Rapid construction method of power data analysis model based on big data

Abstract

The invention relates to the technical field of model construction, in particular to a rapid construction method of a power data analysis model based on big data, which comprises the following steps of: monitoring voltage amplitude, frequency deviation and phase angle error based on voltage, current and power data of a transformer substation, detecting data integrity, removing abnormal data node information, and establishing a power data analysis model; and obtaining a power data processing result. According to the method, the prediction precision of the power demand is optimized by enhancing verification of data integrity and accurate extraction of key features, the reliability of data is ensured by combining data backtracking and Hash comparison, and the quality of data input is improved by deeply analyzing power load features and equipment operation parameters; the analysis of the behavior of the power system is optimized, the actual use trend is matched by dynamically adjusting the weight of the power input data, and the application flexibility and the real-time response capability of a data analysis model are enhanced, so that the aspects of resource allocation and power generation efficiency are optimized.

Description

Technical Field

[0001] The present invention relates to the technical field of model construction, and particularly relates to a method for quickly constructing a power data analysis model based on big data. Background Art

[0002] Model construction refers to creating a mathematical or simulation model to imitate real-world processes, systems, or phenomena, mainly used for prediction, optimization, and decision support. It is a core component in fields such as data science, artificial intelligence, and machine learning. In the power industry, model construction particularly focuses on how to use a large amount of collected power usage data to predict power demand, optimize power generation and distribution resources, and improve the energy efficiency and reliability of the system.

[0003] Among them, the method for quickly constructing a power data analysis model based on big data aims to rapidly develop a model that can process and analyze large-scale power data sets to conduct in-depth research on the energy usage of rural enterprises using power data. It particularly focuses on collecting the power consumption data of enterprises within a specific time period to calculate the output value per kilowatt-hour, which helps to provide necessary data support for the energy-saving transformation of high-energy-consuming equipment and provide intelligent diagnostic services for high-voltage enterprise customers. In addition, this model also supports real-time energy usage analysis, accurately monitors the safety production status of the production area, establishes an electricity consumption tracking ledger, specifically monitors the power load of agricultural product enterprises, and timely assists in meeting the power demand in production.

[0004] The existing technology has deficiencies in data integrity verification and dynamic adjustment mechanisms, which limit the accuracy of power demand prediction and the efficiency of resource allocation. The lack of an effective data verification mechanism makes it difficult to discover and correct data errors in a timely manner, affecting the accuracy of power data analysis. In addition, the existing methods have not fully exerted their potential in optimizing the model using power load characteristics and equipment operation parameters, resulting in an inability to accurately predict power demand during peak demand periods. This lack of prediction easily leads to uneven distribution of power resources, affecting production efficiency and the stability of the power system. Summary of the Invention

[0005] The purpose of the present invention is to solve the deficiencies existing in the prior art, and propose a method for quickly constructing a power data analysis model based on big data.

[0006] To achieve the above purpose, the present invention adopts the following technical solution: A method for quickly constructing a power data analysis model based on big data, including the following steps: S1: Based on the voltage, current, and power data of the substation, monitor the voltage amplitude, frequency deviation, and phase angle error, detect data integrity, and remove abnormal data node information to obtain the power data processing result; S2: According to the power data processing result, collect power load characteristics, equipment operation parameters and time series data, evaluate the mutual dependence between each characteristic, screen the key characteristics of the power system, and obtain the balanced configuration of power characteristics. S3: Based on the balanced configuration of power characteristics, calculate the power change rate within the differential time window, screen the time intervals of power demand growth, analyze the fluctuation offset of the target time interval, and adjust the weight of the power input data to obtain the dynamic index of power growth trend. S4: According to the dynamic index of power growth trend, analyze the load parameters of the transmission line and the power supply situation of the grid sub-region, judge the power supply and demand status of each sub-region, and obtain the power supply and demand balance evaluation data. S5: Call the power supply and demand balance evaluation data, analyze and screen the power generation units with the optimal efficiency, and then combine the real-time load of the grid and the power prediction data to obtain the power analysis and deployment ready model.

[0007] The improvement of the present invention is that the power data processing result includes data integrity records and abnormal node indexes. The balanced configuration of power characteristics is specifically a load characteristic set, an equipment parameter set, and a time series set. The dynamic index of power growth trend includes rate change analysis results, growth interval identification results, and data weight conditions. The power supply and demand balance evaluation data is specifically load status information, prediction error evaluation results, and analysis parameter details. The power analysis and deployment ready model includes power generation efficiency data, cost-benefit data, and load response configuration.

[0008] The improvement of the present invention is that the steps for obtaining the power data processing result are specifically as follows: S111: Based on the voltage, current and power data of the substation, monitor the voltage amplitude, frequency offset and phase angle error, calculate the amplitude offset rate, frequency error amount and phase angle change amount of the data node, screen the data nodes that exceed the standard range, and obtain the abnormal data node set. S112: Call the abnormal data node set, use the data backtracking mechanism to restore the missing data information from the backup system, perform hash value comparison, screen the data nodes with mismatched hash values, and obtain the data integrity abnormal point set. S113: Call the data integrity abnormal point set, remove the abnormal data nodes, evaluate the voltage amplitude, frequency offset and phase angle error, and use the formula: ; Calculate the data quality evaluation value , and screen the qualified data nodes according to the evaluation value to obtain the power data processing result, where represents the voltage amplitude of the th data node, represents the voltage reference value, represents the The frequency of each data node, represents the frequency reference value, represents the phase angle of the th data node, represents the total number of data nodes, represents the standard deviation of voltage data, represents the standard deviation of frequency data, represents the standard deviation of phase angle data.

[0009] The improvement of the present invention is that the obtaining steps of the power feature balanced configuration are specifically as follows: S211: According to the power data processing result, collect power load characteristics, equipment operation parameters and time series data, extract the load power, equipment current and voltage change trends of each data point, and analyze the distribution of characteristics in the time dimension to obtain the power feature time series distribution; S212: Based on the power feature time series distribution, calculate the correlation coefficient between each feature, screen the features with key mutual dependencies, and analyze their relevance to the operation state of the power system to obtain the power key feature set; S213: Call the power key feature set, reformat the feature representation, and according to the standardized mean and feature deviation, use the formula: ; Calculate the feature balance degree , and adjust the feature normalization parameter to obtain the power feature balanced configuration, where represents the th power feature value, represents the mean value of the feature , represents the standard deviation of the feature , represents the total number of feature data points.

[0010] The improvement of the present invention is that the obtaining steps of the power growth trend dynamic index are specifically as follows: S311: Based on the power feature balanced configuration, analyze the power load curve, calculate the power change rate within the differential time window, continuously detect the power change value, and calculate the average change rate of power at time intervals to generate power change data; S312: Based on the power change data, calculate the rate increment value, set the dynamic screening threshold, and screen the time intervals where the rate increment exceeds the threshold to obtain the power demand growth interval set; S313: Call the power demand growth interval set, analyze the fluctuation offset of the target time interval, adjust the weight of the power input data, and match the current power usage trend, using the formula: ; Obtain the dynamic index of power growth trend , where represents the change amount of power input data, represents the total number of power data points within the time interval, represents the weight coefficient of power input data, represents the fluctuation offset of the power load curve within the target time interval, represents the total number of data points in the fluctuation analysis, represents the adjustment factor of the offset trend.

[0011] The improvement of the present invention is that the steps for obtaining the supply-demand balance evaluation data are specifically as follows: S411: Based on the dynamic index of power growth trend, analyze the load parameters of transmission lines, normalize the power transmission ratio of each group of transmission lines, and obtain the transmission line load ratio; S412: Based on the transmission line load ratio, combine the power supply capacity parameters of the power grid partition, calculate the supply-demand deviation value of each partition, and judge the current power supply state to obtain the supply-demand deviation amount of the power grid partition; S413: Based on the supply-demand deviation amount of the power grid partition, determine the supply-demand prediction error, optimize the analysis parameters, and use the formula: ; Calculate the supply-demand balance error value , and compare the adjusted prediction data with the current data to obtain the supply-demand balance evaluation data, where represents the actual load demand of the th power grid partition, represents the power supply capacity of the th power grid partition, represents the total number of power grid partitions.

[0012] The improvement of the present invention is that the steps for obtaining the power analysis deployment ready model are specifically as follows: S511: Invoke the supply-demand balance evaluation data, evaluate the power demand level, obtain the power generation capacity of the power generation unit, and analyze its matching degree with the demand level to obtain the power supply-demand matching degree; S512: Based on the power supply-demand matching degree, combine the power generation cost data of the power generation unit, calculate the unit power generation cost, and screen the power generation unit with the optimal power generation cost, using the formula: ; Calculate the unit efficiency coefficient , where represents the power generation power of the th power generation unit, represents the power generation cost of the th power generation unit, represents the power demand for the th time period, represents the total number of all time periods; S513: Based on the unit efficiency coefficient, combined with the selected power generation units, according to the real-time load of the power grid and power prediction data, create an executable model file to obtain a power analysis deployment ready model.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the present invention, by enhancing the verification of data integrity and accurately extracting key features, the prediction accuracy of power demand is optimized. Combining data backtracking and hash comparison ensures the reliability of data. By deeply analyzing the power load characteristics and equipment operation parameters, the quality of data input is improved, thereby optimizing the analysis of power system behavior. By dynamically adjusting the weights of power input data to match the actual usage trend, the application flexibility and real-time response ability of the data analysis model are enhanced, thus optimizing resource allocation and power generation efficiency. Not only is the utilization efficiency of power resources improved, but also the energy security and efficiency of power production are ensured. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 is a flowchart of a method for quickly constructing a power data analysis model based on big data proposed by the present invention; Figure 2 is a flowchart for obtaining the processing result of power data in the present invention; Figure 3 is a flowchart for obtaining the balanced allocation of power characteristics in the present invention; Figure 4 is a flowchart for obtaining the dynamic index of power growth trend in the present invention; Figure 5 is a flowchart for obtaining the evaluation data of supply-demand balance in the present invention; Figure 6 is a flowchart for obtaining a power analysis deployment ready model in the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0015] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0016] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as a limitation to the present invention. In addition, in the description of the present invention, the meaning of "a plurality of" is two or more, unless otherwise specifically defined. Embodiment

[0017] Please refer to Figure 1 , the present invention provides a technical solution: a method for quickly constructing a power data analysis model based on big data, including the following steps: S1: Based on the voltage, current, and power data of the substation, monitor the voltage amplitude, frequency deviation, and phase angle error, use the data backtracking mechanism to recover the missing data information from the backup system, perform hash value comparison, detect data integrity, and remove the abnormal data node information to obtain the power data processing result; S2: According to the power data processing result, collect the power load characteristics, equipment operation parameters, and time series data, evaluate the mutual dependence between each feature, and screen the key features of the power system, and reformat the feature representation to obtain the power feature balanced configuration; S3: Based on the power feature balanced configuration, analyze the power load curve, calculate the power change rate within the differential time window, screen the time interval of power demand growth, analyze the fluctuation offset of the target time interval, adjust the weight of the power input data, and match the current power usage trend to obtain the dynamic index of power growth trend; S4: According to the dynamic index of power growth trend, analyze the load parameters of the transmission line and the power supply situation of the power grid partition, judge the power supply and demand status of each partition, determine the supply and demand prediction error, optimize the analysis parameters, and then compare the adjusted prediction data with the current data to obtain the supply and demand balance evaluation data; S5: Call the supply and demand balance evaluation data, collect the available power generation parameters and power generation cost data, analyze and screen the power generation unit (power plant or generator set) with the optimal efficiency, and then combine the real-time load of the power grid and the power prediction data to create an executable model file to obtain the power analysis deployment ready model.

[0018] The power data processing results include data integrity records and abnormal node indexes. The power feature balanced configuration specifically refers to the load feature set, equipment parameter set, and time series set. The dynamic indicators of power growth trends include rate change analysis results, growth interval identification results, and data weight conditions. The supply-demand balance assessment data specifically refers to load status information, prediction error assessment results, and analysis parameter details. The power analysis deployment ready model includes power generation efficiency data, cost-benefit data, and load response configuration.

[0019] Please refer to Figure 2 , and the steps for obtaining the power data processing results are specifically as follows: S111: Based on the voltage, current, and power data of the substation, monitor the voltage amplitude, frequency offset, and phase angle error, calculate the amplitude offset rate, frequency error amount, and phase angle change amount of the data nodes, screen the data nodes that exceed the standard range, and obtain the abnormal data node set; For the historical operation data of specific monitoring points, obtain the initial voltage amplitude, frequency, and phase angle data. The initial data of a certain substation is: voltage amplitude 230V, frequency 50.05Hz, and phase angle error 0.5°. Use a high-precision sensor for data collection and continuously monitor at 1s time intervals. The monitoring time is set to 10 minutes, and the number of data points is 600. After collecting the data at each time point, calculate the amplitude offset rate, frequency error amount, and phase angle change amount of each data node. The calculation methods are as follows. The amplitude offset rate Calculate: , where represents the voltage of the th data node, represents the system reference voltage (set to 230V). If the sampled voltage at a certain moment is 235V, the offset rate is calculated as follows: , similarly, the frequency error amount is calculated as follows: , where represents the frequency of the th data node, is set to 50Hz. If at a certain moment , then , and the phase angle change amount is calculated as follows: , where is the phase angle of the th data node, is the reference phase angle (set to 0°). If at a certain moment , then: , if the amplitude offset rate of a certain data point exceeds 5%, the frequency error amount exceeds 0.2Hz, and the phase angle change amount exceeds 1°, then mark this data point as an abnormal data point and record it in the abnormal data node set.

[0020] S112: Invoke the set of abnormal data nodes, use the data backtracking mechanism to recover the missing data information from the backup system, perform hash value comparison, filter out the data nodes with unmatched hash values, and obtain the set of data integrity exception points; Adopt the data backtracking mechanism to retrieve the lost or abnormal data within 10 minutes from the backup system. Assume that the data recovered from the backup database at a certain abnormal time point is: voltage 229V, frequency 50Hz, phase angle 0.3°. Perform hash value calculation on the recovered data, and the calculation method is as follows: After calculating the hash value, compare it with the standard hash value stored in the system. If the hash values do not match, it is considered that the data integrity is abnormal, and record this data point in the data integrity exception node set.

[0021] S113: Invoke the set of data integrity exception points, remove the abnormal data nodes, evaluate the voltage amplitude, frequency deviation, and phase angle error, and use the formula: ; Calculate the data quality evaluation value and filter out the qualified data nodes according to the evaluation value to obtain the power data processing result, where represents the voltage amplitude of the th data node, represents the voltage reference value, represents the th data node's frequency, represents the frequency reference value, represents the th data node's phase angle, represents the phase angle reference value, represents the total number of data nodes, represents the standard deviation of the voltage data, represents the standard deviation of the frequency data, represents the standard deviation of the phase angle data; Remove the abnormal data nodes, filter the remaining data nodes, calculate the voltage amplitude, frequency deviation, and phase angle error of each node, and build a data quality evaluation model based on the calculated data. The number of data points Let the standard deviation , , Take some data examples as shown in the following table: Power Data Monitoring Point Data Table

[0022] As shown in the above table, when calculating , take 4 data points as an example: ; ; If the qualified quality range is 0.8 ≤ ≤ 1.2, then the result meets this range, indicating that the overall deviation of the filtered data nodes is small, belonging to the range where data integrity and accuracy meet the standards. Therefore, the data nodes can be used for subsequent analysis, scheduling, and monitoring of the power system without further data supplementation or correction.

[0023] Please refer to Figure 3 , the specific steps for obtaining the balanced configuration of power characteristics are as follows: S211: According to the power data processing results, collect power load characteristics, equipment operation parameters, and time series data, extract the load power, equipment current, and voltage change trends of each data point, and analyze the distribution of characteristics in the time dimension to obtain the time series distribution of power characteristics; In the data collection stage, it is necessary to monitor different operating states of the power system and record its key operating parameters at different time points, such as the voltage, current, frequency of the substation, and the power consumption of the load equipment. Assuming that the daily power load of the load equipment in a certain substation shows periodic fluctuations, with higher loads during the morning peak (7:00 - 9:00) and evening peak (18:00 - 21:00) periods, and the lowest load at night (23:00 - 5:00), then during the data extraction process, it is necessary to record the power, equipment current, and voltage data of the key time periods respectively, and calculate their change trends at different time points. First, calculate the load power : ; Among them, is the voltage, is the current, is the power factor. If the voltage at a certain moment is 230V, the current is 50A, and the power factor is 0.85, then:

[0024] Similarly, calculate the change rate of equipment current: ; Among them, is the current value at the current time point, is the current value at the previous time point. If the current increases from 40A to 50A at 8:00, then: ; If this value exceeds a certain set threshold (such as 20%), it is considered that there is a significant change in the load at this time point. In addition, the voltage change trend can be calculated by moving average:

[0025] If at a certain moment , then: ; By calculating and comparing the load power, device current, and voltage change trends at different time points, the distribution in the time dimension can be obtained, and the time-series distribution of power characteristics can be acquired.

[0026] S212: Based on the time-series distribution of power characteristics, calculate the correlation coefficients between each feature, screen the features with strong mutual dependence, and analyze their relevance to the operating state of the power system to obtain the set of key power features; During the operation of the power system, there are mutual dependence relationships among multiple characteristic variables. For example, power load, device current, voltage fluctuations, etc. will all affect the system stability. To screen out the key features with strong mutual dependence, it is necessary to statistically analyze the change situations of each characteristic data at different time points and calculate the correlation coefficients between the features. First, obtain the characteristic data at different time points and calculate their mean values and standard deviations. For example, the load power and device current are as follows: At 7:00, the load power is 5000W and the device current is 22A; At 12:00, the load power is 5500W and the device current is 24A; At 18:00, the load power is 6000W and the device current is 27A; At 21:00, the load power is 6200W and the device current is 28A; It can be seen that the device current and the load power show a synchronous change trend. Therefore, it is necessary to calculate the correlation between the two to determine whether there is a strong dependence relationship. The correlation calculation is based on the covariance and standard deviation of the characteristic variables. For example, if the correlation coefficient between the device current and the load power is close to 1, it indicates a high correlation between the two. If it is close to 0, it means almost no correlation. When screening features, a correlation threshold is usually set, such as 0.3. If the correlation coefficient of a certain feature with other features is lower than 0.3, it is considered that its impact on the system operating state is small and it is not retained as a key feature. If the correlation coefficient of a certain feature is higher than 0.8, it indicates that it has a strong dependence on other key features and can be included in the key feature set. In addition, in addition to calculating the numerical correlation between features, it is also necessary to analyze its practical significance in the operation of the power system. For example, if it is found that the load power and the device current are highly correlated, the current change can be directly used to predict the load fluctuation, thereby reducing unnecessary data redundancy and obtaining the set of key power features.

[0027] S213: Call the set of key power features, reformat the feature representation, and adopt the formula based on the standardized mean and feature deviation: ; Calculate the feature balance , and adjust the feature normalization parameters to obtain an equilibrium configuration of power features. Among them, represents the th power feature value, represents the mean value of feature , represents the standard deviation of feature , represents the total number of feature data points; For the selected key feature set, feature equalization is required to make different features have a consistent scale after normalization. The key features of a certain power system are power and current . Part of the sample data is as follows: Key Feature Data Table

[0028] Calculate the mean value based on the data in the above table: ; ; Then calculate the standard deviation: ; ; ; ; ;

[0029] ; Next, calculate the feature equalization degree : ; ; ; Similarly, calculate the current equalization degree: ; ; ; The obtained equalization degree is 0.93 (power) and 0.90 (current). The qualified interval of the equalization degree is 0.85 ≤ ≤ 1.15. The result means that the selected key features have good equalization after normalization, and the difference in data scale between different features is small. Therefore, it can be directly used for further analysis and scheduling of the power system to obtain an equilibrium configuration of power features.

[0030] Please refer to Figure 4 , and the steps for obtaining the dynamic index of the power growth trend are specifically as follows: S311: Based on the balanced configuration of power characteristics, analyze the power load curve, calculate the power change rate within the differential time window, continuously detect the power change value, and calculate the average change rate of power at time intervals to generate power change data; Call the historical power load curve data, select the load power values within the recent one cycle, and combine with the real-time power measurement data. Divide the data set in units of time windows. The length of each time window can be set according to the actual situation, such as 5 minutes or 0 minutes as a unit. Extract the power values of all sampling points within this time window, and calculate its change rate. During the calculation process, use the power change calculation formula: , where represents the power change rate at the current moment, is the power value at the current moment, is the power value at the previous moment, is the time interval between two time points. During the continuous change process of the power load curve, this formula is used to characterize the change trend of the load power over time. Subsequently, perform a continuity detection on the calculated power change rate data, that is, judge whether its change trend is stable within multiple consecutive time windows. Set the power change rate threshold . When the power change rate exceeds for multiple consecutive time windows, mark this time period as a period of drastic change, and focus on the data change in this interval in the next calculation. At the same time, calculate the average change rate of the power values within each time window. The formula is as follows: , where represents the average power change rate within a certain time period, is the instantaneous power change rate within each time window, is the total number of time windows. After obtaining the average change rate of all time windows, form a power change data set and generate power change data.

[0031] S312: Based on the power change data, calculate the rate increment value, set the dynamic screening threshold, and screen the time intervals where the rate increment exceeds the threshold to obtain the power demand growth interval set; First, according to the generated power change data set, extract the average change rate of each time window and calculate the rate increment value. The calculation method of the rate increment value is: , where represents the rate increment value, and respectively represent the average power change rates of two adjacent time windows. When the rate increment value When the rate increment threshold is exceeded it is determined that the power demand growth within this time window is significant. The threshold can be obtained through statistical analysis of historical data. For example, it can be set as the mean of all rate increment values in the past cycle plus the standard deviation, that is: , where represents the mean of the rate increment values of all time windows, represents the standard deviation of the rate increment values. This method can adaptively adjust the screening threshold, thereby screening out the time windows with rate increments exceeding the threshold and marking them as the rapid power demand growth intervals, obtaining the set of power demand growth intervals.

[0032] S313: Invoke the set of power demand growth intervals, analyze the fluctuation offset of the target time interval, adjust the weight of the power input data to match the current power usage trend, using the formula: ; Obtain the dynamic index of power growth trend , where represents the change in power input data, represents the total number of power data points within the time interval, represents the weight coefficient of the power input data, represents the fluctuation offset of the power load curve within the target time interval, represents the total number of data points in the fluctuation analysis, represents the adjustment factor for the offset trend; If within a certain time period, the change in power input data is kW, the total number of data points , the fluctuation offset of the power load curve is kW, the total number of data points , set the weight of the power input data , the offset trend adjustment factor , substitute into the calculation: ; ; This result indicates that the dynamic index of power growth trend is 6.2, representing the comprehensive situation of the current power demand growth trend. If this value exceeds a certain preset interval, for example, higher than 6.0, it can be determined that the power demand growth is significant, and adjustment strategies should be taken to optimize power dispatching.

[0033] Please refer to Figure 5 , the specific steps for obtaining the supply-demand balance assessment data are as follows: S411: Analyze the load parameters of transmission lines based on the dynamic indicators of power growth trends, normalize the power transmission ratios of each group of transmission lines, and obtain the transmission line load ratios. First, obtain the rated power and actual transmission power data of the transmission lines. For example, the rated power of a certain transmission line is 500 MW and the actual transmission power is 350 MW. Calculate the load parameter of this line, that is, the load rate. The calculation formula is: Load rate = (Actual transmission power / Rated power) × 100%. Substituting the data, we get: Load rate = (350 MW / 500 MW) × 100% = 70%, indicating that this line is operating at 70% of its maximum capacity. Next, normalize the power transmission ratios of each group of transmission lines. Suppose there are three transmission lines with actual transmission powers of 350 MW, 275 MW, and 400 MW respectively. First, calculate the total transmission power: 350 MW + 275 MW + 400 MW = 1025 MW. Then, calculate the power transmission ratio of each line: Line 1: 350 MW / 1025 MW ≈ 0.341; Line 2: 275 MW / 1025 MW ≈ 0.268; Line 3: 400 MW / 1025 MW ≈ 0.390; Finally, normalize the ratios so that their sum is 1. The normalized power transmission ratios are: Line 1: 0.341 / (0.341 + 0.268 + 0.390) ≈ 0.341 Line 2: 0.268 / (0.341 + 0.268 + 0.390) ≈ 0.268 Line 3: 0.390 / (0.341 + 0.268 + 0.390) ≈ 0.390 The normalized power transmission ratios reflect the relative contributions of each line to the total transmission power. Finally, obtain the transmission line load ratios to evaluate the load conditions of each line.

[0034] S412: Based on the transmission line load ratios, combined with the power supply capacity parameters of the grid partitions, calculate the supply-demand deviation values of each partition, and judge the current power supply status to obtain the grid partition supply-demand deviation amounts. Based on the transmission line load ratio calculated above, combined with the power supply capacity parameters of the power grid partitions, evaluate the supply-demand balance of each partition. Assume that the power grid is divided into three partitions A, B, and C, with power supply capacities of 50 MW, 60 MW, and 70 MW respectively. Through real-time monitoring, the actual load demands of each partition are 45 MW, 65 MW, and 68 MW respectively. Calculate the supply-demand deviation value of each partition = power supply capacity - actual load demand. Therefore, the supply-demand deviation of partition A is 5 MW (50 MW - 45 MW), that of partition B is -5 MW (60 MW - 65 MW), and that of partition C is 2 MW (70 MW - 68 MW). Judge the current power supply status according to the supply-demand deviation value: when the deviation is positive, it means there is power supply surplus; when it is negative, it means there is power supply shortage. Therefore, partitions A and C have power supply surplus, and partition B has power supply shortage. Through the above analysis, the supply-demand deviation amounts of the power grid partitions are obtained.

[0035] S413: Based on the supply-demand deviation amounts of the power grid partitions, determine the supply-demand prediction error and optimize the analysis parameters. Use the formula: ; Calculate the supply-demand balance error value , and compare the adjusted prediction data with the current data to obtain the supply-demand balance evaluation data. Among them, represents the actual load demand of the th power grid partition, represents the power supply capacity of the th power grid partition, represents the total number of power grid partitions; Determine the supply-demand prediction error, that is, the difference between the actual load demand and the predicted value. Assume that the predicted values of the load demands of partitions A, B, and C are 48 MW, 63 MW, and 69 MW respectively. Then the prediction errors are -3 MW (45 MW - 48 MW), 2 MW (65 MW - 63 MW), and -1 MW (68 MW - 69 MW) respectively. Substitute the above data into the formula: ; Calculate the supply-demand balance error value , which indicates that the overall supply and demand of the current power grid are basically balanced with a small error. By comparing the adjusted prediction data with the current actual data, the supply-demand balance evaluation data is obtained, providing a reference basis for power grid dispatching and planning.

[0036] Please refer to Figure 6 , and the specific steps for obtaining the power analysis deployment ready model are as follows: S511: Call the supply-demand balance evaluation data, evaluate the power demand level, obtain the power generation capacity of the power generation units, and analyze their matching degree with the demand level to obtain the power supply-demand matching degree; Obtain the real-time load data of the power grid, and combine it with the power prediction data to determine the power demand levels for multiple future time periods. On this basis, collect the power generation capacity parameters of each power generation unit, analyze its maximum and minimum power output ranges, and evaluate their respective adjustment rates. Calculate the available power of each power generation unit at different load levels. For different time points, calculate the difference between the power demand and the available power generation to obtain the power profit and loss value at each moment. Set a threshold for supply-demand matching determination. Assume the threshold is ±5%. If the power profit and loss value is within this range, it is considered that the supply-demand matching is reasonable. If it exceeds the threshold, calculate the output power of the power generation unit that needs to be adjusted according to the exceeded part, and use uniform adjustment or preferentially call the power generation unit with low cost for supplementary calculation to adjust the load distribution of each power generation unit. Finally, calculate the supply-demand matching degree, and the matching degree can be calculated by the ratio of the supply-demand difference to the demand value. For example, at a certain time point, the demand is 100MW and the power generation capacity is 95MW, then the supply-demand matching degree is , indicating that the supply and demand are basically matched. If it is 110MW, then the matching degree is , exceeding the threshold range, then the load distribution needs to be adjusted to obtain the power supply-demand matching degree.

[0037] S512: Based on the power supply-demand matching degree, combined with the power generation cost data of the power generation unit, calculate the unit power generation cost, and screen the power generation unit with the optimal power generation cost. Use the formula: ; Calculate the unit efficiency coefficient , where represents the power generation of the th power generation unit, represents the power generation cost of the th power generation unit, represents the power demand in the th time period, represents the total number of all time periods; First, obtain the fuel consumption rate of each power generation unit, and combine it with the unit fuel price to calculate the unit power generation cost. For example, the fuel consumption rate of a coal-fired power generation unit is 0.3t / MWh, and the unit fuel price is 500 yuan / t, then the unit power generation cost is yuan / MWh. At the same time, calculate the unit cost of all power generation units, set the cost threshold to 200 yuan / MWh, screen the power generation units with costs lower than the threshold and meeting the supply-demand matching, and further calculate their unit efficiency parameters. Use the formula:

[0038] The data of the coal-fired power generation unit is as follows: power output: 80MW, 90MW, 100MW (corresponding to three time periods); unit power generation cost: 120 yuan / MWh, 110 yuan / MWh, 105 yuan / MWh; power demand: 100MW, 100MW, 100MW - total number of time periods: 3. Substitute the data into the formula: ; ; ; Comparison result, the of the coal-fired power generation unit, the of the gas-fired power generation unit. Therefore, the unit efficiency coefficient of the coal-fired power generation unit is more excellent. Select it into the power generation unit list to obtain the unit efficiency coefficient.

[0039] S513: Based on the unit efficiency coefficient, combined with the selected power generation units, according to the real-time load of the power grid and the power prediction data, create an executable model file to obtain a power analysis deployment-ready model; Obtain the power output range of the selected optimal power generation unit, calculate its power adjustment amplitude under different load conditions, set the load adjustment period, for example, adjust the power distribution every 15 minutes. Based on the real-time load data, obtain the current load level, and calculate the load trend for the next adjustment period according to the prediction data. For example, if the current load is 200MW and it is predicted that the load will increase to 210MW after 10 minutes, then calculate the additional power generation required, and perform power scheduling according to the power generation unit with the optimal unit efficiency coefficient to ensure that the power distribution is within the adjustable range, generate an executable model file containing the load adjustment plan for all time periods, and obtain a power analysis deployment-ready model.

[0040] The above is only a preferred embodiment of the present invention, and it does not limit the present invention in other forms. Any person skilled in the art may use the disclosed technical content to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for rapidly constructing a power data analysis model based on big data, characterized in that, It includes the following steps: S1: Based on the voltage, current, and power data of the substation, monitor the voltage amplitude, frequency deviation, and phase angle error, detect the data integrity, remove the abnormal data node information, and obtain the power data processing result; S2: According to the power data processing result, collect the power load characteristics, equipment operation parameters, and time series data, evaluate the mutual dependence between each feature, screen the key features of the power system, and obtain the balanced configuration of power features; S3: Based on the balanced configuration of power features, calculate the power change rate within the differential time window, screen the time intervals of power demand growth, analyze the fluctuation offset of the target time interval, and adjust the weight of the power input data to obtain the dynamic index of power growth trend; S4: According to the dynamic index of power growth trend, analyze the load parameters of the transmission line and the power supply situation of the grid partition, judge the power supply and demand status of each partition, and obtain the power supply and demand balance evaluation data; S5: Call the power supply and demand balance evaluation data, analyze and screen the power generation units with the optimal efficiency, and then combine the real-time load of the grid and the power prediction data to obtain the power analysis deployment ready model.

2. The rapid construction method of the power data analysis model based on big data according to claim 1, wherein The power data processing result includes the data integrity record and the abnormal node index. The balanced configuration of power features is specifically the load feature set, equipment parameter set, and time series set. The dynamic index of power growth trend includes the rate change analysis result, growth interval identification result, and data weight situation. The power supply and demand balance evaluation data is specifically the load status information, prediction error evaluation result, and analysis parameter details. The power analysis deployment ready model includes the power generation efficiency data, cost-benefit data, and load response configuration.

3. The rapid construction method of the power data analysis model based on big data according to claim 1, characterized in that The specific steps for obtaining the power data processing result are as follows: S111: Based on the voltage, current, and power data of the substation, monitor the voltage amplitude, frequency deviation, and phase angle error, calculate the amplitude offset rate, frequency error amount, and phase angle change amount of the data node, screen the data nodes that exceed the standard range, and obtain the abnormal data node set; S112: Call the abnormal data node set, use the data backtracking mechanism to restore the missing data information from the backup system, perform the hash value comparison, screen the data nodes with unmatched hash values, and obtain the data integrity abnormal point set; S113: Call the data integrity abnormal point set, remove the abnormal data nodes, evaluate the voltage amplitude, frequency deviation, and phase angle error, and use the formula: ； Calculate the data quality evaluation value , and filter qualified data nodes according to the evaluation value to obtain the power data processing result. Among them, represents the voltage amplitude of the th data node, represents the voltage reference value, represents the frequency of the th data node, represents the frequency reference value, represents the phase angle of the th data node, represents the phase angle reference value, represents the total number of data nodes, represents the standard deviation of voltage data, represents the standard deviation of frequency data, represents the standard deviation of phase angle data.

4. The rapid construction method of the power data analysis model based on big data according to claim 1, characterized in that The specific steps for obtaining the balanced configuration of power features are as follows: S211: According to the power data processing result, collect the power load characteristics, equipment operation parameters, and time series data, extract the load power, equipment current, and voltage change trend of each data point, and analyze the distribution of features in the time dimension to obtain the time series distribution of power features; S212: Based on the time series distribution of power features, calculate the correlation coefficient between each feature, screen the features with key mutual dependence, and analyze its relevance to the operation status of the power system to obtain the power key feature set; S213: Call the power key feature set, reformat the feature representation, and calculate the feature balance degree according to the standardized mean and feature deviation using the formula: Calculate the feature balance degree , and adjust the feature normalization parameter to obtain the power feature balanced configuration, where represents the th power feature value, represents the mean of feature , represents the standard deviation of feature , represents the total number of feature data points.

5. The rapid construction method of the power data analysis model based on big data according to claim 1, characterized in that The specific steps for obtaining the dynamic index of power growth trend are as follows: S311: Based on the balanced configuration of the power characteristics, analyze the power load curve, calculate the power change rate within the differential time window, continuously detect the power change value, calculate the average change rate of power at time intervals, and generate power change data; S312: Based on the power change data, calculate the rate increment value, set the dynamic screening threshold, screen the time intervals where the rate increment exceeds the threshold, and obtain the power demand growth interval set; S313: Call the power demand growth interval set, analyze the fluctuation offset of the target time interval, adjust the weight of the power input data, match the current power usage trend, and use the formula: ; Obtain the dynamic index of the electricity growth trend , where represents the change in electricity input data represents the total number of electricity data points within the time interval represents the weight coefficient of the electricity input data represents the fluctuation offset of the electricity load curve within the target time interval represents the total number of data points in the fluctuation analysis represents the adjustment factor of the offset trend 6. The rapid construction method of the power data analysis model based on big data according to claim 1, characterized in that, The specific steps for obtaining the supply-demand balance evaluation data are as follows: S411: Based on the dynamic index of the power growth trend, analyze the load parameters of the transmission line, normalize the power transmission ratio of each group of transmission lines, and obtain the transmission line load ratio; S412: Based on the transmission line load ratio, combined with the power supply capacity parameters of the power grid partition, calculate the supply-demand deviation value of each partition, and judge the current power supply state to obtain the supply-demand deviation of the power grid partition; S413: Based on the power grid partition supply-demand deviation amount, determine the supply-demand prediction error, optimize the analysis parameters, and use the formula: ; Calculate the supply-demand balance error value , and compare the adjusted predicted data with the current data to obtain the supply-demand balance evaluation data, where represents the actual load demand of the th power grid sub-region, represents the power supply capacity of the th power grid sub-region, represents the total number of power grid sub-regions.

7. The rapid construction method of the power data analysis model based on big data according to claim 1, characterized in that The specific steps for obtaining the power analysis deployment ready model are as follows: S511: Call the supply-demand balance evaluation data, evaluate the power demand level, obtain the power generation capacity of the power generation unit, and analyze its matching degree with the demand level to obtain the power supply-demand matching degree; S512: Based on the power supply-demand matching degree, combined with the power generation cost data of the power generation units, calculate the unit power generation cost, and screen the power generation unit with the optimal power generation cost, using the formula: ; Calculate the unit efficiency coefficient , where represents the power generation of the th power generation unit, represents the power generation cost of the th power generation unit, represents the power demand in the th time period, represents the total number of all time periods; S513: Based on the unit efficiency coefficient, combined with the selected power generation units, create an executable model file according to the real-time load and power prediction data of the power grid, and obtain the power analysis deployment ready model.

Images