Edge-end data fusion analysis method and system based on electric power internet-of-things operation system

Abstract

The invention discloses a side end data fusion analysis method and system based on an electric power Internet of Things operating system, and relates to the technical field of electric power Internet of Things operating systems, and the method specifically comprises the steps: obtaining original data from side end equipment, and carrying out the coding processing of the original data, and obtaining coded data; calculating an abnormal coefficient based on the coded data to identify abnormal data, and repairing the identified abnormal data to obtain repaired data; feature extraction is carried out based on the repair data to construct feature vectors representing data features; calculating a fusion degree index based on the feature vector; judging whether the fusion degree index reaches a preset fusion threshold value or not; if yes, outputting a fusion analysis result; and if not, iterative optimization is carried out until the fusion degree index reaches a fusion threshold value, so that the finally output fusion analysis result can accurately meet the application requirements of the electric power Internet of Things operation system, and the intelligent level and the operation efficiency of the whole electric power system are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electric power Internet of Things operation systems, and particularly relates to a method and system for edge data fusion analysis based on an electric power Internet of Things operation system. BACKGROUND

[0002] With the continuous advancement of smart grid construction in the field of electric power Internet of Things, the number of edge devices has increased dramatically. These devices continuously generate massive amounts of multi-source heterogeneous data, covering current, voltage, power, device status, and other types. However, the current electric power Internet of Things operation system faces many challenges in data fusion analysis. On the one hand, the original data collected by edge devices have problems such as non-uniform format and uneven data quality, with a large amount of noise data and abnormal data mixed in. If directly used for analysis, it will seriously affect the accuracy of the results. On the other hand, different types of data have large differences in characteristics, and there is a lack of effective methods for deep fusion of these data, making it difficult to extract key information that can fully represent the operating state of the power system.

[0003] Traditional data processing methods often focus on a single link and lack a full-process fine processing mechanism from original data encoding, abnormality identification and repair to feature extraction and fusion degree evaluation. In the encoding link, there is no special encoding method designed for the characteristics of electric power data, which limits the efficiency and accuracy of subsequent processing. Abnormal data identification mainly relies on simple threshold judgment, which cannot accurately identify abnormalities in complex scenarios. Feature extraction is based on single features or shallow features, which cannot fully exploit the internal relationships between data. Fusion degree evaluation lacks scientific and effective quantitative indicators, making it difficult to determine whether the data fusion meets the needs of electric power Internet of Things applications.

[0004] Therefore, there is a need for a method and system for edge data fusion analysis based on an electric power Internet of Things operation system. SUMMARY

[0005] To solve the problems in the prior art, the present application provides a method and system for edge data fusion analysis based on an electric power Internet of Things operation system, and the specific technical solutions are as follows: A method for edge data fusion analysis based on an electric power Internet of Things operation system, comprising the following steps: Step S1, obtaining original data from edge devices and performing encoding processing on the original data to obtain encoded data; Step S2, calculating an abnormality coefficient based on the encoded data to identify abnormal data, and repairing the identified abnormal data to obtain repaired data; Step S3, performing feature extraction based on the repaired data to construct a feature vector representing the characteristics of the data; Step S4: Calculate the fusion degree index based on the feature vector; determine whether the fusion degree index has reached the preset fusion threshold; if yes, output the fusion analysis result; if no, iterate and optimize until the fusion degree index reaches the fusion threshold.

[0006] Preferably, step S1 involves acquiring raw data from the edge device and encoding the raw data to obtain encoded data, specifically including: Through the first calculation formula The i-th original data is calculated. Encoded data ; in, This is the data type offset coefficient. This is the scaling factor for the fluctuation range. These are nonlinear correction parameters. This is the adjustment factor for the error function. For amplitude compensation parameters, For data sharpness parameters, For phase modulation coefficients, Angular frequency parameter, These are the initial phase parameters. The logarithmic correction factor is... This is the amplitude scaling sub-parameter.

[0007] Preferably, step S2, which calculates anomaly coefficients based on coded data to identify anomalous data, specifically includes: Through the second calculation formula The anomaly coefficient was calculated. ; in, The mean of the encoded data, For the standard deviation of the encoded data, This is the gradient influence coefficient. For the data gradient value, For periodic disturbance parameters, As a periodic adjustment factor, The fluctuation sensitivity coefficient The difference between adjacent data. The exponential decay coefficient is... This is the decay rate parameter; For the i-th original data Encoded data, This represents the volatility coefficient of the i-th set of edge data.

[0008] Preferably, in step S2, the identified abnormal data is repaired to obtain repaired data, specifically including: Through the third calculation formula The i-th original data is calculated. Repair data ; in, The interpolation slope parameter, For the repair cycle parameters, This is the mean correction factor. This is the hyperbolic secant adjustment parameter.

[0009] Preferably, feature extraction is performed based on the repaired data to construct a feature vector representing the characteristics of the data, specifically as follows: Through the fourth calculation formula Calculate the eigenvectors ; in, T represents the dimension weight coefficients. For periodic parameters, For characteristic phase modulation parameters, For amplitude enhancement parameters, This is the influence coefficient; For the i-th original data The repaired data, where n represents the number of feature dimensions.

[0010] Preferably, it further includes: Through the fifth calculation formula The influence coefficient was calculated. ; in, The arctangent adjustment coefficient, For nonlinear strength parameters, For the coefficients of the complementary error function, This is the attenuation adjustment parameter.

[0011] Preferably, the fusion degree index is calculated based on the feature vector, specifically including: Through the sixth calculation formula The fusion index was calculated. ; in, To integrate the weighting coefficients, These are the sinusoidal modulation parameters. For frequency adjustment parameters, The square root coefficient of the amplitude. For hyperbolic cosine parameters, The influence coefficient of the squared term. For logarithmic correction parameters, is the hyperbolic sine parameter, and m is the number of edge data groups.

[0012] An edge-to-edge data fusion and analysis system based on a power IoT operating system, applied to the methods described, includes: The data acquisition and encoding module is used to acquire raw data from edge devices and encode the raw data to obtain encoded data. The data anomaly repair module is used to calculate anomaly coefficients based on coded data to identify abnormal data, and to repair the identified abnormal data to obtain repaired data. The feature vector construction module is used to extract features from the repaired data to construct feature vectors that characterize the data properties. The data fusion analysis module is used to calculate the fusion degree index based on feature vectors; determine whether the fusion degree index has reached a preset fusion threshold; if so, output the fusion analysis result; if not, iterate and optimize until the fusion degree index reaches the fusion threshold.

[0013] A computer-readable storage medium, characterized in that the computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to execute the edge data fusion analysis method based on the power IoT operating system.

[0014] A processor, characterized in that the processor is used to run a program, wherein the program executes the edge data fusion and analysis method based on the power IoT operating system.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention encodes raw data from edge devices, transforming diverse types and formats of raw data into standardized coded data. This lays a solid foundation for subsequent operations such as anomaly identification and feature extraction, ensuring data consistency and operability in later processes. It effectively solves the processing challenges posed by the diverse and complex data from edge devices in power IoT scenarios. The process of calculating anomaly coefficients based on coded data to identify and repair abnormal data significantly improves data quality. Accurate identification and repair of anomaly data avoids interference from outliers in subsequent analysis results, allowing the data to more accurately reflect the actual operating status of the power system. This provides a reliable data source for subsequent data-driven decision-making and analysis, ensuring the accuracy of data-driven power IoT applications. Feature extraction based on the repaired data to construct feature vectors characterizing data properties extracts key information that best reflects the essential characteristics of the data from massive amounts of repaired data. Constructing feature vectors condenses complex data into a more representative form, facilitating subsequent calculations and analysis of data fusion. The mechanism calculates a fusion degree index based on feature vectors and determines whether to output fusion analysis results or perform iterative optimization based on whether the index reaches a preset fusion threshold. This ensures the effectiveness and reliability of the fusion analysis results. When the fusion degree index does not reach the threshold, iterative optimization continuously adjusts each stage of data processing until the requirements are met. This ensures that the final output fusion analysis results accurately match the application needs of the power IoT operating system. Whether in power equipment status monitoring, fault early warning, or energy optimization and scheduling, it provides strong data support, helping the power IoT system achieve more efficient and intelligent operation and management, and improving the intelligence level and operational efficiency of the entire power system. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0017] Figure 1 This is a flowchart of the method of the present invention.

[0018] Figure 2 This is a system schematic diagram of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] It should be understood that, when used in this specification and the appended claims, the terms "comprising" and "including" indicate the presence of the described features, integrals, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or collections thereof.

[0021] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0022] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0023] Example 1: like Figure 1 As shown, this embodiment provides an edge data fusion and analysis method based on a power IoT operating system, including the following steps: Step S1: Obtain raw data from the edge device and encode the raw data to obtain encoded data. Specifically, this includes: First, raw data is acquired from the edge device; this raw data forms the basis for subsequent analysis. To facilitate encoding operations in later processing, the raw data is converted into encoded data, conforming to the format and requirements of subsequent processing.

[0024] Through the first calculation formula The i-th original data is calculated. Encoded data ; in, This is the data type offset coefficient, with a value range of [0.1, 2.0]. This is the scaling factor for the fluctuation range, with a value range of [0.5, 3.0]. This is a nonlinear correction parameter, with a value range of [0.01, 1.0]. This is the error function adjustment factor, with a value range of [0.1, 5.0]. This is the amplitude compensation parameter, with a value range of [0.8, 1.2]. This is a data sharpness parameter, with a value range of [0.05, 0.5]. The phase modulation coefficient has a value range of [0,π]. ω is the angular frequency parameter, with a value range of [0, 2πf] (f is the power frequency); The initial phase parameter has a value range of [−π,π]. This is the logarithmic correction factor, with a value range of 0.1 to 2.0. The amplitude scaling sub-parameter has a value range of [0.5, 2.0].

[0025] For data with strong periodicity and large fluctuations, such as current and voltage, the data type offset coefficient... It will adapt to its different reference ranges (e.g., when the voltage reference is 10kV and the current reference is 1000A). Adjust to the corresponding offset range respectively); fluctuation range scaling factor It amplifies small voltage fluctuations (such as voltage sags) or reduces current spikes, ensuring the accuracy of feature extraction; angular frequency parameter It directly matches the power frequency (50Hz), which aligns with the sinusoidal periodic characteristics of current and voltage.

[0026] For nonlinear, small-range fluctuating data such as power factor, nonlinear correction parameters It will take a smaller value (e.g., 0.01~0.1) to correct its nonlinear correlation with active / reactive power; phase modulation coefficient It adapts to the phase characteristics of the power factor (usually in the range of 0~π) and accurately reflects the degree of phase shift of the power factor.

[0027] For slowly changing, low-fluctuation data such as equipment temperature, amplitude compensation parameters Use values ​​close to 1 (e.g., 0.8~1.2) to ensure the stability of temperature data; data sharpness parameter It will take a smaller value (e.g., 0.05~0.2) to avoid over-amplifying small temperature fluctuations; logarithmic correction factor This will adjust the logarithmic feature weights of the temperature data to highlight its gradual change trend.

[0028] This invention obtains the i-th original data using a first calculation formula. Encoded data In this formula, the numerator incorporates the original data. Data type offset coefficient Nonlinear correction parameters Error function adjustment factor Participating error function and amplitude compensation parameters and data sharpness parameter Related The original data underwent comprehensive integration and preliminary adjustment based on various characteristics.

[0029] The denominator then introduces a fluctuation range scaling factor. This is used to scale the numerical range of the data to make it more consistent; it also includes the phase modulation coefficients. Angular frequency parameters Initial phase parameters The sine function terms formed To simulate and adapt to the potential periodic fluctuations in the data; and also a logarithmic correction coefficient. and amplitude scaling sub-parameter Logarithmic terms involved It is used to smooth out the influence of extreme values ​​in the data and avoid excessive interference from individual abnormally large or small values ​​on the overall encoding result.

[0030] Through formula Calculate the data type offset coefficient ; in, The average of the original data (e.g., current is taken as [0,1000]A, voltage as [0,10000]V). The standard deviation of the original data is denoted as [0.01]. 0.5 ]; This is the baseline adjustment coefficient, with a value range of [0.5, 2.0]. This is a sign-sensitive parameter, with a value range of [−1.0, 1.0]. This is the mean amplification parameter, with a value range of [1.0, 5.0]. The sine offset coefficient has a value range of [0, π / 2]. This is the angle adjustment parameter, with a value range of [0, 2π].

[0031] For current and voltage, the mean of the raw data It will directly match its actual acquisition range (e.g., current is taken as [0,1000]A, voltage as [0,10000]V), and the standard deviation of the raw data will be used. This corresponds to the degree of fluctuation (e.g., the standard deviation of voltage is taken as 0.01). 0.5 (Adapting to small fluctuations in voltage sags / surges); reference adjustment factor. It will be adjusted to the corresponding range of 0.5~2.0 according to the difference in the reference range of current and voltage (such as 10kV voltage and 1000A current), with a sinusoidal offset coefficient. It matches the power frequency cycle and adapts to the sinusoidal waveform characteristics of current and voltage.

[0032] For sign-sensitive data such as power factor (lagging / leading corresponds to positive and negative values), the sign-sensitive parameter... The range [−1.0, 1.0] can accurately distinguish its positive and negative characteristics; mean amplification parameter It will take a smaller value between 1.0 and 5.0 to avoid over-amplifying small fluctuations in the power factor (usually the power factor is between 0.8 and 1.0).

[0033] For slowly varying, unsigned data such as equipment temperature, sign-sensitive parameters Fixed to a value close to 0 to ignore the effect of sign; angle adjustment parameter To adapt to the non-periodic characteristics of temperature, a fixed value within [0, 2π] is used to simplify calculations for aperiodic data, while the mean of the original data is also taken into account. It will fit the actual temperature range to ensure the consistency of the physical meaning of the parameters and data.

[0034] This invention first clarifies the various parameters involved in the calculation, wherein... It is the mean of the raw data, reflecting the average level of the raw data; The standard deviation of the original data reflects the degree of dispersion of the original data; It is the benchmark adjustment coefficient, used to adjust the calculation benchmark; It is a sign-sensitive parameter, meaning it is highly sensitive to the sign characteristics of the data; This is a mean amplification parameter, which can amplify the degree of influence of the original data mean; This is the sine offset coefficient, used for adjusting the offset of the sine function; This is an angle adjustment parameter used to adjust the angle characteristics of the sine function.

[0035] Calculating data type offset coefficients The formula consists of several parts. First, the mean of the original data is calculated. Subtract the benchmark adjustment factor Standard deviation of the original data The product of the two values ​​is used to determine a basic baseline adjustment based on the mean and standard deviation of the original data, initially reflecting the impact of the overall data distribution on the offset coefficient. Next, a sign-sensitive parameter is added. With symbolic functions The product of, sign function It can reflect the sign characteristics of the original data mean, combined with the mean amplification parameter. The square root operation captures the effect of the sign and magnitude correlation of the original data on the offset coefficient. Then, a sinusoidal offset coefficient is added. and The product of the two values ​​is used here, where the ratio of the original data's mean to its standard deviation is taken as the angle input for the sine function, combined with angle adjustment parameters. and sinusoidal offset coefficient By utilizing the periodicity and volatility of the sine function, the offset coefficient is further enriched in its representation of the original data characteristics, taking into account the periodic influence of the relative relationship between the data mean and standard deviation.

[0036] The system acquires raw data from edge devices distributed throughout the power network, such as smart meters and power sensors. This raw data encompasses various types, including current, voltage, power factor, and equipment temperature, and is characterized by its wide range of sources, inconsistent formats, and large numerical fluctuations. Subsequently, the raw data is encoded using a first calculation formula. Through the coordinated calculation of multiple parameters, such as data type offset coefficients and fluctuation range scaling coefficients, the raw data is transformed into coded data, achieving data format unification and standardization, laying the foundation for subsequent processing.

[0037] Anomaly coefficients are calculated based on coded data, enabling precise identification of anomalous components within the data. In power IoT scenarios, data accuracy is paramount; therefore, after identifying anomalous data, it must be repaired to obtain restored data. This ensures data quality and provides a reliable data source for subsequent operations such as feature extraction.

[0038] Step S2: Calculate the anomaly coefficient based on the encoded data to identify abnormal data, and repair the identified abnormal data to obtain repaired data.

[0039] Specifically, the calculation of anomaly coefficients based on coded data to identify anomalous data includes: Through the second calculation formula The anomaly coefficient was calculated. ; in, The mean of the encoded data is normalized to [0,1]. The standard deviation of the encoded data ranges from [0.05]. 0.3 ]; This is the gradient influence coefficient, with a value range of [0.1, 1.0]. This represents the data gradient value, ranging from [0, 0.5]. This is a periodic interference parameter, with a value range of [0.01, 0.2]. This is a periodic adjustment factor, with a value range of 0 to 2πf (where f is the power frequency). This is the fluctuation sensitivity coefficient, with a value range of [0.5, 2.0]. This represents the difference between adjacent data points, with a value range of [0, 0.1]. The exponential decay coefficient has a value range of [0.05, 0.5]. This is the decay rate parameter, with a value range of [0.1, 1.0]. For the i-th original data Encoded data, The fluctuation coefficient represents the i-th set of edge data, with a value range of [0.1, 1.0].

[0040] For data such as current and voltage, which exhibit periodic fluctuations after normalization encoding, the mean of the encoded data... It will stabilize within the [0,1] interval (fitting its normalized concentrated distribution), with periodic disturbance parameters. A smaller value in the range [0.01, 0.2] is selected to suppress power frequency harmonic interference; period adjustment factor. This directly matches the power frequency (0.2πf), accurately adapting to the sinusoidal periodic characteristics of current and voltage; simultaneously, the data gradient value We take [0, 0.5] to match the fluctuation range of adjacent data after encoding.

[0041] For nonlinear data such as power factor, which exhibits smooth fluctuations after encoding, the gradient influence coefficient... A smaller value in the range [0.1, 1.0] is chosen to weaken the interference of small gradients, thus reducing the fluctuation sensitivity coefficient. Take the median value of [0.5, 2.0] to balance its sensitivity to small fluctuations; the difference between adjacent data... The range [0, 0.1] is chosen to match the gradual variation characteristics of the data after power factor encoding.

[0042] For data like equipment temperature, which is encoded without periodicity or gradual change, periodic interference parameters... The value will be close to 0 to ignore periodic interference, and the exponential decay coefficient will be used. A smaller value in the range [0.05, 0.5] is chosen to reflect the slow decay trend of temperature data; the standard deviation of the encoded data is used. Take [0.05] 0.3 The smaller interval of ] reflects the low dispersion of temperature-coded data.

[0043] This invention first clarifies the various parameters involved in the calculation, wherein It is the mean of the encoded data, which reflects the overall average level of the encoded data; The standard deviation of the encoded data reflects the degree of dispersion of the encoded data.

[0044] In the numerator of the formula, the i-th encoded data... With the mean of the encoded data Subtract the difference, then subtract the gradient influence coefficient. and data gradient values The resulting gradient influence term, and the periodic disturbance parameter. With periodic adjustment factor , coded data mean Related sine function terms This approach comprehensively considers the impact of factors such as deviations from the mean, gradient changes, and periodic disturbances on anomaly detection.

[0045] In the denominator, the standard deviation of the encoded data is first introduced. Then multiply by the fluctuation sensitivity coefficient. Difference between adjacent data Items that constitute And the exponential decay coefficient and decay rate parameter The exponential decay term constituted The fluctuation sensitivity coefficient and the difference between adjacent data are used to capture data fluctuations, while the exponential decay term can be adjusted according to data characteristics, allowing the denominator to reasonably scale the result of the numerator, thereby calculating the outlier coefficient more accurately. Through this calculation, the anomaly coefficient... It can effectively quantify the degree of data anomaly, help the system identify abnormal data, so as to carry out targeted repair processing and ensure the accuracy and reliability of data in the power IoT system.

[0046] The identified anomalous data is repaired to obtain repaired data, specifically including: Through the third calculation formula The i-th original data is calculated. Repair data ; in, This is the interpolation slope parameter, with a value range of [0.1, 3.0]. For the repair cycle parameters, The value range is [1, 10]; This is the mean correction factor, with a value range of [0.8, 1.2]. This is the hyperbolic secant adjustment parameter, with a value range of [0.5, 2.0].

[0047] For example, data with high fluctuation frequencies, such as current and voltage, will correspond to larger values. (Taking values ​​close to 1.3), a larger interpolation slope is used to match its rapidly changing characteristics, while combining a smaller... (Values ​​close to 1) shorten the repair cycle to ensure that such high-frequency fluctuating data can be quickly repaired; while data such as the power factor, which are relatively stable and have small fluctuations, will be paired with smaller values. (Values ​​close to 0) are used to avoid over-correction with a gentle interpolation slope, while allowing for appropriate scaling. (Values ​​close to 10) extend the repair cycle.

[0048] For data such as equipment temperature, which is easily affected by the environment and whose mean is prone to deviation, It will tend towards the upper limit of 1.2, using a stronger mean correction capability to offset numerical shifts caused by environmental interference; while data such as the power factor, which is inherently stable in mean, It will take a smaller value around 0.8 to reduce unnecessary mean adjustment.

[0049] In addition, the hyperbolic secant adjustment parameter Adaptation to different data distribution characteristics: For data that approximates a normal distribution, such as current and voltage. We take a smaller value near 5.0 and use the smoothing properties of the hyperbolic secant function to adapt it to the distribution; however, for asymmetric distribution data such as equipment temperature, which may have extreme values ​​(such as overheating), By moving closer to the lower limit of 2.0, and leveraging the convergence properties of the hyperbolic secant function, more precise constraints and adjustments can be made to extreme temperature data.

[0050] The first part of the formula of the present invention Utilize the preceding data adjacent to the abnormal data and the next data The average value provides a basic reference value based on the trend of adjacent normal data for abnormal data repair, and initially fits the general trend of the data.

[0051] Next, the interpolation slope parameter is introduced. , combined (That is, the difference between the next adjacent data point and the previous data point, reflecting the slope of the data change) and (in To repair the periodic parameters and adapt to the possible periodic characteristics of the data, so that the repair process can conform to the periodic change pattern of the data, this part adjusts the basic reference value by considering the slope change and periodicity of the data, so that the repair result is more in line with the actual change trend of the data.

[0052] Then there's the mean correction part, the mean correction coefficient. and (mean of encoded data) With the current encoded data to be repaired The difference (reflecting the degree of deviation of the data from the mean) is multiplied together, and then combined with the hyperbolic secant adjustment parameter. Participants (The hyperbolic secant function can smoothly adjust the correction amplitude to avoid over- or under-correction.) Further correction of the repaired values ​​ensures that the repaired data can better revert to the mean of the overall data and conform to the overall distribution characteristics of the data.

[0053] Anomaly coefficients are calculated using a second formula based on the encoded data. This formula comprehensively evaluates the data from multiple dimensions, including the mean, standard deviation, and gradient influence coefficient, to accurately identify anomalous data. For the identified anomalous data, a third formula is used to repair it based on interpolation slope parameters and repair cycle parameters, resulting in high-quality repaired data that effectively ensures the accuracy and reliability of the data.

[0054] Step S3: Perform feature extraction based on the repaired data to construct a feature vector representing the characteristics of the data.

[0055] Feature extraction is performed based on the repaired data. The purpose of feature extraction is to extract key information that characterizes the data's properties from massive and complex datasets, thereby constructing feature vectors. These vectors contain the core features of the data, specifically including: Through the fourth calculation formula Calculate the eigenvectors ; in, T represents the dimension weight coefficient, with a value range of [0.1, 1.0]; The period parameter is T, which corresponds to 50Hz. The value range is [1, 5]. The characteristic phase modulation parameter has a value range of [0, 2π]. This is the amplitude enhancement parameter, with a value range of [0.5, 2.0]. This is the influence coefficient, with a value range of [0.5, 2.0]. For the i-th original data The repaired data; n represents the number of feature dimensions, with a value range of [3, 10].

[0056] For AC power data such as current and voltage, which are strongly correlated with the grid frequency, the period parameter T is fixed to match the periodic characteristics of a 50Hz grid, while the number of periods s is determined based on the details of the current / voltage fluctuation period, taking a value between 1 and 5 (for example, the more harmonic components in the voltage waveform, the closer s is to 5). Furthermore, this type of data is a core dimension of power monitoring, and its dimension weighting coefficient... It will take a higher value near 0.1 to strengthen its weight in the feature vector; feature phase modulation parameter This will adapt to the phase difference characteristics of current and voltage (for example, when the voltage phase is 0, the current phase can be adjusted by...). Adjust the phase angle to the corresponding power factor.

[0057] For ratio-based data such as power factor that reflects electrical energy utilization efficiency, its value is relatively stable, and its dimension priority is slightly lower than that of current and voltage. It will take a smaller value below 0.1; at the same time, the power factor is directly related to the phase of the current and voltage. It will be adjusted to the corresponding phase range (such as the lag phase corresponding to the power factor under inductive load), while the amplitude enhancement parameter We will take a smaller value around 0.5 to avoid over-amplifying the magnitude of this ratio-type data.

[0058] For physical quantities like equipment temperature, which are indirectly related to electrical parameters, they inherently lack periodicity; therefore, the period number s is usually taken as 1 (to mitigate the influence of periodicity). Since temperature is a key indicator of equipment condition, A higher value around 0.1 will be chosen to ensure its weight; meanwhile, the amplitude fluctuations of temperature data are relatively independent. It will adapt to the range of temperature changes (e.g., when the equipment temperature rises rapidly). The upper limit of 2.0 is used to enhance its characteristic amplitude; while the influence coefficient X will be related to the coupling relationship between temperature and current (for example, the larger the current, the closer X is to 2.0 to strengthen the influence weight of current on temperature).

[0059] First, this invention requires clarifying the various parameters involved in the calculation, wherein... These are dimensional weight coefficients, used to measure the importance of different dimensional features in the whole; T and S are periodic parameters, corresponding to the different periodic characteristics that the data may have. It is a characteristic phase modulation parameter that can adjust the phase of the feature to match the phase characteristics of the data; This is an amplitude enhancement parameter used to enhance the amplitude performance of the feature. It is the influence coefficient, which can have an additional impact on the overall calculation results.

[0060] In calculating eigenvectors Then, using the fourth calculation formula, the terms from j=0 to n are summed. For each term, the repaired data is first... and Multiplication, this part combines the period parameter T and the characteristic phase modulation parameter. The aim is to capture the sinusoidal fluctuation characteristics of repaired data over a specific period. Simultaneously, and Multiplication, here using the period parameter S and the amplitude enhancement parameter This is used to extract the cosine and amplitude correlation features of the repaired data in another period dimension. Then, these two results are summed, along with the influence coefficient X, and finally multiplied by the dimension weight coefficient. By summing all these terms, and comprehensively considering the characteristics of the repaired data across different periods, phases, amplitudes, and other dimensions, a feature vector that can fully characterize the data's properties is finally obtained. This provides a foundation for subsequent operations such as fusion degree calculation based on feature vectors.

[0061] Through the fifth calculation formula The influence coefficient was calculated. ; in, The arctangent adjustment coefficient has a value range of [0.1, 1.0]. This is a nonlinear intensity parameter, with a value range of [0.05, 0.5]. These are the coefficients of the complementary error function, with values ​​ranging from [0.1, 2.0]. This is the attenuation adjustment parameter, with a value range of [0.01, 0.2].

[0062] For data such as current and voltage that fluctuate dynamically and are prone to nonlinear distortion, the arctangent adjustment coefficient... A higher value near 0.1 will be chosen, utilizing the saturation characteristics of the arctangent function to smooth out extreme fluctuations in current / voltage; simultaneously, if the current exhibits nonlinear fluctuations due to sudden load changes, the nonlinear intensity parameter will be... It will approach the upper limit of 0.5, strengthening the characteristic correction for this type of nonlinear data; while when the current / voltage is in a stable fluctuation range, We then take a smaller value around 0.05 to reduce unnecessary nonlinear intervention.

[0063] For proportional data such as power factor, which has a fixed numerical range (0-1) and changes gradually, the complementary error function coefficients... It will take a higher value around 1.2, utilizing the tail convergence characteristic of the complementary error function to accurately characterize the subtle changes when the power factor is close to 0 or 1; at the same time, since the power factor has no obvious attenuation characteristic, the attenuation adjustment parameter It will take a minimum value near 0.01 to weaken the effect of decay; while Then take a smaller value below 0.1 to avoid excessive compression of its effective value range by the arctangent adjustment.

[0064] For data such as equipment temperature that decays slowly over time (e.g., during the cooling process after equipment shutdown), the decay adjustment parameter... It will adjust towards the upper limit of 0.2 to adapt to the gradual temperature change through its attenuation characteristics; however, temperature is prone to errors due to the coupling effect of the environment and load. It will take a higher value around 1.2, and use a complementary error function to correct the numerical deviation caused by this type of error; while when the temperature is in the stable operating range (weaker nonlinearity), It will take a smaller value around 0.05 to reduce the intervention of nonlinear adjustment.

[0065] This invention specifies the parameters involved in the calculation, wherein It is the arctangent adjustment coefficient, used to adjust the changing trend of the arctangent function; As a nonlinear intensity parameter, it determines the degree of influence of the nonlinear characteristics of the data; These are the coefficients of the complementary error function, used to adjust the effect of the complementary error function. This is an attenuation adjustment parameter that allows for adjustment of the attenuation characteristics of the data.

[0066] The formula for calculating the influence coefficient X consists of two parts. The first part is... The introduction of the arctangent function atan can... The nonlinear relationship is transformed smoothly and boundedly, combined with the arctangent adjustment coefficient. This allows for precise control over the contribution of this part to the influence coefficient X, effectively capturing the nonlinear characteristics in the data and transforming them into a suitable numerical range. The second part is... complementary error function It can reflect the degree to which the data deviates from a certain reference state; the attenuation adjustment parameter and complementary error function coefficients The combined effect adjusts this part of the impact and can be used to reflect the attenuation or distribution characteristics of the data.

[0067] Feature extraction is performed on the repaired data, and a feature vector is constructed using the fourth calculation formula. This process considers dimensional weight coefficients, periodic parameters, feature phase modulation parameters, etc., and combines them with the influence coefficient obtained through the fifth calculation formula to extract key information that can comprehensively characterize the data characteristics from the repaired data, providing accurate feature representations for data fusion.

[0068] Step S4: Calculate the fusion degree index based on the feature vector; determine whether the fusion degree index has reached the preset fusion threshold; if yes, output the fusion analysis result; if no, iterate and optimize until the fusion degree index reaches the fusion threshold.

[0069] The fusion degree index is calculated based on feature vectors, specifically including: Through the sixth calculation formula The fusion index was calculated. ; in, The fusion weighting coefficient has a value range of [0.1, 1.0]. These are the sinusoidal modulation parameters, with values ​​ranging from [0, π / 2]. This is a frequency adjustment parameter, with a value range of [0, 2πf] (where f is the power frequency). This is the square root coefficient of the amplitude, with a value range of [0.5, 2.0]. The hyperbolic cosine parameter has a value range of [0.1, 1.0]. This is the influence coefficient for the squared term, with a value range of [0.1, 2.0]. This is the logarithmic correction parameter, with a value range of [0.05, 0.5]. is the hyperbolic sine parameter, with a value range of [0.1, 1.0]; m is the number of edge data groups, with a value range of [2, 8].

[0070] For electrical parameters such as current and voltage, which are strongly tied to the power frequency, frequency adjustment parameters It will precisely match the power frequency f (e.g., take the value corresponding to 2πf under a 50Hz power grid), and at the same time, the sinusoidal modulation parameters It will adapt to its sinusoidal waveform characteristics (taking the corresponding phase value within [0, π / 2]); as core data of the power system, their fusion weighting coefficients A higher value around 0.1 will be chosen to strengthen its priority in fusion analysis; while the square root coefficient of amplitude... It will be adjusted according to the amplitude range of current / voltage (e.g., the upper limit of 2.0 is taken for high voltage data to amplify its amplitude characteristics).

[0071] For ratio data such as power factor that reflect phase relationships, it is directly related to the phase difference between current and voltage, therefore It will take the value of the corresponding phase difference (such as the lag phase corresponding to the power factor under inductive load); since the power factor value range is stable (0-1), the square root coefficient of the amplitude is... A smaller value around 0.5 will be chosen to avoid excessive amplification of its amplitude; meanwhile, as an auxiliary electrical parameter, its fusion weighting coefficient... It will take a smaller value below 0.1 to balance the weight ratio of the core electrical parameters.

[0072] For non-electrical physical quantities such as equipment temperature, there is no direct correlation with the power frequency, therefore Values ​​deviating from 2πf will be used to weaken the influence of frequency; the temperature data changes relatively smoothly, and the hyperbolic cosine parameter... It will take a higher value around 0.1, utilizing the smoothing property of hyperbolic cosine to adapt to its gradual trend; while the influence coefficient of the squared term... It will adjust according to the coupling relationship between temperature and electrical parameters (e.g., the higher the current, the better). (Use an upper limit of 1.2 to enhance the effect of current on temperature).

[0073] This invention first clarifies the various parameters involved in the calculation, wherein... These are the fusion weight coefficients, used to reflect the importance of different feature vectors in the fusion process; These are the sinusoidal modulation parameters, which can be used to adjust the modulation effect of the sinusoidal function; It is a frequency adjustment parameter used to adjust the frequency characteristics of a sine function; This is the square root coefficient, which is used in the square root operation of the eigenvector; These are the hyperbolic cosine parameters, used for related calculations of the hyperbolic cosine function; This is the influence coefficient of the squared term, which affects the effect of the squared term of the eigenvector. This is a logarithmic correction parameter used to correct logarithmic operations; These are the parameters for the hyperbolic sine function, used in its calculation.

[0074] When calculating the fusion index C, the numerator of the sixth calculation formula is a summation of terms from i=1 to m. For each term, the fusion weight coefficient is first... eigenvectors Multiply by, then multiply by the sinusoidal modulation parameter and frequency adjustment parameters Participants This part uses a sine function to modulate the eigenvector to capture the periodic or fluctuating characteristics of the data; then it is multiplied by the square root coefficient of the amplitude. and hyperbolic cosine parameters Participants The amplitude square root operation, combined with the hyperbolic cosine function, can integrate the amplitude and hyperbolic cosine features of the feature vector.

[0075] The denominator is the summation of the terms from i=1 to m, followed by the square root. In each term, a fusion weight coefficient is included. The square of the eigenvector Multiply the squares together, then multiply by the influence coefficient of the squared terms. Participants And the parameter corrected by logarithm and hyperbolic sine parameters Participants The combination of squared terms, logarithmic operations, and hyperbolic sine functions here comprehensively considers the squared properties, logarithmic properties, and hyperbolic sine properties of the eigenvectors.

[0076] Based on the constructed feature vectors, the fusion degree index is calculated using the sixth calculation formula. Multiple parameters, including the fusion weight coefficient and sinusoidal modulation parameters, work together to quantify the degree of data fusion. The system determines whether the fusion degree index reaches a preset fusion threshold. If it does, the fusion analysis results are output, which can be used in applications such as power equipment status monitoring, fault early warning, and energy optimization scheduling. If not, the system iteratively optimizes steps such as data encoding, anomaly handling, feature extraction, and fusion degree calculation. By adjusting relevant parameters in each calculation formula, such as recalculating the data type offset coefficient (using the seventh calculation formula), the subsequent processes are executed again until the fusion degree index meets the fusion threshold requirements. This ensures that the output fusion analysis results accurately and effectively serve various applications of the power IoT operating system, helping the power system achieve intelligent and efficient operation.

[0077] The fusion degree index is calculated based on the feature vectors, reflecting the degree of data fusion. It is then determined whether this index reaches a preset fusion threshold. If it does, the data fusion effect meets the requirements, and the fusion analysis result is directly output. If it does not, iterative optimization is performed, repeatedly going through the steps of data encoding, anomaly identification and repair, feature extraction, and fusion degree calculation until the fusion degree index meets the fusion threshold. This ensures that the final output fusion analysis result has high accuracy and effectiveness, better serving power IoT-related application scenarios.

[0078] Example 2: Based on the same inventive concept as Embodiment 1, this embodiment provides an edge data fusion and analysis system based on a power IoT operating system, applied to the method described, including: The data acquisition and encoding module is used to acquire raw data from edge devices and encode the raw data to obtain encoded data. The data anomaly repair module is used to calculate anomaly coefficients based on coded data to identify abnormal data, and to repair the identified abnormal data to obtain repaired data. The feature vector construction module is used to extract features from the repaired data to construct feature vectors that characterize the data properties. The data fusion analysis module is used to calculate the fusion degree index based on feature vectors; determine whether the fusion degree index has reached a preset fusion threshold; if so, output the fusion analysis result; if not, iterate and optimize until the fusion degree index reaches the fusion threshold.

[0079] Example 3: Based on the same inventive concept as Embodiment 1, this embodiment provides a computer-readable storage medium, characterized in that the computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to execute the edge data fusion analysis method based on the power IoT operating system.

[0080] Example 4: Based on the same inventive concept as Embodiment 1, this embodiment provides a processor, characterized in that the processor is used to run a program, wherein the program executes the edge data fusion and analysis method based on the power IoT operating system.

[0081] Those skilled in the art will recognize that the modules of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components of the examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of the invention.

[0082] In the embodiments provided by this invention, it should be understood that the division of modules is only a logical functional division. In actual implementation, there may be other division methods, such as multiple modules can be combined into one module, one module can be split into multiple modules, or some features can be ignored.

[0083] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0084] If the integrated module is implemented as a software functional module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the 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 to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0085] Finally, 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 foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for edge-to-edge data fusion and analysis based on a power IoT operating system, characterized in that, Includes the following steps: Step S1: Obtain raw data from the edge device and encode the raw data to obtain encoded data; Step S2: Calculate the anomaly coefficient based on the encoded data to identify abnormal data, and repair the identified abnormal data to obtain repaired data; Step S3: Perform feature extraction based on the repaired data to construct a feature vector representing the characteristics of the data; Step S4: Calculate the fusion degree index based on the feature vector; determine whether the fusion degree index has reached the preset fusion threshold; if yes, output the fusion analysis result; if no, iterate and optimize until the fusion degree index reaches the fusion threshold.

2. The edge data fusion and analysis method based on the power IoT operating system according to claim 1, characterized in that, Step S1 involves acquiring raw data from the edge device and encoding the raw data to obtain encoded data, specifically including: Through the first calculation formula The i-th original data is calculated. Encoded data ; in, This is the data type offset coefficient. This is the scaling factor for the fluctuation range. These are nonlinear correction parameters. This is the adjustment factor for the error function. For amplitude compensation parameters, For data sharpness parameters, For phase modulation coefficients, Angular frequency parameter, These are the initial phase parameters. The logarithmic correction factor is... This is the amplitude scaling sub-parameter.

3. The edge data fusion and analysis method based on the power IoT operating system according to claim 1, characterized in that, Step S2 involves calculating anomaly coefficients based on the encoded data to identify anomalous data, specifically including: Through the second calculation formula The anomaly coefficient was calculated. ; in, The mean of the encoded data, For the standard deviation of the encoded data, This is the gradient influence coefficient. For the data gradient value, For periodic disturbance parameters, As a periodic adjustment factor, The fluctuation sensitivity coefficient The difference between adjacent data. The exponential decay coefficient is... This is the decay rate parameter; For the i-th original data Encoded data, This represents the volatility coefficient of the i-th set of edge data.

4. The edge data fusion and analysis method based on the power IoT operating system according to claim 1, characterized in that, Step S2 involves repairing the identified abnormal data to obtain repaired data, specifically including: Through the third calculation formula The i-th original data is calculated. Repair data ; in, The interpolation slope parameter, For the repair cycle parameters, This is the mean correction factor. This is the hyperbolic secant adjustment parameter.

5. The edge data fusion and analysis method based on the power IoT operating system according to claim 1, characterized in that, Feature extraction is performed based on the repaired data to construct feature vectors that characterize the data properties, specifically: Through the fourth calculation formula Calculate the eigenvectors ; in, T represents the dimension weight coefficients. For periodic parameters, For characteristic phase modulation parameters, For amplitude enhancement parameters, This is the influence coefficient; For the i-th original data The repaired data, where n represents the number of feature dimensions.

6. The edge data fusion and analysis method based on the power IoT operating system according to claim 5, characterized in that, Also includes: Through the fifth calculation formula The influence coefficient was calculated. ; in, The arctangent adjustment coefficient, For nonlinear strength parameters, For the coefficients of the complementary error function, This is the attenuation adjustment parameter.

7. The edge data fusion and analysis method based on the power IoT operating system according to claim 6, characterized in that, The fusion degree index is calculated based on feature vectors, specifically including: Through the sixth calculation formula The fusion index was calculated. ; in, To integrate the weighting coefficients, These are the sinusoidal modulation parameters. For frequency adjustment parameters, The square root coefficient of the amplitude. For hyperbolic cosine parameters, The influence coefficient of the squared term. For logarithmic correction parameters, is the hyperbolic sine parameter, and m is the number of edge data groups.

8. An edge-to-edge data fusion and analysis system based on a power IoT operating system, characterized in that, The method applied to any one of claims 1 to 7 includes: The data acquisition and encoding module is used to acquire raw data from edge devices and encode the raw data to obtain encoded data. The data anomaly repair module is used to calculate anomaly coefficients based on coded data to identify abnormal data, and to repair the identified abnormal data to obtain repaired data. The feature vector construction module is used to extract features from the repaired data to construct feature vectors that characterize the data properties. The data fusion analysis module is used to calculate the fusion degree index based on feature vectors; determine whether the fusion degree index has reached a preset fusion threshold; if so, output the fusion analysis result; if not, iterate and optimize until the fusion degree index reaches the fusion threshold.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to perform the edge data fusion and analysis method based on the power IoT operating system as described in any one of claims 1 to 7.

10. A processor, characterized in that, The processor is used to run a program, wherein the program executes the edge data fusion and analysis method based on the power IoT operating system as described in any one of claims 1 to 7.

Images