Power system multi-source data intelligent fusion method based on soft merging D-S evidence theory

Abstract

The invention discloses a power system multi-source data intelligent fusion method based on a soft merging D-S evidence theory, and relates to the technical field of power system industrial multi-source heterogeneous data usage, comprising: acquiring multi-source data, and calculating a change index of the multi-source data; performing data detection on the multi-source data to obtain a detection result; performing aggregation processing on the multi-source data based on the change index and the detection result to obtain data features; and fusing the data features to obtain a data fusion result. According to the method, abnormal data is detected by aiming at the volatility and change characteristics of the power grid data and combining a Z-score method, an aggregation mode is flexibly selected, and the accuracy of data processing is improved; multi-source heterogeneous data are fused by using a D-S evidence theory, and in the face of evidence conflicts, a soft merging method is used for introducing regulatory factors to relieve conflict influences, so that the problem that time scales are not matched in the use process of industrial source data is effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of the use of industrial multi-source heterogeneous data in the power system, and particularly to an intelligent fusion method for multi-source data in the power system based on the soft combination D-S evidence theory. Background Art

[0002] With the continuous expansion of the power grid scale and the increase in operation complexity, the power grid monitoring system needs to process multi-modal data from different data sources, such as PMU, SCADA or meteorological data; however, there are significant differences in the sampling frequencies and time scales of these data. For example, PMU provides high-frequency data at the millisecond level, while SCADA usually samples at intervals of minutes or longer. This inconsistency in time scale makes it difficult to directly use the data for industrial analysis; to address this issue, it is urgent to dynamically select an appropriate time scale aggregation method according to the data characteristics to uniformly process data from different sources to achieve data consistency and integrity; at the same time, to fully explore the key characteristics of the aggregated data, it is also necessary to further perform feature fusion to uniformly quantify and analyze the dynamic change characteristics of multi-source data. Summary of the Invention

[0003] The purpose of this part is to outline some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this part, as well as in the abstract and title of the present application, to avoid obscuring the purpose of this part, the abstract, and the title, but such simplifications or omissions shall not be used to limit the scope of the present invention.

[0004] In view of the above problems, the present invention is proposed.

[0005] Therefore, the present invention provides an intelligent fusion method for multi-source data in the power system based on the soft combination D-S evidence theory, which can solve the problems mentioned in the background art.

[0006] To solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides an intelligent fusion method for multi-source data in the power system based on the soft combination D-S evidence theory, which includes obtaining multi-source data and calculating the change index of the multi-source data; performing data detection on the multi-source data to obtain a detection result; performing aggregation processing on the multi-source data based on the change index and the detection result to obtain data features; and performing fusion on the data features to obtain a data fusion result.

[0008] As a preferred solution of the intelligent fusion method for multi-source data in the power system based on the soft combination D-S evidence theory of the present invention, wherein: the change index includes volatility and change rate.

[0009] As a preferred embodiment of the intelligent multi-source data fusion method for power systems based on soft combination D-S evidence theory of the present invention, the data detection of the multi-source data includes the following steps: performing abnormal data detection on the multi-source data, and determining whether the multi-source data is an abnormal point according to the mean and standard deviation of the multi-source data to obtain a detection result.

[0010] As a preferred embodiment of the intelligent multi-source data fusion method for power systems based on soft combination D-S evidence theory of the present invention, the aggregation processing includes mean aggregation, extreme value aggregation, and median aggregation; the extreme value aggregation includes maximum aggregation and minimum aggregation.

[0011] As a preferred embodiment of the intelligent multi-source data fusion method for power systems based on soft combination D-S evidence theory of the present invention, the aggregation processing of the multi-source data based on the change index and the detection result includes: making a judgment based on the volatility, change rate, and detection result, and deciding to perform mean aggregation, extreme value aggregation, or median aggregation on the multi-source data according to the judgment result.

[0012] As a preferred embodiment of the intelligent multi-source data fusion method for power systems based on soft combination D-S evidence theory of the present invention, the fusion of the data features includes the following steps: obtaining the data features as evidence sources, performing basic probability assignment on the evidence sources; calculating the conflict degree between the evidence sources; introducing a regulation factor, and combining the evidence sources by using the soft combination D-S evidence theory according to the regulation factor and the conflict degree; and performing normalization processing on the combination result to obtain a data fusion result.

[0013] As a preferred embodiment of the intelligent multi-source data fusion method for power systems based on soft combination D-S evidence theory of the present invention, the core idea of the soft combination D-S evidence theory is to control the influence of the conflict degree K on the combination result by introducing a regulation factor α.

[0014] In a second aspect, to further solve the security problems existing in the use of industrial multi-source heterogeneous data in power systems, the embodiment provides an intelligent multi-source data fusion system for power systems based on soft combination D-S evidence theory, which includes: a data acquisition module for acquiring multi-source data and calculating the change index of the multi-source data; a data detection module for performing data detection on the multi-source data to obtain a detection result; an aggregation processing module for performing aggregation processing on the multi-source data based on the change index and the detection result to obtain data features; and a feature fusion module for fusing the data features to obtain a data fusion result.

[0015] In a third aspect, an embodiment of the present invention provides a computer device, including a memory and a processor, where the memory stores a computer program, and: when the computer program is executed by the processor, any step of the power system multi-source data intelligent fusion method based on soft combination D-S evidence theory as described in the first aspect of the present invention is implemented.

[0016] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, and: when the computer program is executed by the processor, any step of the power system multi-source data intelligent fusion method based on soft combination D-S evidence theory as described in the first aspect of the present invention is implemented.

[0017] Advantages of the present invention: By targeting the volatility and change characteristics of power grid data, combining the Z-score method to detect abnormal data, and flexibly selecting the aggregation method, the accuracy of data processing is greatly improved, ensuring that the aggregation result can truly reflect the state of the power grid system; by using D-S evidence theory to fuse multi-source heterogeneous data, and when facing evidence conflicts, using the soft combination method to introduce a regulation factor to mitigate the impact of conflicts, the state assessment is made more stable and reliable, adapting to the fusion requirements of multi-source data, and effectively solving the problem of time-scale mismatch in the use of industrial source data. Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for description in the embodiments will be briefly introduced below. Obviously, the 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:

[0019] Figure 1 It is the overall flowchart of the power system multi-source data intelligent fusion method based on soft combination D-S evidence theory in Embodiment 1.

[0020] Figure 2 It is the structural schematic diagram of the computer device in Embodiment 3. Detailed Embodiments

[0021] To make the above objects, features, and advantages of the present invention more obvious and understandable, the detailed embodiments of the present invention will be described in detail below with reference to the drawings in the specification.

[0022] Many specific details are set forth in the following description to facilitate a full understanding of the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar extensions without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0023] Second, as used herein, "an embodiment" or "embodiments" refer to specific features, structures, or characteristics that may be included in at least one implementation of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or alternative embodiments that exclude each other.

[0024] Embodiment 1

[0025] Referring to Figure 1 , which is the first embodiment of the present invention. This embodiment provides an intelligent multi-source data fusion method for power systems based on soft combination D-S evidence theory.

[0026] The existing multi-source data fusion methods for power systems mainly have the following problems: With the continuous expansion of the power grid scale and the increase in operation complexity, the power grid monitoring system needs to process multi-modal data from different data sources, such as PMU, SCADA, or meteorological data; however, there are significant differences in the sampling frequencies and time scales of these data. For example, PMU provides high-frequency data in milliseconds, while SCADA usually samples at minute or longer intervals. This inconsistency in time scale makes it difficult to directly use the data for industrial analysis.

[0027] This application provides a solution that can effectively solve the above-mentioned problems. Next, multiple embodiments will be combined to elaborate in detail how to implement the intelligent multi-source data fusion method for power systems based on soft combination D-S evidence theory.

[0028] Figure 1 shows the overall flowchart of the intelligent multi-source data fusion method for power systems based on soft combination D-S evidence theory, including:

[0029] S1: Obtain multi-source data and calculate the change indicators of the multi-source data.

[0030] In an optional embodiment, when solving the intelligent multi-source data fusion of power systems, adaptive aggregation can be used to process multi-source data, where adaptive aggregation can dynamically select the aggregation method according to the volatility, change pattern, and anomalies of the data to better capture the trends and characteristics of the power grid system; first, define and calculate the basic data characteristics for adaptive aggregation. Usually, power grid data (such as node voltage, active power, and reactive power) fluctuates over time, and it is necessary to analyze these volatility characteristics to determine whether adaptive adjustment is required.

[0031] In the embodiments of this application, the change indicators include volatility and change rate.

[0032] In the embodiments of this application, volatility refers to an indicator that measures the degree of data change, usually represented by variance and standard deviation.

[0033] In the embodiments of the present application, the change rate refers to the rate of change of data over time, where the greater the data volatility, the higher the change rate.

[0034] Exemplarily, calculating volatility includes calculating the variance and standard deviation within a time window [t - Δt, t], and the specific formulas are as follows:

[0035]

[0036] where Var(t) is the variance; x i (t) is the i-th data point within the time window; μ(t) is the mean of the data within the window; σ(t) is the standard deviation.

[0037] Furthermore, the specific formula for the change rate is as follows:

[0038]

[0039] where α is the change rate; x(t) is the data at the current time point; x(t - Δt) is the data of the previous time window; Δt is the time difference.

[0040] In an alternative embodiment, various methods and techniques can also be used to calculate volatility and change rate to improve the comprehensiveness of data feature extraction; for example, when performing volatility analysis, in addition to variance and standard deviation, information entropy can be used to measure the uncertainty of data distribution, or fractal dimension can be used to quantify the complexity of fluctuations, or even the sample entropy within a sliding window can be combined to analyze the regularity of the data sequence; when performing change rate analysis, in addition to the formula based on time difference, moving average convergence divergence can be introduced to capture the trend strength, or the dynamic time warping algorithm can be used to quantify the change rate of non-steady data, and for high-frequency noise scenarios, wavelet transform can also be combined to decompose multi-scale change features; in addition, for periodic data, frequency domain features can be extracted through Fourier transform, and the calculation of volatility and change rate can be extended to the frequency domain space to further enrich the dimension of data features, and specific limitations are not made in this embodiment.

[0041] S2: Perform data detection on multi-source data to obtain a detection result.

[0042] It should be noted that in power grid monitoring, abnormal data (such as voltage mutations, current peaks) may be caused by equipment failures or emergencies, and some methods are needed to detect these abnormalities and select appropriate aggregation methods.

[0043] In the embodiments of the present application, performing data detection on multi-source data includes the following steps: using the Z-score method to detect abnormal data in multi-source data, and determining whether the multi-source data is an abnormal point based on the mean and standard deviation of the multi-source data to obtain a detection result, and the specific formula is as follows:

[0044]

[0045] Among them, x(t) is the data at the current time point; μ(t) is the mean value of the data within this time window; σ(t) is the standard deviation; when |Z| > α (usually α = 2 or α = 3 is selected), the data is determined to be an outlier point.

[0046] In an optional embodiment, anomaly detection can also be implemented by selecting various methods according to the characteristics of power grid data. For example, in the basic scenario, the Z-score method is used to judge outlier points through a fixed threshold, which is applicable to scenarios where the data distribution is approximately normal and the noise is low; in complex scenarios, the Isolation Forest or Local Outlier Factor algorithm can be introduced for unsupervised anomaly detection, and outliers are identified through the isolation degree or local density difference of data points; for time series data, an LSTM autoencoder can be used to construct a prediction model, and outlier points are detected through the reconstruction error; in addition, a dynamic threshold mechanism can be combined to adaptively adjust the anomaly determination threshold based on the power grid load cycle or sliding window statistics; in the semi-supervised scenario, a generative adversarial network can be used to learn the normal data distribution, and outliers are identified through the difference between the discriminator or generated data and the real data; for the correlation analysis of multi-source data, an anomaly detection method based on a graph model can be used to mine hidden anomaly patterns through the relationships between nodes. This embodiment does not make specific limitations on this.

[0047] It should be noted that in the existing power grid monitoring and control methods, most of them focus on processing after multi-source data fusion. However, before multi-source data fusion, an adaptive aggregation method is used to process the volatility, change rate, and anomalies of different source data, ensuring the quality and accuracy of the data at the initial stage of data fusion; this method can ensure that when multi-source data fusion is carried out through prior adaptive aggregation selection, the adaptive aggregation method can dynamically select appropriate aggregation strategies (such as mean aggregation, extreme value aggregation, or median aggregation) according to the change patterns of each data source (such as volatility, change rate, and anomaly detection), thereby reducing unnecessary noise and data deviation, guaranteeing the quality and consistency of the fused data, and enabling more stable and meaningful integration results for power grid data from different sources during fusion, laying a foundation for the next feature fusion.

[0048] S3: Aggregate and process the multi-source data based on the change index and detection result to obtain data features.

[0049] In the embodiment of the present application, the aggregation processing includes mean aggregation, extreme value aggregation, and median aggregation.

[0050] In the embodiment of the present application, extreme value aggregation includes maximum value aggregation and minimum value aggregation.

[0051] In the embodiments of the present application, aggregating multi-source data based on change metrics and detection results includes: making a judgment based on volatility, change rate, and detection results, and deciding to perform mean aggregation, extreme value aggregation, or median aggregation on the multi-source data according to the judgment result.

[0052] Exemplarily, in the embodiments of the present application, deciding to perform mean aggregation, extreme value aggregation, or median aggregation on the multi-source data according to the judgment result includes the following steps: making a judgment based on volatility, change rate, and detection results. If the volatility is less than the volatility threshold or the change rate is less than the change threshold, and the detection result is that there is no abnormal data. For example, the variance Var(t) in volatility is less than the threshold τ var , it indicates that the volatility of the data is low and the change is stable. In this case, there will be no extreme changes in the data, so mean aggregation can effectively reflect the overall trend of the data. Use mean aggregation to extract the trend within the time window. The specific formula is as follows:

[0053]

[0054] where Mean(t) is the mean value of the data within the time window t; N is the total number of data points within the time window; x i (t) is the value of the i-th data point at the current time point.

[0055] If the volatility is greater than the volatility threshold or the change rate is greater than the change threshold, and the detection result is that there is no abnormal data. For example, the variance Var(t) in volatility is greater than the threshold τ var , it indicates that the data fluctuates violently, and extreme value aggregation is adopted to capture the extreme changes of the data, and the next step is executed.

[0056] When capturing the maximum voltage peak or current peak event in the power grid, use maximum value aggregation to aggregate the multi-source data, and select the maximum value of the multi-source data within a time window. The specific formula is as follows:

[0057] Max(t) = max(x1(t), x2(t),..., x n (t))

[0058] where Max(t) is the maximum value of the data within the time window t; max is to solve the maximum value of all data points within the time window, indicating to select the maximum value within this time window; N is the total number of data points within the time window; x i (t) is the value of the i-th data point at the current time point.

[0059] When capturing the lowest voltage or current drop event in the power grid, use minimum value aggregation to aggregate the multi-source data and select the minimum value of the multi-source data. The specific formula is as follows:

[0060] Min(t) = min(x1(t), x2(t),..., x n (t))

[0061] Where Min(t) is the minimum value of the data within the time window t; min represents finding the minimum value of all data points within the time window, indicating selecting the smallest value within the time window; N is the total number of data points within the time window; x i (t) is the value of the i-th data point at the current time point.

[0062] Since extreme value aggregation may be overly sensitive to outliers, resulting in distorted data processing, if the volatility is greater than the volatility threshold or the change rate is greater than the change threshold, and the detection result indicates the existence of abnormal data, then median aggregation is used to aggregate the multi-source data. After detecting abnormal data through Z-score, median aggregation is combined to process the time window containing outliers.

[0063] Exemplarily, in the embodiments of the present application, using median aggregation to aggregate multi-source data includes the following steps: For each data point x within the time window i , its standardized value is calculated using the Z-score formula:

[0064]

[0065] Where x i (t) is the value of the i-th data point at the current time point; μ(t) is the mean of the data within the time window; σ(t) is the standard deviation.

[0066] After calculating its standardized value using the Z-score formula, it is determined whether the data point is abnormal. By setting a threshold Z threshold (usually 2 or 3), if the absolute value of the Z-score of a certain data point exceeds the threshold Z threshold , then this data point is considered an outlier. For example, if |Z| > 2 or |Z| > 3, then x i (t) is regarded as an outlier.

[0067] The data point x i (t) detected as abnormal is removed from the data set within the time window. Assuming the original data set within the time window is {x1(t), x2(t),..., x N (t)}, after abnormal data detection, the remaining data is {x i1 (t), x i2 (t),..., x iM (t)}, where M ≤ N and all outliers are excluded.

[0068] Perform median aggregation on the remaining data after excluding abnormal data, and calculate the median of the data set after excluding abnormal points. The median is the middle value in the sorted data set. If the number of data points M is odd, the median is the middle value; if M is even, the median is the average of the two middle values.

[0069] If M is odd, the median is the th element in the sorted data set, and the specific formula is as follows:

[0070]

[0071] where Median(t) is the median aggregation value of the remaining data set after removing abnormal data at time point t; is the th data point in the sorted remaining data set, and t is the time index.

[0072] If M is even, the median is the average of the th and the th elements in the sorted data set, and the specific formula is as follows:

[0073]

[0074] where Median(t) is the median aggregation value of the remaining data set after removing abnormal data at time point t; is the th data point in the sorted remaining data set; is the th data point in the sorted remaining data set; t is the time index.

[0075] The aggregation result is the median after excluding abnormal data, representing the aggregation value of the power grid data within this time window, which can better reflect the data trend under normal operating conditions and avoid the influence of abnormal points on the aggregation result.

[0076] In an optional embodiment, the selection of aggregation processing can be achieved through various methods. For example, based on a reinforcement learning framework, volatility, rate of change, and anomaly detection results are used as state inputs, and the weight distribution of mean, extreme value, and median aggregation is dynamically adjusted through Q-learning or deep reinforcement learning to maximize the robustness of the fusion result. In addition, a dynamic programming algorithm can be introduced to optimize the aggregation strategy in stages according to the data change trend within a time window. For example, when volatility suddenly increases, extreme value aggregation is preferentially adopted, and then smoothly transitioned to mean aggregation. Or combined with online learning technology, the parameters of the aggregation strategy are dynamically updated through real-time feedback. For multi-objective optimization scenarios, a multi-objective evolutionary algorithm can be used to search for Pareto optimal solutions to determine the weight combination of different aggregation methods. In addition, a fuzzy logic controller can be combined to adaptively generate switching rules for the aggregation strategy according to the fuzzy membership functions of data volatility and anomaly degree, etc.

[0077] It should be noted that by analyzing the variance, standard deviation, and rate of change of power grid data, and at the same time using the Z-score method to detect abnormal data, a suitable aggregation method is dynamically selected, thereby improving the accuracy of data processing, ensuring that the aggregation result can truly reflect the state of the power grid system, and at the same time being able to accurately extract the trends and characteristics of the power grid characteristic quantity data, laying a data foundation for subsequent feature fusion.

[0078] S4: Fuse the data features to obtain a data fusion result.

[0079] It should be noted that in power grid monitoring, the features after aggregation processing (such as voltage volatility, active power change rate, or reactive power extreme value) can be regarded as "evidence" from different data sources. These evidences reflect the dynamic changes of different monitoring indicators. Since there may be correlations and conflicts between different indicators, the D-S evidence theory can be further used for fusion to uniformly quantify the reliability and trend of the power grid state.

[0080] In the embodiment of the present application, fusing the data features includes the following steps: obtaining the data features as evidence sources and performing basic probability assignment on the evidence sources.

[0081] Calculate the conflict degree between the evidence sources.

[0082] By introducing an adjustment factor, according to the adjustment factor and the conflict degree, the evidence sources are combined using the soft combination D-S evidence theory.

[0083] Normalize the combined result to obtain a data fusion result.

[0084] In the embodiment of the present application, the basic representation and symbol convention of the evidence source include: assuming that A is a subset in the hypothesis space, the normalization condition is that for all hypotheses A, the following conditions are satisfied:

[0085]

[0086] Among them, Θ is the universal set, representing all possible hypotheses; m(A) is the basic probability assignment attributed to hypothesis A.

[0087] In the embodiments of the present application, the standard Dempster combination rule is used to combine two evidence sources m1 and m2, and the specific formula is as follows:

[0088]

[0089] Among them, is the combined basic probability assignment, representing the confidence level of the occurrence of hypothesis A after combining evidence ε1 and ε2; is the basic probability assignment from evidence source ε1, representing the confidence level that evidence source ε1 believes hypothesis B occurs; is the basic probability assignment from evidence source ε2, representing the confidence level that evidence source ε2 believes hypothesis C occurs; B∩C = A is the intersection between hypotheses B and C from different evidence sources, that is, the part jointly supported by the two pieces of evidence; K is the conflict degree, representing the inconsistent part between the evidences, that is, the situation where there is no intersection between the two evidence sources on B and C.

[0090] In the embodiments of the present application, the specific formula for calculating the conflict degree between evidence sources is as follows:

[0091]

[0092] Among them, K is the conflict degree, representing the inconsistent part between the evidences, that is, the situation where there is no intersection between the two evidence sources on B and C. If the conflict degree K is large, it means that the conflict between the evidence sources is relatively serious, and the standard Dempster combination rule will normalize it to the part of 1 - K, excluding the conflict part.

[0093] Furthermore, the core idea of the soft combination D - S evidence theory is to control the influence of the conflict degree K on the combination result by introducing a regulation factor α. The specific formula of the soft combination D - S evidence theory is as follows:

[0094]

[0095] Among them, α is the regulation factor, used to control the influence of the conflict degree on the final result. When the regulation factor α is very small (close to 0), the soft combination D - S evidence rule is close to the standard D - S evidence combination rule, and the influence of the conflict degree K on the combination result is large; when the regulation factor α is large, the influence of the conflict degree K is weakened, and the combination result depends more on the intersection part of the evidence sources.

[0096] In the embodiments of the present application, normalizing the combination result means that after soft combination, the combination result needs to be normalized to ensure that the sum of probabilities after combination is 1. The specific formula for normalization is as follows:

[0097]

[0098] Wherein, is the unnormalized combination result, and after normalization, it can ensure that the result meets the basic requirements of probability.

[0099] In the embodiments of the present application, the soft combination D-S evidence theory can also handle uncertainties. When the conclusions given by two evidence sources are highly conflicting, the standard D-S rule will return an empty set or empty evidence, which will lead to an inability to make a reasonable decision; the soft combination D-S evidence theory "softens" this phenomenon through an adjustment factor, allowing a certain degree of uncertainty to exist; for example, if in some cases the conflict degree K is greater than the conflict threshold, it means that the conflict is relatively serious, and the soft combination will introduce an adjustment factor α to reduce the impact of the conflict, so that the combination result can continue to exist instead of being completely zeroed.

[0100] In the embodiments of the present application, if there are multiple evidence sources ε1, ε2,..., ε n , the soft combination D-S evidence theory can be gradually applied for combination; for example: first, the first and second evidence sources are soft combined to obtain a combination result Then, the combination result is soft combined with the third evidence source ε3 to obtain a new combination result; and so on, until all evidence sources are combined into a final evidence. This step-by-step combination method can help handle the inconsistencies in multi-source evidence and flexibly adjust the impact of conflicts through the adjustment factor α.

[0101] In the embodiments of the present application, the soft combination D-S evidence theory can also include credibility adjustment. When the conflict degree K between evidence sources is less than the conflict threshold, the credibility of the combination result can be increased; while when the conflict degree K is greater than the conflict threshold, the credibility of the combination result can be reduced to reflect the uncertainty of the evidence sources. The specific formula is as follows:

[0102]

[0103] It should be noted that by using credibility adjustment to reflect the uncertainty of evidence sources, it can be ensured that in the case of high conflicts, the result can still reasonably reflect the contributions of evidence sources and is not overly affected by excessive conflicts.

[0104] It should be noted that the traditional D-S evidence theory is mainly used to fuse information or evidence from different sources. The soft combination technology can be more adaptable to complex and uncertain data characteristics by introducing a certain degree of ambiguity or flexibility during data combination, rather than being static or fixed. Combining these two forms a new fusion framework. By introducing flexible decision-making, conflicts in evidence can be processed more efficiently during the combination process, reducing decision-making errors caused by conflicts and improving the accuracy and fault tolerance of data fusion.

[0105] It should be noted that the D-S evidence theory usually deals with the uncertainty of information by combining evidence from different sources, but it may face conflicts or inconsistencies when dealing with information from multiple sources. By introducing soft combination (such as through fuzzy logic or soft decision-making methods) to perform weighted fusion on different information sources, different evidence sources can, to a certain extent, retain their original information characteristics without having to enforce a unified decision. This synergy can improve the quality of decision-making in complex and unstable environments, especially in systems such as power grids, where it can better adapt to fluctuations, changes, and uncertainties. Soft combination can better resolve information conflicts and optimize the handling of uncertainties in the application of evidence theory.

[0106] In an optional embodiment, the conflict handling and fusion process of the soft combination D-S evidence theory can also be implemented using various strategies. For example, in the traditional framework, the Yager rule or Dubois-Prade rule can be introduced to replace the standard Dempster combination rule. By redistributing the degree of conflict to the universal set or specific hypotheses, counterintuitive results caused by normalization can be avoided. It is also possible to combine fuzzy integrals or possibility theory, model the uncertainty of evidence sources as fuzzy measures, and achieve smooth handling of conflicts through non-linear integration. In addition, a dynamic adjustment factor optimization method can be adopted, such as through neural networks or reinforcement learning models, to adaptively generate the value of the adjustment factor according to the degree of conflict, data source reliability, and real-time state of the power grid. Secondly, a hierarchical fusion architecture can be constructed. First, perform local soft combination on similar evidence sources, and then integrate the results through the global fusion layer to reduce the impact of high-dimensional conflicts. No matter which strategy is adopted, it is necessary to ensure the normalization constraint and physical interpretability of the fusion result. For example, in power grid state assessment, it is necessary to ensure that the fused evidence can reflect the practical significance of core indicators such as voltage stability or power balance.

[0107] In summary, the present invention is directed at the volatility and changing characteristics of power grid data, combines the Z-score method to detect abnormal data, flexibly selects the aggregation method, greatly improves the accuracy of data processing, and ensures that the aggregation result can truly reflect the state of the power grid system; by using the D-S evidence theory to fuse multi-source heterogeneous data, and when facing evidence conflicts, the soft combination method is used to introduce a regulation factor to alleviate the impact of conflicts, so that the state assessment is more stable and reliable, adapts to the fusion requirements of multi-source data, and effectively solves the problem of mismatched time scales in the use of industrial source data.

[0108] Embodiment 2 is an embodiment of the present invention, which provides an intelligent fusion system for multi-source data of a power system based on the soft combination D-S evidence theory, including: a data acquisition module for acquiring multi-source data and calculating the change index of the multi-source data; a data detection module for performing data detection on the multi-source data to obtain a detection result; an aggregation processing module for performing aggregation processing on the multi-source data based on the change index and the detection result to obtain data features; a feature fusion module for fusing the data features to obtain a data fusion result.

[0109] Embodiment 3 is an embodiment of the present invention, which is different from the previous embodiment in that:

[0110] As Figure 2 shown, if the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or a part of this technical solution can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc that can store program codes.

[0111] The logic and / or steps represented in the flowchart or otherwise described herein can, for example, be considered as a definitional sequence list of executable instructions for implementing logical functions, which can be embodied specifically in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or used in conjunction with these instruction execution systems, apparatuses, or devices. For the 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.

[0112] More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion having one or more wirings (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing it as appropriate, and then storing it in a computer memory.

[0113] 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 or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0114] Example 4, an embodiment of the present invention, provides a multi-source data intelligent fusion method for power systems based on soft combination D-S evidence theory. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0115] In this embodiment, by classifying the feature types into unaggregated features and aggregated features, and the fusion methods into direct D-S evidence theory and soft combination D-S evidence theory, simulation experiments were carried out under the stable and unstable states; the experimental results are shown in Table 1. Among them, the soft combination D-S evidence theory performs excellently under the stable state, with a trust degree as high as 0.91, a doubt degree of 0.96, and an uncertainty of only 0.05, indicating that the fusion result is highly reliable and conflicts and noises are almost eliminated; under the unstable state, the trust degree and the doubt degree are reduced to 0.09 and 0.14 respectively, and the uncertainty is 0.05, showing a strong conflict suppression and noise mitigation ability and significantly improving the fusion effect under the unstable condition; this method can provide a higher trust degree and a lower uncertainty, performs excellently in multi-source data fusion, and verifies the effectiveness of the present invention.

[0116] Table 1 Evaluation Table of Multi-source Data Fusion Effect for Power Grid Monitoring

[0117]

[0118] 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. An intelligent fusion method for multi-source data in a power system based on soft combination D-S evidence theory, characterized in that: include: Obtain multi-source data and calculate the change indicators of multi-source data; Performing data detection on the multi-source data to obtain a detection result; Aggregate the multi-source data based on the change index and the detection result to obtain data features; The data features are fused to obtain a data fusion result.

2. The intelligent fusion method for multi-source data of power system based on soft combination D-S evidence theory according to claim 1, characterized in that: The change indicators include volatility and rate of change.

3. The method for intelligent fusion of multi-source data in power systems based on soft-merging DS evidence theory according to claim 2, characterized in that: Performing data detection on the multi-source data includes the following steps: Perform abnormal data detection on multi-source data, determine whether the multi-source data is an abnormal point based on the mean and standard deviation of the multi-source data, and obtain the detection results.

4. The intelligent fusion method for multi-source data of power system based on soft combination D-S evidence theory according to claim 3, characterized in that: The aggregation processing includes mean aggregation, extreme value aggregation and median aggregation; The extreme value aggregation includes maximum value aggregation and minimum value aggregation.

5. The intelligent fusion method for multi-source data of power system based on soft combination D-S evidence theory according to claim 4, characterized in that: Aggregating the multi-source data based on the change indicator and the detection result includes: A judgment is made based on the volatility, rate of change and detection results, and according to the judgment result, it is decided to use mean aggregation, extreme value aggregation or median aggregation to aggregate the multi-source data.

6. The intelligent fusion method for multi-source data of power system based on soft combination D-S evidence theory according to claim 5, characterized in that: The data features are integrated, including the following steps: Acquire the data features and use them as evidence sources, and perform basic probability assignment on the evidence sources; Calculating the degree of conflict between the evidence sources; By introducing an adjustment factor, the evidence sources are merged using the soft merge DS evidence theory according to the adjustment factor and the conflict degree; The merged results are normalized to obtain the data fusion results.

7. The intelligent fusion method for multi-source data of power system based on soft combination D-S evidence theory according to claim 6, characterized in that: The core idea of the soft merge DS evidence theory is to control the impact of the conflict degree K on the merge result by introducing a regulation factor α.

8. An intelligent multi-source data fusion system for power systems based on the soft combination D-S evidence theory, based on the intelligent multi-source data fusion method for power systems based on the soft combination D-S evidence theory according to any one of claims 1 to 7, characterized in that: include, The data acquisition module is used to acquire multi-source data and calculate the change index of multi-source data; The data detection module is used to perform data detection on multi-source data and obtain detection results; Aggregation processing module, used to aggregate multi-source data based on change indicators and detection results to obtain data features; The feature fusion module is used to fuse data features to obtain data fusion results.

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 method for intelligent fusion of multi-source data of a power system based on soft merging DS evidence theory according to any one of claims 1 to 7 are implemented.

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 method for intelligent fusion of multi-source data of a power system based on soft merging DS evidence theory according to any one of claims 1 to 7 are implemented.

Images