Power load optimal distribution method and system based on intelligent algorithm

Abstract

The invention relates to the technical field of intelligent power system optimization scheduling, and discloses a power load optimization distribution method and system based on an intelligent algorithm, and the method comprises the steps: collecting historical load data of a power system, and carrying out the preprocessing; the fractal dimension of the load time sequence is calculated, a load change critical point is identified, and the size of a time window for load prediction is dynamically adjusted. And inputting the dynamically adjusted time window data into an LSTM model for load prediction. And according to a load prediction result, carrying out intelligent optimization distribution on the power load by adopting an optimization algorithm. According to the method, the optimal distribution of the power load not only has high precision, but also can flexibly cope with various sudden fluctuations and complex constraint conditions in a power system. Through the steps, the operation efficiency, the stability and the economical efficiency of the electric power system are improved, meanwhile, the uncertainty and the risk in load distribution of the electric power system are reduced, and a new thought and technical support are provided for a future intelligent power grid and an electric power dispatching system.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent power system optimal dispatching, and specifically to a method and system for optimizing power load distribution based on intelligent algorithms. Background Art

[0002] With the continuous development of intelligence and informatization, power load optimal distribution, as an important research field in power system dispatching, has received extensive attention in recent years. Traditional load forecasting methods mostly rely on statistical models such as linear regression and time series analysis. However, with the variability and complexity of power demand, these traditional methods are gradually becoming inadequate. In recent years, with the rapid development of artificial intelligence technology, especially the application of deep learning and intelligent optimization algorithms, the methods for power load forecasting and optimal distribution have gradually transformed from traditional statistical models to optimization methods based on intelligent algorithms. For example, the load forecasting method based on the long short-term memory (LSTM) network, with its powerful ability to model time series data, can effectively capture the non-linear change trend of power load. In addition, intelligent optimization algorithms such as particle swarm optimization (PSO) and genetic algorithm (GA) have also been widely applied to the distribution and dispatching of power load in order to improve the efficiency and stability of the power system.

[0003] Although the existing methods for optimizing power load distribution have introduced intelligent algorithms, especially the combination of load forecasting based on machine learning and intelligent optimization algorithms, there are still the following deficiencies and cannot completely solve the problem of power load optimization. First of all, traditional load forecasting methods mostly rely on fixed time windows and cannot dynamically adjust the forecasting granularity, resulting in low forecasting accuracy when the load fluctuates violently, thus affecting subsequent load distribution decisions. This is because the fixed time window cannot adapt to the rapid change of load demand, especially at the critical point of load change, and cannot accurately capture the short-term fluctuations.

[0004] Secondly, although the existing intelligent optimization algorithms can solve the problem of power load distribution, they generally rely on fixed parameter settings and simple constraint conditions and lack the ability to flexibly adjust according to the complex requirements in the actual power system. When facing complex power network structures and various real-time constraints (such as supply-demand balance, power system capacity, etc.), many algorithms are prone to falling into local optimal solutions and cannot effectively find the global optimal solution.

[0005] In addition, although some evaluation indexes of forecasting accuracy and load distribution efficiency are adopted in the existing methods, most methods do not fully consider the direct impact of load forecasting errors on power load distribution. Since the errors of load forecasting may affect the accuracy of power load distribution to a certain extent, the existing technologies often do not effectively constrain and optimize this problem, resulting in uncertainty in the load distribution process and potential risks of the power system.

[0006] Therefore, the existing technologies have obvious deficiencies in accurately predicting load fluctuations, dynamically adjusting the time window, dealing with complex constraint conditions, ensuring global optimization, and controlling prediction errors. Based on these deficiencies, the present invention proposes a method for optimizing the distribution of electric power loads based on intelligent algorithms, aiming to effectively improve the accuracy and efficiency of load distribution by dynamically adjusting the time window, combining fractal dimension analysis and intelligent optimization algorithms, and further improve the reliability and safety of the distribution through load prediction accuracy constraints. Summary of the Invention

[0007] In view of the above problems, the present invention is proposed.

[0008] Therefore, the technical problems solved by the present invention are: the existing methods for optimizing the distribution of electric power loads have problems such as inaccurate capture of load fluctuations, the optimization algorithm being prone to falling into local optimal solutions, and the lack of effective control of load prediction errors, as well as the problem of how to accurately distribute electric power loads and ensure the reliability of optimization by dynamically adjusting the time window, introducing fractal dimension analysis, and combining intelligent optimization algorithms.

[0009] To solve the above technical problems, the present invention provides the following technical solution: a method for optimizing the distribution of electric power loads based on intelligent algorithms, including: collecting historical load data of the power system and performing preprocessing.

[0010] Calculating the fractal dimension of the load time series, identifying the critical points of load changes, and dynamically adjusting the size of the time window for load prediction.

[0011] Using the data of the dynamically adjusted time window as input to an LSTM model for load prediction.

[0012] According to the load prediction results, an optimization algorithm is used to intelligently optimize the distribution of electric power loads.

[0013] As a preferred embodiment of the method for optimizing the distribution of electric power loads based on intelligent algorithms according to the present invention, wherein: the collecting historical load data of the power system and performing preprocessing includes:

[0014] Real-time collecting electric power load data from the SCADA system of the power system, including the electricity consumption, load change trend, and power factor of each substation, distribution network, and load point in the power system.

[0015] Preprocessing includes removing missing data, removing outliers, data smoothing, and data normalization.

[0016] Checking the collected data for missing data, and if missing data is found, filling the missing data by linear interpolation.

[0017] Remove outliers. Use methods such as z-score to identify and remove obvious outliers.

[0018] Data smoothing. Smooth the collected data, use the moving average method to remove short-term fluctuations and retain long-term trends.

[0019] Data normalization. Normalize the load data, adopt the min-max normalization method, and standardize the load data to the interval [0, 1].

[0020] As a preferred solution of the power load optimal allocation method based on intelligent algorithms described in the present invention, wherein: calculating the fractal dimension of the load time series includes:

[0021] The original fractal dimension calculation formula of the load time series is expressed as:

[0022]

[0023] Among them, D represents the fractal dimension, N(∈) represents the number of boxes covering the data points with a size of ∈, and ∈ represents the box size.

[0024] To accurately capture load fluctuations, introduce a time window and a weighting coefficient. The optimized fractal dimension formula is expressed as:

[0025]

[0026] Among them, D adj (t) represents the dynamic fractal dimension at time t, considering the weighted influence of load fluctuations. N i (∈) represents the calculation of self-similarity in the i-th box, w i (t) represents the weighting coefficient, representing the weighted influence of load changes, and is expressed as:

[0027]

[0028] Among them, α represents the weighting coefficient adjustment parameter, controlling the sensitivity of load fluctuations. μ represents the threshold of load fluctuations, and y i (t) represents the difference of the load sequence at time t, which is the load change amount between two adjacent times.

[0029] As a preferred solution of the power load optimal allocation method based on intelligent algorithms described in the present invention, wherein: identifying the critical point of load change and dynamically adjusting the time window size of load prediction includes:

[0030] Based on the fractal dimension, identify the critical point of the load by observing the speed and amplitude of load changes, which is expressed as:

[0031]

[0032] Among them, ΔD adj (t) represents the change rate of the optimized fractal dimension, indicating the amplitude of load fluctuation. D adj (t - 10) represents the dynamic fractal dimension at time t - 1.

[0033] If ΔD adj (t) > T thresh , it is considered that the critical point of load change occurs at time t. Among them, the threshold T thresh is the key fluctuation point obtained through historical data analysis, used to distinguish important fluctuations from ordinary fluctuations.

[0034] Dynamically adjust the time window of load forecasting, expressed as:

[0035]

[0036] Among them, ∈ adj (t) represents the dynamic time window at time t, and β represents the adjustment parameter, controlling the response sensitivity of the time window.

[0037] As a preferred scheme of the power load optimal allocation method based on intelligent algorithms described in the present invention, among them: the use of the time window data after dynamic adjustment and inputting into the LSTM model for load forecasting includes:

[0038] The calculation of each time step in the LSTM network includes the calculation of the forget gate, input gate, candidate layer, and output gate.

[0039] As a preferred scheme of the power load optimal allocation method based on intelligent algorithms described in the present invention, among them: the LSTM model includes:

[0040] According to the output h of the LSTM t , the final power load forecasting value is expressed as:

[0041]

[0042] Among them, represents the forecasting load value at time t, representing the forecasting result output by the LSTM model. W y represents the forecasting weight, and b y represents the forecasting bias.

[0043] As a preferred scheme of the power load optimal allocation method based on intelligent algorithms described in the present invention, among them: the intelligent optimal allocation of power load according to the load forecasting result by using an optimization algorithm includes:

[0044] Use P pred,t to represent the final power load forecasting value of the LSTM model It is confirmed that the optimization objective is to minimize the power distribution cost, expressed as:

[0045]

[0046] where F opt (x) represents the optimization objective function, which represents the total cost of the power load optimization problem. Ch t represents the power cost at time period t, which is calculated based on the power market price and the load demand fluctuation. P t (x t ) represents the amount of power load allocated at time period t. P pred,t represents the power load demand predicted by the LSTM model at time period t. λ t represents the penalty coefficient, which is used to balance the error between the allocated amount and the predicted value, and reflects the cost when the allocated amount deviates from the predicted value. γ t represents the penalty coefficient, which is used to handle the power overload situation. represents the power overload indicator function, which takes the value of 1 when P t (x t ) exceeds the maximum load P max,t of the system, and 0 otherwise.

[0047] As a preferred solution of the power load optimal allocation method based on the intelligent algorithm described in the present invention, wherein: the intelligent optimal allocation of the power load by using the optimization algorithm according to the load prediction result further includes: using the particle swarm optimization algorithm to optimize the objective function through the position and velocity of the particles, and each particle in the particle swarm represents a power allocation scheme, and the update rule of the particles is expressed as:

[0048]

[0049] where represents the velocity of particle i at time period t. represents the amount of power allocation of particle i at time period t. w represents the inertia weight, which controls the influence degree of the previous velocity of the particle. c1 and c2 represent the learning factors, which respectively represent the learning degrees of the particle for the individual optimum and the global optimum. r1 and r2 represent random numbers, and the value range is [0,1], which is used to introduce randomness. represents the personal optimum solution of particle i. g best represents the global optimum solution.

[0050] The constraint conditions of the particle swarm optimization algorithm include the power compliance limit and the load balance requirement.

[0051] The power load limit is that the amount of power allocation in each time period should be within the maximum load range, and the load balance requirement is that the total power demand should be equal to the total power allocation.

[0052] As a preferred solution of the power load optimal allocation method based on intelligent algorithms according to the present invention, wherein: the constraint conditions further include:

[0053] New load prediction accuracy constraint, expressed as:

[0054]

[0055] Wherein, P t (x t ) represents the power load allocation amount at time period t, which is the power load value calculated in the particle swarm optimization algorithm. P pred,t represents the load prediction value at time period t, which is calculated by the LSTM model. represents the maximum allowable prediction error threshold at time period t, which is defined as the maximum difference between the load allocation amount and the load prediction value in the power load at time period t, and is determined by the tolerance error of the power system and the change range of the power market.

[0056] A power load optimal allocation system based on intelligent algorithms, characterized by comprising:

[0057] A preprocessing module, which collects the historical load data of the power system and performs preprocessing.

[0058] A prediction module, which calculates the fractal dimension of the load time series, identifies the critical points of load change, and dynamically adjusts the time window size of load prediction. Using the dynamically adjusted time window data, it is input into the LSTM model for load prediction.

[0059] A calculation module, which performs intelligent optimal allocation of the power load according to the load prediction result by using an optimization algorithm.

[0060] A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of the method described above are implemented.

[0061] A computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method described above are implemented.

[0062] Beneficial effects of the present invention: By introducing the step of dynamically adjusting the time window, the prediction granularity is dynamically adjusted according to the actual changes in load fluctuations, and the instantaneous fluctuations of the power load can be accurately captured. In traditional methods, the time window size is fixed and cannot be flexibly adjusted according to changes in actual load demand, resulting in a decrease in prediction accuracy during periods of drastic load fluctuations, affecting subsequent allocation decisions. By dynamically adjusting the time window, the system can reduce the window during periods of large load fluctuations and expand the window during periods of small fluctuations, thereby improving the accuracy of load forecasting. The role of this step is to optimize the response speed of the prediction model, so that it can adapt to changes in load demand in a timely manner, and ensure the efficiency and accuracy of power load distribution. This adjustment method has important practical value in real-time scheduling of power systems, and can reduce the risk of power waste and overload, thereby improving the stability and economy of the system.

[0063] By introducing fractal dimension analysis to evaluate the self-similarity of load time series, the present invention can more accurately identify the critical points of load changes. Traditional load forecasting methods often ignore the complexity of load fluctuations and easily miss some important changes in the short term, especially when the load fluctuates violently. Using fractal dimension analysis, the changing pattern of the load series can be measured more finely, especially in periods of strong load volatility, which plays an important role in improving the accuracy of the prediction. By analyzing the self-similarity of the load time series, this step can automatically capture the moment when the fluctuation amplitude suddenly changes, so that the load forecasting model can identify and adapt to emergencies in advance, thereby effectively avoiding the system's load overload or power shortage problems caused by prediction lags. This technological innovation has significant practical value for improving the accuracy of load forecasting, especially the prediction accuracy during periods of high volatility.

[0064] The present invention effectively limits the error between the prediction result and the actual load by introducing load prediction accuracy constraints, thereby ensuring the reliability of the power load distribution process. Traditional methods usually ignore the risks that may be caused by load prediction errors and rely too much on the accuracy of the prediction results, which easily leads to problems such as unbalanced power distribution and waste of resources. By setting the prediction error tolerance, the present invention ensures the accuracy of load distribution and ensures the balance and stability of the power system even in the presence of prediction errors. The innovation of this step is that it not only improves the stability of the load forecast of the power system, but also reduces the uncertainty and risk in power dispatching by controlling the prediction error, further improving the safety and reliability of the power system.

[0065] By adopting the Particle Swarm Optimization (PSO) algorithm for intelligent optimization allocation of power load, the system can find the optimal power distribution scheme under multiple constraints. Traditional optimization algorithms usually adopt static and fixed-parameter allocation strategies, lacking flexibility and adaptability. The Particle Swarm Optimization algorithm can dynamically adjust the allocation strategy according to the actual demand of power load and system constraints, thereby maximizing the efficiency of power distribution and reducing operating costs. The role of this step is that PSO can comprehensively consider grid load balance, power demand, cost, and various constraints to achieve the global optimal solution under more efficient conditions, avoiding the trap of local optimal solutions. Through Particle Swarm Optimization, the allocation of power load can be more flexible and accurate, greatly improving the economic benefits and operating stability of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings. Among them:

[0067] Figure 1 FIG. is the overall flowchart of a power load optimization allocation method and system based on an intelligent algorithm provided by the first embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0068] To make the above objects, features, and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings of the specification. Obviously, the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0069] Example 1, referring to Figure 1 , which is an embodiment of the present invention, provides a power load optimization allocation method based on an intelligent algorithm, including:

[0070] S1: Collect historical load data of the power system and perform preprocessing.

[0071] Real-time collect power load data from the SCADA system of the power system, including the electricity consumption, load change trend, and power factor of each substation, distribution network, and load point in the power system.

[0072] It should be noted that the real-time data of the power system comes from substations, distribution networks, and various load points, mainly including data such as power consumption, load change trends, and power factors. High-precision smart meters, load monitoring devices, and distribution automation systems are used to collect relevant data through data acquisition terminals.

[0073] Furthermore, in order to capture the rapid fluctuations of the load, the sampling frequency is set to per second to ensure the real-time nature and accuracy of the data. The real-time transmission of data is achieved through modern communication networks to ensure the fast and accurate transfer of the data stream to the centralized data processing platform.

[0074] The preprocessing includes removing missing data, removing outliers, data smoothing, and data normalization.

[0075] To remove missing data, the collected data is checked. If missing data is found, the missing data is filled by linear interpolation.

[0076] To remove outliers, methods such as z-score are used to identify and remove obvious outliers.

[0077] For data smoothing, the collected data is smoothed. The moving average method is used to remove short-term fluctuations and retain long-term trends.

[0078] For data normalization, the load data is normalized. The min-max normalization method is adopted to standardize the load data to the interval [0, 1].

[0079] S2: Calculate the fractal dimension of the load time series, identify the critical points of load change, and dynamically adjust the time window size of load forecasting.

[0080] The original fractal dimension calculation formula of the load time series is expressed as:

[0081]

[0082] Among them, D represents the fractal dimension, N(∈) represents the number of boxes covering the data points with a size of ∈, and ∈ represents the box size.

[0083] To accurately capture the load fluctuations, a time window and a weighting coefficient are introduced. The optimized fractal dimension formula is expressed as:

[0084]

[0085] Among them, D adj (t) represents the dynamic fractal dimension at time t, considering the weighted influence of load fluctuations. N i (∈) represents the calculation of self-similarity in the i-th box, w i (t) represents the weighting coefficient, indicating the weighted influence of load changes, and is expressed as:

[0086]

[0087] Among them, α represents the weighted coefficient adjustment parameter, which controls the sensitivity of load fluctuations. μ represents the threshold of load fluctuations, and y i (t) represents the difference of the load sequence at time t, which is the load change amount between two adjacent times.

[0088] It should be noted that in order to capture the changing trend of load fluctuations, a time window is introduced in the calculation of the original fractal dimension of the load time series. The size of this time window is automatically adjusted according to the load fluctuation characteristics, historical data, and the change of the fractal dimension. By using the time window to focus on the data within different time ranges, the detection accuracy of load fluctuations can be improved. The introduction of the weighted coefficient is to weight the calculation results according to the data at different time points, so as to enhance the sensitivity to some important time periods (such as peak load periods or load mutation periods). The weighted coefficient is set according to factors such as the amplitude of load change and the urgency of the time period.

[0089] Furthermore, by introducing a time window, irrelevant or noisy data can be effectively removed, ensuring that the calculation of the fractal dimension is more accurate. The setting of the weighted coefficient can be flexibly adjusted according to the influence of different data, avoiding excessive deviation of the overall load prediction caused by a single data point. The adaptive adjustment in the design idea can automatically optimize the size of the time window when the load fluctuates greatly, ensuring adaptation to the changes in different time periods.

[0090] Based on the fractal dimension, the critical point of the load is identified by observing the speed and amplitude of the load change, which is expressed as:

[0091]

[0092] Among them, ΔD adj (t) represents the change rate of the optimized fractal dimension, which represents the amplitude of load fluctuations. D adj (t - 1) represents the dynamic fractal dimension at time t - 1.

[0093] If ΔD adj (t) > T thresh , it is considered that the critical point of load change occurs at time t. Among them, the threshold T thresh is the key fluctuation point obtained through historical data analysis, which is used to distinguish important fluctuations from ordinary fluctuations.

[0094] It should be noted that by calculating the change rate of load data within a short period of time (i.e., the increase or decrease rate of the load), if the change speed suddenly increases, it may mean that the system will enter a new load state or approach the critical point of load fluctuation. By observing the fluctuation amplitude of the load data, if the load fluctuation amplitude is extremely large within a certain period (beyond the normal fluctuation range), it indicates that the load has a critical change during this period, which may lead to overload of the system load or power supply imbalance. Based on the calculation of the fractal dimension, by observing the self-similarity change of the load time series, the mutation of load fluctuation can be detected. If the fractal dimension curve shows mutation or non-stationary characteristics, it can be used as an early warning signal for the critical point.

[0095] Dynamically adjust the time window of load forecasting, expressed as:

[0096]

[0097] where, ∈ adj (t) represents the dynamic time window at time t, and β represents the adjustment parameter, which controls the response sensitivity of the time window.

[0098] It should be noted that when the load critical point is identified, based on the suddenness and violent fluctuation of the load change, the size of the time window during prediction is automatically adjusted. In the period near the critical point, the time window will be narrowed to more precisely capture the load fluctuation. While in the load stable period, the time window will be appropriately widened to avoid over-response to short-term fluctuations.

[0099] As the load data is continuously input, the system will monitor the fluctuation trend of the load in real time and dynamically adjust the size of the time window. For example, if the load changes rapidly and has a large fluctuation amplitude, the system will shorten the time window to ensure a faster response in prediction. On the contrary, if the load changes smoothly, the window can be widened to improve the calculation efficiency.

[0100] The size and adjustment frequency of the time window depend on the law of load change. The system will automatically optimize the adjustment process of the time window through load forecasting accuracy and error to maintain the real-time performance and accuracy of the system.

[0101] S3: Use the time window data after dynamic adjustment and input it into the LSTM model for load forecasting.

[0102] The calculation of each time step in the LSTM network includes the calculation of the forget gate, input gate, candidate layer, and output gate.

[0103] The calculation process of the forget gate is expressed as:

[0104] f t = σ(W f ·[h t-1 , x t +bf )

[0105] The calculation process of the input gate is expressed as:

[0106] i t = σ(W i |·[h t-1 , x t +b i )

[0107] The calculation process of the candidate layer is expressed as:

[0108]

[0109] The update of the cell state is expressed as:

[0110]

[0111] The output gate is expressed as:

[0112] o t = σ(W o ·[h t-1 , x t +b o )

[0113] The final output is expressed as:

[0114] h t = o t ·tanh(C t )

[0115] Among them, f t represents the output of the forget gate, which determines the influence of the previous cell state on the current moment; i t represents the output of the input gate, which controls the influence degree of the current input data; represents the candidate cell state, which represents the candidate update value at the current moment; C t represents the cell state, which stores the long-term memory of the LSTM network; o t represents the output of the output gate, which determines the influence of the cell state on the current moment's output; h t represents the output at the current moment, which represents the prediction of the LSTM network for the current load; σ represents the activation function; tanh represents the hyperbolic tangent activation function, which is used for the update of the cell state and the calculation of the output; x t represents the input data at time t, which includes the dynamically adjusted time window data; W f , W i , W C , W o represent the weight matrices, which represent the training parameters of each gating operation; b f , bi , b C , b o represents the bias term, which is the training bias for each gating operation;

[0116] According to the output h of the LSTM t , the final power load prediction value is expressed as:

[0117]

[0118] Among them, represents the predicted load value at time t, which is the prediction result output by the LSTM model. W y represents the prediction weight, and b y represents the prediction bias.

[0119] S4: According to the load prediction result, use an optimization algorithm to intelligently optimize the distribution of the power load.

[0120] Use P pred,t to represent the final power load prediction value of the LSTM model Confirm that the optimization objective is to minimize the power distribution cost, which is expressed as:

[0121]

[0122] Among them, F opt (x) represents the optimization objective function, which is the total cost of the power load optimization problem. Ch t represents the power cost at time period t, which is calculated based on the power market price and the load demand fluctuation. P t (x t ) represents the amount of power load allocated at time period t. P pred,t represents the power load demand predicted by the LSTM model at time period t. λ t represents the penalty coefficient, which is used to balance the error between the allocated amount and the predicted value, and reflects the cost when the allocated amount deviates from the predicted value. γ t represents the penalty coefficient, which is used to handle the power overload situation. represents the power overload indicator function, when P t (x t ) exceeds the system maximum load P max,t , its value is 1, otherwise it is 0.

[0123] It should be noted that introducing the dynamic adjustment of the time window of load forecasting into the optimization process enables the system to automatically adjust the calculation granularity during different load demand periods, thereby more precisely controlling the distribution cost. Compared with the traditional fixed-time window optimization method, the present invention makes the optimization of power distribution cost more flexible and efficient. The dynamic time window can accurately adjust the calculation granularity under different load demand scenarios, thus reducing unnecessary calculations and resource waste. The objective function can automatically adjust the optimization strategy with the change of load demand, and is more adaptable to the complex power load management requirements.

[0124] Using the particle swarm optimization algorithm, the objective function is optimized through the positions and velocities of particles. Each particle in the particle swarm represents a power distribution scheme, and the update rule of the particle swarm is expressed as:

[0125]

[0126] Among them, represents the velocity of particle i at time period t. represents the power distribution amount of particle i at time period t. w represents the inertia weight, which controls the influence degree of the previous velocity of the particle. c1 and c2 represent the learning factors, which respectively represent the learning degrees of the particle for the individual optimum and the global optimum. r1 and r2 represent random numbers, and the value range is [0,1], which is used to introduce randomness. represents the personal optimum solution of particle i. g best represents the global optimum solution.

[0127] The constraint conditions of the particle swarm optimization algorithm include power compliance limitations and load balance requirements.

[0128] The power load limitation is that the power distribution amount in each time period should be within the maximum load range, and the load balance requirement is that the total power demand should be equal to the total power distribution amount.

[0129] A new load forecasting accuracy constraint is added, which is expressed as:

[0130]

[0131] Among them, P t (x t ) represents the power load distribution amount in time period t, which is the power load value calculated in the particle swarm optimization algorithm. P pred,t represents the load forecasting value in time period t, which is calculated by the LSTM model. represents the maximum allowable prediction error threshold in time period t, which is defined as the maximum difference between the load distribution amount and the load forecasting value of the power load in time period t, and is determined by the tolerance error of the power system and the change range of the power market.

[0132] It should be noted that the load forecasting accuracy constraint ensures that the difference between the load forecasting result and the actual load demand is within an acceptable range by restricting the upper limit of the load forecasting error. This constraint is used to control the impact of forecasting errors in power load distribution and avoid unreasonable distribution caused by excessive errors.

[0133] The load forecasting accuracy constraint guarantees that the forecasting value on which the load distribution is based is as accurate as possible, avoiding system instability caused by excessive forecasting errors. By setting the accuracy constraint, more refined load scheduling can be achieved, reducing the uncertainty in power load distribution and enhancing the robustness of the system.

[0134] Furthermore, the present invention particularly relies on the LSTM model for power load forecasting, and the output of the LSTM model may be affected by training data and model optimization. Therefore, additional constraints need to be imposed through the error tolerance to ensure that the accuracy of load forecasting meets the actual requirements. In contrast, other traditional load forecasting methods (such as linear regression, support vector machines, etc.) usually do not have such refined error control requirements. Therefore, this constraint condition is adapted to the deep learning load forecasting process in the present invention.

[0135] In the above embodiments, there is also a power load optimal distribution system based on intelligent algorithms, specifically:

[0136] A preprocessing module that collects historical load data of the power system and performs preprocessing.

[0137] A forecasting module that calculates the fractal dimension of the load time series, identifies the critical points of load changes, and dynamically adjusts the time window size of load forecasting. Using the dynamically adjusted time window data, it is input into the LSTM model for load forecasting.

[0138] A calculation module that intelligently optimizes the distribution of power loads using an optimization algorithm based on the load forecasting results.

[0139] The computer device can be a server. The computer device includes a processor, a memory, an input / output interface (Input / Output, abbreviated as I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the data cluster data of the power monitoring system. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with external terminals through a network connection. When the computer program is executed by the processor, it realizes a power load optimal allocation method based on an intelligent algorithm.

[0140] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memories. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The databases involved in the embodiments provided in the present application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the embodiments provided in the present application can be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, data processing logics based on quantum computing, etc., without limitation.

[0141] Embodiment 2 is an embodiment of the present invention, which provides a method and system for optimizing the allocation of electric power loads based on intelligent algorithms. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0142] This embodiment verifies the effect of the method for optimizing the allocation of electric power loads based on intelligent algorithms through experimental data. The experiments mainly focus on load forecasting through the LSTM model and combining with the particle swarm optimization (PSO) algorithm for load allocation to evaluate the advantages of the method of the present invention in the optimization of electric power load allocation.

[0143] In this experiment, assuming the power load distribution problem in a certain area, the time period is set to 24 hours, and the load demand per hour is predicted by an LSTM model. To verify the effectiveness of the method of the present invention, the experiment is divided into two groups: a control group (traditional load forecasting method) and an experimental group (LSTM prediction combined with PSO optimization allocation method). The control group uses the traditional regression analysis method for load forecasting, and does not adopt dynamic adjustment of the time window. The load allocation is directly optimized based on the forecasting results. The experimental group adopts the LSTM prediction model in the present invention, dynamically adjusts the time window, and combines the particle swarm optimization algorithm for load optimization allocation.

[0144] Collect power load data in a certain area, including load demand, load change trend, power factor, etc. The data collection frequency is set to once per hour, and preprocessing (denoising, missing value filling, etc.) is performed.

[0145] Use the LSTM model to predict the power load demand for the next 24 hours. The control group uses the traditional linear regression method for prediction, and the experimental group uses the LSTM model.

[0146] For the experimental group, use the fractal dimension analysis method to ensure that the prediction accuracy is improved when the load fluctuates greatly by dynamically adjusting the size of the time window.

[0147] Adopt the particle swarm optimization algorithm (PSO) to perform load allocation on the prediction results of the experimental group and the control group. During the allocation process, consider factors such as power demand, cost, and constraint conditions, and seek the optimal load allocation scheme through the optimization algorithm.

[0148] Compare the load forecasting errors, allocation costs, and system load balance conditions of the two groups. The experimental results are shown in Table 1.

[0149] Table 1

[0150]

[0151] It can be seen from the table that the error between the load values predicted by the experimental group and the actual load is smaller, and the error rate is continuously lower than that of the control group. This shows that the load forecasting method based on LSTM can more accurately capture the non-linear change trend of the power load, and the prediction accuracy in high-fluctuation periods is significantly improved compared with the traditional regression method. Especially in the periods of large load fluctuations (such as periods 7, 9, and 14), the error is small, indicating that the LSTM model can effectively cope with this complex fluctuation.

[0152] In addition, the effect of dynamically adjusting the time window is also reflected in the prediction accuracy. As the load change progresses, the adjustment of the time window ensures that the system can quickly respond to sudden load changes in a short period. This dynamic adjustment mechanism is crucial for the scheduling of power systems because it can significantly reduce the lag in load fluctuation prediction, thereby improving the accuracy of load distribution.

[0153] In terms of the load distribution cost, the distribution costs of the experimental group are generally lower, reflecting that the optimization of load distribution by the Particle Swarm Optimization (PSO) algorithm can effectively reduce the operating cost of the power system while meeting the load demand and system constraints. Although there is a slight gap in the distribution cost between the experimental group and the control group, the overall difference is very small, and the system operation becomes more stable through the optimization algorithm.

[0154] In this embodiment, by comparing the prediction errors and load distribution costs of the experimental group and the control group, the innovation and effectiveness of the power load optimal distribution method based on intelligent algorithms are verified. The combination of the LSTM model, dynamic time window adjustment, and the Particle Swarm Optimization algorithm enables higher power load prediction accuracy, lower load distribution costs, and better coping with the challenges brought by load fluctuations. Through these beneficial effects, the present invention demonstrates stronger adaptability and accuracy than traditional methods, especially in dealing with complex scenarios with large power demand fluctuations, showing significant advantages.

[0155] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for optimizing power load distribution based on intelligent algorithm, characterized in that: include: Collect historical load data of the power system and perform preprocessing; Calculate the fractal dimension of the load time series, identify the critical point of load change, and dynamically adjust the time window size of load forecasting; Use the dynamically adjusted time window data and input it into the LSTM model for load forecasting; According to the load forecast results, the optimization algorithm is used to intelligently optimize the distribution of power load.

2. The method for optimizing power load distribution based on intelligent algorithm according to claim 1, characterized in that: The collecting of historical load data of the power system and preprocessing includes: Collect power load data in real time from the power system's SCADA system, including power consumption, load change trends, and power factors of each substation, distribution network, and load point in the power system; Preprocessing includes removing missing data, removing outliers, data smoothing, and data normalization; Removal of missing data The collected data were checked and if missing data were found, the missing data were filled by linear interpolation; Remove outliers Use z-score to identify and remove obvious outliers; Data smoothing smoothes the collected data, using the sliding average method to remove short-term fluctuations and retain long-term trends; Data normalization is to normalize the load data to the interval [0,1] using the minimum-maximum normalization method.

3. The method for optimizing power load distribution based on intelligent algorithm according to claim 2, characterized in that: The fractal dimension of the calculated load time series includes: The calculation formula of the original fractal dimension of the load time series is expressed as: Where D represents the fractal dimension, N(∈) represents the number of data points covered by boxes of size ∈, and ∈ represents the box size; In order to accurately capture load fluctuations, the time window and weighting coefficient are introduced, and the optimized fractal dimension formula is expressed as: Among them, D adj (t) represents the dynamic fractal dimension at time t, taking into account the weighted influence of load fluctuation; N i (∈) represents the calculation of self-similarity in the i-th box, w i (t) represents the weighting coefficient, which represents the weighted impact of load change and is expressed as: Among them, α represents the weighted coefficient adjustment parameter, which controls the sensitivity of load fluctuation; μ represents the threshold of load fluctuation, y i (t) represents the difference of the load sequence at time t, which is the load change between two adjacent moments.

4. The method for optimizing power load distribution based on intelligent algorithm according to claim 3, characterized in that: The identifying of the critical point of load change and dynamically adjusting the time window size of load forecasting includes: Based on the fractal dimension, the critical point of the load is identified by observing the speed and fluctuation amplitude of the load change, which is expressed as: Where, ΔD adj (t) represents the change rate of the optimized fractal dimension, which indicates the amplitude of load fluctuation; D adj (t-1) represents the dynamic fractal dimension at time t-1; If ΔD adj (t)>T thresh , it is considered that the critical point of load change occurs at time t; the threshold value T thresh It is the key fluctuation point obtained through historical data analysis, which is used to distinguish important fluctuations from ordinary fluctuations; Dynamically adjust the time window of load forecasting, expressed as: Among them, ∈ adj (t) represents the dynamic time window at time t, and β represents the adjustment parameter that controls the response sensitivity of the time window.

5. The method for optimizing power load distribution based on intelligent algorithm according to claim 4, characterized in that: The method of using the dynamically adjusted time window data and inputting the LSTM model to perform load forecasting includes: The calculation of each time step in the LSTM network includes the calculation of the forget gate, input gate, candidate layer, and output gate.

6. The method for optimizing power load distribution based on intelligent algorithm according to claim 5, characterized in that: The LSTM model includes: According to the output h of LSTM t , the final power load forecast value is expressed as: in, represents the predicted load value at time t, which represents the prediction result output by the LSTM model; W y represents the prediction weight, b y Represents the prediction bias.

7. The method for optimizing power load distribution based on intelligent algorithm according to claim 6, characterized in that: The method of using an optimization algorithm to intelligently optimize the distribution of power loads according to the load forecast results includes: Use P pred,t Represents the final power load forecast value of the LSTM model Confirm that the optimization goal is to minimize the power distribution cost, which is expressed as: Among them, F opt (x) represents the optimization objective function and the total cost of the power load optimization problem; Ch t P represents the electricity cost in period t, which is calculated based on the electricity market price and load demand fluctuations; t (x t ) represents the power load allocated in time period t; P pred,t represents the power load demand predicted by the LSTM model in time period t; t represents the penalty coefficient, which is used to balance the error between the allocated amount and the predicted value, reflecting the cost when the allocated amount deviates from the predicted value; t represents the penalty coefficient, which is used to deal with power overload conditions; Represents the power overload indication function. When P t (x t ) exceeds the maximum system load P max,t The value is 1 when , otherwise it is 0.

8. The method for optimizing power load distribution based on intelligent algorithm according to claim 7, characterized in that: The method of using an optimization algorithm to intelligently optimize the distribution of power load according to the load forecast result also includes: The particle swarm optimization algorithm is used to optimize the objective function through the position and velocity of the particles. Each particle in the particle swarm represents a power distribution plan, and the particle update rule is expressed as: in, represents the velocity of particle i in time period t; represents the power distribution of particle i in time period t; w represents the inertia weight, which controls the influence of the particle's previous speed; c1 and c2 represent learning factors, which represent the learning degree of the particle for individual optimality and group optimality respectively; r1 and r2 represent random numbers with a value range of [0,1], which are used to introduce randomness; represents the personal optimal solution of particle i; g best represents the optimal solution of the group; The constraints of the particle swarm optimization algorithm include power compliance limits and load balancing requirements; The power load limit is that the power allocation in each period should be within the maximum load range, and the load balance requirement is that the total power demand should be equal to the total power allocation.

9. The method for optimizing power load distribution based on intelligent algorithm according to claim 8, characterized in that: The constraints also include: Add a new load forecast accuracy constraint, expressed as: Among them, P t (x t ) represents the power load distribution in time period t, which is the power load value calculated in the particle swarm optimization algorithm; P pred,t represents the load forecast value for time period t, calculated by the LSTM model; The maximum allowable forecast error threshold for time period t is defined as the maximum difference between the load distribution and the load forecast value in time period t, which is determined by the tolerance error of the power system and the change range of the power market.

10. An intelligent algorithm-based power load optimization distribution system using the method according to any one of claims 1 to 9, characterized in that: The preprocessing module collects historical load data of the power system and performs preprocessing; The prediction module calculates the fractal dimension of the load time series, identifies the critical point of load change, and dynamically adjusts the time window size of the load prediction; Use the dynamically adjusted time window data and input it into the LSTM model for load forecasting; The calculation module uses an optimization algorithm to intelligently optimize the distribution of power load based on the load forecast results.

Images