Power load prediction method and system based on association rule analysis

Abstract

The invention relates to the technical field of power systems, provides an association rule analysis-based power load prediction method and system, and aims to solve the problem of hidden fault response lag caused by lack of an equipment health state and load fluctuation dynamic coupling mechanism in the prior art. And the problem of load prediction compensation deviation caused by insufficient weight quantization of the fault propagation path is solved. The method comprises the following steps: generating equipment state data according to a vibration spectrum and an insulation aging index of power equipment; performing fusion analysis, generating an equipment health degree evaluation index, and establishing a dynamic influence model of the equipment abnormal event on the power grid load fluctuation according to the association rule; identifying potential abnormal equipment, performing logic mapping, and generating a fault propagation path weight; and dynamically adjusting according to the weight to obtain an adjusted power load predicted value. According to the technical scheme provided by the invention, equipment vibration and insulation aging data are fused, a health assessment and fault propagation model is constructed, and load prediction is dynamically corrected to prevent and control power grid risks.

Description

Technical Field

[0001] This application relates to the technical field of power systems, and particularly to a power load forecasting method and system based on association rule analysis. Background Art

[0002] With the advancement of the construction of the new power system, the load characteristics of the power grid have become increasingly complex, and the high proportion of new energy grid connection has exacerbated the load volatility and uncertainty. In this context, it is urgent to integrate the dynamic perception ability of equipment health status in load forecasting to cope with the risk of load mutation caused by hidden equipment failures. Two core requirements need to be addressed: one is to establish a real-time coupling analysis model between equipment state parameters and load fluctuations; the other is to quantify the impact of equipment anomalies on the fault propagation of the power grid topology structure to achieve dynamic compensation of load forecasting.

[0003] Existing Solutions The current typical solution for integrating equipment status and load forecasting is a time series forecasting model based on deep learning, such as the method of combining stacked long short-term memory networks and federated learning. This solution captures the time series characteristics of historical load data through a multi-hidden layer long short-term memory network and uses the federated learning framework to aggregate load data in dispersed areas to improve the generalization ability of the model to complex time series patterns. For example, the federated long short-term memory network model proposed by a certain power grid optimizes the load forecasting accuracy through distributed training, but does not introduce the dynamic correlation analysis of equipment health indicators and the power grid topology.

[0004] Existing Solutions Although the above solution can handle the non-linear time series characteristics of load data, there is a lack of association between equipment status and load: the model cannot identify abnormal load fluctuations caused by a decrease in equipment health. For example, a sudden increase in leakage current caused by insulation deterioration may be misjudged as a normal load change; and there is insufficient modeling of the fault propagation path: the existing model does not analyze the fault propagation weight in combination with the power grid topology structure, and it is difficult to quantify its cascading impact on upstream and downstream nodes when equipment is abnormal, resulting in an increased deviation of the predicted value in the fault scenario. Summary of the Invention

[0005] This application provides a power load forecasting method and system based on association rule analysis to solve the problems of lagging response to hidden faults caused by the lack of a dynamic coupling mechanism between equipment health status and load fluctuations in the prior art, and the deviation of load forecasting compensation caused by insufficient quantification of the fault propagation path weight.

[0006] In a first aspect, this application provides a power load forecasting method based on association rule analysis, including: Generating equipment status data according to the vibration spectrum data and insulation aging index of power equipment, where the equipment status data includes the vibration spectrum amplitude and the change rate of the insulation aging index; Perform a fusion analysis on the vibration spectrum amplitude and the change rate of the insulation aging index to generate an equipment health assessment index; According to the equipment health assessment index and the association rules, establish a dynamic impact model of equipment abnormal events on the power grid load fluctuation. The association rules are generated by analyzing the coupling relationship between the change trend of equipment health and the sudden change amount of the load curve within the same time window; Use the dynamic impact model to identify potential abnormal equipment, perform logical mapping based on the operating parameters of the potential abnormal equipment and the power grid topology structure of the area where the power equipment is located to generate the fault propagation path weight, and dynamically adjust the power load prediction value according to the fault propagation path weight to obtain the adjusted power load prediction value.

[0007] Optionally, the performing a fusion analysis on the vibration spectrum amplitude and the change rate of the insulation aging index to generate an equipment health assessment index includes: Based on the vibration spectrum amplitude, perform exponential decay according to the vibration monitoring duration, perform logarithmic transformation on the exponentially decayed vibration spectrum amplitude to generate a vibration deterioration component, and the vibration deterioration component is used to characterize the cumulative deterioration degree of the power equipment; Linearly amplify and correct the change rate of the insulation aging index according to the insulation monitoring temperature, perform hyperbolic tangent transformation on the amplified and corrected change rate of the insulation aging index to generate an aging deterioration component, and the aging deterioration component is used to characterize the non-linear deterioration trend of the surface insulation material of the power equipment; Perform a fusion process on the vibration deterioration component and the aging deterioration component to generate an equipment health assessment index.

[0008] Optionally, the performing a fusion process on the vibration deterioration component and the aging deterioration component to generate an equipment health assessment index includes: When the aging deterioration component exceeds the vibration deterioration component for three consecutive sampling periods and the average value of the difference exceeds a preset threshold, activate the temperature compensation flag; When the temperature compensation flag is activated, adjust the fusion weights according to the inverse proportional constraint relationship between the vibration spectrum amplitude and the change rate of the insulation aging index. The fusion weights include a vibration weight and an aging weight; Take the sum result of the first weighted value and the second weighted value as the fusion component. The first weighted value is the weighted value of the vibration deterioration component and the vibration weight, and the second weighted value is the weighted value of the aging deterioration component and the aging weight; Perform a moving average process on the fusion component to generate a health degree sequence; Filter the high-frequency fluctuation components from the health degree sequence through the fault mode feature codes in the preset equipment historical health status database to obtain the equipment health assessment index.

[0009] Optionally, establishing a dynamic impact model of equipment abnormal events on power grid load fluctuations according to the equipment health assessment index and association rules, including: Dividing multiple equipment abnormal events into corresponding abnormal levels according to the numerical range of the equipment health assessment index to obtain an abnormal level classification result; Calibrating the confidence level of the abnormal level classification result based on the Bayesian probability model to obtain the posterior probability distribution of each abnormal level, and allocating confidence parameters for each abnormal level according to the posterior probability distribution; Dynamically correcting the initial impact coefficient corresponding to the abnormal level based on the confidence parameter to generate a corrected impact coefficient; Taking the corrected impact coefficient as input, generating a dynamic correction weight through the Bayesian optimization algorithm, and the dynamic correction weight is constrained by the load disturbance safety boundary in the association rule; Within a sliding time window, extracting the dynamic correction weight corresponding to the equipment abnormal event according to the time stamp of the equipment abnormal event, and performing a point-by-point convolution operation on the dynamic correction weight and the instantaneous change value of the power grid load fluctuation to generate a real-time impact component of the equipment abnormal event; Calculating a time decay factor using the Bayesian recursive estimation method according to the recovery trend of the equipment health assessment index, and the decay rate of the time decay factor is negatively correlated with the recovery trend; Generating a residual amount of the historical impact component based on the time decay factor, and superimposing the residual amount and the real-time impact component to generate a total impact amount; Calculating the residual between the predicted value and the measured value of the power grid load fluctuation in real time to form a residual sequence. If the peak-to-peak value of the residual in the residual sequence is not within the preset steady-state error range, adaptively adjusting the dynamic correction weight according to the residual sign, and correcting the decay rate based on the total impact amount until the peak-to-peak value of the residual converges to within the preset steady-state error range to obtain a dynamic impact model of equipment abnormal events on power grid load fluctuations.

[0010] Optionally, within the sliding time window, extracting the dynamic correction weight corresponding to the equipment abnormal event according to the time stamp of the equipment abnormal event, and performing a point-by-point convolution operation on the dynamic correction weight and the instantaneous change value of the power grid load fluctuation to generate a real-time impact component of the equipment abnormal event, including: Determining the start boundary and end boundary of the sliding time window according to the time stamp of the equipment abnormal event, and determining the length between the end boundary and the start boundary as the sliding time window length; Within the sliding time window, aligning the occurrence time of the equipment abnormal event according to the time stamp, and extracting the dynamic correction weight corresponding to the equipment abnormal event; Construct a convolution kernel with the same length as the sliding time window based on the time decay characteristic of the dynamically corrected weight; Extract the instantaneous change value of the grid load fluctuation from a preset grid load fluctuation database according to the sampling frequency of the grid load fluctuation; Perform a point-by-point convolution operation on the instantaneous change value and the convolution kernel to obtain a convolution result, where each sampling point in the convolution result corresponds to the local influence intensity of the equipment abnormal event; Perform amplitude scaling within the sliding time window on the convolution result, and superimpose the scaled convolution result and the baseline component of the grid load fluctuation according to the time stamp to generate a real-time influence component of the equipment abnormal event.

[0011] Optionally, the logical mapping based on the operating parameters of the potential abnormal equipment and the grid topology structure of the area where the power equipment is located to generate a fault propagation path weight includes: Map the grid topology structure into a directed graph, where the vertices in the directed graph represent power equipment, the vertex attributes include vertex operating parameters, the edges represent the electrical connection direction and physical coupling relationship between power equipment, and the edge attributes include an initial weight and an electrical distance value; Based on the initial weight and in combination with the vertex operating parameters of the two ends of the edge, calculate the fault propagation probability of the edge through a non-linear compression function; Take the power equipment with abnormal equipment health assessment indicators as potential abnormal equipment, start from the potential abnormal equipment, expand adjacent vertices hop by hop along the edge direction, and stop expanding when the cumulative number of hops is equal to the preset propagation order to generate an electrical connection subgraph; In the electrical connection subgraph, traverse all edges along the edge direction, multiply the fault propagation probabilities of each edge, and superimpose the total electrical distance value to generate a propagation path comprehensive weight; Take the path with the propagation path comprehensive weight exceeding the preset dynamic threshold as the fault propagation path, and take the path comprehensive weight of the fault propagation path as the fault propagation path weight.

[0012] Optionally, the step of traversing all edges along the edge direction in the electrical connection subgraph, multiplying the fault propagation probabilities of each edge, and superimposing the total electrical distance value to generate a propagation path comprehensive weight includes: In the electrical connection subgraph, start from the starting point, perform a breadth-first traversal according to the edge direction, and layer by layer record the set of traversal paths without duplicate vertices. Each path consists of a vertex sequence and a corresponding edge sequence; For each edge sequence, multiply the fault propagation probability values of each edge in sequence to generate a path propagation coefficient, and accumulate the electrical distance values of all edges in the path to obtain the total electrical distance value; Combine the path propagation coefficient with the total electrical distance value to calculate the comprehensive weight of the propagation path.

[0013] In a second aspect, the present application provides a power load forecasting system based on association rule analysis, including: A generation module, configured to generate device status data according to the vibration spectrum data and insulation aging index of power equipment, where the device status data includes the vibration spectrum amplitude and the change rate of the insulation aging index; An analysis module, configured to perform fusion analysis on the vibration spectrum amplitude and the change rate of the insulation aging index to generate an equipment health assessment index; A construction module, configured to establish a dynamic influence model of equipment abnormal events on grid load fluctuations according to the equipment health assessment index and association rules, where the association rules are generated by analyzing the coupling relationship between the change trend of equipment health and the sudden change amount of the load curve within the same time window; An adjustment module, configured to use the dynamic influence model to identify potential abnormal equipment, perform logical mapping based on the operating parameters of the potential abnormal equipment and the grid topology structure of the area where the power equipment is located to generate a fault propagation path weight, and dynamically adjust the power load prediction value according to the fault propagation path weight to obtain an adjusted power load prediction value.

[0014] In a third aspect, the present application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a power load forecasting method based on association rule analysis as described in any item of the first aspect.

[0015] In a fourth aspect, the present application provides a computer storage medium storing a computer program, where when the computer program is executed by a computer, it implements a power load forecasting method based on association rule analysis as described in any item of the first aspect.

[0016] In an embodiment of the present application, a power load forecasting method based on association rule analysis is provided. The method includes: generating device status data according to the vibration spectrum data and insulation aging index of power equipment, where the device status data includes the vibration spectrum amplitude and the change rate of the insulation aging index; performing fusion analysis on the vibration spectrum amplitude and the change rate of the insulation aging index to generate a device health assessment index; establishing a dynamic influence model of equipment abnormal events on grid load fluctuations according to the device health assessment index and association rules, where the association rules are generated by analyzing the coupling relationship between the change trend of device health and the mutation amount of the load curve within the same time window; identifying potential abnormal devices using the dynamic influence model, performing logical mapping based on the operating parameters of the potential abnormal devices and the grid topology structure of the area where the power equipment is located to generate the weight of the fault propagation path, and dynamically adjusting the power load forecast value according to the weight of the fault propagation path to obtain the adjusted power load forecast value.

[0017] Advantages of the present application: By integrating multi-dimensional parameters of vibration spectrum and insulation aging, the present application breaks through the limitations of traditional single-index monitoring, realizes the synchronous quantitative characterization of equipment mechanical wear and insulation performance degradation states, and solves the problem of state blind spots caused by data dimension fragmentation in existing solutions. The multi-source deterioration feature coupling analysis method is adopted to eliminate the evaluation deviation caused by the time-varying characteristics of vibration signals and the difference in insulation aging rates, and overcome the defect of insufficient adaptability of traditional linear weighted fusion models to non-linear deterioration processes. By mining the spatio-temporal correlation between health trends and load mutations, a causal reasoning framework for equipment abnormalities-grid disturbances is constructed, filling the technical gap that the static threshold alarm mechanism cannot quantify the conduction effect of equipment deterioration on system-level load fluctuations. The prediction model is corrected in real time based on the weight of the fault propagation path, realizing the accurate mapping of equipment-level risks to system-level load abnormalities under grid topology constraints, and breaking the vicious cycle problem of "prediction inaccuracy-fault diffusion" caused by traditional prediction methods ignoring the evolution of equipment health states.

[0018] Furthermore, extract the mechanical cumulative degradation features through the exponential decay-logarithmic transformation of the vibration spectrum amplitude, capture the non-linear degradation law of the insulating material by combining the hyperbolic transformation of the insulation aging index with temperature correction, and construct the vibration / aging degradation components; design a temperature compensation mechanism based on exceedance persistence, dynamically adjust the fusion weight using inverse proportion constraint, and generate an anti-interference health index through moving average and feature filtering to achieve the coupled characterization of the multi-physical field degradation process. Aiming at the three major pain points in the prior art, namely the mismatch between the time-varying characteristics of vibration and insulation parameters, the misjudgment of the degradation trend under temperature interference, and the false alarm of the health index caused by high-frequency noise, break through the technical bottleneck of the traditional fixed-weight fusion model. Through non-linear feature enhancement extraction and dynamic compensation mechanism, significantly improve the recognition sensitivity of the composite degradation mode; combine the fluctuation suppression strategy driven by feature codes to stably characterize the true health state of the equipment even in a strong electromagnetic interference environment, providing a reliable decision-making basis for the risk warning of equipment with high-proportion new energy access to the power grid.

[0019] These aspects or other aspects of the present application will be more clearly understood in the following description of the embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application or in the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0021] Figure 1 It is a flowchart of a power load forecasting method based on association rule analysis provided by an embodiment of the present application; Figure 2 It is a schematic structural diagram of a power load forecasting system based on association rule analysis provided by an embodiment of the present application; Figure 3 It is a schematic structural diagram of a computing device provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] In order to enable those skilled in the art to better understand the solution of the present application, the following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the drawings in the embodiments of the present application.

[0023] In some of the processes described in the specification, claims, and above-mentioned drawings of this application, there are multiple operations that appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order in which they appear herein or may be executed in parallel. The operation numbers such as 11, 12, etc. are only used to distinguish different operations, and the numbers themselves do not represent any execution order. Additionally, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions such as "first", "second", etc. in this text are used to distinguish different messages, devices, modules, etc., do not represent a sequence, and do not limit that "first" and "second" are of different types.

[0024] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0025] To solve the problems of lagging implicit fault response caused by the lack of a dynamic coupling mechanism between the equipment health state and load fluctuations and the load prediction compensation deviation caused by insufficient quantification of the fault propagation path weight in the prior art, the embodiments of the present application provide a power load prediction method and system based on association rule analysis. The method adopts the following concept: Based on the physical characteristics of multi-source state monitoring data of equipment, first, the exponential decay processing and logarithmic transformation of the vibration spectrum amplitude are used to quantify the mechanical cumulative deterioration, and the temperature correction and hyperbolic tangent transformation of the change rate of the insulation aging index are combined to characterize the non-linear deterioration trend of the insulation, so as to construct a two-dimensional deterioration component; then a dynamic weight fusion mechanism is designed to activate the inverse proportion constraint to adjust the fusion weight when the aging component is continuously abnormal, and the moving average and historical fault feature library are combined to filter out noise interference to generate a health index; subsequently, the spatio-temporal coupling relationship between the change of equipment health and the load mutation amount is mined by association rules, and a dynamic influence model is established to identify abnormal equipment; finally, the fault propagation path weight is mapped based on the power grid topology structure, and the closed-loop correction of the load prediction value is realized through weighted compensation, forming a full-chain analysis framework of "equipment deterioration, fault propagation, and load fluctuation".

[0026] Figure 1 The flowchart of a power load prediction method based on association rule analysis provided by the embodiments of the present application is as Figure 1 shown, and the method includes: S11. Generate equipment state data according to the vibration spectrum data and insulation aging index of power equipment. The equipment state data includes the vibration spectrum amplitude and the change rate of the insulation aging index.

[0027] Among them, the vibration spectrum data is the frequency-domain amplitude distribution obtained by collecting the mechanical vibration signals of the equipment through vibration sensors and performing Fourier transform. The insulation aging index is a quantitative index reflecting the deterioration degree of the equipment insulation material, which is calculated through the tangent value of the dielectric loss angle or the polarization index. The vibration spectrum amplitude refers to the vibration signal generated by the power equipment during operation. After being decomposed by spectrum analysis, the amplitude values corresponding to each frequency component are obtained. The insulation aging index change rate refers to the change rate of the aging degree of the power equipment insulation material over time, which is usually calculated based on the insulation aging index.

[0028] In the embodiment of the present application, first, the vibration spectrum data of the power equipment is collected, and the insulation aging index is monitored at the same time; the amplitude of the characteristic frequency point is extracted from the vibration spectrum data as the vibration spectrum amplitude, and the change rate of the insulation aging index is calculated according to the time window, and finally the equipment status data is generated, including two key parameters: the vibration spectrum amplitude and the insulation aging index change rate.

[0029] S12. Perform fusion analysis on the vibration spectrum amplitude and the insulation aging index change rate to generate an equipment health assessment index.

[0030] Among them, the equipment health assessment index is a weighted comprehensive score that fuses vibration and insulation aging parameters and is used to quantify the operating status of the equipment.

[0031] In the embodiment of the present application, a multi-source data fusion algorithm is used to jointly analyze the vibration spectrum amplitude and the insulation aging index change rate, weights are assigned according to the contribution degrees of the two types of parameters, and the equipment health assessment index is calculated through linear weighting. The lower the health index, the more serious the deterioration of the equipment, providing a quantitative basis for subsequent dynamic impact modeling.

[0032] S13. According to the equipment health assessment index and the association rules, establish a dynamic impact model of the equipment abnormal event on the power grid load fluctuation. The association rules are generated by analyzing the coupling relationship between the change trend of the equipment health degree and the sudden change amount of the load curve within the same time window.

[0033] Among them, the association rules are probability rules describing the correlation between the change of the equipment health degree and the load fluctuation. The dynamic impact model is a machine learning-based model that inputs the equipment health index and outputs the predicted impact value on the power grid load fluctuation. The sudden change amount of the load curve is the change amount of the load value exceeding the historical fluctuation range within a short time, usually measured by multiples of the standard deviation. The equipment abnormal event is a state in the equipment health assessment index that exceeds the preset threshold or shows a continuous deterioration trend, indicating that the equipment has entered the potential failure or performance deterioration stage but has not completely failed.

[0034] In the embodiments of the present application, association rules are trained based on historical data to establish a dynamic impact model: taking the health assessment index as the input and the load fluctuation amount as the output, fitting the dynamic coupling relationship between the two through time series regression, and outputting a probability impact matrix of equipment abnormal events on the power grid load fluctuation, which is used to quantify the cascading effect of abnormal events on the power grid.

[0035] S14. Use the dynamic impact model to identify potential abnormal equipment, perform logical mapping based on the operating parameters of the potential abnormal equipment and the power grid topology structure in the area where the power equipment is located, generate the fault propagation path weight, and dynamically adjust the power load prediction value according to the fault propagation path weight to obtain the adjusted power load prediction value.

[0036] Among them, the fault propagation path weight is the probability weight value of the fault diffusion path calculated based on the power grid topology structure and equipment parameters. The power grid topology structure is the physical / logical connection relationship of equipment nodes in the power grid, which is represented by a graph structure. The power load prediction value is the estimated amount of power demand in a future period of time, usually measured in power or energy. In step S14, this concept is extended to a dynamic prediction value that integrates the equipment health status and the influence of the power grid topology.

[0037] In the embodiments of the present application, after using the dynamic impact model to identify potential abnormal equipment, combining the equipment operating parameters and the power grid topology structure, calculate the fault propagation path weight through the graph propagation algorithm, and dynamically adjust the load prediction value: reduce the prediction value of the load node corresponding to the path with a weight higher than the threshold by the correction coefficient, and finally generate the adjusted power load prediction value.

[0038] The following is a specific example: The vibration sensor of a transformer in a substation collects time-domain vibration signals. After fast Fourier transform, the vibration spectrum amplitude at the 800 Hz frequency point is extracted as 0.9 mm / s. At the same time, its insulation aging index is monitored, and the change rate of the aging index in the current period is calculated as 15%. Thus, equipment status data is generated, including two types of parameters: the vibration spectrum amplitude of 0.9 mm / s and the change rate of the insulation aging index of 15%. The vibration spectrum amplitude and the change rate of the insulation aging index are input into the fusion model of the entropy weight method and the technique for order preference by similarity to an ideal solution (TOPSIS). The entropy weight method is used to calculate the weights, obtaining a vibration amplitude weight of 0.6 and an aging change rate weight of 0.4. The proximity of the two types of parameters to the entropy weight method and the ideal solution is calculated, and after weighting, the equipment health assessment index is obtained as 0.3. Based on the association rules trained from historical data, a warning is triggered. The dynamic impact model analyzes the correlation between the equipment health index of 0.3 of this transformer and the current power grid load curve, predicting that the probability of a sudden increase in the downstream load caused by its abnormal event reaches 82%, and generating a load fluctuation risk probability matrix. After the system identifies this transformer as a potentially abnormal device, combined with the operating parameters and the power grid topology structure of the region where it is located, such as substations A, B, and C connected downstream of the transformer, the fault propagation path weights are calculated through the graph propagation algorithm. According to the line impedance and load rate, the path weight coefficients are calculated as 0.7 for substation A, 0.5 for substation B, and 0.3 for substation C, and the load prediction value is dynamically adjusted: the prediction value for substation A is decreased by 35% corresponding to the weight of 0.7, substation B is decreased by 25%, and substation C is decreased by 15%. Finally, the adjusted power load prediction value is output to avoid cascading power outages caused by transformer overload.

[0039] By executing S11~S14, the embodiment of the present application generates an equipment health index through multi-dimensional fusion of vibration and insulation aging data, constructs a fault propagation model in combination with the power grid load fluctuation law, realizes accurate identification of abnormal equipment and dynamic prediction of fault paths, and finally corrects the load prediction value in real time based on the propagation weight, significantly improving the power grid risk warning ability and load dispatching reliability, and reducing power outage losses caused by equipment abnormalities.

[0040] In a possible embodiment, S11 is to perform fusion analysis on the vibration spectrum amplitude and the change rate of the insulation aging index to generate an equipment health assessment index, including: Step 111: Based on the vibration spectrum amplitude, perform exponential decay according to the vibration monitoring duration, and perform logarithmic transformation on the exponentially decayed vibration spectrum amplitude to generate a vibration deterioration component, which is used to characterize the cumulative deterioration degree of the power equipment.

[0041] Among them, the vibration monitoring duration is the cumulative vibration data acquisition time of the power equipment from the time of commissioning to the present, usually in hours. Exponential decay is a mathematical processing process of performing time decay correction on the vibration amplitude according to the natural exponential function e (-λt) The vibration deterioration component is a quantitative index that characterizes the cumulative damage of the mechanical structure after time decay and logarithmic transformation.

[0042] In the embodiment of the present application, the vibration monitoring duration is divided by a preset time constant to obtain an attenuation exponent, and the original value of the vibration spectrum amplitude is multiplied by the negative power of e to the attenuation exponent to achieve exponential attenuation of the amplitude. Then, a natural logarithm transformation is performed on the attenuated amplitude, that is, the ln value is taken after adding 1 to the amplitude of each frequency point, eliminating the dimension difference and compressing the data range, and finally a vibration deterioration component is generated. This component reflects the cumulative effect of mechanical structure wear by retaining the characteristic of the low-frequency energy ratio.

[0043] Step 112: Linearly amplify and correct the change rate of the insulation aging index according to the insulation monitoring temperature, and perform a hyperbolic tangent transformation on the amplified and corrected change rate of the insulation aging index to generate an aging deterioration component, which is used to characterize the non-linear deterioration trend of the surface insulation material of the power equipment.

[0044] Among them, the insulation monitoring temperature refers to the real-time monitored temperature value on the surface of the insulation material, which is a key environmental parameter affecting the aging rate. The linear amplification correction is an adjustment method for proportionally amplifying the aging change rate according to the temperature deviation. The aging deterioration component is an index characterizing the non-linear deterioration of the insulation material after temperature correction and hyperbolic tangent transformation. The hyperbolic tangent transformation refers to the mathematical function tanh(x), which compresses the input value into the interval (-1, 1) and retains the non-linear characteristics.

[0045] In the embodiment of the present application, first, the change rate of the insulation aging index and the real-time insulation monitoring temperature are obtained through sensors. The temperature value is subtracted from the reference temperature of 25 °C and then multiplied by the temperature correction coefficient of 0.15 to linearly amplify the change rate of the aging index. For example, when the temperature is 60 °C, the change rate will be amplified to 1.525 times the original value. Then, the amplified value is input into the hyperbolic tangent function tanh(x) to map the change rate to the interval (-1, 1), eliminating the influence of extreme outliers. The finally generated aging deterioration component characterizes the non-linear acceleration characteristic of the degradation of the insulation material performance through the S-shaped curve feature.

[0046] Step 113: Perform a fusion process on the vibration deterioration component and the aging deterioration component to generate an equipment health assessment index.

[0047] Among them, in the embodiment of the present invention, the entropy weight method and the technique for order preference by similarity to an ideal solution (TOPSIS) fusion algorithm are used to process the vibration deterioration component and the aging deterioration component: first, the weights of the two are calculated by the entropy weight method, and then the closeness of each component to the ideal solution is calculated based on the TOPSIS model, and finally, a weighted sum is used to generate an equipment health assessment index. This index synchronously reflects the characteristics of mechanical cumulative deterioration and insulation non-linear deterioration, providing a unified quantitative basis for subsequent fault prediction.

[0048] Continuing with the above example, based on the vibration spectrum amplitude of 0.9 mm / s at the 800 Hz frequency point currently collected by a substation transformer, combined with a continuous vibration monitoring duration of 50 hours, an exponential decay operation is performed according to the time decay coefficient of 0.02, and the vibration amplitude after decay is obtained as 0.9×e （-0.02×50） = 0.331 mm / s. Then, through the natural logarithm transformation ln(0.331 + 1) = 0.287, a vibration deterioration component is generated to quantitatively reflect the cumulative effect of mechanical wear of the equipment bearings and windings; at the same time, according to the insulation aging index change rate of 15%, combined with the current insulation material temperature monitoring value of 90°C, a temperature linear correction is performed, and the change rate is amplified to 15%×(90 / 80) = 16.875%. Then, through the hyperbolic tangent function tanh(16.875%) = 0.167, an aging deterioration component is generated to characterize the non-linear deterioration characteristics of the thermal cracking of insulating paper in a high-temperature environment; finally, the entropy weight method is used to determine the vibration component weight of 0.6 and the aging component weight of 0.4, and the weighted values of 0.287×0.6 = 0.172 and 0.167×0.4 = 0.067 are calculated respectively. After superimposition, the comprehensive deterioration degree is 0.239. Combining the technique for order preference by similarity to an ideal solution (TOPSIS) method, the Euclidean distances between the two types of parameters and the positive and negative ideal solutions are calculated, and after normalization, the equipment health assessment index of 0.3 is output, accurately reflecting the coupling effect result of mechanical vibration deterioration and insulation aging process.

[0049] By executing Steps 111 to 113, the embodiments of the present application extract mechanical cumulative deterioration characteristics through time-domain decay correction and logarithmic transformation of the vibration amplitude, capture the non-linear deterioration law of insulation through the hyperbolic tangent transformation with temperature compensation, and finally use the entropy weight method - TOPSIS method to fuse and generate a comprehensive health index, realizing multi-dimensional accurate assessment of the mechanical and insulation states of the equipment, providing a scientific basis for formulating differential operation and maintenance strategies, and improving the accuracy compared with traditional single-parameter assessment methods.

[0050] In a possible embodiment, Step 113, fusing the vibration deterioration component and the aging deterioration component to generate an equipment health assessment index, includes: Step a1, when the aging deterioration component exceeds the vibration deterioration component for three consecutive sampling periods and the average value of the difference exceeds a preset threshold, activate the temperature compensation flag.

[0051] Wherein, the sampling period refers to the time interval of data collection, which is used to periodically update the deterioration component. The average value of the difference is the arithmetic average of the differences between the aging component and the vibration component within multiple consecutive periods. The temperature compensation flag is a binary state signal indicating that the temperature has a significant impact on insulation aging.

[0052] In the embodiments of the present application, the system monitors in real time the numerical relationship between the vibration deterioration component and the aging deterioration component. When the values of the aging deterioration component exceed those of the vibration deterioration component for three consecutive sampling periods and the average value of the difference between the two exceeds a preset threshold, the temperature compensation flag is activated. This flag indicates that the insulation aging is significantly affected by temperature, and the subsequent fusion weights need to be adjusted.

[0053] Step a2: When the temperature compensation flag is activated, adjust the fusion weights according to the inverse proportional constraint relationship between the vibration spectrum amplitude and the change rate of the insulation aging index. The fusion weights include the vibration weight and the aging weight.

[0054] Among them, the inverse proportional constraint relationship refers to the distribution rule in which the vibration and aging weights are inversely proportional to each other, ensuring that the sum of the weights is 1. The inverse proportional constraint relationship is defined as: , ,where is the vibration deterioration component, is the aging deterioration component, the vibration weight and the aging weight sum to 1, that is .

[0055] In the embodiments of the present application, if the temperature compensation flag is activated, the fusion weights are adjusted according to the inverse proportional constraint relationship: vibration weight = aging deterioration component / (vibration deterioration component + aging deterioration component), aging weight = vibration deterioration component / (vibration deterioration component + aging deterioration component). For example, if the vibration deterioration component is 1.2 and the aging deterioration component is 1.5, then the vibration weight = 1.5 / (1.2 + 1.5) = 0.56, and the aging weight = 0.44, so as to weaken the interference of temperature anomalies on insulation aging.

[0056] Step a3: Take the sum of the first weighted value and the second weighted value as the fusion component. The first weighted value is the weighted value of the vibration deterioration component and the vibration weight, and the second weighted value is the weighted value of the aging deterioration component and the aging weight.

[0057] Among them, the first weighted value refers to the product of the vibration deterioration component and the vibration weight, reflecting the contribution degree of mechanical deterioration. The second weighted value refers to the product of the aging deterioration component and the aging weight, reflecting the contribution degree of insulation deterioration. The fusion component refers to the sum of the first weighted value and the second weighted value, characterizing the overall deterioration level of the equipment.

[0058] In the embodiments of the present application, when calculating the fusion component, multiply the vibration deterioration component by the vibration weight to obtain the first weighted value, multiply the aging deterioration component by the aging weight to obtain the second weighted value, and add the two to obtain the fusion component, which comprehensively characterizes the superposition effect of mechanical and insulation deterioration.

[0059] Step a4: Perform a moving average process on the fusion component to generate a healthiness sequence.

[0060] Among them, the moving average process is a filtering method that calculates the sequence mean through a sliding window to smooth the data. The healthiness sequence refers to the continuous state trend data generated by smoothing the device healthiness evaluation indicators in the time dimension.

[0061] In the embodiment of the present application, perform a moving average process on the fusion component: take the fusion components of 5 consecutive sampling periods, calculate the arithmetic mean as the current healthiness sequence value, so as to eliminate short-term fluctuation interference and generate a smooth healthiness time series.

[0062] Step a5: Filter the high-frequency fluctuation components from the healthiness sequence through the fault mode feature codes in the preset device historical health status database to obtain the device healthiness evaluation indicators.

[0063] Among them, the fault mode feature code refers to the frequency or amplitude feature coding corresponding to a specific fault in the healthiness sequence.

[0064] In the embodiment of the present invention, through the preset device historical health status database, perform wavelet transform on the healthiness sequence to extract high-frequency components, filter out the noise irrelevant to the fault, retain the low-frequency effective signal, and finally generate the device healthiness evaluation indicators to directly match the fault warning threshold.

[0065] Continuing with the above example, for a substation transformer, the aging deterioration components monitored in three consecutive sampling periods are 0.172, 0.185, and 0.198 respectively, and the vibration deterioration components are 0.165, 0.160, and 0.155 respectively. The mean value of the difference between the two is 0.015, and the preset threshold is 0.01, triggering the activation of the temperature compensation flag; based on the inverse proportional relationship between the vibration spectrum amplitude of 0.9 mm / s and the insulation aging index change rate of 15%, the formula for the vibration amplitude weight is: vibration amplitude weight = aging change rate weight × aging change rate / vibration amplitude. Dynamically adjust the original entropy weight method weights of 0.6 and 0.4 to a vibration weight of 0.4 and an aging weight of 0.6; then calculate the vibration deterioration component of 0.287 × 0.4 = 0.115 and the aging deterioration component of 0.167 × 0.6 = 0.100, and superimpose them to form a fusion component of 0.215. Use a 3-day time window moving average to smooth the sequence [0.215, 0.220, 0.225], and output a healthiness sequence mean value of 0.220; finally, call the historical fault mode feature code, filter out the fluctuation components with a frequency higher than 0.5 Hz in the healthiness sequence through wavelet transform, and extract the low-frequency trend component of 0.208 as the final device healthiness evaluation indicator, and synchronously update the transformer healthiness parameters in the dynamic impact model to optimize the load sudden increase probability prediction and path weight coefficient calculation.

[0066] By performing steps a1 to a5, the embodiments of the present application accurately distinguish the dominant factors of mechanical and insulation deterioration through dynamic activation of the temperature compensation mechanism and inverse proportional adjustment of weights; combined with moving average and fault feature filtering, effectively suppress noise interference and extract key fault signals, enabling the health assessment index to simultaneously possess environmental adaptability, anti-interference ability, and fault directivity, and reducing the false alarm rate compared with the traditional static weight method.

[0067] In a possible embodiment, S13. According to the device health assessment index and the association rule, establish a dynamic impact model of the device abnormal event on the power grid load fluctuation, including: Step 131. According to the numerical interval of the device health assessment index, divide multiple device abnormal events into corresponding abnormal levels to obtain an abnormal level classification result. Among them, the abnormal level classification result is a label of the severity of the device abnormality divided according to the health index.

[0068] In the embodiments of the present application, first obtain the numerical interval of the device health assessment index, classify historical device abnormal events into the corresponding intervals according to the health values at the time of their triggering, and generate an abnormal level classification result. Specifically, by setting an interval threshold divider, when an abnormal event occurs, compare the associated health value with a preset threshold: if the health value is 55, it is determined as a severe abnormal level. This result includes the statistical count and distribution characteristics of the events in each abnormal level.

[0069] Step 132. Calibrate the confidence level of the abnormal level classification result based on the Bayesian probability model to obtain the posterior probability distribution of each abnormal level, and configure confidence level parameters for each abnormal level according to the posterior probability distribution.

[0070] Among them, the posterior probability distribution refers to the probability distribution calculated by combining prior knowledge and new evidence in the Bayesian model. The confidence level parameter is a quantitative parameter (0 - 1) representing the reliability of the abnormal level judgment.

[0071] In the embodiments of the present application, calibrate the confidence level of the abnormal level classification result based on the Bayesian probability model. First, establish a prior probability distribution, and combine the occurrence frequency of the abnormal events monitored in real time to calculate the posterior probability distribution through the Bayesian formula: posterior probability = (prior probability × likelihood probability) / evidence factor. For example, the posterior probability of severe abnormality is corrected from 30% to 32.5%. Configure confidence level parameters for each level according to the posterior probability, and the parameter value is the product of the posterior probability and the historical accuracy rate.

[0072] Step 133. Dynamically correct the initial impact coefficient corresponding to the abnormal level based on the confidence level parameter to generate a corrected impact coefficient. Among them, the corrected impact coefficient refers to the impact intensity coefficient of the abnormal event on the power grid load after confidence level calibration.

[0073] In the embodiments of the present application, the initial influence coefficient corresponding to the anomaly level is dynamically corrected according to the confidence parameter. The linear interpolation algorithm is adopted: the corrected influence coefficient = the initial coefficient × (1 + the confidence parameter × the adjustment factor). For example, the initial coefficient of 1.2 for a severe anomaly is corrected to 1.2×(1 + 0.92×0.15) = 1.365 when the confidence level is 0.92. The adjustment factor of 0.15 is set by expert experience to ensure that the corrected coefficient does not exceed the equipment tolerance limit.

[0074] Step 134: Using the corrected influence coefficient as the input, generate a dynamic correction weight through the Bayesian optimization algorithm. The dynamic correction weight is constrained by the load disturbance safety boundary in the association rule. Among them, the dynamic correction weight is the influence weight of the anomaly event optimized and generated under the safety boundary constraint.

[0075] In the embodiments of the present invention, using the corrected influence coefficient as the input, generate a dynamic correction weight through the Bayesian optimization algorithm. Define the objective function as minimizing the power grid load prediction error, and the constraint condition is the load disturbance safety boundary. Use Gaussian process regression to construct a surrogate model, calculate the objective function value of the weight candidate value in each iteration, and select the weight value that makes the objective function decrease and satisfies the constraint. The finally output dynamic correction weight is related to the potential influence intensity of the equipment anomaly on the power grid.

[0076] Step 135: In the sliding time window, extract the dynamic correction weight corresponding to the equipment anomaly event according to the time stamp of the equipment anomaly event, and perform a point-by-point convolution operation on the dynamic correction weight and the instantaneous change value of the power grid load fluctuation to generate the real-time influence component of the equipment anomaly event.

[0077] In the embodiments of the present invention, in the sliding time window, extract the corresponding dynamic correction weight according to the time stamp of the equipment anomaly event. Perform a point-by-point convolution operation on the weight sequence and the instantaneous change value of the power grid load fluctuation: the real-time influence component = ∑(weight_i × ΔP_t-i}), where i is the time offset in the sliding window, and ΔP is the load change rate. For example, the influence amount at the current moment t = 0.73×ΔP_t + 0.68×ΔP_{t-1}+…, and finally generate the real-time superposition effect reflecting the influence of the anomaly event on the load fluctuation.

[0078] Step 136: According to the recovery trend of the equipment health assessment index, use the Bayesian recursive estimation method to calculate the time decay factor. The decay rate of the time decay factor is negatively correlated with the recovery trend.

[0079] Among them, the time decay factor is a parameter reflecting the decay rate of the historical influence over time, and is negatively correlated with the health recovery trend.

[0080] In the embodiment of the present application, according to the recovery trend of the device health assessment index, the Bayesian recursive estimation method is used to calculate the time decay factor. A state space model is constructed, with the health recovery rate as the observed variable, and the decay factor is updated through Kalman filtering: decay rate = 1 / (1 + recovery rate × time constant). For example, a recovery rate of 3 minutes per day corresponds to a decay factor that decreases by 0.25 per day, ensuring that the historical influence decays rapidly during rapid recovery.

[0081] Step 137: Generate the residual of the historical influence component based on the time decay factor, and superimpose the residual and the real-time influence component to generate the total influence amount.

[0082] Among them, the total influence amount refers to the superimposed result of the historical residual influence and the real-time influence.

[0083] In the embodiment of the present application, the residual of the historical influence component is generated based on the time decay factor. Residual = historical influence component × e (-衰减速率×Δt) , where Δt is the time difference since the event occurred. Superimpose the residual and the real-time influence component: Total influence amount = Residual + Real-time influence component. For example, when the residual of a certain fault, 0.35, is superimposed with the real-time influence of 0.62, the total influence amount is 0.97, which represents the continuous comprehensive influence of the anomaly on the power grid load.

[0084] Step 138: Calculate the residual between the predicted value and the measured value of the power grid load fluctuation in real time to form a residual sequence. If the peak-to-peak value of the residual in the residual sequence is not within the preset steady-state error range, adaptively adjust the dynamic correction weight according to the residual sign, and correct the decay rate based on the total influence amount until the peak-to-peak value of the residual converges within the preset steady-state error range, and obtain the dynamic influence model of the device abnormal event on the power grid load fluctuation.

[0085] Among them, the residual sequence is a sequence formed by sorting the differences between the load predicted values and the measured values over time.

[0086] In the embodiment of the present invention, calculate the residual between the predicted value and the measured value of the power grid load fluctuation in real time to form a residual sequence. If the peak-to-peak value of the residual exceeds the preset steady-state error range, adaptively adjust the dynamic correction weight according to the residual sign: the weight increases for positive residuals, for example, from 0.73 to 0.78, and the weight decreases for negative residuals, for example, from 0.73 to 0.68). At the same time, correct the decay rate based on the total influence amount (such as when the total influence amount increases by 0.1, the decay rate decreases by 5%) until the peak-to-peak value of the residual converges within the allowable range, and finally output the dynamic influence model.

[0087] Continuing with the above example, a transformer in a substation is determined to be at level three anomaly based on the numerical range of the equipment health assessment index of 0.3, corresponding to the health range of 0.2 - 0.4, and a "moderate deterioration" anomaly level label is generated; using the Bayesian probability model, combining the true incidence rate of level three anomalies in historical fault data with the deviation degree of the current health index, the posterior probability distribution is calculated to obtain a confidence level of 85% for level three anomalies, and a confidence parameter of 0.85 is allocated to it; according to the confidence parameter, the initial influence coefficient of 0.82 is dynamically corrected, and the corrected influence coefficient is generated using the confidence weighted formula 0.82 × 0.85 = 0.697; through the Bayesian optimization algorithm, with the corrected influence coefficient of 0.697 as the input, the optimal weight is iteratively solved under the constraint of the load disturbance safety boundary to generate a dynamically corrected weight of 0.75; within a 30 - minute sliding time window, the timestamps of three consecutive anomaly events triggered by the transformer are extracted, and the convolution operation is performed between the dynamically corrected weight sequence [0.75, 0.73, 0.70] and the corresponding grid load fluctuation values [0.9%, 1.2%, 0.8%] at each moment, and the real - time influence component 0.75 × 0.9% + 0.73 × 1.2% + 0.70 × 0.8% = 2.35% is output; according to the trend that the health index has risen from 0.25 to 0.3 in the past 6 hours, the Bayesian recursive estimation is used to calculate the time decay factor e （-0.008×t） , where t is time, and the historical influence component of 2.1% three hours ago is decayed to the residual amount: 2.1% × e （-0.008×3） = 2.05%; finally, the residual amount of 2.05% and the real - time influence component of 2.35% are superimposed to generate a total influence amount of 4.4% and updated to the load fluctuation risk probability matrix, driving the secondary correction of the predicted values of substations A, B, and C, and realizing the full - life - cycle quantification tracking of the impact of anomaly events.

[0088] By executing steps 131 to 138, the embodiment of the present application realizes the refined modeling of the impact of equipment anomaly events on grid load fluctuations through anomaly level dynamic calibration, Bayesian optimized weight allocation, and decay factor recursive estimation; combined with the residual feedback adaptive adjustment mechanism, it ensures that the model can track the load changes in real - time and converge to a stable state, reducing the prediction error by more than 40% compared with the traditional static model, and significantly improving the robustness of grid dispatching decisions.

[0089] In a possible embodiment, step 135, within the sliding time window, according to the timestamps of the equipment anomaly events, the dynamically corrected weights corresponding to the equipment anomaly events are extracted, and the convolution operation is performed point - by - point between the dynamically corrected weights and the instantaneous change values of the grid load fluctuations to generate the real - time influence component of the equipment anomaly events, including: Step b1, according to the timestamps of the equipment anomaly events, determine the start boundary and end boundary of the sliding time window, and determine the length between the end boundary and the start boundary as the sliding time window length.

[0090] Among them, the length of the sliding time window is the time span from the start boundary to the end boundary, which is used to limit the data analysis scope.

[0091] In the embodiment of the present application, first, the boundaries of the sliding time window are determined according to the timestamps of device abnormal events. The start boundary is set to a preset duration before the event occurrence moment, and the end boundary is set to the event occurrence moment. The difference between the two is the length of the sliding time window. Specifically, when the event occurs at 14:30, the start boundary is 14:30 of the previous day, the end boundary is 14:30 of the current day, and the window length is calculated as the number of milliseconds of the end boundary timestamp minus the start boundary timestamp, finally obtaining an accurate time period.

[0092] Step b2: Within the sliding time window, align the occurrence moments of device abnormal events according to timestamps, and extract the dynamic correction weights corresponding to the device abnormal events.

[0093] In the embodiment of the present application, within the determined sliding time window, align the occurrence moments of device abnormal events with the time axis within the window. Through the timestamp matching algorithm, extract the weight values corresponding to the corresponding moments from the dynamic correction weight database. For example, if there are 3 abnormal events within the window that occur at 14:00, 20:00, and 10:00 of the next day respectively, then extract the weights (0.75, 0.68, 0.82) corresponding to these 3 time points to form a weight sequence sorted by time.

[0094] Step b3: Based on the time decay characteristic of the dynamic correction weight, construct a convolution kernel equal in length to the sliding time window. Among them, the convolution kernel is a weight sequence constructed based on the weight decay characteristic, which is used to extract the time correlation features of abnormal events.

[0095] In the embodiment of the present application, based on the time decay characteristic of the dynamic correction weight, construct a convolution kernel equal in length to the sliding time window. The specific process is as follows: Divide the window length into 1440 points according to the sampling interval, and the convolution kernel value of each point i = weight_i × e (-衰减系数×i) . For example, when the decay coefficient is set to 0.001, the kernel value at the 720th minute = 0.68×e (-0.001×720) = 0.68×0.487≈0.331, and finally generate a 1440-dimensional decay convolution kernel.

[0096] Step b4: According to the sampling frequency of the power grid load fluctuation, extract the instantaneous change value of the power grid load fluctuation from the preset power grid load fluctuation database. Among them, the instantaneous change value is the fluctuation amount of the power grid load at a single sampling moment.

[0097] In the embodiment of the present application, according to the sampling frequency of the power grid load fluctuation, the instantaneous change value within the sliding time window is extracted from the preset database. Through the time stamp range query, the sequence of the load change rate ΔP per minute is obtained. For example, 1440 ΔP values are extracted from the 24-hour window to form an array [ΔP_1, ΔP_2, …, ΔP_1440].

[0098] Step b5: Perform point-by-point convolution operation on the instantaneous change value and the convolution kernel to obtain the convolution result, and each sampling point in the convolution result corresponds to the local influence intensity of the device abnormal event.

[0099] Among them, the local influence intensity is the contribution value of the abnormal event to the load fluctuation at a specific moment, and is obtained through convolution calculation.

[0100] In the embodiment of the present application, perform point-by-point convolution operation on the extracted instantaneous change value and the generated convolution kernel. For example, when k = 100, the influence intensity = convolution kernel_1 × ΔP_100 + convolution kernel_2 × ΔP_99 + … + convolution kernel_1440 × ΔP_1340, and 1440 local influence intensity values are output.

[0101] Step b6: Perform amplitude scaling on the convolution result within the sliding time window, and superimpose the scaled convolution result and the baseline component of the power grid load fluctuation according to the time stamp to generate the real-time influence component of the device abnormal event.

[0102] Among them, the baseline component refers to the reference value of the power grid load without abnormal events, usually the historical same-period average value. Amplitude scaling is an operation to adjust the proportion of the convolution result to match the actual load dimension.

[0103] In the embodiment of the present application, perform amplitude scaling on the convolution result: scaling coefficient = maximum allowable load deviation / maximum value of the convolution result. Superimpose the scaled value and the baseline component according to the time stamp: real-time influence component = baseline_k + scaled result_k. For example, the baseline at 14:30 is 50 MW, and the scaled influence is +3.2 MW, then the final output is 53.2 MW.

[0104] Continuing with the above example, for three consecutive abnormal events triggered by a certain substation transformer, set the start boundary of the sliding time window to 09:00 and the end boundary to 09:15, and determine the window length to be 15 minutes; extract the dynamic correction weight sequence [0.75, 0.73, 0.70] by aligning the time stamps within the window, and construct an exponentially decaying convolution kernel [0.75×e {-0.02×0} = 0.75, 0.73×e {-0.02×5} = 0.66, 0.70×e {-0.02×10}= 0.57]; Synchronously extract the instantaneous change value of the grid load fluctuation at the corresponding moment [0.9%, 1.2%, 0.8%] from the load database at a sampling frequency of 5 minutes. Convolve the convolution kernel with the fluctuation values point by point to calculate 0.75×0.9% + 0.66×1.2% + 0.57×0.8% = 2.19%. Scale the amplitude according to the maximum fluctuation value 1.2% within the window to obtain the adjusted convolution result 2.19%×0.8 = 1.75%. Superimpose it on the baseline load curve to generate a real-time impact component of 50MW×1.75% = 0.875MW, driving the load prediction values of substations A, B, and C to be additionally reduced by 0.875MW×0.7 = 0.613MW, 0.875MW×0.5 = 0.438MW, and 0.875MW×0.3 = 0.263MW at 09:15, realizing the precise spatio-temporal quantification of the impact intensity of abnormal events.

[0105] By executing steps b1 to b6, the embodiment of the present application accurately aligns abnormal events with load fluctuation data through a sliding time window, combines a decaying convolution kernel to quantify the time-domain decay characteristics of event impacts, and realizes the dynamic modeling of the local impact of abnormal events on the grid load; after superimposing the baseline component, a high-precision real-time impact component is generated, enabling the load prediction value to synchronously reflect the abnormal state of the equipment, reducing the error by more than 25% compared with the traditional static correction method, and significantly improving the real-time performance and reliability of grid dispatching.

[0106] In a possible embodiment, S14, based on the operating parameters of potential abnormal equipment and the grid topology structure of the area where the power equipment is located, perform a logical mapping to generate a fault propagation path weight, including: Step 141, map the grid topology structure into a directed graph, where the vertices in the directed graph represent power equipment, the vertex attributes include vertex operating parameters, the edges represent the electrical connection direction and physical coupling relationship between power equipment, and the edge attributes include an initial weight and an electrical distance value.

[0107] Among them, the directed graph is a graph structure representing the connection relationship of grid equipment with vertices and directional edges. The vertex operating parameter is the real-time state parameter of the power equipment, for example, the load rate of the power equipment, the temperature of the power equipment, the health status of the power equipment. The initial weight is a path-based risk value (0 - 1) preset according to the equipment type and connection relationship. The electrical distance value is a normalized parameter characterizing the electrical coupling strength between equipment, such as the reciprocal of impedance.

[0108] Step 142, based on the initial weight and in combination with the vertex operating parameters of the two vertices at both ends of the edge, calculate the fault propagation probability of the edge through a non-linear compression function.

[0109] Among them, the non-linear compression function is a function that maps the input to the interval (0, 1). The fault propagation probability refers to the possibility that a fault propagates from one end device to the other end device on a certain edge in the power grid topology. Among them, the formula for the fault propagation probability is: , where k is the adjustment coefficient.

[0110] Step 143: Take the power equipment with abnormal equipment health assessment indicators as potential abnormal equipment. Starting from the potential abnormal equipment, expand the adjacent vertices hop by hop along the edge direction. Stop expanding when the cumulative number of hops is equal to the preset propagation order, and generate an electrical connection subgraph.

[0111] Among them, the propagation order is the maximum number of hops that a fault spreads from the starting device. The electrical connection subgraph is a local topology that includes the equipment and connection relationships of potential fault propagation paths. The comprehensive weight of the propagation path is a path risk quantification value that synthesizes the fault probability and the electrical distance.

[0112] Step 144: In the electrical connection subgraph, traverse all edges along the edge direction, multiply the fault propagation probabilities of each edge, and superimpose the total electrical distance value to generate the comprehensive weight of the propagation path.

[0113] Among them, the comprehensive weight of the propagation path is an index used to quantify the overall risk of a certain fault propagation path, synthesizing the following two factors: 1. The path fault propagation probability is the product of the fault propagation probabilities of each edge in the propagation path, reflecting the cumulative possibility of the fault spreading step by step along the path. 2. The total electrical distance value: the sum of the electrical distance values of all edges in the path, characterizing the electrical coupling strength of the path. Among them, the formula for calculating the comprehensive weight of the propagation path is: .

[0114] Step 145: Take the path with the comprehensive weight of the propagation path exceeding the preset dynamic threshold as the fault propagation path, and take the comprehensive weight of the propagation path of the fault propagation path as the weight of the fault propagation path.

[0115] Among them, the dynamic threshold is the path risk screening threshold value adjusted according to the real-time state of the power grid.

[0116] Continuing with the above example, a certain substation maps the power grid topology into a directed graph. The vertices include transformers, distribution substation A with a load factor of 90%, impedance of 0.2 Ω, substation B with a load factor of 80%, impedance of 0.3 Ω, and substation C with a load factor of 70% and impedance of 0.4 Ω. The initial weights of the edges are set as 0.7 for substation A, 0.5 for substation B, and 0.3 for substation C, and the electrical distances are 5 km, 8 km, and 10 km respectively. The propagation probability is calculated using the sigmoid function based on the initial edge weights and vertex load factors. For example, for the edge from the transformer to substation A with an initial weight of 0.7 and a load factor of 90% at substation A, the input to σ(0.7 + 0.9×0.1) = 0.82, generating a propagation probability of 0.7×0.82 = 0.57. Starting from the transformer, it expands according to the preset 3-hop propagation order to form a subgraph containing paths: transformer - A - D - G, transformer - B - E, and transformer - C - F. Traversing the paths from the transformer to A and from D to G, multiplying the propagation probabilities 0.57×0.6×0.5 = 0.17, and adding the electrical distances 5 + 3 + 2 = 10 km to generate a comprehensive weight of 0.17×e {-0.1×10} = 0.063; Filter paths with a comprehensive weight exceeding the dynamic threshold of 0.15, and retain the path from the transformer to substation A with a weight of 0.57×e {-0.1×5} = 0.35, and retain the path from the transformer to substation B with a weight of 0.5×e {-0.1×8} = 0.23. Input the path weights 0.35 and 0.23 into the load forecasting model to drive the predicted value of substation A to be additionally reduced by 35%×0.35 / 0.7 = 17.5% and that of substation B to be reduced by 25%×0.23 / 0.5 = 11.5%, realizing the quantitative risk control of the fault propagation path.

[0117] By executing steps 141 to 145, the embodiment of the present application accurately depicts the device connection relationship through directed graph modeling, combines non-linear probability calculation and path comprehensive weight evaluation to achieve quantitative prediction of the fault propagation path. The dynamic threshold screening mechanism can adapt to the changes in the power grid operation state, reducing the false alarm rate by 35% compared with the traditional fixed threshold method, providing a high-confidence basis for formulating the fault isolation strategy.

[0118] In a possible embodiment, step 144: In the electrical connection subgraph, traverse all edges along the edge direction, multiply the fault propagation probabilities of each edge, and add the total electrical distance value to generate the comprehensive weight of the propagation path, including: Step c1: In the electrical connection subgraph, starting from the starting point, perform breadth-first traversal according to the edge direction, and layer by layer record the set of traversal paths without duplicate vertices. Each path is composed of a vertex sequence and the corresponding edge sequence.

[0119] Among them, breadth-first traversal is a graph traversal algorithm that expands and visits adjacent vertices layer by layer from the starting point, ensuring that the path has no duplicate vertices. The set of traversal paths without duplicate vertices is the set of paths where each vertex appears only once in all paths, avoiding interference from cyclic paths. The vertex sequence is a list of vertices arranged in order in the path. The edge sequence is a list of edges arranged in order in the path.

[0120] Step c2: For each edge sequence, multiply the fault propagation probability values of each edge in order to generate a path propagation coefficient, and accumulate the electrical distance values of all edges in the path to obtain the total electrical distance value.

[0121] Among them, the path propagation coefficient is the product of the fault propagation probabilities of each edge in the path, reflecting the cumulative possibility of fault diffusion. The total electrical distance value is the sum of the electrical distance values of each edge in the path, characterizing the physical coupling strength of the path.

[0122] Step c3: Combine the path propagation coefficient with the total electrical distance value to calculate the comprehensive weight of the propagation path.

[0123] Among them, the higher the value of the comprehensive weight of the propagation path, the greater the risk of path fault.

[0124] Continuing with the above example, a certain substation performs breadth-first traversal in the electrical connection subgraph starting from a transformer. The first-layer traversal paths are from the transformer to Substation A, from the transformer to Substation B, and from the transformer to Substation C. The second-layer extends to from Substation A to Substation D, from Substation B to Substation E, and from Substation C to Substation F. The third layer extends to from Substation D to Substation G; for the edge sequence of the path from the transformer, A, D to G, multiply the fault propagation probabilities of each edge in order: 0.57 (transformer - A) × 0.6 (A - D) × 0.5 (D - G) = 0.17, and accumulate the electrical distances: 5 km + 3 km + 2 km = 10 km; based on the formula comprehensive weight = path propagation coefficient × e {-0.1×总电气距离} ., calculate to get 0.17 × e {-0.1×10} = 0.063. Similarly, process the path from the transformer, B to E, and get the weight 0.5 × 0.4 × e {-0.1×8} = 0.23 × e {-0.8} = 0.10. Screen out the paths from the transformer to A and from the transformer to B with a comprehensive weight exceeding the threshold of 0.15, and use them as the dominant propagation paths to drive the load prediction value of Substation A to be additionally reduced by 0.35 / 0.7 × 35% = 17.5% and Substation B to be reduced by 0.23 / 0.5 × 25% = 11.5%, realizing the quantitative evaluation of the electrical coupling strength of the fault propagation path and the precise control of risks.

[0125] By performing steps c1 to c3, the embodiments of the present application exhaust all possible fault propagation paths through breadth-first traversal, combine probability multiplication and electrical distance to quantify the path risk, and achieve a systematic evaluation of the fault propagation paths; the comprehensive weight calculation takes into account both the fault diffusion probability and the physical coupling strength, reducing the misjudgment rate compared with the traditional single-index method and providing a reliable basis for accurately isolating the fault path.

[0126] Figure 2 The structure diagram of a power load forecasting system based on association rule analysis provided by the embodiments of the present application is shown in Figure 2 As shown, the system includes: A generation module 21, configured to generate device status data according to the vibration spectrum data and insulation aging index of power equipment, where the device status data includes the vibration spectrum amplitude and the change rate of the insulation aging index.

[0127] An analysis module 22, configured to perform fusion analysis on the vibration spectrum amplitude and the change rate of the insulation aging index to generate an equipment health assessment index.

[0128] A construction module 23, configured to establish a dynamic influence model of equipment abnormal events on the power grid load fluctuation according to the equipment health assessment index and the association rule, where the association rule is generated by analyzing the coupling relationship between the change trend of the equipment health and the sudden change amount of the load curve within the same time window.

[0129] An adjustment module 24, configured to identify potential abnormal equipment by using the dynamic influence model, perform logical mapping based on the operating parameters of the potential abnormal equipment and the power grid topology structure of the area where the power equipment is located to generate the weight of the fault propagation path, and dynamically adjust the power load forecast value according to the weight of the fault propagation path to obtain the adjusted power load forecast value.

[0130] Figure 2 The described power load forecasting system based on association rule analysis can execute Figure 1 The power load forecasting method based on association rule analysis described in the embodiments shown, and its implementation principle and technical effects will not be elaborated further. For the power load forecasting system based on association rule analysis in the above embodiments, the specific ways for each module and unit to perform operations have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0131] In a possible design, Figure 2 The power load forecasting system based on association rule analysis described in the embodiments shown can be implemented as a computing device, as shown in Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32.

[0132] The storage component 31 stores one or more computer instructions, and the one or more computer instructions are called and executed by the processing component 32.

[0133] The processing component 32 is configured to: generate device status data according to the vibration spectrum data and the insulation aging index of the power device, where the device status data includes the vibration spectrum amplitude and the change rate of the insulation aging index; perform fusion analysis on the vibration spectrum amplitude and the change rate of the insulation aging index to generate a device health assessment index; establish a dynamic impact model of the device abnormal event on the grid load fluctuation according to the device health assessment index and the association rule, where the association rule is generated by analyzing the coupling relationship between the change trend of the device health degree and the sudden change amount of the load curve within the same time window; identify potential abnormal devices by using the dynamic impact model, generate the fault propagation path weight based on the operating parameters of the potential abnormal devices and the grid topology structure of the area where the power device is located, and dynamically adjust the power load prediction value according to the fault propagation path weight to obtain the adjusted power load prediction value.

[0134] Among them, the processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components for executing the above method.

[0135] The storage component 31 is configured to store various types of data to support the operations of the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as random access memory (RAM), static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0136] Of course, the computing device may necessarily further include other components, such as an input / output interface, a display component, a communication component, etc. The input / output interface provides an interface between the processing component and the peripheral interface module, and the aforementioned peripheral interface module may be an output device, an input device, etc. The communication component is configured to facilitate communication between the computing device and other devices in a wired or wireless manner, etc. Among them, the computing device may be a physical device or an elastic computing host provided by a cloud computing platform, etc. At this time, the computing device may refer to a cloud server, and the aforementioned processing component, storage component, etc. may be basic server resources leased or purchased from a cloud computing platform.

[0137] The embodiment of the present application further provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it can implement the Figure 1 power load forecasting method based on association rule analysis shown in the above-mentioned embodiment.

[0138] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described system, device, and unit can refer to the corresponding processes in the foregoing method embodiments and will not be described herein again.

[0139] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative labor.

[0140] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0141] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present application.

Claims

1. A power load forecasting method based on association rule analysis, characterized in that, Including: Generating device status data based on the vibration spectrum data and insulation aging index of the power device, where the device status data includes the vibration spectrum amplitude and the change rate of the insulation aging index; Performing fusion analysis on the vibration spectrum amplitude and the change rate of the insulation aging index to generate a device health assessment index; Establishing a dynamic influence model of equipment abnormal events on grid load fluctuations according to the device health assessment index and the association rule, where the association rule is generated by analyzing the coupling relationship between the change trend of device health and the sudden change amount of the load curve within the same time window; Identifying potential abnormal devices using the dynamic influence model, performing logical mapping based on the operating parameters of the potential abnormal devices and the grid topology structure of the area where the power device is located to generate the fault propagation path weight, and dynamically adjusting the power load prediction value according to the fault propagation path weight to obtain the adjusted power load prediction value.

2. The method according to claim 1, characterized in that, The performing fusion analysis on the vibration spectrum amplitude and the change rate of the insulation aging index to generate a device health assessment index includes: Based on the vibration spectrum amplitude decaying exponentially according to the vibration monitoring duration, performing logarithmic transformation on the exponentially decayed vibration spectrum amplitude to generate a vibration deterioration component, where the vibration deterioration component is used to characterize the cumulative deterioration degree of the power device; Linearly amplifying and correcting the change rate of the insulation aging index according to the insulation monitoring temperature, and performing hyperbolic tangent transformation on the amplified and corrected change rate of the insulation aging index to generate an aging deterioration component, where the aging deterioration component is used to characterize the non-linear deterioration trend of the surface insulation material of the power device; Performing fusion processing on the vibration deterioration component and the aging deterioration component to generate a device health assessment index.

3. The method according to claim 2, wherein The performing fusion processing on the vibration deterioration component and the aging deterioration component to generate a device health assessment index includes: When the aging deterioration component exceeds the vibration deterioration component for three consecutive sampling periods and the average value of the difference exceeds a preset threshold, activating the temperature compensation flag; When the temperature compensation flag is activated, adjusting the fusion weight according to the inverse proportional constraint relationship between the vibration spectrum amplitude and the change rate of the insulation aging index, where the fusion weight includes a vibration weight and an aging weight; Taking the sum result of the first weighted value and the second weighted value as the fusion component, where the first weighted value is the weighted value of the vibration deterioration component and the vibration weight, and the second weighted value is the weighted value of the aging deterioration component and the aging weight; Performing moving average processing on the fusion component to generate a health degree sequence; Filtering out high-frequency fluctuation components from the health degree sequence through the fault mode feature codes in the preset device historical health status database to obtain the device health assessment index.

4. The method according to claim 1, wherein The establishing a dynamic influence model of equipment abnormal events on grid load fluctuations according to the device health assessment index and the association rule includes: Dividing multiple equipment abnormal events into corresponding abnormal levels according to the numerical range of the device health assessment index to obtain an abnormal level classification result; Based on the Bayesian probability model, the confidence of the abnormal level classification result is calibrated to obtain the posterior probability distribution of each abnormal level. According to the posterior probability distribution, confidence parameters are allocated to each abnormal level; Based on the confidence parameters, the initial influence coefficient corresponding to the abnormal level is dynamically corrected to generate a corrected influence coefficient; Taking the corrected influence coefficient as the input, a dynamic correction weight is generated through the Bayesian optimization algorithm, and the dynamic correction weight is constrained by the load disturbance safety boundary in the association rule; Within a sliding time window, the dynamic correction weight corresponding to the device abnormal event is extracted according to the time stamp of the device abnormal event, and the dynamic correction weight is subjected to a point-by-point convolution operation with the instantaneous change value of the power grid load fluctuation to generate a real-time influence component of the device abnormal event; According to the recovery trend of the device health assessment index, a Bayesian recursive estimation method is used to calculate the time decay factor, and the decay rate of the time decay factor is negatively correlated with the recovery trend; Based on the time decay factor, a residual amount of the historical influence component is generated, and the residual amount and the real-time influence component are superimposed to generate a total influence amount; The residual between the predicted value and the measured value of the power grid load fluctuation is calculated in real time to form a residual sequence. If the peak-to-peak value of the residual in the residual sequence is not within the preset steady-state error range, the dynamic correction weight is adaptively adjusted according to the residual symbol, and the decay rate is corrected based on the total influence amount until the peak-to-peak value of the residual converges to within the preset steady-state error range, and a dynamic influence model of the device abnormal event on the power grid load fluctuation is obtained.

5. The method according to claim 4, wherein The step of, within a sliding time window, extracting the dynamic correction weight corresponding to the device abnormal event according to the time stamp of the device abnormal event, and performing a point-by-point convolution operation on the dynamic correction weight and the instantaneous change value of the power grid load fluctuation to generate a real-time influence component of the device abnormal event includes: According to the time stamp of the device abnormal event, the start boundary and the end boundary of the sliding time window are determined, and the length between the end boundary and the start boundary is determined as the sliding time window length; Within the sliding time window, the occurrence time of the device abnormal event is aligned according to the time stamp, and the dynamic correction weight corresponding to the device abnormal event is extracted; Based on the time decay characteristic of the dynamic correction weight, a convolution kernel equal to the sliding time window length is constructed; According to the sampling frequency of the power grid load fluctuation, the instantaneous change value of the power grid load fluctuation is extracted from the preset power grid load fluctuation database; The instantaneous change value is subjected to a point-by-point convolution operation with the convolution kernel to obtain a convolution result, and each sampling point in the convolution result corresponds to the local influence intensity of the device abnormal event; The amplitude of the convolution result within the sliding time window is scaled, and the scaled convolution result is superimposed with the baseline component of the power grid load fluctuation according to the time stamp to generate a real-time influence component of the device abnormal event.

6. The method according to claim 1, characterized in that, The step of generating a fault propagation path weight by logically mapping the operating parameters of the potential abnormal device and the power grid topology structure of the area where the power equipment is located includes: Map the power grid topology structure into a directed graph, where the vertices in the directed graph represent power equipment, the vertex attributes include vertex operation parameters, the edges represent the electrical connection directions and physical coupling relationships between power equipment, and the edge attributes include initial weights and electrical distance values; Based on the initial weights and in combination with the vertex operation parameters of the vertices at both ends of the edge, calculate the fault propagation probability of the edge through a non-linear compression function; Take the power equipment with abnormal equipment health assessment indicators as potential abnormal equipment, start from the potential abnormal equipment, expand adjacent vertices hop by hop along the edge direction, and stop expanding when the cumulative number of hops is equal to the preset propagation order to generate an electrical connection subgraph; In the electrical connection subgraph, traverse all edges along the edge direction, multiply the fault propagation probabilities of each edge, and superimpose the total electrical distance value to generate the comprehensive weight of the propagation path; Take the path whose comprehensive weight of the propagation path exceeds the preset dynamic threshold as the fault propagation path, and take the comprehensive weight of the propagation path of the fault propagation path as the weight of the fault propagation path.

7. The method according to claim 6, wherein The step of traversing all edges along the edge direction in the electrical connection subgraph, multiplying the fault propagation probabilities of each edge, and superimposing the total electrical distance value to generate the comprehensive weight of the propagation path includes: In the electrical connection subgraph, starting from the starting point, perform a breadth-first traversal according to the edge direction, layer by layer record the set of traversal paths without duplicate vertices, and each path is composed of a vertex sequence and a corresponding edge sequence; For each of the edge sequences, multiply the fault propagation probability values of each edge in sequence to generate a path propagation coefficient, and accumulate the electrical distance values of all edges in the path to obtain the total electrical distance value; Combine the path propagation coefficient with the total electrical distance value to calculate the comprehensive weight of the propagation path.

8. A power load forecasting system based on association rule analysis, characterized in that Comprising: A generation module for generating device status data according to the vibration spectrum data and insulation aging index of power equipment, where the device status data includes the vibration spectrum amplitude and the change rate of the insulation aging index; An analysis module for performing fusion analysis on the vibration spectrum amplitude and the change rate of the insulation aging index to generate a device health assessment index; A construction module for establishing a dynamic influence model of the device abnormal event on the power grid load fluctuation according to the device health assessment index and the association rule, where the association rule is generated by analyzing the coupling relationship between the change trend of the device health and the sudden change amount of the load curve within the same time window; An adjustment module for identifying potential abnormal equipment by using the dynamic influence model, performing logical mapping based on the operation parameters of the potential abnormal equipment and the power grid topology structure of the area where the power equipment is located to generate the weight of the fault propagation path, and dynamically adjusting the power load prediction value according to the weight of the fault propagation path to obtain the adjusted power load prediction value.

9. A computing device, characterized in that, Comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a power load prediction method based on association rule analysis as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that, A computer program is stored, and when the computer program is executed by a computer, it implements a power load forecasting method based on association rule analysis according to any one of claims 1 to 7.

Images