Power system monitoring data analysis method and system based on big data

Abstract

The invention discloses an electric power system monitoring data analysis method and system based on big data, relates to the technical field of electric power system monitoring, ensures comprehensiveness and accuracy of data through real-time and multi-dimensional data acquisition, and provides a solid foundation for subsequent data processing and analysis. According to the method, the problems of data lag and incomplete information in the traditional data acquisition process are effectively solved, the timeliness and accuracy of the data are improved, and the load dispatching optimization module dispatches power to an area with a high load value according to the comparison of a load predicted value LT and a load critical value LLT. And intelligent scheduling of electric power resources is realized. The waste of electric power resources is effectively avoided, the electric power demand of a high-load area is ensured to be preferably met, and the efficiency and fairness of electric power distribution are improved. And the real-time adjustment module adjusts the power dispatching scheme according to the actual load prediction value LT to ensure dynamic optimal distribution of power resources.

Description

Technical Field

[0001] The present invention relates to the technical field of power system monitoring, and in particular to a power system monitoring data analysis method and system based on big data. Background Art

[0002] The combination of big data technology and artificial intelligence is gradually changing the way power systems are monitored and managed. In the specific field of smart grids, power load forecasting has become one of the key tasks. Power load forecasting involves accurately predicting power demand to optimize power distribution and scheduling. Specifically at the application level of power load forecasting, the dynamic load forecasting model can predict changes in power demand under different time and weather conditions by analyzing power load data in real time, thereby optimizing load scheduling and ensuring efficient use of power resources.

[0003] Traditional load forecasting methods usually rely on static data and fixed models, which are incapable of dealing with dynamically changing load demands. In particular, the volatility of power demand increases under different time periods and weather conditions, making it difficult for traditional models to adapt to actual conditions.

[0004] Existing forecasting models often fail to update and adjust in real time, resulting in the inability of power dispatch plans to respond to changes in demand in a timely manner. In addition, many systems have problems with low computational efficiency and insufficient accuracy when processing large-scale data, and are unable to fully utilize real-time data for accurate forecasting and dispatch optimization. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a method and system for analyzing power system monitoring data based on big data, which solves the problems mentioned in the background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a power system monitoring data analysis system based on big data, wherein the data acquisition module collects data information of the power system, including power load data, weather data and time information data, to form a first data set X C ;

[0007] The data processing module collects the first data set X C Perform cleaning and normalization to generate data set X;

[0008] The data training module trains the dynamic load forecasting model, reshapes the data set X through the reshape function reshape(), and obtains the reshaped data set X R , use the reshaped data set X R Train the LSTM model;

[0009] The load prediction module predicts the future power load based on the LSTM model trained by the data training module, and reshapes the data set X R Make predictions and generate future data X F , the load forecast value L for the future time period is predicted by the LSTM model T ;

[0010] The load scheduling optimization module calculates the load forecast value L according to the load forecast module. T and load critical value L LT进行 By comparison, we can determine the area with high load value and dispatch the power;

[0011] The real-time adjustment module is based on the load forecast value L predicted by the load scheduling optimization module. T Dispatch electricity, adjust the dispatch plan in real time, and dispatch electricity to areas where the load forecast value LT is greater than the load critical value LLT.

[0012] Preferably, the data acquisition module includes a data acquisition unit;

[0013] The data collection unit is responsible for collecting power load data, weather data and time information data in real time, and integrating them into a first data set X C Among them, the power load data is obtained from the power monitoring system, including the load at different time points, the weather data is obtained through the meteorological data source, including temperature, humidity and wind speed, and the time information data is obtained from the recorded timestamp.

[0014] Preferably, the data processing module includes a data cleaning unit and a data normalization processing unit;

[0015] The data cleaning unit is responsible for collecting the first data set X C Perform cleaning to handle missing values ​​and outliers;

[0016] Cleaning involves using the Pandas library to perform C Perform data cleaning;

[0017] In the formula, X fi Indicates the data after filling, X i Represents a valid data point, specifically the first data set X C Including power load data, weather data and time information data, n represents the number of valid data points;

[0018] In the formula, Z represents the standardized value, and σ represents the first data set X C The standard deviation of the first data set X C The mean of

[0019] The data normalization processing unit cleans and processes the first data set X with missing values ​​and abnormal values. C Normalize the data to a standard range to facilitate model training and comparison;

[0020] In the formula, X min Represents the first data set X C The valley value, X max Represents the first data set X C The peak value of

[0021] The first data set X after data normalization C , integrated into the data set X.

[0022] Preferably, the data training module includes a data preparation unit and a model training unit;

[0023] The data preparation unit converts the data set X into data input to the LSTM model, reshapes the data set X, and obtains the reshaped data set X. R :

[0024] The reshaped data set X R By reshaping the function to the number of samples n s , time step t s and the number of features n fe Reshape acquisition;

[0025] Generate time series samples using sliding windows;

[0026] The model training unit is responsible for training the LSTM model for load forecasting, including model construction, training and evaluation: the LSTM model is constructed through the hidden state of the time step and the input of the current time step, and the weight matrix of the forget gate, input gate and candidate state is obtained, and the weight matrix of the forget gate, input gate and candidate state is activated using an activation function.

[0027] Preferably, the load forecasting module includes a data input unit and a load forecasting unit;

[0028] The data input unit will reshape the data set X R Make predictions and generate future data X F :

[0029] The future data X F By reshaping the data set X R Get, Future DataX F Represents the predicted data, reflecting the changes in future data, and converting future data into F Input into the LSTM model, it can judge and predict future situations;

[0030] The load forecasting unit uses the trained LSTM model to predict the input future data X F Carry out load forecasting and obtain the load forecast value L for the future time period T ;

[0031] The load forecast value L in the future time period T Obtained through the weight matrix of the LSTM model and the internal transformation matrix of the LSTM, the future data X is processed using the gated recurrent unit F Processing, and then through the internal transformation matrix and weight matrix to the future data X F Perform permutations and combinations to finally obtain the load forecast value L for the future time period T .

[0032] Specifically, through the prediction function of the LSTM model, we can accurately predict future power load conditions and support more reasonable power dispatch. By predicting future loads, we can adjust the power distribution strategy in real time, give priority to meeting the needs of high-load areas, and optimize resource utilization.

[0033] Preferably, the load dispatch optimization module includes a load comparison unit and a dispatch optimization unit;

[0034] The load comparison unit compares the load prediction value L T and load critical value L LT , determine the area for dispatching power; the load forecast value L T and load critical value L LT The comparison is obtained by:

[0035] When the load forecast value of area i is L Ti >Load critical value L LT When , it means that area i needs to be dispatched;

[0036] When the load forecast value of area i is L Ti <Load critical value L LT When , it means that area i does not need to be scheduled;

[0037] Fit the area i that needs to be scheduled and obtain the scheduling priority Priority of area i i ;

[0038] For all Priorities i Arrange and generate a priority list of all areas LIST ;

[0039] When compared with the load forecast value L Ti >Load critical value L LT , then the power is dispatched to area i first;

[0040] When compared with the load forecast value L Ti ≤Load critical value L LT , then assigned to the priority list Priority LIST Waiting for processing.

[0041] Preferably, the dispatch optimization unit includes power distribution optimization and power dispatch planning;

[0042] The method of power distribution optimization is as follows:

[0043] When the priority in area i is equal to 1 and the peak power demand P in area i max The total available power P in the system to If the power is within the allowed range, the power is allocated to area i;

[0044] When the priority in area i is not equal to 1, no power is allocated;

[0045] When the priority in region i is equal to 1 and the peak power demand P in region i max Greater than the total available power P to , no power is allocated.

[0046] The method of the power dispatch plan is as follows:

[0047] Within the permitted range of the power dispatch plan, power is dispatched in sequence from low to high.

[0048] Preferably, the real-time adjustment module includes a real-time scheduling unit and a scheduling feedback unit;

[0049] The real-time adjustment module includes a real-time scheduling unit and a scheduling feedback unit;

[0050] The real-time dispatching unit dispatches the power according to the power dispatching plan, and calculates the power allocation amount P of area i. i Make the actual allocation;

[0051] Specifically, electricity is distributed through the distribution function F(.), and the electricity consumption of each area is distributed according to the scheduling priority of each area.

[0052] Preferably, the dispatch feedback unit monitors and records the actual power distribution, adjusts the dispatch plan according to the actual situation, and records the actual power distribution amount P i,Allocate :

[0053] Actual power distribution P i,Allocate represents the actual amount of electricity allocated to area i and is related to the power allocation amount P of area i. i Compare: Get the actual power distribution Pi,Allocate The power distribution amount P of area i i The difference DCZ i ;

[0054] When the difference DCZ i =0, indicating that the power distribution is normal, the system maintains the existing dispatching plan, records and monitors;

[0055] When the difference DCZ i >0, indicating excess power. The system needs to adjust the scheduling plan to avoid wasting resources and modify the power scheduling plan to allocate the remaining power to other high-priority areas. Adjusting the scheduling plan includes adjusting the priority list Priority LIST , set Priority 1 Priority 2 Make adjustments and scheduling;

[0056] When the difference DCZ i <0, indicating insufficient power. The system adjusts the scheduling plan and adjusts the power supply to the area to meet the demand of area i.

[0057] By adjusting the power dispatch plan in real time, we can respond quickly to load changes, ensure the effective use of power resources, give priority to areas where load forecast values ​​exceed critical values, improve the accuracy and effectiveness of power distribution, and continuously optimize the power dispatch strategy through feedback from actual power distribution conditions to improve the overall operating efficiency and stability of the system.

[0058] A method for analyzing power system monitoring data based on big data comprises the following steps:

[0059] Step 1: The data acquisition module collects data information of the power system, including power load data, weather data and time information data, to form the first data set X C ;

[0060] Step 2: The data processing module collects the first data set X C Perform cleaning and normalization to generate data set X;

[0061] Step 3: The data training module trains the dynamic load forecasting model. The data set X is reshaped by the reshape function reshape() to obtain the reshaped data set X. R , use the reshaped data set X R Train the LSTM model;

[0062] Step 4: The load forecasting module predicts the future power load based on the LSTM model trained by the data training module, and reshapes the data set X R Make predictions and generate future data X F, the load forecast value L for the future time period is predicted by the LSTM model T ;

[0063] Step 5: The load scheduling optimization module calculates the load forecast value L according to the load forecast module T and load critical value L LT Compare and determine the area with high load value and dispatch electricity;

[0064] Step 6: The real-time adjustment module adjusts the load forecast value L according to the load scheduling optimization module T Dispatch power, adjust the dispatch plan in real time, and dispatch power to the load forecast value L T Greater than the load critical value L LT area.

[0065] The present invention provides a method and system for analyzing power system monitoring data based on big data, which has the following beneficial effects:

[0066] (1) An LSTM dynamic load forecasting model is introduced, which is able to handle the time dependency and change patterns in power load data. This model can predict power demand in real time, taking into account load fluctuations under different time and weather conditions. This improves the accuracy and timeliness of load forecasting, enabling the power system to better cope with load changes and emergencies.

[0067] The load dispatch optimization module is based on the load forecast value L T and load critical value L LT The real-time adjustment module adjusts the load according to the actual load forecast value L. T Adjust the power dispatch plan to ensure dynamic and optimal allocation of power resources.

[0068] It improves the system's responsiveness to sudden load changes, reduces the risk of power shortage or surplus due to forecast errors, and enhances the system's flexibility and stability.

[0069] (2) Through real-time and multi-dimensional data collection, the comprehensiveness and accuracy of the data are ensured, providing a solid foundation for subsequent data processing and analysis. This method effectively solves the problems of data lag and incomplete information in the traditional data collection process, improves the timeliness and accuracy of the data, and adapts to the changing needs of the dynamic power system. Data cleaning and normalization ensure the quality and consistency of the data and reduce the impact of data noise and outliers on model training. This process significantly improves the reliability of the data, allowing the model to use cleaner and more standardized data during training and prediction, improving the accuracy of the prediction results and the generalization ability of the model.

[0070] (3) Through LSTM model training and load forecasting methods, accurate prediction and optimized scheduling of power load are achieved. In the data training module, the data preparation unit ensures the time series characteristics of the data by reshaping the data group X, providing a reliable input format for the LSTM model. The model training unit uses a detailed LSTM construction formula to conduct in-depth learning and training of power load, thereby improving the prediction accuracy.

[0071] The load forecasting module further uses the trained model to predict future data, thereby accurately predicting the power load in future time periods.

[0072] (4) By introducing high-precision load forecasting and real-time adjustment mechanisms, the optimal configuration and dynamic adjustment of power resources are achieved. This method significantly improves the power system's response speed to load changes and resource utilization efficiency, and reduces uneven power supply and resource waste caused by forecast errors. The system can better adapt to load changes under different time and weather conditions, realize the precision and intelligence of power dispatch, and improve the overall performance and stability of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 A block diagram of a power system monitoring data analysis system based on big data according to the present invention;

[0074] Figure 2 The present invention is a schematic diagram of the steps of a method for analyzing power system monitoring data based on big data. DETAILED DESCRIPTION

[0075] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0076] Example 1

[0077] The present invention provides a method and system for analyzing power system monitoring data based on big data. Figure 1 , including data acquisition module, data processing module, data training module, load forecasting module, load dispatch optimization module and real-time adjustment module;

[0078] The data acquisition module collects data information of the power system, including power load data, weather data and time information data, to form a first data set X C ;

[0079] The data processing module collects the first data set X C Perform cleaning and normalization to generate data set X;

[0080] The data training module trains the dynamic load forecasting model, reshapes the data set X through the reshape function reshape(), and obtains the reshaped data set X R , use the reshaped data set X R Train the LSTM model;

[0081] The load prediction module predicts the future power load based on the LSTM model trained by the data training module, and reshapes the data set X R Make predictions and generate future data X F , the load forecast value L for the future time period is predicted by the LSTM model T ;

[0082] The load scheduling optimization module calculates the load forecast value L according to the load forecast module. T and load critical value L LT Compare and determine the area with high load value and dispatch electricity;

[0083] The real-time adjustment module is based on the load forecast value L predicted by the load scheduling optimization module. T Dispatch electricity, adjust the dispatch plan in real time, and dispatch electricity to areas where the load forecast value LT is greater than the load critical value LLT.

[0084] In this embodiment, an LSTM dynamic load forecasting model is introduced to handle the time dependency and change pattern in power load data. This model can predict power demand in real time, taking into account load fluctuations under different time and weather conditions. It improves the accuracy and timeliness of load forecasting, enabling the power system to better cope with load changes and emergencies.

[0085] The load dispatch optimization module is based on the load forecast value L T and load critical value L LTThe real-time adjustment module adjusts the load according to the actual load forecast value L. T Adjust the power dispatch plan to ensure dynamic and optimal allocation of power resources.

[0086] It improves the system's responsiveness to sudden load changes, reduces the risk of power shortage or surplus due to forecast errors, and enhances the system's flexibility and stability.

[0087] Example 2

[0088] This embodiment is explained in Example 1, please refer to Figure 1 ,Specifically: the data acquisition module includes a data acquisition unit;

[0089] The data collection unit is responsible for collecting power load data, weather data and time information data in real time, and integrating them into a first data set X C ; Among them, the power load data is obtained from the power monitoring system, including the load at different time points, the weather data is obtained through the meteorological data source, including temperature, humidity and wind speed, and the time information data is obtained from the recorded timestamp; the data processing module includes a data cleaning unit and a data normalization processing unit;

[0090] The data cleaning unit is responsible for collecting the first data set X C Perform cleaning to handle missing values ​​and outliers;

[0091] Cleaning involves using the Pandas library to perform C Perform data cleaning;

[0092] The formula for handling missing values ​​is as follows:

[0093]

[0094] In the formula, X fi Indicates the data after filling, X i Represents a valid data point, specifically the first data set X C Including power load data, weather data and time information data, n represents the number of valid data points;

[0095] The formula for processing outliers is as follows:

[0096]

[0097] In the formula, Z represents the standardized value, and σ represents the first data set XC The standard deviation of the first data set X C The mean of

[0098] The data normalization processing unit cleans and processes the first data set X with missing values ​​and abnormal values. C Normalize the data to a standard range to facilitate model training and comparison;

[0099] The formula for data normalization is as follows:

[0100]

[0101] In the formula, X min Represents the first data set X C The valley value, X max Represents the first data set X C The peak value of

[0102] The first data set X after data normalization C , integrated into the data set X.

[0103] In this embodiment, through real-time and multi-dimensional data collection, the comprehensiveness and accuracy of the data are ensured, providing a solid foundation for subsequent data processing and analysis. This method effectively solves the problems of data lag and incomplete information in the traditional data collection process, improves the timeliness and accuracy of the data, adapts to the demand changes of the dynamic power system, and data cleaning and normalization ensure the quality and consistency of the data, reducing the impact of data noise and outliers on model training. This process significantly improves the reliability of the data, allowing the model to use cleaner and more standardized data during training and prediction, improving the accuracy of the prediction results and the generalization ability of the model.

[0104] The data acquisition module ensures the comprehensiveness and timeliness of the data by acquiring power load data, weather data and time information in real time. The data processing module cleans and normalizes the data, effectively eliminating the interference of missing values ​​and outliers and improving the quality of the data. Data normalization scales the data to a standard range, improving the effect of model training. Combining these measures not only improves the accuracy of power load forecasting, but also optimizes the real-time adjustment capability of the power dispatching plan, significantly improving the overall efficiency and stability of the power system.

[0105] Example 3

[0106] This embodiment is explained in Example 2. Please refer to Figure 1 ,Specifically: the data training module includes a data preparation unit and a model training unit;

[0107] The data preparation unit converts the data set X into data input to the LSTM model, reshapes the data set X, and obtains the reshaped data set X. R :

[0108] The reshaped data set X R Obtained by the following formula:

[0109] X R = reshape(X, shape = (n s ,t s ,n fe ));

[0110] In the formula, reshape represents the reshaping function, n s represents the sample size, t s represents the time step, n fe Indicates the number of features;

[0111] Generate time series samples using a sliding window:

[0112] Windowed X ={X G [i:i+t s ]i∈[0,n s -t s ]};

[0113] In the formula, Windowed X Represents a time series sample generated based on the time step;

[0114] The model training unit is responsible for training the LSTM model for load forecasting, including model construction, training and evaluation: The LSTM model construction formula is as follows:

[0115]

[0116] In the formula, h t-1 represents the hidden state of the previous time step, X t represents the input of the current time step, W f , W i , W c Represents the weight matrices of the forget gate, input gate and candidate state, respectively, b f , b i , b c Respectively represent the bias items of the forget gate, the input gate and the candidate state, f, sigmoid and tanh represent activation functions, specifically including Swish, HardSigmoid and PReLU activation functions; the load forecasting module includes a data input unit and a load forecasting unit;

[0117] The data input unit will reshape the data set X R Make predictions and generate future data X F :

[0118] The future data X F Obtained by the following formula:

[0119]

[0120] In the formula, Represents the normalization factor, which is used to adjust the impact range, X Rmax Represents the reshaped data set X R The peak value, X Rmin Represents the reshaped data set X R The valley value of , γ represents a constant parameter;

[0121] The load forecasting unit uses the trained LSTM model to predict the input future data X F Carry out load forecasting and obtain the load forecast value L for the future time period T ;

[0122] The load forecast value L T Obtained by the following formula:

[0123]

[0124] Where W LS Represents the weight matrix of the LSTM model, V LS represents the internal transformation matrix of LSTM, U LS represents the transformation matrix for input data, GRU(.) represents the gated recurrent unit, specifically including the dependencies of time series samples in processing sequence data, B LS Represents the bias vector of the LSTM model, X F Represents future data.

[0125] In this embodiment, accurate prediction and optimized scheduling of power load are achieved through LSTM model training and load forecasting method. In the data training module, the data preparation unit ensures the time series characteristics of the data by reshaping the data group X, providing a reliable input format for the LSTM model. The model training unit conducts in-depth learning and training of power load through detailed LSTM construction formulas, thereby improving the prediction accuracy.

[0126] The load forecasting module further uses the trained model to predict future data, thereby accurately predicting the power load in the future time period. This embodiment effectively solves the problems of insufficient load forecasting accuracy and untimely scheduling caused by improper data processing in traditional methods, improves the dynamic scheduling capability and resource allocation efficiency of the power system, thereby ensuring the stability and reliability of power supply, optimizing the overall performance of the power system, and meeting the demand changes in complex power environments.

[0127] Example 4

[0128] This embodiment is explained in Example 3, please refer to Figure 1 ,Specifically: the load scheduling optimization module includes a load comparison unit and a scheduling optimization unit;

[0129] The load comparison unit compares the load prediction value L T and load critical value L LT , determine the area for dispatching power; load forecast value L T and load critical value L LT The comparison formula is as follows:

[0130]

[0131] In the formula, Priority i represents the scheduling priority of area i, L Ti represents the load forecast value of area i;

[0132] For all Priorities i Arrange and generate a priority list of all areas LIST ;

[0133] Priority LIST Obtained by the following formula:

[0134] Priority LIST ={Priority 1 ,Priority 2 ,...,Priority m};

[0135] In the formula, m represents the number of regions;

[0136] If we compare the load forecast value L Ti >Load critical value L LT , then the power is dispatched to area i first;

[0137] If we compare the load forecast value L Ti ≤ load critical value L LT, then assigned to the priority list Priority LIST Waiting for processing.

[0138] Specifically, the load comparison unit calculates the load prediction value L Ti and load critical value L LT Compare and accurately determine the priority of power dispatch. This process ensures that high-load areas get power supply first, effectively avoiding waste and over-allocation of power resources.

[0139] The dispatch optimization unit includes power distribution optimization and power dispatch planning;

[0140] The calculation formula for power distribution optimization is as follows:

[0141]

[0142] Where P i represents the power distribution in area i, P max represents the peak power demand in the region, P to Indicates the total available power of the system. If Priority=1andavailable P to allows means that if the priority is equal to 1, power is allowed to be obtained;

[0143] The calculation formula of the power dispatch plan is as follows:

[0144] SPdPlan={P 1 ,P 2 ,...,P n};

[0145] Where, SPdPlan represents the power dispatch plan;

[0146] Specifically, the dispatch optimization unit uses the power distribution optimization formula and the power dispatch plan calculation formula to reasonably allocate regional power distribution to ensure that each region can still meet its needs under the condition of limited power resources. The optimized power distribution plan can balance the power demand of different regions, avoid uneven power supply caused by insufficient or excessive power, and ensure the stability of the system and the reliability of power supply.

[0147] The real-time adjustment module includes a real-time scheduling unit and a scheduling feedback unit;

[0148] The real-time dispatching unit dispatches the power according to the power dispatching plan, and calculates the power allocation amount P of area i. i Make the actual allocation;

[0149] Power distribution through the distribution function F(.)

[0150] P i,Allocate =F(P i , Priority i );

[0151] Where P i,Allocate Indicates the amount of electricity actually allocated to area i, Priority i represents the scheduling priority of area i;

[0152] The dispatch feedback unit monitors and records the actual power distribution, adjusts the dispatch plan according to the actual situation, and records the actual power distribution amount P i,Allocate :

[0153] P i,Allocate =actual power allocate to based on area i;

[0154] In the formula, actual power allocated to based area i represents the actual power allocated to area i;

[0155] The power distribution amount P of area i i For comparison:

[0156] DCZ i =P i,Allocate -P i ;

[0157] Where DCZ i Indicates the actual power distribution P i,Allocate The power distribution amount P of area i i The difference between

[0158] When the difference DCZ i =0, indicating that the power distribution is normal, the system maintains the existing dispatching plan, records and monitors;

[0159] When the difference DCZ i >0, indicating excess power. The system needs to adjust the scheduling plan to avoid wasting resources and modify the power scheduling plan to allocate the remaining power to other high-priority areas. Adjusting the scheduling plan includes adjusting the priority list Priority LIST , set Priority 1 Priority 2 Make adjustments and scheduling;

[0160] When the difference DCZ i <0, indicating insufficient power. The system adjusts the scheduling plan and adjusts the power supply to the area to meet the demand of area i.

[0161] In this embodiment, through accurate priority sorting, power resources can be more reasonably allocated to areas with high demand, thereby optimizing the overall efficiency of power supply and improving the system's response speed to sudden load demands. This embodiment not only improves the accuracy of power forecasting, but also significantly enhances the dynamic dispatching capability of the power system. It enables power resources to be quickly adjusted according to actual demand, reducing resource waste or insufficient supply caused by power demand forecast deviations, thereby optimizing the overall performance and resource utilization efficiency of the power system.

[0162] Example 5

[0163] A method and system for analyzing power system monitoring data based on big data, please refer to Figure 2 , specifically: including the following steps:

[0164] Step 1: The data acquisition module collects data information of the power system, including power load data, weather data and time information data, to form the first data set X C ;

[0165] Step 2: The data processing module collects the first data set X C Perform cleaning and normalization to generate data set X;

[0166] Step 3: The data training module trains the dynamic load forecasting model. The data set X is reshaped by the reshape function reshape() to obtain the reshaped data set X. R , use the reshaped data set X R Train the LSTM model;

[0167] Step 4: The load forecasting module predicts the future power load based on the LSTM model trained by the data training module, and reshapes the data set X R Make predictions and generate future data X F , the load forecast value L for the future time period is predicted by the LSTM model T ;

[0168] Step 5: The load scheduling optimization module calculates the load forecast value L according to the load forecast module T and load critical value L LT Compare and determine the area with high load value and dispatch electricity;

[0169] Step 6: The real-time adjustment module adjusts the load forecast value L according to the load scheduling optimization module T Dispatch power, adjust the dispatch plan in real time, and dispatch power to the load forecast value L T Greater than the load critical value L LT area.

[0170] In this embodiment, steps 1 to 6 realize accurate dynamic load forecasting and optimized scheduling through comprehensive data collection, processing, training and prediction of the power system. The system includes a data collection module, a data processing module, a data training module, a load forecasting module, a load scheduling optimization module and a real-time adjustment module, and each module plays an important role in the overall architecture.

[0171] The data acquisition module is responsible for collecting power load data, weather data and time information data to form the first data set X C , providing a comprehensive data foundation for subsequent analysis. The data processing module generates data set X by cleaning and normalizing the first data set, ensuring the quality and consistency of the data. The data training module reshapes the data set X and uses it to train the LSTM model, improving the model's ability to predict changes in power load. The load forecasting module accurately predicts future power loads based on the trained LSTM model and generates load forecast values ​​L for future time periods. T Then, the load dispatch optimization module converts the predicted value L T and load critical value L LT By comparing the loads, the power dispatching strategy is optimized to prioritize power distribution to areas with high demand. Finally, the real-time adjustment module adjusts the power dispatching plan in real time according to the predicted load value, ensuring the timeliness and effectiveness of power supply.

[0172] This embodiment achieves optimal configuration and dynamic adjustment of power resources by introducing high-precision load forecasting and real-time adjustment mechanisms. This method significantly improves the power system's response speed to load changes and resource utilization efficiency, and reduces uneven power supply and resource waste caused by forecast errors. The system can better adapt to load changes under different time and weather conditions, realizes precise and intelligent power dispatching, and improves the overall performance and stability of the power system.

[0173] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A power system monitoring data analysis system based on big data, characterized by: It includes data acquisition module, data processing module, data training module, load forecasting module, load dispatch optimization module and real-time adjustment module; The data acquisition module collects data information of the power system, including power load data, weather data and time information data, to form a first data set X C ; The data processing module collects the first data set X C Perform cleaning and normalization to generate data set X; The data training module trains the dynamic load forecasting model, reshapes the data set X through the reshape function reshape(), and obtains the reshaped data set X R , use the reshaped data set X R Train the LSTM model; The load prediction module predicts the future power load based on the LSTM model trained by the data training module, and reshapes the data set X R Make predictions and generate future data X F , the load forecast value L for the future time period is predicted by the LSTM model T ; The load scheduling optimization module calculates the load forecast value L according to the load forecast module. T and load critical value L LT Compare and determine the area with high load value and dispatch electricity; The real-time adjustment module is based on the load forecast value L predicted by the load scheduling optimization module. T Dispatch power, adjust the dispatch plan in real time, and dispatch power to the load forecast value L T Greater than the load critical value L LT area.

2. The power system monitoring data analysis system based on big data according to claim 1, characterized in that: The data acquisition module includes a data acquisition unit; The data collection unit is responsible for collecting power load data, weather data and time information data in real time, and integrating them into a first data set X C Among them, the power load data is obtained from the power monitoring system, including the load at different time points, the weather data is obtained through the meteorological data source, including temperature, humidity and wind speed, and the time information data is obtained from the recorded timestamp.

3. The power system monitoring data analysis system based on big data according to claim 1 is characterized in that: The data processing module includes a data cleaning unit and a data normalization processing unit; The data cleaning unit is responsible for collecting the first data set X C Perform cleaning to handle missing values ​​and outliers; Cleaning involves using the Pandas library to perform C Perform data cleaning; The data normalization processing unit cleans and processes the first data set X with missing values ​​and abnormal values. C Normalize the data to a standard range; normalize the data to the first data set X C , integrated into the data set X.

4. The power system monitoring data analysis system based on big data according to claim 1 is characterized by: The data training module includes a data preparation unit and a model training unit; The data preparation unit converts the data set X into data input to the LSTM model, reshapes the data set X, and obtains the reshaped data set X. R : The reshaped data set X R By reshaping the function to the number of samples n s , time step t s and the number of features n fe Reshape acquisition; Generate time series samples using sliding windows; The model training unit is responsible for training the LSTM model for load forecasting, including model construction, training and evaluation: the LSTM model is constructed through the hidden state of the time step and the input of the current time step, and the weight matrix of the forget gate, input gate and candidate state is obtained, and the weight matrix of the forget gate, input gate and candidate state is activated using an activation function.

5. The power system monitoring data analysis system based on big data according to claim 4 is characterized in that: The load prediction module includes a data input unit and a load prediction unit; The data input unit will reshape the data set X R Make predictions and generate future data X F : The future data X F By reshaping the data set X R Get, Future DataX F Represents the predicted data, reflecting the changes in future data, and converting future data into F Input into the LSTM model, it can judge and predict future situations; The load forecasting unit uses the trained LSTM model to predict the input future data X F Carry out load forecasting and obtain the load forecast value L for the future time period T ; The load forecast value L for the future time period T Obtained through the weight matrix of the LSTM model and the internal transformation matrix of the LSTM, the future data X is processed using the gated recurrent unit F Processing, and then through the internal transformation matrix and weight matrix to the future data X F Perform permutations and combinations to finally obtain the load forecast value L for the future time period T .

6. The power system monitoring data analysis system based on big data according to claim 1, characterized in that: The load scheduling optimization module includes a load comparison unit and a scheduling optimization unit; The load comparison unit compares the load prediction value L T and load critical value L LT , determine the area for dispatching power; the load forecast value L T and load critical value L LT The comparison is obtained by: When the load forecast value of area i is L Ti >Load critical value L LT When , it means that area i needs to be dispatched; When the load forecast value of area i is L Ti <Load critical value L LT When , it means that area i does not need to be scheduled; Fit the area i that needs to be scheduled and obtain the scheduling priority Priority of area i i ; For all Priorities i Arrange and generate a priority list of all areas LIST ; When compared with the load forecast value L Ti >Load critical value L LT , then the power is dispatched to area i first; When compared with the load forecast value L Ti ≤Load critical value L LT , then assigned to the priority list Priority LIST Waiting for processing.

7. The power system monitoring data analysis system based on big data according to claim 1, characterized in that: The dispatch optimization unit includes power distribution optimization and power dispatch planning; The method of power distribution optimization is as follows: When the priority in area i is equal to 1 and the peak power demand P in area i max The total available power P in the system to If the power is within the allowed range, the power is allocated to area i; When the priority in area i is not equal to 1, no power is allocated; When the priority in region i is equal to 1 and the peak power demand P in region i max Greater than the total available power P to , then no electricity is allocated; The method of the power dispatch plan is as follows: Within the permitted range of the power dispatch plan, power is dispatched in sequence from low to high.

8. The power system monitoring data analysis system based on big data according to claim 7 is characterized by: The real-time adjustment module includes a real-time scheduling unit and a scheduling feedback unit; The real-time dispatching unit dispatches the power according to the power dispatching plan, and calculates the power allocation amount P of area i. i Make the actual allocation; Specifically, electricity is distributed through the distribution function F(.), and the electricity consumption of each area is distributed according to the scheduling priority of each area.

9. The power system monitoring data analysis system based on big data according to claim 8, characterized in that: The dispatch feedback unit monitors and records the actual power distribution, adjusts the dispatch plan according to the actual situation, and records the actual power distribution amount P i,Allocate : Actual power distribution P i,Allocate represents the actual amount of electricity allocated to area i and is related to the power allocation amount P of area i. i Compare: Get the actual power distribution P i,Allocate The power distribution amount P of area i i The difference DCZ i ; When the difference DCZ i =0, indicating that the power distribution is normal, the system maintains the existing dispatching plan, records and monitors; When the difference DCZ i >0, indicating excess power, the system adjusts the dispatching plan to avoid wasting resources and modifies the power dispatching plan to allocate the remaining power to other high-priority areas; Adjusting the scheduling plan includes adjusting the priority list Priority LIST , adjust and schedule Priority1 to Priority2; When the difference DCZ i <0, indicating insufficient power. The system adjusts the scheduling plan and adjusts the power supply to the area to meet the demand of area i.

10. A method for analyzing power system monitoring data based on big data, applied to a power system monitoring data analysis system based on big data as claimed in any one of claims 1 to 9, characterized in that: The following steps are involved: Step 1: The data acquisition module collects data information of the power system, including power load data, weather data and time information data, to form the first data set X C ; Step 2: The data processing module collects the first data set X C Perform cleaning and normalization to generate data set X; Step 3: The data training module trains the dynamic load forecasting model. The data set X is reshaped by the reshape function reshape() to obtain the reshaped data set X. R , use the reshaped data set X R Train the LSTM model; Step 4: The load forecasting module predicts the future power load based on the LSTM model trained by the data training module, and reshapes the data set X R Make predictions and generate future data X F , the load forecast value L for the future time period is predicted by the LSTM model T ; Step 5: The load scheduling optimization module calculates the load forecast value L according to the load forecast module T and load critical value L LT Compare and determine the area with high load value and dispatch electricity; Step 6: The real-time adjustment module adjusts the load forecast value L according to the load scheduling optimization module T Dispatch power, adjust the dispatch plan in real time, and dispatch power to the load forecast value L T Greater than the load critical value L LT area.

Images