Power system load prediction and scheduling method based on optimization algorithm

Abstract

The invention discloses a power system load prediction and scheduling method based on an optimization algorithm, and particularly relates to the field of power engineering, and the method comprises the steps: S1, monitoring a power system in real time, S2, preprocessing a first power load, S3, analyzing a first power key data set corresponding to the first power load, and obtaining a first load demand, S4, analyzing the first load demand, and S5, obtaining a second power key data set corresponding to the first power load; s5, obtaining a second load demand; S6, obtaining a power load adjustment model; S7, optimizing a power system scheduling strategy module; and S8, outputting a power system scheduling strategy, monitoring the performance of the power system, and performing regular maintenance. According to the power system load prediction and scheduling method based on the optimization algorithm, the first load demand is calculated through the optimization algorithm, the calculation efficiency of a model is improved, and the deviation between a prediction result and an actual load is reduced; the influence of prediction deviation on a scheduling decision is reduced by adjusting a load prediction and scheduling strategy in time according to actual conditions.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power engineering, and more specifically, to a method for predicting and dispatching load of an electric power system based on an optimization algorithm. Background Art

[0002] With the transformation of the global energy structure and the rapid development of smart grid technology, load forecasting and dispatching of power systems have become increasingly important. As a key energy form supporting industrial production, commercial operations and residents' lives, the stable supply and efficient use of electricity have become the cornerstone for ensuring the sustainable development of the social economy. Accurate load forecasting and efficient dispatching strategies are the core to ensure the stable operation of the power system, optimize the operation mode of the power grid, avoid insufficient or excessive power supply, and thus improve the reliability and economy of the power system.

[0003] Traditional load forecasting methods are difficult to cope with complex and changeable power load demands due to limited data volume and insufficient computing power. Combined with big data and artificial intelligence technologies, power system load forecasting and scheduling methods can significantly improve the operating efficiency and reliability of the power system. By using big data technology and artificial intelligence for load forecasting, it helps to improve the stability and reliability of the power grid, optimize the scheduling method to balance power supply and demand, reduce the peak-to-valley difference of the power grid, reduce the risk of overload of transmission lines, and avoid power outages caused by excessive load fluctuations.

[0004] However, there are still some shortcomings in its actual use. For example, the scheduling strategy fails to fully integrate emerging loads, resulting in inefficient resource allocation; data is missing, erroneous or inconsistent, affecting the accuracy and reliability of the prediction model; the existing load forecasting and scheduling models are not adaptable enough when faced with rapidly changing operating conditions; the update and adjustment of the model often requires a certain amount of time and computing resources, and it is difficult to respond to new situations in a timely and accurate manner in a short period of time, resulting in deviations between the prediction results and the actual load, which in turn affects the effectiveness of the scheduling decision; the current load forecasting and scheduling technology is insufficient in considering the coordinated operation of multiple energy systems, and lacks an effective information interaction and coordination mechanism across energy systems. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method and system for power system load forecasting and scheduling based on an optimization algorithm, and solves the problems raised in the above-mentioned background technology through the following scheme.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for power system load forecasting and scheduling based on an optimization algorithm, characterized by comprising:

[0008] S1: Real-time monitoring of the power system: real-time data collection and monitoring of the power system are performed through sensors and monitoring equipment to obtain the first power load;

[0009] S2: preprocessing the first power load: performing a preprocessing operation on the first power load, the preprocessing operation is used to obtain a first power key data group corresponding to the first power load;

[0010] S3: Analyze the first power key data group corresponding to the first power load and obtain the first load demand: calculate the first load demand based on the first power key data group corresponding to the first power load, and the first load demand is used to express the operation efficiency of the power system dispatch;

[0011] S4: Analyze the first load demand and obtain a power load demand analysis report: Analyze the resource allocation in the power system based on the first load demand, identify potential resource waste and unreasonable allocation problems in the power system, and obtain a power load demand analysis report;

[0012] S5: Obtaining a second load demand: Obtaining a second power load and a load prediction model, and obtaining a second load demand corresponding to the second power load through the second power load according to the load prediction model;

[0013] S6: Obtaining a power load adjustment model: Obtaining a power load adjustment model, and dynamically adjusting parameters related to the power load demand analysis report according to the power load adjustment model through the second load demand to obtain a power load adjustment report;

[0014] S7: Optimizing the power system dispatching strategy module: Based on the real-time operation data corresponding to the first power key data group, the load forecast result corresponding to the second load demand, the power load demand analysis report, and the power load adjustment report, optimizing the allocation of power generation resources associated with the power system to obtain the power system dispatching strategy;

[0015] S8: Output power system dispatching strategy, monitor power system performance, and perform regular maintenance: Provide a power system interactive interface to set parameters of power equipment associated with the power system, view load forecast results and dispatching strategies, monitor power system performance in real time, and perform regular system maintenance and upgrades.

[0016] Preferably, the first power load in S1 includes real-time load data, power generation data, equipment status data, meteorological data, time data, and load conditions in the power grid.

[0017] Preferably, in said S2, a first power key data group corresponding to the first power load is obtained, wherein the preprocessing steps are as follows:

[0018] A1: Data cleaning: remove abnormal values ​​in the first power load that are obviously deviated from the normal range due to sensor failure, data transmission error, etc.;

[0019] A2: Missing value processing: For the missing values ​​in the first power load, the missing values ​​are supplemented according to the characteristics of the time series. The supplementation methods include using linear interpolation, nearest neighbor filling method, and mean filling method based on historical data of the same period;

[0020] A3: Data standardization: convert various physical quantity units in the first power load data into standard units, and perform normalization processing on the first power load so that the value range is between [-1, 1];

[0021] A4: Feature extraction: cluster analysis method is used to extract the key features of the first power load;

[0022] A5: Data dimensionality reduction: Use the principal component analysis method to calculate the covariance matrix of the extracted feature data, solve the eigenvalues ​​and eigenvectors, and perform dimensionality reduction.

[0023] Preferably, the method for obtaining the first load demand in S3 is as follows:

[0024] Based on the actual operation conditions and constraints of the power system, a load demand calculation model is constructed. The load demand calculation model is a pre-constructed learning model. By inputting the first power key data group into the load demand calculation model, the load demand calculation model obtains the first load demand according to the first power key data group.

[0025] Preferably, the steps of constructing the load demand calculation model in S3 are as follows:

[0026] B1: Sort out the correspondence between the first power load and the first power key data group, and construct a load demand training set containing various features, recorded as {(z1, y1), (z2, y2), (z3, y3), ..., (z n ,y n )}, where z i Represented as the i-th key feature in the first power key data group, y i is the first power load value corresponding to the key feature, i represents the sequence number of the key feature in the first power key data group, and n represents the total number of key features in the first power key data group;

[0027] B2: Select a training kernel function, where the training kernel function includes a linear kernel function, a polynomial kernel function, and a radial basis function kernel function;

[0028] B3: Determine the penalty parameter C and kernel function parameter γ in the load demand calculation model by cross-validation method;

[0029] B4: training the load demand calculation model to obtain the optimal parameters involved in the load demand calculation model;

[0030] B5: Obtain a load demand calculation model by inputting the first power key data group into the load demand calculation model, so that the load demand calculation model obtains the first load demand according to the first power key data group.

[0031] Preferably, the load demand calculation model in S3 is specifically expressed as:

[0032] EIR=ω T ×φ(z i )+b,

[0033] Where EIR is the first load demand, ω is the characteristic weight vector, and ω T Represented as the transposed vector of ω, z i It is represented as the i-th key feature in the first power key data group, φ(z i ) is represented by z through the kernel function i The feature vector mapped to the high-dimensional space, b is represented as the feature bias.

[0034] Preferably, obtaining the second load demand corresponding to the second power load in S5 includes:

[0035] Select a load forecasting model, which includes autoregressive integrated moving average model ARIMA, seasonal ARIMA, support vector regression SVR, random forest regression, and deep learning model;

[0036] The integrated second power load is input into the load forecasting model to perform load forecasting calculations and output the predicted value of future load, i.e., the second load demand. The second load demand includes the predicted value of power load in a future period of time, load change trend, and uncertainty information.

[0037] Preferably, the step of obtaining the power load adjustment report in S6 includes:

[0038] Input the acquired second load demand data into the power load adjustment model; dynamically adjust the power generation plan, load distribution, demand response strategy, and interruptible load arrangement according to the characteristics and scale of the power load adjustment model using an optimization algorithm;

[0039] According to the adjusted parameters and the model solution results, a power load adjustment report is generated, and the power load adjustment report includes the reason for the power load adjustment, the adjustment target, the second load demand, and the load adjustment plan.

[0040] Technical effects and advantages of the present invention:

[0041] 1. The present invention calculates the first load demand by reasonably selecting model parameters and optimizing algorithms, thereby improving the calculation efficiency of the model, being able to respond promptly and accurately to new situations in a short time, reducing the deviation between the prediction results and the actual load, and improving scheduling decisions;

[0042] 2. The present invention dynamically adjusts the relevant parameters of the power load demand analysis report through the second load demand, obtains the power load adjustment report, and timely adjusts the load forecast and scheduling strategy according to the actual situation, improves the adaptability and response speed of the model, and reduces the impact of prediction deviation on scheduling decisions;

[0043] 3. The present invention formulates a scheduling strategy by considering the integration of emerging loads and the coordinated operation of multi-energy systems to improve resource allocation efficiency and system stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a diagram of the method steps of the present invention.

[0045] Figure 2 The figure is a flow chart of the method of the present invention.

[0046] Figure 3 It is a schematic diagram of the structure of the method of the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification of the present application, the singular expressions "one", "a kind of", "said", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to and includes any or all possible combinations of one or more listed items.

[0049] In the following, the terms "first", "second", and "third" are used only for descriptive purposes and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first", "second", and "third" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, the meaning of "multiple" is two or more.

[0050] As attached Figure 1 A method for load forecasting and scheduling of a power system based on an optimization algorithm is shown, comprising S1: real-time monitoring of the power system, S2: preprocessing the first power load, S3: analyzing the first power key data group corresponding to the first power load, and obtaining the first load demand, S4: analyzing the first load demand, and obtaining a power load demand analysis report, S5: obtaining the second load demand, S6: obtaining a power load adjustment model, S7: optimizing the power system scheduling strategy, and S8: outputting the power system scheduling strategy, monitoring the power system performance, and performing regular maintenance.

[0051] S1: Real-time monitoring of the power system: Real-time data collection and monitoring of the power system are performed through sensors and monitoring equipment to obtain the first power load.

[0052] In one possible implementation, various types of high-performance sensors are provided in terminal devices connected to the power system, and the types of sensors include but are not limited to: smart meters, current transformers, voltage transformers, etc.; the sensors sense the electrical parameters of the terminal devices in the power system in real time, and the electrical parameters include but are not limited to: current, voltage, power, etc.; in this embodiment, the terminal devices collaborate with the monitoring equipment through sensors, and communication technology is used when monitoring the terminal devices. The communication technology includes but is not limited to: high-speed Ethernet, wireless communication, etc. When the monitoring equipment detects a fault in the sensor or itself, it immediately issues an alarm and activates the corresponding fault-tolerant mechanism.

[0053] It should be noted that the sensor samples data at a high frequency, and preliminarily processes and integrates the collected power load fluctuation data to obtain the first power load, which includes real-time load data, power generation data, equipment status data, meteorological data, time data, and load conditions in the power grid.

[0054] S2: Preprocessing the first power load: performing a preprocessing operation on the first power load, where the preprocessing operation is used to obtain a first power key data group corresponding to the first power load.

[0055] Specifically, in the power system load forecasting and scheduling method, a preprocessing operation is performed on the acquired first power load to obtain a first power key data group corresponding to the first power load, wherein the preprocessing steps are as follows:

[0056] A1: Data cleaning: remove abnormal values ​​in the first power load that are obviously deviated from the normal range due to sensor failure, data transmission error, etc.;

[0057] In one possible implementation method, the load data at a certain moment is subjected to the triple standard deviation method to filter out transient interference data in the power system, and a time window and fluctuation threshold are set to perform smoothing processing to retain data reflecting the real change trend of the power load;

[0058] A2: Missing value processing: For the missing values ​​in the first power load, the missing values ​​are supplemented according to the characteristics of the time series. The supplementation methods include using linear interpolation, nearest neighbor filling method, and mean filling method based on historical data of the same period;

[0059] In this embodiment, if the load data at a certain moment is missing, a linear interpolation calculation is performed with reference to the load data of the adjacent time periods before and after the moment to estimate the missing value;

[0060] A3: Data standardization: convert various physical quantity units in the first power load data into standard units, and perform normalization processing on the first power load so that the value range is between [-1, 1];

[0061] In a possible implementation, the method for normalizing the first power load includes: minimum-maximum normalization and Z-score standardization, wherein the minimum-maximum normalization formula is:

[0062] X ia =(XX min ) / (X max -X min ),

[0063] Among them, X ia represents the first power load after minimum-maximum normalization, X represents the first power load before normalization, and X min Expressed as the minimum value in the first power load sequence, X max is represented as the maximum value in the first power load sequence;

[0064] The Z-score normalization formula is:

[0065] Z=(X ia -μ) / σ,

[0066] Among them, Z represents the first power load after Z-score standardization, X1 represents the first power load after minimum-maximum normalization, μ represents the mean of data with the same nature in the first power load, and σ represents the standard deviation of data with the same nature in the first power load.

[0067] In this embodiment, the power data is uniformly converted from kilowatts (kW) to megawatts (MW), and the current data is converted from amperes (A) to kiloamperes (kA);

[0068] A4: Feature extraction: cluster analysis method is used to extract the key features of the first power load;

[0069] In a possible implementation, a method for extracting the key features of the first power load is as follows:

[0070] A41: Take hour, day, week, month, and season time information as feature variables and calculate time interval features, including the load difference between the current moment and the previous moment, and the load difference between the current moment and the previous day at the same moment of the day;

[0071] A42: Take temperature, humidity, wind speed, and sunshine intensity as characteristic variables, further associate them with power load data, and analyze the correlation between load change trends and mutations and meteorological data;

[0072] In this embodiment, for the temperature feature, the temperature change rate, high temperature duration, and low temperature duration are calculated; for the humidity feature, the relative change of humidity is considered;

[0073] A43: performing pattern recognition on the first power load data to obtain a load pattern category, where the load pattern category includes a weekday load pattern, a weekend load pattern, a summer peak load pattern, and a winter valley load pattern, and marking the load pattern category of the first power load at each moment to obtain a load pattern feature;

[0074] A5: Data dimensionality reduction: Use the principal component analysis method to calculate the covariance matrix of the extracted feature data, solve the eigenvalues ​​and eigenvectors, and perform dimensionality reduction.

[0075] Specifically, after the preprocessing step, the first power load obtains a first power key data group corresponding to the first power load, and the first power key data group includes key features after cleaning, standardization, feature extraction and dimensionality reduction processing.

[0076] S3: Analyze the first power key data group corresponding to the first power load, and obtain the first load demand: calculate the first load demand based on the first power key data group corresponding to the first power load, and the first load demand is used to express the operating efficiency of the power system scheduling.

[0077] Specifically, the method for obtaining the first load demand in step S3 is as follows:

[0078] Based on the actual operation conditions and constraints of the power system, a load demand calculation model is constructed. The load demand calculation model is a pre-constructed learning model. By inputting the first power key data group into the load demand calculation model, the load demand calculation model obtains the first load demand according to the first power key data group.

[0079] In a possible implementation, a method for constructing a load demand calculation model is as follows:

[0080] B1: Sort out the correspondence between the first power load and the first power key data group, and construct a load demand training set containing various features, recorded as {(z1, y1), (z2, y2), (z3, y3), ..., (z n ,y n )}, where z i Represented as the i-th key feature in the first power key data group, y i is the first power load value corresponding to the key feature, i represents the sequence number of the key feature in the first power key data group, and n represents the total number of key features in the first power key data group;

[0081] B2: Select a training kernel function, where the training kernel function includes a linear kernel function, a polynomial kernel function, and a radial basis function kernel function;

[0082] In this embodiment, taking the radial basis function kernel function as an example, it is specifically expressed as:

[0083] K(z i , z j ) = exp(-γ×‖z i -z j ‖ 2 ),

[0084] Among them, K(z i , z j ) represents the key feature data point z i and z j In the similarity value after the radial basis function kernel function conversion, γ is represented as the kernel function parameter, z i Represented as the i-th key feature in the first power key data group, z j represents the jth key feature in the first power key data group, i represents the sequence number of the key feature in the first power key data group, and j represents the sequence number of the key feature in another first power key data group different from i;

[0085] It should be noted that K(z i , z j ) value, the higher the similarity between the two data points in the high-dimensional space; γ determines the scope of the radial basis function kernel function. A larger γ value makes the data points more concentrated in the high-dimensional space and more sensitive to the slight differences between the sample points. A smaller γ value makes the data points more dispersed in the high-dimensional space and insensitive to the differences between the sample points.

[0086] B3: Determine the penalty parameter C and kernel function parameter γ in the load demand calculation model by cross-validation method;

[0087] B4: training the load demand calculation model to obtain the optimal parameters involved in the load demand calculation model;

[0088] In a possible implementation, the load demand calculation model training step specifically includes:

[0089] B41: Define the objective function of load demand;

[0090] In a possible implementation, the objective function is specifically expressed as:

[0091]

[0092] Wherein, ω represents the feature weight vector, C represents the penalty parameter, i represents the sequence number of the key feature in the first power key data group, n represents the total number of key features in the first power key data group, ξ i and i * is represented as a slack variable;

[0093] It should be noted that ω determines the importance of each feature in the model, and C is used to balance the complexity of the model and the training error. A larger C will penalize the error more and tend to fit the training data. A smaller C allows more errors and tends to generalize. i and i * It is used to measure the deviation between the predicted value and the actual value from different directions, so that the load demand calculation model will not excessively pursue completely accurate fitting and cause overfitting when facing noisy data or data points that are difficult to fit completely. By adjusting these two slack variables, the tolerance of the load demand calculation model to deviations in different directions can be controlled to a certain extent, thereby balancing the accuracy and generalization ability of the model;

[0094] B42: Determine the constraints on load demand;

[0095] In a possible implementation, the objective function is specifically expressed as:

[0096] y i -ω T ×φ(z i )-b≤∈+ξ i ,ω T ×φ(z i )+by i ≤∈+ξ i * ,

[0097] Among them, yi is represented by the first power load value corresponding to the key feature, ω is represented by the feature weight vector, ω T Represented as the transposed vector of ω, z i It is represented as the i-th key feature in the first power key data group, φ(z i ) is represented by z through the kernel function i The feature vector mapped to the high-dimensional space, b is represented as the feature bias, ∈ is represented as the insensitive loss parameter, ξ i and i * is represented as a slack variable;

[0098] B43: Calculate the optimal solution of ω and b under the objective function and constraints by sequential minimum optimization method;

[0099] B5: obtaining a load demand calculation model, by inputting the first power key data group into the load demand calculation model, so that the load demand calculation model obtains a first load demand according to the first power key data group;

[0100] In a possible implementation, the load demand calculation model in S3 is specifically expressed as follows:

[0101] EIR=ω T ×φ(z i )+b,

[0102] Where EIR is the first load demand, ω is the characteristic weight vector, and ω T Represented as the transposed vector of ω, z i It is represented as the i-th key feature in the first power key data group, φ(z i ) is represented by z through the kernel function i The feature vector mapped to the high-dimensional space, b is represented as the feature bias.

[0103] S4: Analyze the first load demand and obtain a power load demand analysis report: Analyze the resource allocation in the power system based on the first load demand, identify potential resource waste and unreasonable allocation problems in the power system, and obtain a power load demand analysis report.

[0104] In a possible implementation, obtaining the power load demand analysis report in step S4 includes: calculating the load mean, peak value and valley value of different time periods based on the historical data of the first load demand, understanding the load fluctuation range analysis, and analyzing the seasonal and cyclical changes of the load.

[0105] In a possible implementation, obtaining the power load demand analysis report in step S4 also includes: comparing the actual output of different types of generator sets with the first load demand to evaluate the utilization efficiency of power generation resources; calculating the load rate of the power grid, that is, the ratio of actual transmission power to transmission capacity; analyzing the degree of matching between the charging and discharging strategy of the energy storage device and the first load demand; and identifying potential resource waste and unreasonable configuration problems in the power system through a comprehensive analysis of the first load demand and resource allocation.

[0106] In a possible implementation, obtaining the power load demand analysis report in step S4 also includes: the content of the power load demand analysis report includes: a summary of the changing trend and characteristics of the first load demand; an evaluation result of the configuration of power generation resources, grid transmission and energy storage equipment, and an analysis of the advantages and disadvantages; clarifying the resource waste and unreasonable configuration problems in the power system, and analyzing the causes and impacts; and proposing improvement suggestions and countermeasures for potential problems.

[0107] S5: Obtaining a second load demand: Obtain a second power load and a load prediction model, and obtain a second load demand corresponding to the second power load through the second power load according to the load prediction model.

[0108] Specifically, the load prediction model is a pre-constructed learning model, and by inputting the second power load into the load prediction model, the load prediction model obtains the second load demand according to the second power load.

[0109] It should be noted that the second power load is used to predict the power load, which is composed of key features extracted from the first load demand data.

[0110] In one possible implementation, obtaining the second load demand corresponding to the second power load in step S5 includes: selecting a load forecasting model, the load forecasting model includes an autoregressive integrated moving average model ARIMA, seasonal ARIMA, support vector regression SVR, random forest regression, and a deep learning model; inputting the integrated second power load into the load forecasting model, performing load forecasting calculations, and outputting a predicted value for the future load, i.e., the second load demand, the second load demand including a predicted value for the power load within a period of time in the future, a load change trend, and uncertainty information.

[0111] S6: Obtaining a power load adjustment model: Obtaining a power load adjustment model, and dynamically adjusting the power load demand analysis report related parameters according to the power load adjustment model through the second load demand to obtain a power load adjustment report.

[0112] Specifically, the power load adjustment model is a pre-constructed learning model. By inputting the second load demand into the power load adjustment model, the power load adjustment model dynamically adjusts the relevant parameters of the power load demand analysis report according to the second power load.

[0113] It should be noted that the power load adjustment model selects existing mature models, including linear programming models, integer programming models, and dynamic programming models.

[0114] In one possible implementation, obtaining the power load adjustment report in step S6 includes: inputting the obtained second load demand data into the power load adjustment model; using an optimization algorithm to dynamically adjust the power generation plan, load distribution, demand response strategy, and interruptible load arrangement according to the characteristics and scale of the power load adjustment model; generating a power load adjustment report based on the adjusted parameters and model solution results, the power load adjustment report including the reason for the power load adjustment, the adjustment target, the second load demand, and the load adjustment plan.

[0115] S7: Optimize the power system dispatching strategy: Based on the real-time operating data corresponding to the first power key data group, the load forecast results corresponding to the second load demand, the power load demand analysis report, and the power load adjustment report, optimize the allocation of power generation resources associated with the power system to obtain the power system dispatching strategy.

[0116] In one possible implementation, based on the real-time operating data corresponding to the first power key data group and the load forecast results corresponding to the second load demand and with reference to the power load demand analysis report and the power load adjustment report, the optimization objectives and constraints are determined through an optimization algorithm; wherein the optimization objectives include minimizing the power generation cost, maximizing the system reliability, and minimizing the environmental impact, and the constraints include the output limit of the power generation equipment, the capacity limit of the transmission line, and the voltage and frequency stability range.

[0117] In one possible implementation, a power system dispatching strategy is generated based on the solution results of the optimization algorithm, and the power system dispatching strategy includes the start and stop plans, output distribution, and load adjustment measures of each power generation resource; the generated dispatching strategy is implemented in the power system, the operating status of the power generation resources is adjusted through the automatic control system, and the implemented power system is monitored in real time and the power load demand is analyzed.

[0118] S8: Output the power system dispatching strategy, monitor the power system performance, and provide a power system interactive interface for regular maintenance to set the parameters of the power equipment associated with the power system, view the load forecast results and dispatching strategy, monitor the performance of the power system in real time, and perform regular system maintenance and upgrades.

[0119] In a possible implementation, the operator understands the operation status of the power system through the power system interactive interface and manually adjusts and optimizes the dispatching strategy according to actual needs.

[0120] As attached Figure 2 The power system load forecasting and dispatching system based on the optimization algorithm shown includes a system operation database, a system central processing module and a user information terminal, and also includes: a power system real-time monitoring module, a first power load preprocessing module, a first power key data group calculation module, a first load demand analysis module, a second load demand acquisition module, a power load adjustment model acquisition module, a power system dispatching strategy optimization module, and a power system dispatching strategy output module.

[0121] Power system real-time monitoring module: used to collect and monitor the power system in real time through sensors and monitoring equipment to obtain the first power load;

[0122] A first power load preprocessing module: used for performing a preprocessing operation on the first power load, wherein the preprocessing operation is used for obtaining a first power key data group corresponding to the first power load;

[0123] A first power key data group calculation module: used to calculate a first load demand based on a first power key data group corresponding to a first power load, wherein the first load demand is used to represent the operation efficiency of the power system dispatch;

[0124] The first load demand analysis module is used to analyze the resource configuration in the power system based on the first load demand, identify potential resource waste and unreasonable configuration problems in the power system, and obtain a power load demand analysis report;

[0125] A second load demand acquisition module: used to acquire a second power load and a load prediction model, and acquire a second load demand corresponding to the second power load through the second power load according to the load prediction model;

[0126] The power load adjustment model acquisition module is used to obtain the power load adjustment model, and dynamically adjust the parameters related to the power load demand analysis report according to the power load adjustment model through the second load demand to obtain the power load adjustment report;

[0127] Power system dispatch strategy optimization module: used to optimize the allocation of power generation resources associated with the power system based on the real-time operation data corresponding to the first power key data group, the load forecast result corresponding to the second load demand, the power load demand analysis report, and the power load adjustment report, so as to obtain the power system dispatch optimization strategy;

[0128] Power system dispatch strategy output module: used to provide a power system interactive interface to set parameters of power equipment associated with the power system, view load forecast results and dispatch strategies, monitor the performance of the power system in real time, and perform regular system maintenance and upgrades;

[0129] The system operation database includes all data texts of the power system load forecasting and dispatching system, and collects the information texts output by each module in real time. The system central processing module is used for the information text instructions output by each module in the central control method. The user information terminal is an information output device for receiving the power system load forecasting and dispatching system.

[0130] Secondly: In the drawings of the embodiments disclosed in the present invention, only the structures related to the embodiments disclosed in the present invention are involved, and other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of the present invention can be combined with each other;

[0131] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for load forecasting and dispatching of a power system based on an optimization algorithm, characterized in that: include: S1: Real-time monitoring of the power system: real-time data collection and monitoring of the power system are performed through sensors and monitoring equipment to obtain the first power load; S2: preprocessing the first power load: performing a preprocessing operation on the first power load, the preprocessing operation is used to obtain a first power key data group corresponding to the first power load; S3: Analyze the first power key data group corresponding to the first power load and obtain the first load demand: calculate the first load demand based on the first power key data group corresponding to the first power load, and the first load demand is used to express the operation efficiency of the power system dispatch; S4: Analyze the first load demand and obtain a power load demand analysis report: Analyze the resource allocation in the power system based on the first load demand, identify potential resource waste and unreasonable allocation problems in the power system, and obtain a power load demand analysis report; S5: Obtaining a second load demand: Obtaining a second power load and a load prediction model, and obtaining a second load demand corresponding to the second power load through the second power load according to the load prediction model; S6: Obtaining a power load adjustment model: Obtaining a power load adjustment model, and dynamically adjusting parameters related to the power load demand analysis report according to the power load adjustment model through the second load demand to obtain a power load adjustment report; S7: Optimizing the power system dispatching strategy module: Based on the real-time operation data corresponding to the first power key data group, the load forecast result corresponding to the second load demand, the power load demand analysis report, and the power load adjustment report, optimizing the allocation of power generation resources associated with the power system to obtain the power system dispatching strategy; S8: Output power system dispatching strategy, monitor power system performance, and perform regular maintenance: Provide a power system interactive interface to set parameters of power equipment associated with the power system, view load forecast results and dispatching strategies, monitor power system performance in real time, and perform regular system maintenance and upgrades.

2. The method for load forecasting and dispatching of a power system based on an optimization algorithm according to claim 1, characterized in that: The first power load in S1 includes real-time load data, power generation data, equipment status data, meteorological data, time data, and load conditions in the power grid.

3. The method for load forecasting and dispatching of a power system based on an optimization algorithm according to claim 1, characterized in that: In S2, a first power key data group corresponding to the first power load is obtained, wherein the preprocessing steps are as follows: A1: Data cleaning: remove abnormal values ​​in the first power load that are obviously deviated from the normal range due to sensor failure, data transmission error, etc.; A2: Missing value processing: For the missing values ​​in the first power load, the missing values ​​are supplemented according to the characteristics of the time series. The supplementation methods include using linear interpolation, nearest neighbor filling method, and mean filling method based on historical data of the same period; A3: Data standardization: convert various physical quantity units in the first power load data into standard units, and perform normalization processing on the first power load so that the value range is between [-1, 1]; A4: Feature extraction: cluster analysis method is used to extract the key features of the first power load; A5: Data dimensionality reduction: Use the principal component analysis method to calculate the covariance matrix of the extracted feature data, solve the eigenvalues ​​and eigenvectors, and perform dimensionality reduction.

4. The method for load forecasting and dispatching of a power system based on an optimization algorithm according to claim 1, characterized in that: The method for obtaining the first load demand in S3 is as follows: Based on the actual operation conditions and constraints of the power system, a load demand calculation model is constructed. The load demand calculation model is a pre-constructed learning model. By inputting the first power key data group into the load demand calculation model, the load demand calculation model obtains the first load demand according to the first power key data group.

5. The method for power system load forecasting and dispatching based on optimization algorithm according to claim 4, characterized in that: The steps for constructing the load demand calculation model in S3 are as follows: B1: Sort out the correspondence between the first power load and the first power key data group, and construct a load demand training set containing various features, recorded as {(z1, y1), (z2, y2), (z3, y3), ..., (z n ,y n )}, where z i Represented as the i-th key feature in the first power key data group, y i is the first power load value corresponding to the key feature, i represents the sequence number of the key feature in the first power key data group, and n represents the total number of key features in the first power key data group; B2: Select a training kernel function, where the training kernel function includes a linear kernel function, a polynomial kernel function, and a radial basis function kernel function; B3: Determine the penalty parameter C and kernel function parameter γ in the load demand calculation model by cross-validation method; B4: training the load demand calculation model to obtain the optimal parameters involved in the load demand calculation model; B5: Obtain a load demand calculation model by inputting the first power key data group into the load demand calculation model, so that the load demand calculation model obtains the first load demand according to the first power key data group.

6. The method for power system load forecasting and dispatching based on optimization algorithm according to claim 4, characterized in that: The load demand calculation model in S3 is specifically expressed as follows: EIR=ω T ×φ(z i )+b, Where EIR is the first load demand, ω is the characteristic weight vector, and ω T Represented as the transposed vector of ω, z i It is represented as the i-th key feature in the first power key data group, φ(z i ) is represented by z through the kernel function i The feature vector mapped to the high-dimensional space, b is represented as the feature bias.

7. The method for power system load forecasting and dispatching based on optimization algorithm according to claim 1, characterized in that: The obtaining of the second load demand corresponding to the second power load in S5 includes: Select a load forecasting model, which includes autoregressive integrated moving average model ARIMA, seasonal ARIMA, support vector regression SVR, random forest regression, and deep learning model; The integrated second power load is input into the load forecasting model to perform load forecasting calculations and output the predicted value of future load, i.e., the second load demand. The second load demand includes the predicted value of power load in a future period of time, load change trend, and uncertainty information.

8. The method for power system load forecasting and dispatching based on optimization algorithm according to claim 1, characterized in that: The obtaining of the power load adjustment report in S6 includes: Input the acquired second load demand data into the power load adjustment model; dynamically adjust the power generation plan, load distribution, demand response strategy, and interruptible load arrangement according to the characteristics and scale of the power load adjustment model using an optimization algorithm; According to the adjusted parameters and the model solution results, a power load adjustment report is generated, and the power load adjustment report includes the reason for the power load adjustment, the adjustment target, the second load demand, and the load adjustment plan.

Images