and should not be construed as limiting the scope of this patent. To better illustrate this embodiment, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual dimensions of the product.

[0018] It will be understood by those skilled in the art that certain well-known structures and their descriptions may be omitted in the accompanying drawings. The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0019] A data-driven intelligent load control method for distribution networks includes the following steps: By collecting load-related characteristics of the current power distribution network through multiple load monitoring devices, the transmission quality coefficient of each load monitoring device is calculated. According to the transmission quality coefficient Calculate the quality timeliness score based on data update time. If the quality timeliness score If the data falls below the preset timeliness threshold, the corresponding load monitoring equipment will re-collect data. The calculated quality timeliness score will be determined based on the re-collected data. If the timeliness threshold is lower than the preset timeliness threshold, alternative data is obtained through data substitution. When the quality aging score When the load does not meet the standard, the corresponding load monitoring equipment is triggered to re-collect data, which can ensure that the subsequent load forecast value can be calculated based on the load-related characteristics with high accuracy and timeliness.

[0020] As a further improvement to the present invention, the transmission quality coefficient of each load monitoring device is calculated. The specific steps include: , in, It is the fault sensitivity factor of the i-th load monitoring device. Is the i-th load monitoring device in the most recent time? Total number of failures occurring within the period It is the packet loss penalty coefficient for the i-th load monitoring device. It is the most recent time The number of packets lost within, It is the most recent time The number of packets sent within, and It is a transmission quality weighting factor. It is a packet loss penalty amplification factor.

[0021] As one embodiment of the present invention It is 0.7. It is 0.3. Assuming the load monitoring equipment is a temperature sensor used to detect real-time temperature values, and since there are generally multiple temperature sensors in the same distribution network, its fault sensitivity factor is set to 6. Therefore, if a fault occurs once a day, then... Assuming a packet loss rate of 3%, or 0.03, If we set it to 3, then The value of the load monitoring device is 0.813, approximately 0.81. If the load monitoring device is a load sensor used to detect the current real-time load value, since the real-time load value is the core feature for calculating the load forecast, the fault sensitivity factor is set between 6 and 12, depending on the number of load sensors installed in the same distribution network. If there are many load sensors, the fault sensitivity factor can be set to 12. When a fault occurs, the subsequent calculated quality timeliness score will be affected. Typically, the data will be below the preset timeliness threshold, requiring the load sensor to retest. If the load sensor still fails to exceed the timeliness threshold after updating its data, its data will not be used. Instead, a data substitution method will be employed to obtain replacement data. This process continues until the load sensor no longer malfunctions during a continuous day of data acquisition, or its packet loss rate decreases, gradually improving its transmission quality coefficient. Only the data collected by the load detection sensor is incorporated into the subsequent load forecast calculation formula.

[0022] As a further improvement of the present invention, based on the transmission quality coefficient Calculate the quality timeliness score based on data update time. The specific steps include: , in, It is the timeliness impact factor of the i-th load monitoring device. It is the current time point. This is the time of the most recent data update. It is a timeliness adjustment factor. It is a dynamic quality weighting coefficient. It is the base weight value of the dynamic quality weight coefficient. It is a transmission quality reward / penalty adjustment factor. It is the basic value for transmission quality rewards and penalties; The specific steps of the data substitution method include: calculating the quality timeliness score of the nearest similar load monitoring device to the load monitoring device. If the time limit is higher than the preset time threshold, the data collected by the same type of load monitoring equipment will be extracted.

[0023] The data substitution method is used when multiple load monitoring devices for collecting the same data are typically installed in the same distribution network. If one of these devices experiences at least one failure or a high packet loss rate within a day, the quality and timeliness score calculated after re-collecting the data may be affected. If the data is still below the timeliness threshold, the data collected by the load monitoring equipment is not trusted, and another quality timeliness score is selected. Data exceeding the timeliness threshold is used to replace the erroneous data, thus ensuring that zero erroneous data participates in subsequent load forecast calculations and improving the accuracy of load forecast calculations. It is important to note that, due to the load forecast... The calculation is performed independently for each forecast period; therefore, the data collected within that period is used to input the load forecast value. Calculations almost always involve averaging. For example, load forecasts. It is calculated on an hourly basis, while the data collected by the load monitoring equipment is further segmented into different time periods according to different data types, which will be referred to as hourly segments, such as real-time load values. The average temperature value is calculated by averaging over two-minute intervals, which is considered as one hour. The average value is calculated by dividing the time interval into five-minute intervals, which is used to calculate the temperature forecast. The most recently collected data is used directly without averaging. Therefore, if erroneous data is included in the load forecast, the error could occur. If the calculation is flawed, all data for that period will be contaminated with erroneous data. Meanwhile, the quality and timeliness score... The purpose of introducing a data update time point is to avoid the data collected by the load monitoring device in the past few minutes being used in the averaging of two-hour periods. After the load monitoring device that has not experienced any failures and has a low packet loss rate re-collects data, the averaging of the current hour period is performed again.

[0024] As an embodiment of the present invention, to facilitate subsequent calculations, the preset time threshold is set to 1. To standardize the correlation between the various parameter formulas and the range of numerical definitions, the present invention specifies the transmission quality coefficient for normally functioning load monitoring equipment. It should be between 0.8 and 1, therefore Set it to 0.8. The values ​​are determined based on the update time requirements for each type of data, such as the weather forecast temperature value. Due to the meteorological forecast temperature value Its update frequency is relatively slow, the predicted temperature values ​​have small deviations, and it collects meteorological forecast temperature values. Load monitoring equipment is usually only available in one distribution network, therefore this timeliness factor... It can be set to 0.4, the current time point. and the time of the most recent data update The unit is uniformly set to minutes. It is 0.5. If it is not updated within 10 minutes, its value will be 0.5. The value is approximately 0.13. When the quality aging score... If the temperature is below the preset time threshold, re-collect the meteorological forecast temperature value. , The minimum is 1, its The value is approximately 0.81. For 1.1, Set to 1, assuming the transmission quality coefficient of this load monitoring equipment is... It is 0.5. After re-collecting data, the quality and timeliness score The equation 0.335 + 0.81 = 1.145 > 1, so this data can be used. It should be noted again that in the transmission quality coefficient... In the case of a lower value, its quality timeliness score The value is still greater than 1, primarily because the collected meteorological forecast temperature values ​​are... There is usually only one load monitoring device in the same distribution network. If data from other similar load monitoring devices is used as a substitute, the error will be significant. Regarding real-time temperature values... and real-time load value Real-time load value Timeliness Influence Factor Greater than the real-time temperature value Timeliness Influence Factor Real-time temperature value Timeliness Influence Factor Temperature greater than the weather forecast value Timeliness Influence Factor .

[0025] The system identifies the duration of historical data for the current distribution network. If this duration exceeds a preset reference duration, the distribution network is considered a source region. Based on load-related characteristics and historical load information, the load forecast for the next time period is calculated. And upload load-related characteristics and historical load information to the cloud database; If the load is less than the preset reference duration, the distribution network is considered the target domain. Load-related characteristics are uploaded to the cloud database to match the source domain, historical load information of the source domain is extracted, and the load forecast value for the next time period of the target domain is calculated. ; Based on load forecast values Dynamically regulate the load of the distribution network.

[0026] Based on the historical data duration of the distribution network, the system can intelligently divide the data into source domains with abundant historical data and target domains with scarce data, and implement targeted prediction strategies. For the target domain, it can compensate for its own lack of historical data by borrowing historical load information from similar source domains, achieving cross-distribution network data synergy. Specifically, the preset reference duration is at least one month, depending on subsequent temperature and load variation coefficients. The calculation requires reference to the number of days of temperature load change data of the same type within the same temperature range. Therefore, it is recommended to preset the reference period to 1 year.

[0027] Dynamic control measures are quite common, so this invention will not elaborate further. Please refer to "GBT+15148-2024 Technical Specification for Power Load Management System".

[0028] As a further improvement of the present invention, the distribution network is equipped with a historical database storing historical load information and historical temperature information, and the load-related features include real-time load values. Real-time temperature value and weather forecast temperature values ; Calculate the load forecast for the next period based on load-related characteristics and historical load information. The specific steps include: ， ， ， ， ， in, It is a dynamic weighting coefficient for temperature changes. It is the deviation between historical predicted temperature values ​​and historical actual temperature values. It refers to the number of time periods for which historical temperature data was collected. This is the maximum allowable predicted temperature error value. It is the minimum temperature load change weighting factor. It is the temperature load variation coefficient, used to convert temperature changes into load changes. It predicts changes in temperature values. The next forecast period +1 is the weather forecast temperature value. The current time period The real-time temperature value, It is the sign correction factor. It is a data quality and security amplification factor. It is the transmission quality factor of the load monitoring equipment that collects temperature data. It is the smallest quality and safety amplification factor. It is a dynamic weighting coefficient based on historical load data trust and temperature changes. It is a load reference value for the same historical period. It is the total number of weeks collected. It is the time decay weighting coefficient. It was before Historical load changes for the same time period of the week compared to the previous period. It is the time decay factor. It is a dynamic base threshold. It is the load temperature influencing factor. It is the first The temperature difference between the same time period of the week and the current time period. It is an optimized control deviation compensation item, calculated based on historical load re-prediction values ​​and historical actual load values.

[0029] The purpose of this is not to correct for changes in predicted temperature values, but rather to increase the predicted load values. The load margin, when the transmission quality coefficient The closer the value is to 1, the smaller the increase in load margin; conversely, the smaller the value is, the higher the transmission quality factor. The closer the load margin is to 0, the greater the increase in load margin. However, if the increase in load margin is too large, the calculated load forecast value will be completely lost. The significance is that, under any circumstances, the load forecast value All of these values ​​are far greater than the actual required load values, therefore a minimum quality safety amplification factor is set. This can prevent the load margin from increasing indefinitely. Specifically, The value ranges from 0.8 to 0.9.

[0030] For dynamic weighting coefficients of temperature change By calculating the deviation between historical predicted temperature values ​​and historical actual temperature values, the weight of temperature influence can be dynamically adjusted. When historical predicted temperature values ​​generally deviate significantly from historical actual temperature values, the impact of temperature influence can be reduced. The value of this item should be kept in order to avoid affecting the load forecast value. For accurate calculation, of course, to avoid the weight of temperature influence being too small, or even negative, a minimum temperature load change weighting factor is set. The specific value can be between 0.05 and 0.2.

[0031] Temperature load variation coefficient Because each distribution network operates in a different environment with numerous influencing factors, calculations based on historical load and temperature information are the most accurate. The impact of temperature on load can be broadly categorized into two types: first, the effect of temperature on human activity, such as the load generated by air conditioning and heating; and second, the effect of temperature on resistance, as the resistivity of metals increases with temperature, leading to increased power loss at high temperatures. The values ​​are calculated based on the first type of effect of temperature on humans and the second type of effect of temperature on electrical resistance. Therefore, the subsequent... The calculation, The data used is the number of days with the same temperature range and the same type of temperature load change data, which enhances the load impact caused by human factors.

[0032] The calculation logic of the formula lies in extracting the load value change characteristics of adjacent time periods. For example, from 5 pm to 8 pm on a weekday, the actual load value will continuously rise. However, since temperature will have some influence on the actual load value, it is necessary to... The index dynamically captures the changing characteristics of load values, increasing the weight of historical load changes on the most recent date. To reduce the impact of personnel movement on actual load values ​​on different dates within a week, calculations are performed on a weekly basis. It is important to note that... The values ​​are adopted The formula is used for dynamic adjustment. When the deviation between historical predicted temperature values ​​and historical actual temperature values ​​is generally small... The value increases, thus decreasing. The value. Additionally, when the temperature changes significantly, it also decreases substantially. The value of the load is adjusted, thereby reducing the weight of historical load reference values ​​for the same period. The logic is that when there is a short-term, rapid rise or fall in temperature, such as a cold wave, the reference significance of historical load reference values ​​for the same period is significantly reduced. Therefore, in this formula, in the event of a short-term, rapid rise or fall in temperature, the weight of the load reference value for the same period is reduced. The value will increase significantly, thus lowering the [value]. The weight values ​​make the load forecast value The numerical values ​​take into account multiple factors to obtain predictions that are close to reality.

[0033] Specifically, It can be set to 1 / 5. Set to 1, and it can be adjusted and optimized according to specific actual scenarios.

[0034] As a further improvement to the present invention, the temperature load variation coefficient The calculation expression is as follows: ， in, This refers to the number of days containing temperature load variation data of the same type within the same temperature range. The data type for temperature load variation includes weekdays and rest days. It is the current time period on day i. The actual load value, It is the time period preceding the current time period on day i. The actual load value, Current time period on day i Historical actual temperature values, The previous time period on day i The historical actual temperature value.

[0035] Specifically, the same temperature range refers to the range where people commonly use air conditioning for cooling or electric heating for heating, while the rest is the range where air conditioning or electric heating is generally not used. This allows for the precise capture of the temperature load variation coefficient during hot and cold weather. Numerical changes, for example, when the temperature drops by one degree when the electric heater reaches the heating range, its actual load value increases significantly. Conversely, when the temperature rises by one degree when the air conditioner reaches the cooling range, its actual load value increases significantly. The number of days for this value can span a year. After the calculation is completed, it can be fixed, or it can be set to be updated regularly, such as once a day or even once a week, taking into account factors such as material aging.

[0036] As a further improvement to the present invention, the control deviation compensation term is optimized. The calculation expression is as follows: ， , in, It is the number of recent historical load re-prediction values ​​and historical actual load values ​​collected. It is the most recent A historical load true value, It is the most recent Historical load re-prediction value Used to reserve margin settings; The calculation steps for historical load reprediction are as follows: based on the real-time temperature values ​​for the current period... ,renew and The value, based on the updated and The value, and the previous Calculation of load-related characteristics and historical load information for each time period The load forecast value for the +1 time period is obtained. Historical load re-prediction values.

[0037] It should be noted that when calculating again, These values ​​are calculated based on the data and circumstances of that specific time period and will not change. If updates are scheduled regularly, such as once a day or even once a week, then the updated values ​​should be used. This avoids multiple parameter adjustments that could affect the load forecast values. Fluctuating back and forth. (Through settings) To avoid optimizing the control deviation compensation item Load forecast Calculation in progress Eliminate the margin setting.

[0038] As a further improvement of the present invention, the load-related features include the shape of the daily load curve and the main load types; The specific steps for matching the source domain include: the cloud database divides the daily load curves of the source and target domains into multiple time periods, calculates the average historical actual load value for each time period, calculates the peak-to-valley difference, morning peak percentage, evening peak percentage, and load factor based on the average historical actual load value for each time period, selects the corresponding source domain as a candidate reference object based on the main load type of the target domain, and calculates the daily load characteristic difference between the target domain and each candidate reference object. , , in, It represents the proportion of the early peak in the target domain. It represents the early peak percentage of the source domain. It represents the proportion of late-peak hours in the target domain. It is the proportion of the late peak in the source domain. It is the load rate of the target domain. It is the load rate of the source domain. and It is a weighting adjustment factor; Matching daily load characteristic gap The smallest source domain.

[0039] Peak-to-valley difference is the difference between the daily maximum load and the daily minimum load, divided by the daily maximum load. Morning peak percentage is the difference between the average load during the morning peak period and the daily minimum load, divided by the peak-to-valley difference. Evening peak percentage is the difference between the average load during the evening peak period and the daily minimum load, divided by the peak-to-valley difference. Load factor is the daily average load divided by the daily maximum load. Major load types can be categorized into industrial areas, commercial areas, residential areas, and mixed residential-commercial areas. Because different major load types exhibit significantly different load variation patterns, there is no meaningful matching between different types of major load types, thus reducing the discrepancy in subsequent daily load characteristics. The computational load of the calculation.

[0040] The difference between the morning peak percentage and the evening peak percentage reflects the shift in peak electricity consumption time, such as differences in factory morning shift start-up and differences in residents turning off high-load appliances. The difference in load factor can measure the overall load balance. This formula can be adapted to scenarios with different main load types. The difference between the morning peak percentage and the evening peak percentage is first summed and then the weights are adjusted to reflect the overall matching degree between the peak periods of the source domain and the target domain.

[0041] and It can be determined according to different main load types. As an example of the present invention, in an industrial area, its It's 12. It is 8, the proportion of early peak in the target domain. It is 0.28, the proportion of evening peak. It is 0.32, load factor The value is 0.75. There are two candidate references, let's say A and B. The proportion of the early peak in A is... It is 0.25, the proportion of evening peak. It is 0.30, load factor It is 0.8, its .

[0042] B's early peak proportion It is 0.31, the proportion of evening peak. It is 0.37, load factor It is 0.75, its .

[0043] As a further improvement of the present invention, the historical load information includes the value of the temperature load change coefficient, the value of the optimization control deviation compensation term coefficient, the historical load change value, and time data. Calculate the load forecast for the target domain in the next time period. The specific expression is as follows: ， ， ， ， ， ， ， in ， It is a dynamic weighting coefficient for temperature changes. This represents the deviation between historical predicted temperature values ​​and historical actual temperature values. If no historical predicted or actual temperature values ​​exist, this value will be 0. This represents the number of time periods for which historical temperature data was collected. If no historical temperature data exists, this value is 1. This is the maximum allowable predicted temperature error value. It is the minimum temperature load change weighting factor. It is the temperature load variation coefficient of the source domain, used to convert temperature changes into load changes. It predicts changes in temperature values. The next forecast period +1 is the weather forecast temperature value. The current time period The real-time temperature value, It is the sign correction factor. It is a data quality and security amplification factor. It is the transmission quality factor of the load monitoring equipment that collects temperature data. It is the smallest quality and safety amplification factor. It is a dynamic weighting coefficient based on historical load data trust and temperature changes. It is a reference value for mixed loads from the source and target domains during the same historical period. This refers to the total number of weeks the target domain was collected. It is the time decay weighting coefficient. Is the target domain before Historical load changes for the same time period of the week compared to the previous period. It is the source domain trust weight coefficient. The daily load characteristic difference conversion factor, This is the total number of weeks of data collection from the source domain. It is the time decay weighting coefficient. It is the source domain before Historical load changes for the same time period of the week compared to the previous period. It is the time decay factor. It is a dynamic base threshold. It is the load temperature influencing factor. It is the first The temperature difference between the same time period of the week and the current time period, if there is no first time period. If the temperature value is the same during the same period of the week, then this item is 1. It is an optimized control deviation compensation item, calculated based on historical load forecast values ​​and historical actual load values.

[0044] Although historical load information from the source domain is extracted after matching the source domain, not all historical load information from the source domain is used. For example, for mixed load reference values... When the target domain has a sufficient number of weeks of historical load change values, the historical load change value information of the source domain will not be considered. Specifically, the present invention... and The recommended range is between 4 and 8. If the target domain has 8 weeks of historical load change values, then the historical load change value information of the source domain will not be considered.

[0045] for The formula does not reference the source domain because the target domain is a newly built power distribution network, and the supporting load monitoring equipment, such as temperature sensors, is also mostly new. The deviation between the historical predicted temperature values ​​and the historical actual temperature values ​​of newly built temperature sensors is generally small. In contrast, the temperature sensors in the source domain have typically been in use for a longer period, resulting in larger deviations and rendering them meaningless for reference. Therefore, for the target domain, if there are no historical predicted or actual temperature values, this term is 0; otherwise, it represents the number of time periods for which historical temperature data was collected. The coefficient is 1. This represents the temperature load variation coefficient of the source domain. Since the optimal solution for the data time span required for this numerical calculation is equal to the preset reference duration, this value is a direct reference source domain. Specifically, the daily load characteristic difference conversion factor... The values ​​are based on the differences in daily load characteristics. The average value is determined, preferably by the daily load characteristic difference conversion factor. The value is 0.3.

[0046] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.

Claims

1. A data-driven intelligent load control method for distribution networks, characterized in that, Includes the following steps: By collecting load-related characteristics of the current distribution network through multiple load monitoring devices, the transmission quality coefficient of each load monitoring device is calculated. According to the transmission quality coefficient Calculate the quality timeliness score based on data update time. If the quality timeliness score If the data is below the preset timeliness threshold, the corresponding load monitoring equipment will re-collect data. The calculated quality timeliness score will be determined based on the re-collected data. If the timeliness threshold is lower than the preset threshold, alternative data is obtained through data substitution. The system identifies the duration of historical data for the current distribution network. If this duration exceeds a preset reference duration, the distribution network is considered a source region. Based on load-related characteristics and historical load information, the load forecast for the next time period is calculated. And upload load-related characteristics and historical load information to the cloud database; If the load is less than the preset reference duration, the distribution network is considered the target domain. Load-related characteristics are uploaded to the cloud database to match the source domain, historical load information of the source domain is extracted, and the load forecast value for the next time period of the target domain is calculated. ; Based on load forecast values Dynamically regulate the load of the distribution network.

2. The intelligent load control method for distribution networks based on data collaboration according to claim 1, characterized in that, Calculate the transmission quality coefficient for each load monitoring device. The specific steps include: ， in, It is the fault sensitivity factor of the i-th load monitoring device. Is the i-th load monitoring device in the most recent time? Total number of failures occurring within the period It is the packet loss penalty coefficient for the i-th load monitoring device. It is the most recent time The number of packets lost within, It is the most recent time The number of packets sent within, and It is a transmission quality weighting factor. It is a packet loss penalty amplification factor.

3. The intelligent load control method for distribution networks based on data collaboration according to claim 2, characterized in that, Based on transmission quality coefficient Calculate the quality timeliness score based on data update time. The specific steps include: ， in, It is the timeliness impact factor of the i-th load monitoring device. It is the current time point. This is the time of the most recent data update. It is a timeliness adjustment factor. It is a dynamic quality weighting coefficient. It is the base weight value of the dynamic quality weight coefficient. It is a transmission quality reward / penalty adjustment factor. It is the basic value for transmission quality rewards and penalties; The specific steps of the data substitution method include: calculating the quality timeliness score of the nearest similar load monitoring device to the load monitoring device. If the time limit is higher than the preset time threshold, then the data collected by the same type of load monitoring equipment will be extracted.

4. The intelligent load control method for distribution networks based on data collaboration according to claim 2, characterized in that, The distribution network has a historical database storing historical load and temperature information. Load-related characteristics include real-time load values. Real-time temperature value and weather forecast temperature values ; Calculate the load forecast for the next period based on load-related characteristics and historical load information. The specific steps include: ， ， ， ， ， in, It is a dynamic weighting coefficient for temperature changes. It is the deviation between historical predicted temperature values ​​and historical actual temperature values. It refers to the number of time periods for which historical temperature data was collected. This is the maximum allowable predicted temperature error value. It is the minimum temperature load change weighting factor. It is the temperature load variation coefficient, used to convert temperature changes into load changes. It predicts changes in temperature values. The next forecast period +1 is the weather forecast temperature value. The current time period The real-time temperature value, It is the sign correction factor. It is a data quality and security amplification factor. It is the transmission quality factor of the load monitoring equipment that collects temperature data. It is the smallest quality and safety amplification factor. It is a dynamic weighting coefficient based on historical load data trust and temperature changes. It is a load reference value for the same historical period. It is the total number of weeks collected. It is the time decay weighting coefficient. It was before Historical load changes for the same time period of the week compared to the previous period. It is the time decay factor. It is a dynamic base threshold. It is the load temperature influencing factor. It is the first The temperature difference between the same time period of the week and the current time period. It is an optimized control deviation compensation item, calculated based on historical load re-prediction values ​​and historical actual load values.

5. The intelligent load control method for distribution networks based on data collaboration according to claim 4, characterized in that, Temperature load variation coefficient The calculation expression is as follows: ， in, This refers to the number of days containing temperature load variation data of the same type within the same temperature range. The data type for temperature load variation includes weekdays and rest days. It is the current time period on day i. The actual load value, It is the time period preceding the current time period on day i. The actual load value, Current time period on day i Historical actual temperature values, The previous time period on day i The historical actual temperature value.

6. The intelligent load control method for distribution networks based on data collaboration according to claim 4, characterized in that, Optimize the control deviation compensation item The calculation expression is as follows: ， ， in, It is the number of recent historical load re-prediction values ​​and historical actual load values ​​collected. It is the most recent A historical load true value, It is the most recent Historical load re-prediction value Used to reserve margin settings; The calculation steps for historical load reprediction are as follows: based on the real-time temperature values ​​for the current period... ,renew and The value, based on the updated and The value, and the previous Calculation of load-related characteristics and historical load information for each time period The load forecast value for the +1 time period is obtained. Historical load re-prediction values.

7. The intelligent load control method for distribution networks based on data collaboration according to claim 1, characterized in that, The load-related characteristics include the shape of the daily load curve and the main load types; The specific steps for matching the source domain include: the cloud database divides the daily load curves of the source and target domains into multiple time periods, calculates the average historical actual load value for each time period, calculates the peak-to-valley difference, morning peak percentage, evening peak percentage, and load factor based on the average historical actual load value for each time period, selects the corresponding source domain as a candidate reference object based on the main load type of the target domain, and calculates the daily load characteristic difference between the target domain and each candidate reference object. , ， in, It represents the proportion of the early peak in the target domain. It represents the early peak percentage of the source domain. It represents the proportion of late-peak hours in the target domain. It is the proportion of the late peak in the source domain. It is the load rate of the target domain. It is the load rate of the source domain. and It is a weighting adjustment factor; Matching daily load characteristic gap The smallest source domain.

8. The intelligent load control method for distribution networks based on data collaboration according to claim 7, characterized in that, Historical load information includes the value of the temperature load change coefficient, the value of the optimization control deviation compensation term coefficient, historical load change values, and time data; Calculate the load forecast for the target domain in the next time period. The specific expression is as follows: ， ， ， ， ， ， ， in ， It is a dynamic weighting coefficient for temperature changes. This represents the deviation between historical predicted temperature values ​​and historical actual temperature values. If no historical predicted or actual temperature values ​​exist, this value will be 0. This represents the number of time periods for which historical temperature data was collected. If no historical temperature data exists, this value is 1. This is the maximum allowable predicted temperature error value. It is the minimum temperature load change weighting factor. It is the temperature load variation coefficient of the source domain, used to convert temperature changes into load changes. It predicts changes in temperature values. The next forecast period +1 is the weather forecast temperature value. The current time period The real-time temperature value, It is the sign correction factor. It is a data quality and security amplification factor. It is the transmission quality factor of the load monitoring equipment that collects temperature data. It is the smallest quality and safety amplification factor. It is a dynamic weighting coefficient based on historical load data trust and temperature changes. It is a reference value for mixed loads from the source and target domains during the same historical period. This refers to the total number of weeks the target domain was collected. It is the time decay weighting coefficient. Is the target domain before Historical load changes for the same time period of the week compared to the previous period. It is the source domain trust weight coefficient. The daily load characteristic difference conversion factor, This is the total number of weeks of data collection from the source domain. It is the time decay weighting coefficient. It is the source domain before Historical load changes for the same time period of the week compared to the previous period. It is the time decay factor. It is a dynamic base threshold. It is the load temperature influencing factor. It is the first The temperature difference between the same time period of the week and the current time period, if there is no first time period. If the temperature value is the same during the same period of the week, then this item is 1. It is an optimized control deviation compensation item, calculated based on historical load forecast values ​​and historical actual load values.

Images