Power demand data analysis method and system based on artificial intelligence

Abstract

The invention discloses a power demand data analysis method and system based on artificial intelligence, and belongs to the technical field of intelligent power grid and power system automation, and the method comprises the steps: collecting multi-source heterogeneous data, and carrying out the preprocessing of the data; analyzing a causal relationship among the multi-source data by using a causal discovery algorithm, eliminating irrelevant factors, selecting a final variable, and outputting a causal reasoning result; constructing a deep learning model, and outputting a prediction result; and optimizing a power dispatching strategy according to a prediction result. According to the method, through the time alignment and anomaly detection mechanism of the multi-source heterogeneous data, the problem of time sequence dislocation during multi-protocol data fusion is effectively solved, and the data quality is improved; in combination with a causal discovery algorithm and statistical test, a real causal relationship is accurately identified, and false correlation interference in traditional correlation analysis is eliminated; the time sequence neural network and the causal reasoning model are fused, the interpretability of the model is enhanced while historical data time sequence features are reserved, and the prediction robustness under the extreme working condition is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of smart grid and power system automation technology, and in particular to an artificial intelligence-based power demand data analysis method and system. Background Art

[0002] Against the backdrop of the rapid development of smart grids, the accuracy of power demand forecasting directly impacts grid dispatch efficiency and operational economics. Traditional methods typically rely on statistical analysis of historical load data and incorporate environmental factors such as temperature and season to construct forecasting models (such as ARIMA and support vector machines). However, with the explosive growth of multi-source, heterogeneous data, existing technologies face significant bottlenecks. First, regarding data integration, multi-protocol data generated by smart meters, SCADA systems, and IoT sensors exhibit varying temporal granularity. Existing time alignment methods often employ fixed-frequency resampling, resulting in loss of detail in high-frequency data and, in particular, synchronization errors during periods of sudden load fluctuations. Second, regarding feature selection, traditional Pearson correlation coefficient-based variable selection is susceptible to confounding factors (such as spurious correlations between economic indicators and electricity consumption). While mainstream LSTM forecasting models can capture time series features, they lack causal inference mechanisms. Consequently, forecast errors can fluctuate by over 25% during extreme weather events or sudden power outages. Finally, in the scheduling optimization phase, existing gradient descent methods often employ fixed penalty coefficients, making it difficult to dynamically adapt to load fluctuations, leading to increased supply-demand imbalances during peak hours. Therefore, there is an urgent need for an intelligent analysis system that integrates causal discovery and dynamic optimization to solve systemic problems such as multi-source data fusion, false correlation of variables, insufficient model generalization ability, and rigid scheduling strategies. Summary of the Invention

[0003] To solve the above technical problems, an artificial intelligence-based power demand data analysis method is proposed, which includes collecting multi-source heterogeneous data and preprocessing the data;

[0004] Based on the pre-processed data, a causal discovery algorithm is used to analyze the causal relationship between multi-source data, eliminate irrelevant factors, select the final variables, and output the causal reasoning results;

[0005] Build a deep learning model, optimize it based on causal reasoning results, and output prediction results;

[0006] Optimize power dispatch strategies based on forecast results;

[0007] The use of the causal discovery algorithm to analyze the causal relationship between multi-source data includes performing a variable independence test using a PC algorithm, eliminating redundant variables when the conditional independence value is lower than a threshold; using a linear non-Gaussian model to infer the causal direction, and combining greedy search to optimize the causal structure;

[0008] The optimized power dispatch strategy includes monitoring the continuous deviation between actual power supply and predicted value, triggering multi-source data recalibration and causal graph update, using LSTM to verify abnormal data to generate revised predictions, and optimizing and issuing new dispatch instructions through gradient descent.

[0009] As a preferred solution of the power demand data analysis method based on artificial intelligence of the present invention, wherein: the collecting of multi-source heterogeneous data and preprocessing of the data includes:

[0010] Load data is collected through industrial communication protocols, received in real time using a distributed stream processing platform, and stored in a time series database. The electricity consumption growth rate of industrial and commercial users is obtained through power company reports.

[0011] Obtain grid status data from the SCADA system via the IEC61850 communication protocol and use synchronized phasor measurement devices for power monitoring;

[0012] The preprocessing includes synchronizing the time of multi-source data to a unified time step through time alignment technology, synchronizing the low-frequency data with the linear interpolation method to make the time step of each data source consistent, processing outliers based on the statistical 3σ criterion, and then normalizing each data point through Min-Max.

[0013] As a preferred solution of the power demand data analysis method based on artificial intelligence of the present invention, wherein: the use of a causal discovery algorithm based on pre-processed data to analyze the causal relationship between multi-source data includes:

[0014] Based on the pre-processed multi-source heterogeneous data, candidate variables related to power demand are preliminarily selected through physical mechanisms. The PC algorithm is used to conduct independence tests of causal relationships and eliminate redundant variables.

[0015] The candidate variables include environmental variables and power grid operation variables;

[0016] Environmental variables,grid operation variables include temperature, humidity, grid load, and power supply status;

[0017] When the value of the conditional independence test is less than the set threshold, the current events are independent of each other under the given conditions. The linear non-Gaussian model is used to infer the direction of the causal relationship and is optimized in combination with the causal intervention strategy.

[0018] Through greedy equivalent search, the optimal structure is searched in all causal relationships, and a maximized causal graph is output. The formula for optimizing the causal structure is expressed as:

[0019]

[0020] Among them, P(Di |G) represents data D i The probability under the causal graph G, D i represents the i-th data, G represents the causal graph, argmax G It means that among all causal graphs G, select G that maximizes the following expression.

[0021] As a preferred solution of the power demand data analysis method based on artificial intelligence of the present invention, wherein: the selection of the final variable by eliminating irrelevant factors includes:

[0022] Use defined intervention operations to eliminate false correlation factors, determine the expected value after intervention through intervention operations, calculate the influence between cause and effect, output the causal influence with low correlation, eliminate variables with low correlation but non-causal relationship through Pearson correlation coefficient, and set the correlation threshold to eliminate data with correlation below the threshold;

[0023] The variables with causal relationships after removing false correlations are scored, and the importance scoring formula is expressed as:

[0024] S(X)=w1·|ρ(X,Y)|+w2·I(Y|do(X))

[0025] Among them, S(X) represents the importance score of variable X, w1 and w2 represent weight parameters, and I(Y|do(X)) represents the amount of information that variable X affects Y through causal intervention. The top k variables with the highest S(X) are selected to form the variable dataset X. * , X * ={X1,X2,...,X k}.

[0026] As a preferred solution of the power demand data analysis method based on artificial intelligence described in the present invention, wherein: the deep learning model is constructed, optimized based on the causal reasoning results, and the output prediction results include:

[0027] New coefficients are introduced to adjust the tolerance of scheduling errors, and a deep learning model is constructed. The deep learning model optimizes power scheduling based on power demand values, minimizes power supply costs, and reduces power demand forecast errors. By combining power supply costs with a scheduling error penalty component and a scheduling flexibility control component, the scheduling of power demand is optimized.

[0028] A deep learning model is constructed using a time series neural network model. The causal reasoning results obtained are combined with the deep learning model to capture the time series pattern of electricity demand and enhance the predictive ability of the deep learning model using causal relationships.

[0029] As a preferred solution of the power demand data analysis method based on artificial intelligence of the present invention, wherein: the optimization of the power dispatching strategy according to the prediction results includes:

[0030] The power dispatch optimization problem is solved by combining nonlinear programming with gradient descent method. The constraints are set as the capacity limit of power supply and the balance constraint of system load, and the power dispatch is implemented according to the optimized power supply plan, so that the difference between actual power supply and predicted power demand approaches zero.

[0031] After the dispatch is implemented, the actual operation of the power system is continuously monitored. If the error is greater than the set threshold, the dispatch strategy is adjusted based on the real-time feedback information.

[0032] As a preferred solution of the power demand data analysis method based on artificial intelligence of the present invention, wherein: the adjusting the scheduling strategy according to the real-time feedback information includes:

[0033] Real-time monitoring of the deviation between the actual power supply of the power system and the predicted demand value. When the deviation percentage exceeds the threshold of the corresponding area type for a consecutive preset number of time periods, a dynamic adjustment mechanism is triggered;

[0034] Recalibrate the current real-time data, use linear interpolation to calibrate the time synchronization of multi-source data, and use long-short-term memory networks to reconstruct and verify abnormal power supply data;

[0035] Based on the recalibrated data, the causal relationship diagram is updated, and the newly added causal variables are screened through the causal discovery algorithm. The updated causal variables are input into the demand forecast model to generate a revised power demand forecast value. The revised forecast value is dispatched and optimized using the gradient descent method to generate a new power supply plan. The optimized supply plan is sent to the distribution automation equipment for execution through the power system communication protocol.

[0036] Another object of the present invention is to provide an artificial intelligence-based power demand data analysis system, which can solve the problems of low data integration efficiency, inaccurate causal inference, poor model generalization ability and delayed dispatch response in existing power forecasting through a multi-module collaborative intelligent analysis architecture.

[0037] To solve the above technical problems, the present invention provides the following technical solutions: an artificial intelligence-based power demand data analysis system, comprising: a multi-source data governance module, a causal reasoning engine module, a time series-causal fusion prediction module, and an adaptive scheduling optimization module;

[0038] The multi-source data governance module uses a time alignment engine and anomaly detection unit to achieve cross-source data synchronization and quality improvement;

[0039] The causal inference engine module is a hybrid architecture based on the PC algorithm and the linear non-Gaussian model. It performs variable independence test and causal direction inference, combines greedy search to generate the optimal causal graph, and eliminates false correlation interference.

[0040] The time series-causal fusion prediction module extracts historical load time series features through the LSTM network and dynamically integrates them with the key variable weights output by causal reasoning to form an interpretable composite prediction model.

[0041] The adaptive scheduling optimization module integrates a nonlinear programming solver and a real-time feedback mechanism, dynamically adjusts the power supply strategy according to the prediction results, iteratively optimizes the supply and demand matching through a gradient descent algorithm, and triggers the system self-repair process based on deviation monitoring.

[0042] A computer device includes a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the power demand data analysis method based on artificial intelligence when executing the computer program.

[0043] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the power demand data analysis method based on artificial intelligence.

[0044] The beneficial effects of the present invention are as follows: the present invention effectively solves the problem of timing misalignment during multi-protocol data fusion and improves data quality through the time alignment and anomaly detection mechanism of multi-source heterogeneous data; combines causal discovery algorithms with statistical tests to accurately identify true causal relationships and eliminate false correlation interference in traditional correlation analysis; integrates time series neural networks and causal reasoning models to enhance the interpretability of the model while retaining the time series characteristics of historical data, and significantly improves the prediction robustness under extreme working conditions; introduces a dynamic scheduling optimization strategy, and adaptively adjusts the prediction model and scheduling parameters through a real-time feedback mechanism, breaking through the limitations of traditional fixed threshold control and realizing intelligent closed-loop control of power supply and demand balance. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0046] Figure 1 An overall flow chart of an artificial intelligence-based power demand data analysis method provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0047] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention.

[0048] Example 1, reference Figure 1 , which is the first embodiment of the present invention, provides an artificial intelligence-based power demand data analysis method, comprising:

[0049] Collect multi-source heterogeneous data and preprocess the data.

[0050] Based on the preprocessed data, the causal relationship between multi-source data is analyzed using a causal discovery algorithm, irrelevant factors are eliminated to select the final variables, and the causal reasoning results are output.

[0051] Build a deep learning model, optimize it based on causal reasoning results, and output prediction results.

[0052] Optimize power dispatch strategies based on prediction results.

[0053] Load data is read from smart meters, substation monitoring equipment, and distribution automation systems through the MODBUS / IEC104 protocol. The Kafka stream processing framework is used for real-time data reception and stored in a time series database.

[0054] Deploy IoT sensors in industrial and residential areas to collect meteorological data in real time.

[0055] Obtain the electricity consumption growth rate of industrial and commercial users through power company reports.

[0056] Grid status data is acquired from the SCADA system using the IEC61850 communication protocol, and synchronized phasor measurement devices are used for high-precision power monitoring. Due to the heterogeneity of the collected data, standardization is required. Time alignment technology is used to synchronize multi-source data to a unified time step, ensuring data consistency and avoiding data errors caused by time mismatches.

[0057] In multi-source data collection scenarios, different data sources have different time granularities. Smart meter data is collected at a higher frequency, while meteorological data is collected at a lower frequency, resulting in different data time series. To ensure temporal consistency between different data sources, linear interpolation is used to synchronize low-frequency data, ensuring consistent time steps across all data sources.

[0058] This processing method can ensure the temporal consistency of multi-source data used in subsequent analysis, avoid the impact of time dislocation on analysis results, and improve data quality and analysis accuracy.

[0059] Outlier processing detects extreme values ​​in the data through the 3σ criterion based on statistics, effectively removes erroneous data and improves data quality.

[0060] The mean and standard deviation of the dataset were calculated. The experimenters set a threshold of ±3 times the standard deviation of the mean, with data outside this range being considered outliers. An LSTM autoencoder was used to model the data after removing outliers to further verify the validity of the data.

[0061] Outlier processing can not only remove abnormal data points, but also further model and verify the rationality of the data through LSTM autoencoders, improve the accuracy of the data, and provide more reliable input for subsequent analysis.

[0062] Normalization can convert data into the same dimensional range, so that the influence of all features on the model is relatively balanced.

[0063] Each data point is normalized by Min-Max.

[0064] The normalized data range is unified, which helps improve the training efficiency and accuracy of machine learning models and avoids deviations caused by inconsistent feature dimensions. Especially when using deep learning models, normalization helps speed up convergence and improve stability.

[0065] Based on the collected multi-source heterogeneous data, a causal relationship diagram is constructed to conduct causal reasoning. First, expert knowledge and physical mechanisms are used to preliminarily select candidate variables related to power demand. Using the PC algorithm, causal independence tests are performed, redundant variables are eliminated, and a more accurate causal relationship structure is obtained.

[0066] The relevant candidate variables include environmental factors, socio-economic factors, and power grid operation factors.

[0067] The environmental factors, socio-economic factors and grid operation factors include temperature, humidity, economic growth rate, grid compliance and power supply status.

[0068] When the conditional independence obtained from the independence test is close to 0, X and Y are considered independent under the given conditions, which helps to remove indirect influencing variables. Next, a linear non-Gaussian model is used to further infer the direction of the causal relationship and optimize it in combination with the causal intervention strategy. The specific formula for the independence test is expressed as:

[0069] CI(X,Y|Z)=P(X|Y,Z)-P(X|Z)

[0070] Among them, X and Z represent influencing variables, and Y represents the target variable. When the value of CI(X,Y|Z) is approximately 0, it is determined that variables X and Y are independent given Z.

[0071] The obtained causal relationship is further inferred using a linear non-Gaussian model. The specific regression model formula is expressed as:

[0072] Y=βX+∈

[0073] Among them, ∈ represents the error term, β represents the causal effect coefficient, if the error term ∈ is not independent of the influencing variable X, then X→Y is the causal direction.

[0074] Through greedy equivalent search, the optimal structure is searched in all possible causal graphs to find a maximized causal graph. The formula for optimizing the causal structure is expressed as:

[0075] Among them, P(D i |G) represents data D i The probability under the causal graph G, D i represents the i-th data, G represents the causal graph, argmax G It means that among all causal graphs G, select G that maximizes the following expression.

[0076] In actual data analysis, correlation between variables is not equivalent to causality. False correlation factors are usually caused by potential common causes or other external factors, rather than direct causal influences. These false correlation factors need to be eliminated from the obtained causal relationship. First, define the intervention operation. The specific formula for the intervention operation is:

[0077] P(Y|do(X))

[0078] Where do(X) represents the artificial control of X, and the expected value of Y after the given intervention is obtained through the intervention operation formula.

[0079] Calculate the causal influence to obtain the causal influence with lower correlation. The specific formula for calculating the causal influence is expressed as:

[0080] ΔY=E[Y|do(X1)]-E[Y|do(X2)]

[0081] Among them, ΔY represents the causal change effect of variable Y, do(X) represents the intervention operation in causal reasoning, and E represents the expectation symbol E[Y|do(X)] represents conditional expectation.

[0082] The Pearson correlation coefficient was calculated and used to eliminate variables with low correlation but non-causal relationships. Experimenters set a correlation threshold based on data analysis and experience to eliminate data with low correlation.

[0083] The variables with causal relationships after removing false correlations are scored, and the importance scoring formula is expressed as:

[0084] S(X)=w1·|ρ(X,Y)|+w2·I(Y|do(X))

[0085] Among them, S(X) represents the importance score of variable X, w1 and w2 represent weight parameters, and I(Y|do(X)) represents the amount of information that variable X affects Y through causal intervention. The top k variables with the highest S(X) are selected to form the variable dataset X. * , X * ={X1,X2,...,X k}.

[0086] A new coefficient is introduced to adjust the tolerance of scheduling errors, and a deep learning model is constructed to enhance the flexibility of the model. The deep learning model optimizes power scheduling according to the power demand value, minimizes the power supply cost, and reduces the prediction error of power demand. By combining the power supply cost with the scheduling error penalty part and the scheduling flexibility control part, the scheduling optimization of power demand is achieved. The specific formula of the power supply cost and scheduling error penalty part is expressed as:

[0087] The penalty part that mainly deals with the cost of power supply and scheduling error ensures that when minimizing the objective function, both the actual cost of power supply and reasonable penalty for prediction error are taken into account.

[0088] The specific formula of the scheduling flexibility control part is expressed as:

[0089] F=θ·(P a (t)-P p (t)))

[0090] The dispatch flexibility control part introduces a dispatch flexibility control coefficient, which is adjusted to control the impact of dispatch error on the objective function and further optimize the power dispatch process.

[0091] The specific calculation formula of the scheduling flexibility control coefficient is expressed as:

[0092]

[0093] The scheduling flexibility control coefficient θ is calculated by weighting the historical scheduling data to reflect the relationship between the prediction error and the power supply cost. The introduction of the flexibility control coefficient can make the scheduling optimization process more flexible in scenarios where the power demand fluctuates greatly.

[0094] The specific formula for power demand dispatch optimization is expressed as:

[0095] L=E+F

[0096] Where L represents the total loss function of power dispatch optimization, P a (t) represents the actual power supply at time t, P p (t) represents the predicted power demand value at time t, c t represents the electricity supply cost at time t, d t represents the error penalty coefficient at time t, and θ represents the scheduling flexibility control coefficient.

[0097] A deep learning model is constructed using a long short-term memory network. The causal reasoning results are combined with the deep learning model to capture the time series pattern of electricity demand and enhance the model's predictive ability by leveraging causal relationships. Causal reasoning is used to determine which causal variables have a significant impact on electricity demand, and the impact is quantified through causal intervention. These causal variables are used as input and combined with the deep learning model for time series prediction. The specific formula for calculating causal variables is expressed as:

[0098]

[0099] Here, A represents the weighted combination of causal variables and historical data. By weighting causal variables, it captures the impact of causal relationships on electricity demand. Combining the output of the causal inference model with the time series learning results of the LSTM network further optimizes the power forecasting process. The specific forecast optimization formula is expressed as:

[0100] B=γ·f LSTM X * δ·P c X *

[0101] Where B represents the output of the LSTM and causal inference model. This part mainly combines the output of the causal inference model with the time series learning results of the LSTM network to further optimize the prediction of power demand. The specific formula of the combined model is expressed as:

[0102] Y p,t =A+B

[0103] Among them, Y p,t represents the predicted power demand value at time t, X c,i,t-krepresents the data of the i-th causal variable at time tk, α i represents the influence coefficient of the i-th causal variable on electricity demand, β k represents the weighting coefficient at time tk in the historical data, f LSTM represents the output of the LSTM model, P c Represents the output of a causal inference model. This combined model extracts temporal features from historical data and adjusts input variables based on causal relationships. This allows for a stronger focus on actual causal relationships during prediction, enhancing the model's interpretability and accuracy.

[0104] To ensure the accuracy and feasibility of power dispatch optimization and its applicability to actual power system dispatch problems, a nonlinear programming method combined with gradient descent is used to solve the power dispatch optimization problem. Nonlinear programming methods are suitable for optimization problems involving square terms and nonlinear constraints. Since the total loss function includes the square term of the dispatch error, the nonlinear programming method can effectively handle these nonlinear terms and find the optimal power dispatch solution. The constraints are set as the balance constraint between the power supply capacity limit and the system load. The balance constraint between the power supply capacity limit and the system load is specifically expressed as:

[0105]

[0106] Among them, P min represents the minimum power supply at time t, P max P min Indicates the maximum power supply at time t.

[0107]

[0108] This constraint ensures that the actual electricity supply is consistent with the forecasted electricity demand.

[0109] Gradient descent is a commonly used optimization algorithm. Especially when the objective function has a continuous and differentiable property, the gradient descent method can converge quickly and find the optimal solution. For nonlinear optimization in power dispatching problems, the gradient descent method can effectively adjust the decision variables and reduce the objective function value. First, the gradient of the loss function with respect to the decision variable is calculated. The gradient calculation formula is specifically expressed as:

[0110]

[0111] The partial derivative of the force supply.

[0112] The power supply value is adjusted by updating the rules. The rule updating formula is specifically expressed as:

[0113]

[0114] Among them, η represents the learning rate, Pa (t) old Indicates the power supply value of the previous iteration, P a (t) new Indicates the updated power supply value.

[0115] Power dispatch is implemented based on the optimized power supply plan to minimize the discrepancy between actual power supply and forecasted power demand. After the dispatch is implemented, the actual operation of the power system is continuously monitored. If significant discrepancies occur, the dispatch strategy is adjusted based on real-time feedback.

[0116] It should be further explained that the real-time feedback information adjustment scheduling strategy includes real-time monitoring of the deviation between the actual power supply of the power system and the predicted demand value. When the deviation percentage of the consecutive preset time periods exceeds the threshold of the corresponding regional type, the dynamic adjustment mechanism is triggered; the current real-time data is recalibrated, and the multi-source data is time-synchronized and calibrated using linear interpolation method, and the abnormal power supply data is reconstructed and verified using long and short-term memory network; the causal relationship diagram is updated based on the recalibrated data, and the newly added causal variables are screened by the causal discovery algorithm, and the updated causal variables are input into the demand forecast model to generate a revised power demand forecast value, and the revised forecast value is optimized by the gradient descent method to generate a new power supply plan, and the optimized supply plan is sent to the distribution automation equipment for execution through the power system communication protocol.

[0117] Example 2 is the second embodiment of the present invention:

[0118] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0119] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0120] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic devices), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering, or processing in another suitable manner as necessary, and then stored in a computer memory.

[0121] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0122] Example 3 is the third embodiment of the present invention. This embodiment provides an artificial intelligence-based power demand data analysis system, including a multi-source data governance module, a causal reasoning engine module, a time series-causal fusion prediction module, and an adaptive scheduling optimization module.

[0123] The multi-source data governance module uses a time alignment engine and anomaly detection unit to achieve cross-source data synchronization and quality improvement.

[0124] The causal reasoning engine module is a hybrid architecture based on the PC algorithm and the linear non-Gaussian model. It performs variable independence tests and causal direction inference, combines greedy search to generate the optimal causal graph, and eliminates false correlation interference.

[0125] The time series-causal fusion prediction module extracts historical load time series features through the LSTM network, and dynamically integrates them with the key variable weights output by causal reasoning to form an interpretable composite prediction model.

[0126] The adaptive scheduling optimization module integrates a nonlinear programming solver and a real-time feedback mechanism. It dynamically adjusts the power supply strategy according to the prediction results, iteratively optimizes the supply and demand matching through the gradient descent algorithm, and triggers the system self-repair process based on deviation monitoring.

[0127] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. The power demand data analysis method based on artificial intelligence is characterized by: include, Collect multi-source heterogeneous data and pre-process the data; Based on the pre-processed data, a causal discovery algorithm is used to analyze the causal relationship between multi-source data, eliminate irrelevant factors, select the final variables, and output the causal reasoning results; Build a deep learning model, optimize it based on causal reasoning results, and output prediction results; Optimize power dispatch strategies based on forecast results; The use of the causal discovery algorithm to analyze the causal relationship between multi-source data includes performing a variable independence test using a PC algorithm and eliminating redundant variables when the conditional independence value is lower than a threshold; A linear non-Gaussian model is used to infer causal direction, combined with greedy search to optimize the causal structure; The optimized power dispatch strategy includes monitoring the continuous deviation between actual power supply and predicted value, triggering multi-source data recalibration and causal graph update, using LSTM to verify abnormal data to generate revised predictions, and optimizing and issuing new dispatch instructions through gradient descent.

2. The power demand data analysis method based on artificial intelligence according to claim 1, characterized in that: The collecting of multi-source heterogeneous data and preprocessing of the data includes: Load data is collected through industrial communication protocols, received in real time using a distributed stream processing platform, and stored in a time series database. The electricity consumption growth rate of industrial and commercial users is obtained through power company reports. Obtain grid status data from the SCADA system via the IEC61850 communication protocol and use synchronized phasor measurement devices for power monitoring; The preprocessing includes synchronizing the time of multi-source data to a unified time step through time alignment technology, synchronizing the low-frequency data with the linear interpolation method to make the time step of each data source consistent, processing outliers based on the statistical 3σ criterion, and then normalizing each data point through Min-Max.

3. The power demand data analysis method based on artificial intelligence according to claim 2, characterized in that: The method of analyzing the causal relationship between multi-source data using a causal discovery algorithm based on pre-processed data includes: Based on the pre-processed multi-source heterogeneous data, candidate variables related to power demand are preliminarily selected through physical mechanisms. The PC algorithm is used to conduct independence tests of causal relationships and eliminate redundant variables. The candidate variables include environmental variables and power grid operation variables; Environmental variables,grid operation variables include temperature, humidity, grid load, and power supply status; When the value of the conditional independence test is less than the set threshold, the current events are independent of each other under the given conditions. The linear non-Gaussian model is used to infer the direction of the causal relationship and is optimized in combination with the causal intervention strategy. Through greedy equivalent search, the optimal structure is searched in all causal relationships, and a maximized causal graph is output. The formula for optimizing the causal structure is expressed as: Among them, P(D i |G) represents data D i The probability under the causal graph G, D i represents the i-th data, G represents the causal graph, argmax G It means that among all causal graphs G, select G that maximizes the following expression.

4. The power demand data analysis method based on artificial intelligence according to claim 3, characterized in that: The selection of the final variables by eliminating irrelevant factors includes: Use defined intervention operations to eliminate false correlation factors, determine the expected value after intervention through intervention operations, calculate the influence between cause and effect, output the causal influence with low correlation, eliminate variables with low correlation but non-causal relationship through Pearson correlation coefficient, and set the correlation threshold to eliminate data with correlation below the threshold; The variables with causal relationships after removing false correlations are scored, and the importance scoring formula is expressed as: Among them, S(X) represents the importance score of variable X, W1 and w2 represent weight parameters, and I(Y|do(X)) represents the amount of information that variable X affects Y through causal intervention. The top k variables with the highest S(X) are selected to form the variable dataset X. * , X * ={X1,X2,...,X k }.

5. The power demand data analysis method based on artificial intelligence according to claim 4, characterized in that: The deep learning model is constructed and optimized based on the causal reasoning results, and the output prediction results include: New coefficients are introduced to adjust the tolerance of scheduling errors, and a deep learning model is constructed. The deep learning model optimizes power scheduling based on power demand values, minimizes power supply costs, and reduces power demand forecast errors. By combining power supply costs with a scheduling error penalty component and a scheduling flexibility control component, the scheduling of power demand is optimized. A deep learning model is constructed using a time series neural network model. The causal reasoning results obtained are combined with the deep learning model to capture the time series pattern of electricity demand and enhance the predictive ability of the deep learning model using causal relationships.

6. The power demand data analysis method based on artificial intelligence according to claim 5, characterized in that: The optimization of the power dispatch strategy according to the prediction results includes: The power dispatch optimization problem is solved by combining nonlinear programming with gradient descent method. The constraints are set as the capacity limit of power supply and the balance constraint of system load, and the power dispatch is implemented according to the optimized power supply plan, so that the difference between actual power supply and predicted power demand approaches zero. After the dispatch is implemented, the actual operation of the power system is continuously monitored. If the error is greater than the set threshold, the dispatch strategy is adjusted based on the real-time feedback information.

7. The power demand data analysis method based on artificial intelligence according to claim 6, characterized in that: The adjusting of the scheduling strategy according to the real-time feedback information includes: Real-time monitoring of the deviation between the actual power supply of the power system and the predicted demand value. When the deviation percentage exceeds the threshold of the corresponding area type for a consecutive preset number of time periods, a dynamic adjustment mechanism is triggered; Recalibrate the current real-time data, use linear interpolation to calibrate the time synchronization of multi-source data, and use long-short-term memory networks to reconstruct and verify abnormal power supply data; Based on the recalibrated data, the causal relationship diagram is updated, and the newly added causal variables are screened through the causal discovery algorithm. The updated causal variables are input into the demand forecast model to generate a revised power demand forecast value. The revised forecast value is dispatched and optimized using the gradient descent method to generate a new power supply plan. The optimized supply plan is sent to the distribution automation equipment for execution through the power system communication protocol.

8. A system using the artificial intelligence-based power demand data analysis method according to any one of claims 1 to 7, characterized in that: It includes multi-source data governance module, causal reasoning engine module, time series-causal fusion prediction module, and adaptive scheduling optimization module; The multi-source data governance module uses a time alignment engine and anomaly detection unit to achieve cross-source data synchronization and quality improvement; The causal inference engine module is a hybrid architecture based on the PC algorithm and the linear non-Gaussian model. It performs variable independence test and causal direction inference, combines greedy search to generate the optimal causal graph, and eliminates false correlation interference. The time series-causal fusion prediction module extracts historical load time series features through the LSTM network and dynamically integrates them with the key variable weights output by causal reasoning to form an interpretable composite prediction model. The adaptive scheduling optimization module integrates a nonlinear programming solver and a real-time feedback mechanism, dynamically adjusts the power supply strategy according to the prediction results, iteratively optimizes the supply and demand matching through a gradient descent algorithm, and triggers the system self-repair process based on deviation monitoring.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the power demand data analysis method based on artificial intelligence are implemented as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the power demand data analysis method based on artificial intelligence according to any one of claims 1 to 7 are implemented.

Images