Power equipment operation safety state prediction method and system based on monitoring data

Abstract

The application relates to the technical field of data processing, and discloses a power equipment operation safety state prediction method and system based on monitoring data, which comprises the following steps: acquiring two continuous monitoring periods before the current moment as a first monitoring period and a second monitoring period, acquiring multiple state data, and acquiring a to-be-predicted time period; acquiring a first data sequence, a first fluctuation trend index, and a first correlation coefficient; acquiring a second data sequence, a second fluctuation trend index, and a second correlation coefficient between any two kinds of state data; screening out corresponding two kinds of state data with a difference between the first correlation coefficient and the second correlation coefficient less than a preset change threshold value as to-be-predicted data, and acquiring data change of the to-be-predicted data in the prediction time period according to the second data sequence of the to-be-predicted data. The application can improve prediction accuracy, enhance the explainability of prediction results, and improve adaptability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, in particular to a power equipment operation safety state prediction method and system based on monitoring data. BACKGROUND

[0002] In the field of power equipment operation and maintenance, the power equipment operation safety state data is often predicted to facilitate early understanding of the operation state of the power equipment.

[0003] At present, the power equipment operation safety state data is mainly predicted by using the state data of the power equipment, wherein a part of the technology only relies on the time sequence change of a single state data itself for prediction, and ignores the physical correlation and collaborative evolution law between multi-source state data. This kind of fragmented prediction mode violates the actual operation mechanism of the equipment. For example, the sudden rise of the environmental temperature causes the short-term rise of the equipment shell temperature, which is misjudged as internal failure (actual current, vibration data has no abnormality). Another part of the existing technology does not establish a correlation stability quantification mechanism between multi-source state data, and cannot distinguish between false fluctuations caused by random disturbances and real failure precursors. At the same time, the random working condition fluctuations of the state data detected in the monitoring period (such as load mutation, instantaneous electromagnetic interference) will cause false correlation between the data. For example, the temperature and vibration data may present high correlation due to random disturbance in a certain monitoring period, but the correlation sharply decays in the next period. If the temperature and vibration are predicted recently, the predicted values of the two deviate from the actual trajectory.

[0004] In summary, the prediction accuracy of the power equipment operation safety state data is poor at present, and there are many defects. SUMMARY

[0005] In view of the defects in the prior art, the present application provides a power equipment operation safety state prediction method and system based on monitoring data.

[0006] According to an aspect of the embodiment of the present application, a power equipment operation safety state prediction method based on monitoring data is provided, comprising: S1: obtaining two continuous periods before the current time as a first monitoring period and a second monitoring period, obtaining a plurality of state data in the power equipment operation in the monitoring period, and obtaining a to-be-predicted time period; S2: obtaining a first data sequence of the various state data in the first monitoring period, obtaining a first fluctuation trend index corresponding to the various state data according to the first data sequence of the various state data, and obtaining a first correlation coefficient between any two state data according to the first fluctuation trend index of the various state data; S3: obtaining a second data sequence of the various state data in the second monitoring period, obtaining a second fluctuation trend index corresponding to the various state data according to the second data sequence of the various state data, and obtaining a second correlation coefficient between any two state data according to the second fluctuation trend index of the various state data; S4: screening out corresponding two state data with a difference value of the first correlation coefficient and the second correlation coefficient less than a preset change threshold as to-be-predicted data, and obtaining a data change of the to-be-predicted data in the prediction time period according to the second data sequence of the to-be-predicted data.

[0007] In the scheme, a cross-period physical correlation stability mechanism is established, the various state data of the two continuous monitoring periods are obtained, the fluctuation trend index of the corresponding state parameters in each period and the correlation coefficient of any two parameters are obtained, the combination of any two state parameters is screened, the combination with a cross-period correlation coefficient difference less than the threshold is screened through the preset threshold, the state data is set as a prediction parameter through the correlation stability of at least one other data, the prediction confidence of the prediction data is improved, the real fault barrier before the fault and the short-time noise fluctuation can be effectively distinguished through the screening of the correlation coefficient, the false correlation caused by the random disturbance such as load mutation is effectively avoided, and the prediction accuracy is improved. Finally, based on the linear extrapolation prediction of the recent behavior rule, for the stable data, the prediction value of the future time point is generated directly according to the change direction and rate of the sequence at the end of the second monitoring period and the current value.

[0008] Optionally, the first fluctuation trend index corresponding to the various state data is obtained according to the first data sequence of the various state data, comprising: the difference between the adjacent two monitoring values in the first data sequence of the state data is obtained as a first relative change amount, the theoretical maximum value and the theoretical minimum value corresponding to the state data are obtained, and the first absolute change amount corresponding to the first relative change amount is obtained according to the first relative change amount, the theoretical maximum value and the theoretical minimum value; the absolute change amount corresponding to each first absolute change amount in the first data sequence of the state data is accumulated as the first fluctuation trend index. In the scheme, the first absolute change amount is the ratio of the difference between the first relative change amount and the theoretical maximum value and the theoretical minimum value.

[0009] Optionally, the first correlation coefficient between any two state data according to the first fluctuation trend index of various state data is represented as: ; wherein, is the first correlation coefficient between the i-th and j-th state data, is the first fluctuation trend index of the i-th state data, is the first fluctuation trend index of the j-th state data, is a scaling coefficient. When the first fluctuation trend indexes of two state data are highly consistent in value and direction, such as A temperature rises and B current increases simultaneously, it means that there is a strong coupling relationship in physical mechanism between the two, at this time the correlation coefficient tends to the maximum value, otherwise the correlation coefficient decreases significantly, even the value is zero.

[0010] Optionally, the second fluctuation trend index of various state data according to the second data sequence of various state data includes: according to the second data sequence of state data, the difference between adjacent two monitoring values in the second data sequence is obtained as the second relative change, and the theoretical maximum value and the theoretical minimum value corresponding to the state data are obtained, and the second absolute change corresponding to the second relative change is obtained according to the second relative change, the theoretical maximum value and the theoretical minimum value; the absolute change corresponding to each second absolute change in the second data sequence of state data is accumulated and taken as the second fluctuation trend index. In the scheme, the second absolute change is the ratio of the difference between the second relative change and the theoretical maximum value and the theoretical minimum value.

[0011] Optionally, the second correlation coefficient between any two state data according to the second fluctuation trend index of various state data is represented as: ; wherein, is the second correlation coefficient between the i-th and j-th state data, is the second fluctuation trend index of the i-th state data, is the second fluctuation trend index of the j-th state data, is a scaling coefficient. When the second fluctuation trend indexes of two state data are highly consistent in value and direction, such as A temperature rises and B current increases simultaneously, it means that there is a strong coupling relationship in physical mechanism between the two, at this time the correlation coefficient tends to the maximum value, otherwise the correlation coefficient decreases significantly, even the value is zero.

[0012] Optionally, the data change of the to-be-predicted data in the prediction time period is obtained according to the second data sequence of the to-be-predicted data, including: obtaining the change rate of the to-be-predicted data in the prediction time period according to the second data sequence of the to-be-predicted data; generating the predicted value of the to-be-predicted data at multiple time points in the prediction time period according to the monitoring value of the to-be-predicted data at the current time and the change rate in the prediction time period.

[0013] According to another aspect of the embodiments of the present application, there is provided a power equipment operation safety state prediction system based on monitoring data, the system comprising: an acquisition module configured to acquire two continuous monitoring periods of the power equipment before a current time as a first monitoring period and a second monitoring period, acquire a plurality of state data in the operation of the power equipment, and acquire a to-be-predicted time period; a first data processing module configured to acquire a first data sequence of the various state data in the first monitoring period, acquire a first fluctuation trend index corresponding to the various state data according to the first data sequence of the various state data, and acquire a first correlation coefficient between any two state data according to the first fluctuation trend index of the various state data; a second data processing module configured to acquire a second data sequence of the various state data in the second monitoring period, acquire a second fluctuation trend index corresponding to the various state data according to the second data sequence of the various state data, and acquire a second correlation coefficient between any two state data according to the second fluctuation trend index of the various state data; and a data prediction module configured to screen out corresponding two state data with a difference between the first correlation coefficient and the second correlation coefficient less than a preset change threshold as to-be-predicted data, and acquire a data change of the to-be-predicted data in the to-be-predicted time period according to the second data sequence of the to-be-predicted data.

[0014] Optionally, the first data processing module is further configured to: acquire, according to the first data sequence of the state data, a difference between adjacent two monitoring values in the first data sequence as a first relative change, acquire a theoretical maximum value and a theoretical minimum value corresponding to the state data, and acquire a first absolute change corresponding to the first relative change according to the first relative change, the theoretical maximum value, and the theoretical minimum value; and accumulate absolute changes corresponding to each first absolute change in the first data sequence of the state data as the first fluctuation trend index.

[0015] According to another aspect of the embodiments of the present application, there is provided a power equipment operation safety state prediction device based on monitoring data, characterized by comprising: a memory having a computer program stored thereon; and a processor configured to execute the computer program in the memory to implement the power equipment operation safety state prediction method based on monitoring data.

[0016] According to still another aspect of the embodiments of the present application, there is provided a non-transitory computer readable storage medium having a computer program stored thereon, the program being executed by a processor to implement the power equipment operation safety state prediction method based on monitoring data.

[0017] The present application has the following beneficial effects:

[0018] 1. To overcome the defect of ignoring the collaborative evolution of multi-source data in the prior art, a cross-cycle physical correlation stability mechanism is established, and a stable correlation data combination is screened by comparing the correlation coefficients of two monitoring cycles. The fluctuation trend indicators of the first monitoring cycle (medium-term history) and the second monitoring cycle (recent dynamics) are used to quantify the correlation strength (the stronger the trend directionality and the closer the amplitude, the higher the correlation coefficient) of any two state data, and the combination with a cross-cycle correlation coefficient difference less than the threshold value is screened by a pre-set threshold value. This mechanism ensures that the data included in the prediction has long-term stability supported by physical mechanism, effectively avoids false correlation caused by random disturbances such as load mutation, and fundamentally distinguishes real fault precursors from noise fluctuations, making the prediction value more accurate.

[0019] 2. Design layered data prediction admission rules to avoid misjudgment of single data prediction. If the correlation of a certain state data with at least one other data is stable, it is listed as a prediction object. When a certain state data meets the stability with multiple data at the same time, its prediction confidence is further improved, and the state data that is completely disconnected is completely excluded. This rule retains the simplicity of independent prediction of single data, strengthens the reliability of data through cross-validation of the correlation network, and avoids misjudgment of internal faults caused by sudden temperature rise and other disturbances.

[0020] Based on the linear extrapolation prediction of recent behavior rules, for stable data, the future prediction value is generated directly according to the change direction and rate of the sequence at the end of the second monitoring cycle, combined with the current value, which can improve the prediction accuracy of the result, enhance the interpretability of the prediction result, and improve the adaptability. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the specific embodiments of the present application or the technical solutions in the prior art, the following will briefly introduce the drawings needed to be used in the specific embodiments or prior art description. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, each element or part is not necessarily drawn according to the actual proportion.

[0022] Figure 1 Part of the flowchart of the embodiment of the power equipment operation safety state prediction method based on monitoring data provided by the present application;

[0023] Figure 2 Another part of the flowchart of the embodiment of the power equipment operation safety state prediction method based on monitoring data provided by the present application;

[0024] Figure 3 The step diagram of the embodiment of the power equipment operation safety state prediction method based on monitoring data provided by the present application;

[0025] Figure 4A structural schematic diagram of an embodiment of a power equipment operation safety state prediction device based on monitoring data provided by the present application is shown. DETAILED DESCRIPTION

[0026] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings can be arranged and designed in various different configurations.

[0027] Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the claimed present application, but only represents selected embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative labor are within the scope of protection of the present application.

[0028] It should be noted that: similar reference numerals and letters represent similar items in the following drawings, therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. In addition, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0029] Figure 1 、 Figure 2 and Figure 3 A flowchart and step diagram of an embodiment of a power equipment operation safety state prediction method based on monitoring data of the present application are shown, which is executed by a power equipment operation safety state prediction system based on monitoring data. As shown in Figures 1 to 3 the method comprises the following steps:

[0030] S1: obtaining two continuous monitoring periods for the power equipment before the current time as a first monitoring period and a second monitoring period, obtaining various state data in the operation of the power equipment, and obtaining a to-be-predicted time period;

[0031] S2: obtaining a first data sequence of the various state data in the first monitoring period, obtaining a first fluctuation trend index corresponding to the various state data according to the first data sequence of the various state data, and obtaining a first correlation coefficient between any two state data according to the first fluctuation trend index of the various state data.

[0032] This step specifically comprises:

[0033] S21: According to the first data sequence of the state data, the difference between the adjacent two monitoring values in the first data sequence is sequentially obtained as the first relative change, and the theoretical maximum value and the theoretical minimum value corresponding to the state data are obtained, and the first absolute change corresponding to the first relative change is obtained according to the first relative change, the theoretical maximum value and the theoretical minimum value.

[0034] In this step, a normalized index reflecting the micro fluctuation of data is constructed. For the time sequence of each state data in the first monitoring period, such as the temperature data of the switch cabinet, the difference between the adjacent two monitoring points is calculated in sequence, that is, the monitoring value at the next time minus the value at the previous time, and the sign needs to be kept to reflect the change direction, to generate a signed relative change sequence, for example, the change amount of the temperature sequence of two points: +0.3℃, -0.2℃. Then, combined with the theoretical extreme value range of the state in the current operating environment, such as the temperature safety threshold of the switch cabinet shell 10~85℃, the relative change is normalized, that is, (first relative change) / (theoretical maximum value-theoretical minimum value) = first absolute change; the difference between the theoretical maximum value and the minimum value is used as a reference to convert each relative change into an absolute change without unit, such as the +0.3℃ change, the normalization formula is 0.3 / (85-10)=0.004. This operation not only eliminates the incomparability between different dimensional data (such as temperature ℃ and current A), but also retains the original fluctuation direction (positive value rising / negative value falling) and relative amplitude, forming a standardized absolute change sequence.

[0035] S22: The absolute change corresponding to each first absolute change in the first data sequence of the state data is accumulated and taken as the first fluctuation trend index.

[0036] In this step, based on the absolute change sequence generated in S21, a comprehensive trend index is constructed by accumulating the absolute change values (including positive and negative signs) at each time. For example, the absolute change sequence of the switch cabinet temperature in a certain monitoring period is [+0.004, -0.003, +0.005], and the cumulative result is +0.006, which is the first fluctuation trend index. This index quantifies the net change direction and intensity of the state data in the entire monitoring period: if the cumulative value is a significant positive number (such as +0.15), it indicates that the data has an overall upward trend and a large fluctuation amplitude; if the cumulative value is a significant negative number (such as -0.12), it indicates an overall downward trend; a value close to zero (such as +0.002) indicates weak fluctuation and no significant directionality in the period. Through this index, the multi-dimensional state characteristics of the equipment are condensed into a trend scalar that can be compared horizontally, providing a unified input for subsequent cross-data correlation analysis.

[0037] Furthermore, let's take transformer winding temperature data as an example. In S21, the relative changes in the time series within the monitoring period are obtained by differentiating adjacent points [-1.5℃, +2.2℃, -0.8℃]; the theoretical temperature range is 0~120℃, and the normalized reference is 120-0=120℃; the absolute changes are calculated as: [-1.5 / 120≈-0.0125, +2.2 / 120≈+0.0183, -0.8 / 120≈-0.0067]. In S22, the cumulative absolute changes are: -0.0125+0.0183-0.0067=-0.0009, and the first fluctuation trend index is -0.0009, reflecting that the winding temperature fluctuates very little and has no significant upward or downward trend within this period, which is consistent with the steady-state operation characteristics of the equipment.

[0038] S23: Obtain the first correlation coefficient between any two sets of state data based on the first fluctuation trend index of various state data.

[0039] The first correlation coefficient in this step is represented as:

[0040] ;in,

[0041] Let be the first correlation coefficient between the i-th and j-th state data. This is the first fluctuation trend indicator for the i-th state data. This is the first fluctuation trend indicator for the j-th state data. This is the scaling factor.

[0042] In this embodiment, it should be noted that throughout the entire expression, As a directional term, when the fluctuation trend index of two state data (such as temperature T and current C) is... , Output 1 when the signs are the same (both rising / falling), and output -1 when the signs are opposite (one rising and one falling); output 0 if either indicator is zero. The correlation must conform to the directional coordination of the physical mechanism; for example, if the ambient temperature rises suddenly, and the equipment casing temperature rises (…). However, the internal current is stable. ),but =0 sets the correlation coefficient to zero, directly avoiding the problem of misjudging internal operational failures by a single data point caused by non-operational anomalies or environmental factors. At the same time, it ensures that a correlation is established only when the two data points evolve in the same direction within the monitoring period (such as when the fan current and temperature increase synchronously when the cooling system is working), strictly following the physical laws of equipment operation.

[0043] Furthermore, To obtain the associated items, calculate the absolute difference between the fluctuation trend indicators of the two states. And multiplied by Nonlinear decay by exponential function. If the two data trends are significantly different in intensity (e.g. temperature sharply increases I=0.3, current slightly increases I=0.05), Time e^{-1.25}≈0.28, relative low correlation coefficient, reflecting the lack of strong matching physical coupling. When random conditions such as transient electromagnetic interference cause the two data trends to accidentally approach each other (e.g. temperature increases I=0.3, current slightly increases I=0.05), ), although e^{0}=1 will output a high correlation, subsequent S4 cross-period stability verification (e.g. T-V combined correlation coefficient drops from 0.9 to 0.2) can filter out such false correlations. At the same time, the sensitive characteristics of the exponential function (small changes cause dramatic changes in correlation coefficient) ensure that only when the fluctuation trend amplitude is highly matched can a high correlation value be output (e.g. when the first fluctuation trend index of temperature is 0.15 and the first fluctuation trend index of current is 0.18, , e^{-0.15}≈0.86), capturing the physical proportional relationship between current load and temperature rise.

[0044] Further, strong screening mechanism, the product of the sign function and the exponential term is truncated to the non-negative interval. If the result is negative, it is forced to output 0. Pseudo-correlation is completely excluded when the trend deviates. For example, when the device is in abnormal working condition (e.g. mechanical jam causes temperature rise but vibration decreases), even extremely small, the overall result is negative, it is set to zero. To some extent, it solves the pain point of false correlation caused by random disturbance. This mechanism directly filters abnormal combinations through physical law inversion verification (temperature rise is often accompanied by increased vibration in real faults), avoiding fault misjudgment.

[0045] Further, the dynamic adaptability of the scaling coefficient , as an adjustable parameter controls the decay rate, the greater the thermal inertia of the device (e.g. transformer), the smaller the , and the more sensitive the device (e.g. lightning arrester), the larger the . Overcomes the generalization defects of traditional models in multi-element device scenarios. For example, the trend amplitude of transformer oil temperature and current naturally approaches due to large heat capacity, small value ( ) can relax the tolerance; while pressure and current in GIS devices need to be strictly matched in proportion, large value ( 0) enhances amplitude sensitivity. Through adaptive device characteristics, it ensures accurate expression of the physical mechanisms of different power devices, solving the problem of one-size-fits-all correlation models in complex scenarios.

[0046] S3: Obtain a second data sequence of the various state data in a second monitoring period, and obtain a second fluctuation trend index corresponding to the various state data according to the second data sequence of the various state data, and obtain a second correlation coefficient between any two kinds of state data according to the second fluctuation trend index of the various state data.

[0047] This step specifically includes:

[0048] S31: Obtain, according to the second data sequence of the state data, a difference between adjacent two monitoring values in the second data sequence as a second relative change, and obtain a theoretical maximum value and a theoretical minimum value corresponding to the state data, and obtain a second absolute change corresponding to the second relative change according to the second relative change, the theoretical maximum value and the theoretical minimum value.

[0049] In this step, the microscopic fluctuation of the state data in the second monitoring period (the latest period) is quantified. For the time sequence of each kind of state data, such as the switch cabinet current data, the signed difference between adjacent monitoring points is calculated, that is, the value at the next time point is subtracted from the value at the previous time point, and the sign is kept to identify the change direction, to generate a second relative change sequence, that is, (second relative change) / (theoretical maximum value-theoretical minimum value)=second absolute change; for example, the current sequence change is +0.5A and -0.3A. Then, the theoretical extreme value range of the state in the current operating environment is combined, such as the maximum allowable current of the switch cabinet 100A, and the difference between the theoretical maximum value and the minimum value (such as 100A-0A=100A) is used to normalize the relative change (such as +0.5A normalized to +0.005). This operation eliminates the dimensional difference while strictly preserving the original fluctuation direction (positive value for upward trend, negative value for downward trend) and relative amplitude, forming a standardized second absolute change sequence that can be compared across data types.

[0050] S32: Accumulate the absolute change corresponding to each second absolute change in the second data sequence of the state data as a second fluctuation trend index.

[0051] In this step, based on the absolute change sequence generated in S31, the signed absolute change at each time point is accumulated (such as the sequence [+0.004, -0.002, +0.006] accumulated to +0.008) to generate a second fluctuation trend index. This index comprehensively reflects the overall change intensity and directionality of the data in the latest period: a significant positive value (such as +0.25) indicates a dramatic increase in the data; a significant negative value (such as -0.18) indicates a sustained decline; a value close to zero (such as +0.003) reflects stable and no fluctuation. This index provides a standardized input of recent dynamics for subsequent correlation analysis.

[0052] For example, taking the transformer current data in the second monitoring period as an example. In S31, the relative change amount of the time sequence difference is [-2.1A, +1.8A, -0.5A]; the theoretical current range is 0-500A, and the normalized reference is 500A; the absolute change amount is calculated: [-2.1 / 500=-0.0042, +1.8 / 500=+0.0036, -0.5 / 500=-0.001]. In S32, the cumulative absolute change amount is -0.0042+0.0036-0.001=-0.0016, the second fluctuation trend index is -0.0016, indicating that the current overall weakly decreases, which is consistent with the recent light load operation condition of the equipment.

[0053] S33: obtaining a second correlation coefficient between any two kinds of state data according to the second fluctuation trend indexes of the various state data.

[0054] The second correlation coefficient in this step is represented as:

[0055] ; wherein,

[0056] is the second correlation coefficient between the ith and jth kinds of state data, is the second fluctuation trend index of the ith kind of state data, is the second fluctuation trend index of the jth kind of state data, is a scaling coefficient.

[0057] In this embodiment, it is necessary to note that the calculation logic is the same as that of S2. Specifically, first, is the second fluctuation trend index of the ith kind of state data, achieves a dynamic verification mechanism, which forcibly requires the fluctuation trend of two kinds of state data in the current period (the second monitoring period) to evolve in the same direction, and ensures that only the physical coordinated change (such as the synchronous increase of current and temperature when the cooling system is started) that occurs in real time is responded. Further, is also a correlation acquisition item, and the fluctuation trend difference is exponentially attenuated. Wherein, The value needs to be obtained through a multi-level adaptive mechanism driven by the characteristics of the equipment, and the core logic thereof is directly linked with the physical characteristics of the power equipment and the data noise level. In the specific acquisition process of the value, first, based on the baseline setting of the physical response characteristics of the equipment, for the equipment with large thermal inertia (such as the transformer oil temperature), because the state changes slowly, the fluctuation trend index difference tolerance is high, and a small value (such as between [2, 5]) needs to be set to prevent excessive sensitivity from causing correlation misjudgment; for the responsive sensitive equipment (such as the GIS mechanism current), because the data is easily affected by instantaneous disturbance, a large value (such as Between [8, 15]), the reinforcement amplitude matching requirement. Further, The value can be corrected by the device nameplate parameters (such as rated current, thermal capacity), first calculate the theoretical fluctuation correlation strength S: S={theoretical change threshold of state quantity A} / {theoretical change threshold of state quantity B}, if S>1 (such as temperature change threshold is much larger than vibration), then Proportionally reduced, such as S>1, Reduced by 10%.

[0058] Further, the second correlation coefficient output by the expression in the embodiment The first correlation coefficient in the foregoing embodiment Cross-cycle stability comparison is performed in S4 to form the final prediction scheme: first, false correlation is screened out, a lightning stroke causes temperature and vibration to present strong correlation in the second cycle ( ), but no such correlation in the first cycle ( ), if the preset threshold is 0.2, it is determined that it is a transient interference combination and is excluded from prediction; second, physical correlation stability capture is realized, in the early aging stage of the cooling pump, the correlation between current and temperature in cycle 1 is 0.82→the correlation in cycle 2 is 0.79, the difference is 0.03<0.2, triggering prediction, placing short-time data correlation (S3 expression output) in the long-time stability verification (S4) framework, solving the defect of false correlation caused by random working condition fluctuation within the monitoring period.

[0059] S4: screen out the corresponding two kinds of state data whose difference between the first correlation coefficient and the second correlation coefficient is less than the preset change threshold, and use the two kinds of state data as predicted data, and obtain the data change of the predicted data within the prediction time period according to the second data sequence of the predicted data.

[0060] This step specifically includes:

[0061] S41: obtaining the change rate of the predicted data within the prediction time period according to the second data sequence of the predicted data;

[0062] S42: generating the predicted values of the predicted data at multiple time points within the prediction time period according to the monitoring value of the predicted data at the current time and the change rate within the prediction time period.

[0063] In this embodiment, it should be noted that in S41, the core is to infer future trends based on the latest data behavior. For each data to be predicted selected in S4, such as the temperature T of the switch cabinet, the complete time sequence in the second monitoring period (the most recent period), i.e., the second data sequence, is extracted. By analyzing the change rule of the continuous monitoring points at the end of the sequence, such as the rising and falling mode of the last several monitoring values, the change rate of the data in the prediction time period is inferred using existing algorithms. Specifically: if the end of the sequence shows a continuous rising trend, such as three consecutive monitoring points increasing, it is determined that the change rate is positive; if it is continuously decreasing, the rate is negative; if it is stable, the rate tends to zero. This rate quantifies the evolution inertia of the data under the current operating state, while avoiding the introduction of noise by a complex model - because S4 has selected data with stable physical mechanisms, the short-term evolution rule has high reliability.

[0064] In S42, linear extrapolation multi-time point prediction is implemented. With the change rate calculated in S41 as the core driving, the real-time monitoring value at the current time, such as the measured value of temperature T at 14:00, is superimposed to generate the predicted values of multiple time points in the prediction time period at equal time intervals. The calculation formula is simplified as: future time point predicted value = current monitoring value + change rate x time span.

[0065] In this embodiment, it should be noted that in S1, the main purpose is to establish a time framework and a basic data source for subsequent data analysis. Specifically, when implementing prediction, it is necessary to first obtain the historical data time sequence from the monitoring of the power equipment (such as the sensor network): two consecutive monitoring periods before the current time are selected, such as periodic data based on fixed sampling intervals, both monitoring periods are close to the current time, and the period before is defined as the first monitoring period, i.e., the second period close to the current time, which represents the previous historical trend of the equipment state; and the period after is defined as the second monitoring period, i.e., the period closest to the current time, which represents the subsequent historical trend of the equipment state. The length of these monitoring periods is set according to the characteristics of the equipment, such as minute-level or hour-level windows. At the same time, S1 obtains multiple state data in the operation of the equipment, such as the temperature of the equipment shell, the internal current or the vibration amplitude, which are derived from the operating mechanism of the equipment, ensuring comprehensive coverage of the prediction. Finally, S1 determines the prediction time period, such as the next few minutes or hours, which defines the prediction target range, so that the subsequent steps can output the state change prediction value at a specific time point, thereby supporting operation and maintenance decisions.

[0066] Further, for example, assume that in a switch cabinet monitoring scenario of a 110 kV substation, the operation and maintenance sets the monitoring period to be 2 hours. The first monitoring period is the data of the previous 4 to 2 hours before the current time, for example, temperature, vibration, and current sequence, which reflects the running state of the equipment under the previous 4 to 2 hours environment; the second monitoring period is the data of the previous 2 hours, for example, temperature sequence from yesterday to this morning, which captures the changes of the equipment under the latest conditions. The to-be-predicted time period is set to be 1 hour in the future, which means that the operation and maintenance personnel need to be warned of potential problems of the equipment at night. At this time, S1 automatically extracts these period data and state types to provide structured input for subsequent correlation stability analysis.

[0067] In S2, the multi-source state data is analyzed in the first monitoring period (i.e., the second closest historical period to the current time), and the core goal is to reveal the physical correlation between the data by quantifying the fluctuation trend of individual data. Specifically, first, for each state data such as temperature, current, and vibration, the time sequence thereof in the period is extracted, i.e., the first data sequence, and the relative change amount sequence is obtained by calculating the difference between adjacent monitoring values (retaining the positive and negative signs of the change direction); then, the relative change amount is normalized to an absolute change amount (eliminating the influence of different dimensions while retaining the trend direction information) that can be compared across data in combination with the theoretical extreme value range of the state under the current operating environment, such as the temperature safety threshold in the equipment design specification; finally, the absolute change amounts are summed to obtain a "first fluctuation trend index", which captures both the overall change amplitude and directionality of the data, with a positive value indicating an upward trend and a negative value indicating a downward trend. On this basis, the first correlation coefficient between any two state data is obtained: when the first fluctuation trend indices of the two states are highly consistent in value and direction, such as A temperature rising and B current increasing simultaneously, it indicates that they have a strong coupling relationship in physical mechanism, and the correlation coefficient tends to be the maximum value; otherwise, the correlation coefficient significantly decreases or even becomes zero.

[0068] Further, for example, taking transformer monitoring as an example, the oil temperature and the cooling fan current data are analyzed in the first monitoring period (12:00-14:00). In S21, the oil temperature sequence is point-by-point differentiated to obtain the relative change amount, such as +3℃, -2℃, etc., which is normalized to the absolute change amount sequence in combination with the theoretical range of 10 to 85, such as 、 ; accumulated oil temperature fluctuation trend index 0.013, overall weak rise. At the same time, fan current data is processed: relative change sequence, such as +0.1A, -0.05A; normalized to absolute change sequence according to the theoretical range of 0 to 10A, such as +0.01, -0.005; accumulated fluctuation trend index +0.015, synchronized weak rise. Since the trends of the two are in the same direction and the amplitudes are close, it is assumed that the first correlation coefficient calculated in step S22 is 0.89, representing strong correlation, reflecting the normal response mechanism of the heat dissipation system to temperature rise; while the fluctuation trend index of the winding vibration data during the same period is -0.32, showing a downward trend, and its correlation with the oil temperature is only 0.03, with no significant correlation between the two.

[0069] In S3, the state data analysis in the second monitoring period (i.e. the period closest to the current time) is focused on, and its logical framework is completely symmetrical but the object of action is different from S2: through the real-time evaluation of the fluctuation trend and correlation of the latest data, the latest dynamic coordination of the device state is captured. Specifically: first, extract the time series of each state data in this period (second data sequence), obtain the second relative change sequence (positive value indicates rise, negative value indicates fall) by calculating the signed difference between adjacent monitoring values point by point; then, based on the theoretical extreme range of the state under the current working condition (such as the maximum vibration amplitude or current value allowed by the device), the relative change is normalized to the standardized second absolute change (eliminate dimensional difference and retain trend direction); finally, all absolute changes in the sequence are accumulated to obtain the second fluctuation trend index - this index quantifies the overall change intensity and directionality of the data in the latest period. On this basis, the second correlation coefficient between any two state data is calculated: if the fluctuation trends of the two in the current period are highly synchronized (such as A temperature rapidly rising and B current synchronously steeply increasing), the correlation coefficient will significantly approach the upper limit value, reflecting that there is strong physical coupling between the two; if the trends are divergent or the amplitude difference is significant (such as temperature rising but vibration unchanged), the correlation coefficient will sharply decrease or even zero, revealing the risk of correlation failure.

[0070] In S4, which is the core decision-making stage of the technical solution, the cross-period correlation stability of multi-source state data is verified, and reliable data combinations with physical mechanism support are selected for final prediction. The specific operation includes two key steps: First, compare the correlation coefficients of each data combination in the first monitoring period (mid-term history) and the second monitoring period (recent dynamics) (for example, the coefficient of temperature-current in period 1 is 0.85, and the coefficient in period 2 is 0.82). If the difference between the two is less than the preset threshold (for example, the threshold is set to 0.1), it is determined that the correlation of the data combination has cross-time stability (false correlation caused by random disturbance), and it is listed as the data to be predicted. Otherwise, if the difference exceeds the threshold (for example, the coefficient of vibration-temperature is 0.9 in a certain period, and drops to 0.2 in the next period), it indicates that the correlation is caused by transient disturbance, and the two are not included in the prediction range (which does not affect the judgment of other data combinations). Subsequently, for the stable data combinations selected, based on the time sequence in the latest period (second monitoring period) (i.e. the second data sequence), the data change rate is deduced by analyzing the value change characteristics at the end of the sequence (such as recent continuous rise or oscillation convergence), and combined with the real-time monitoring value at the current time, the prediction values at multiple time points in the predicted time period are generated by linear extrapolation.

[0071] It should also be noted that in S4, the to-be-predicted data is selected through multi-level correlation stability verification, which includes three layers of progressive logic: First, if the correlation coefficient difference between a certain state data (such as temperature T) and any other state data (such as current C) in the first monitoring period and the second monitoring period is less than the preset threshold (such as 0.1), T and C are listed as to-be-predicted data. This mechanism ensures that as long as there is at least one stable physical correlation support, the data combination is considered reliable; the second layer, when the correlation coefficient difference of a certain data (such as current C) and multiple data (such as temperature T, voltage U) is less than the threshold at the same time, it represents that the data maintains a coordinated evolution rule in multiple physical dimensions (such as the thermal effect correlation of C and T, the power balance correlation of C and U), and its cross-period behavior consistency is higher, and the prediction confidence is significantly improved; the third layer, the complete exclusion mechanism, if the correlation coefficient difference between a certain state data (such as vibration V) and all other data exceeds the threshold (such as 0.1), it indicates that its fluctuation is completely out of the framework of device physical mechanism (such as random electromagnetic interference), and the state data is directly excluded as a prediction object to avoid noise interference.

[0072] Further, the monitoring scene of the switch cabinet is continued (three state data: temperature T, current C, and vibration V), and a preset correlation stability threshold is 0.1. The first correlation coefficient of the temperature T-current C combination in period 1 is 0.18, and the second correlation coefficient in period 2 is 0.15 (a difference of 0.03 < 0.1), and it is determined to be a stable correlation combination. At this time, the T and C sequences in the second period (the last 2 hours) are extracted: the end of the T sequence shows an upward trend of x units / hour, and the end of the C sequence shows a downward trend of y units / hour. Based on the current T value and C value, the system predicts the T and C values every 10 minutes in the next 1 hour according to the respective change rates. The first correlation coefficient of the temperature T-vibration V combination in period 1 is 0.75, but the second correlation coefficient in period 2 suddenly drops to 0.15 (a difference of 0.6 > 0.1), and it is determined to be a false correlation. Although the vibration data is monitored, it is excluded from the prediction range because the correlation is unstable, avoiding misjudgment caused by a certain transient disturbance (such as switch operation vibration). This mechanism accurately distinguishes between real failure precursors and noise fluctuations, improving prediction reliability.

[0073] In summary, in the above-mentioned entire power equipment operation safety state prediction method based on monitoring data, first, a cross-period physical correlation stability mechanism is established, which addresses the defect of existing technologies that ignore the collaborative evolution of multiple source data. A stable correlation data combination is selected by comparing the correlation coefficients of the first monitoring period (medium-term history) and the second monitoring period (recent dynamics). The fluctuation trend indicators of any two state data are used to quantify the correlation strength (the stronger the trend directionality and the closer the amplitude, the higher the correlation coefficient), and the combinations with a cross-period correlation coefficient difference less than the threshold are selected by a preset threshold. This mechanism ensures that the data included in the prediction has long-term stability supported by physical mechanisms, effectively avoiding false correlations caused by random disturbances such as load surges, and distinguishing between real failure precursors and noise fluctuations from the root cause. Further, a layered data prediction access rule is designed to avoid misjudgment in single data prediction. If the correlation of a certain state data with at least one other data is stable, it is listed as a prediction object. When a certain state data meets the stability with multiple data, its prediction confidence is further improved, and a completely disconnected state data is completely excluded. This rule preserves the simplicity of independent single data prediction while enhancing data reliability through cross-validation of the correlation network, avoiding misjudgment of internal faults caused by sudden environmental temperature rises. Finally, linear extrapolation prediction based on recent behavior patterns is used for stable data. The future prediction value at a certain time point is generated directly based on the change direction and rate of the sequence at the end of the second monitoring period and the current value. In summary, the entire technical solution couples the physical correlation stability verification and behavior linear extrapolation links in a closed loop, achieving reduced failure misjudgment rate, enhanced prediction result interpretability, and improved long-term and short-term working condition adaptability.

[0074] The scheme also provides a system for implementing the above method, comprising an acquisition module, a first data processing module, a second data processing module and a data prediction module,

[0075] The acquisition module is configured to acquire two continuous monitoring periods of the power equipment before the current time as a first monitoring period and a second monitoring period, acquire a plurality of state data in the operation of the power equipment, and acquire a to-be-predicted time period.

[0076] The first data processing module is configured to acquire a first data sequence of the various state data in the first monitoring period, acquire a first fluctuation trend index corresponding to the various state data according to the first data sequence of the various state data, and acquire a first correlation coefficient between any two state data according to the first fluctuation trend index of the various state data.

[0077] The second data processing module is configured to acquire a second data sequence of the various state data in the second monitoring period, acquire a second fluctuation trend index corresponding to the various state data according to the second data sequence of the various state data, and acquire a second correlation coefficient between any two state data according to the second fluctuation trend index of the various state data.

[0078] The data prediction module is configured to screen out corresponding two state data with a difference between the first correlation coefficient and the second correlation coefficient less than a preset change threshold as to-be-predicted data, and acquire a data change of the to-be-predicted data in the to-be-predicted time period according to the second data sequence of the to-be-predicted data.

[0079] In an optional manner, the first data processing module is further configured to: acquire, according to the first data sequence of the state data, a difference between adjacent two monitoring values in the first data sequence as a first relative change, acquire a theoretical maximum value and a theoretical minimum value corresponding to the state data, and acquire a first absolute change corresponding to the first relative change according to the first relative change, the theoretical maximum value and the theoretical minimum value; and accumulate absolute changes corresponding to each first absolute change in the first data sequence of the state data as the first fluctuation trend index.

[0080] In an optional manner, the second data processing module is further configured to: acquire, according to the second data sequence of the state data, a difference between adjacent two monitoring values in the second data sequence as a second relative change, acquire a theoretical maximum value and a theoretical minimum value corresponding to the state data, and acquire a second absolute change corresponding to the second relative change according to the second relative change, the theoretical maximum value and the theoretical minimum value; and accumulate absolute changes corresponding to each second absolute change in the second data sequence of the state data as the second fluctuation trend index.

[0081] The acquisition module acquires two continuous monitoring periods before the current time as a first monitoring period and a second monitoring period, and obtains a plurality of state data and a to-be-predicted time period in the two periods. The first data processing module acquires a first data sequence of various state data in the first monitoring period, and acquires a first fluctuation trend index corresponding to the various state data according to the first data sequence of the various state data, and acquires a first correlation coefficient between any two state data according to the first fluctuation trend index of the various state data. The second data processing module acquires a second data sequence of various state data in the second monitoring period, and acquires a second fluctuation trend index corresponding to the various state data according to the second data sequence of the various state data, and acquires a second correlation coefficient between any two state data according to the second fluctuation trend index of the various state data. The data prediction module screens out corresponding two state data whose difference between the first correlation coefficient and the second correlation coefficient is less than a preset change threshold as to-be-predicted data, and acquires a data change of the to-be-predicted data in the prediction time period according to the second data sequence of the to-be-predicted data. The scheme couples the physical correlation stability verification and the behavior linear extrapolation link in a closed loop, reduces the fault misjudgment rate, enhances the interpretability of the prediction result, and improves the adaptability of long-term and short-term working conditions.

[0082] In the embodiment, it should be noted that, regarding the above-mentioned power equipment operation safety state prediction system based on monitoring data, the specific manner of performing operation has been described in detail in the embodiment of the power equipment operation safety state prediction method based on monitoring data, and will not be described in detail here.

[0083] Figure 4 The structure schematic diagram of the embodiment of the power equipment operation safety state prediction device based on monitoring data of the application is shown, and the specific embodiment of the application does not limit the specific implementation of the power equipment operation safety state prediction device based on monitoring data.

[0084] As Figure 4 shown, the power equipment operation safety state prediction device based on monitoring data 700 can include a processor 701, a memory 702. The electronic device 700 can also include one or more of a multimedia component 703, an I / O interface 704 (input / output interface), and a communication component 705.

[0085] The processor 701 is configured to control overall operations of the electronic device 700 to complete all or part of the steps of the power equipment operation safety state prediction method based on monitoring data described above. The memory 702 is configured to store various types of data to support operations of the electronic device 700, which can include, for example, instructions for any application or method operating on the electronic device 700, and application-related data, such as contact data, sent and received messages, pictures, audio, video, and the like. The memory 702 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as 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. The multimedia component 703 can include a screen and an audio component. The screen can be, for example, a touch screen, and the audio component is configured to output and / or input audio signals. For example, the audio component can include a microphone configured to receive external audio signals. The received audio signals can be further stored in the memory 702 or transmitted through the communication component 705. The audio component also includes at least one speaker configured to output audio signals. The I / O interface 704 provides an interface between the processor 701 and other interface modules, which can be a keyboard, a mouse, a button, and the like. The buttons can be virtual buttons or physical buttons. The communication component 705 is configured to perform wired or wireless communication between the electronic device 700 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, near field communication (NFC), 2G, 3G, 4G, NB-IOT, eMTC or other 5G, and the like, or a combination of one or more of them, is not limited herein. Therefore, the corresponding communication component 705 can include a Wi-Fi module, a Bluetooth module, an NFC module, and the like.

[0086] In an exemplary embodiment, the electronic device 700 can 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, micro-controllers, microprocessors, or other electronic elements for performing the above-mentioned power equipment operation safety state prediction method based on monitoring data.

[0087] In another exemplary embodiment, a computer readable storage medium including program instructions is also provided, which, when executed by a processor, implement the steps of the above-mentioned power equipment operation safety state prediction method based on monitoring data. For example, the computer readable storage medium can be the above-mentioned memory 702 including program instructions, and the above-mentioned program instructions can be executed by the processor 701 of the electronic device 700 to complete the above-mentioned power equipment operation safety state prediction method based on monitoring data.

[0088] In another exemplary embodiment, a computer program product is also provided, which contains a computer program capable of being executed by a programmable device, and the computer program has code portions for performing the above-mentioned power equipment operation safety state prediction method based on monitoring data when executed by the programmable device.

[0089] The preferred embodiments of the present disclosure are described in detail above with reference to the accompanying drawings, but the present disclosure is not limited to the specific details in the above-described embodiments. Within the technical concept scope of the present disclosure, various simple modifications can be made to the technical solutions of the present disclosure, and these simple modifications all belong to the protection scope of the present disclosure.

[0090] In addition, it should be noted that each specific technical feature described in the above-described specific embodiments can be combined in any appropriate manner without contradiction. In order to avoid unnecessary repetition, various possible combinations are not described again in the present disclosure.

[0091] In addition, any combination of various different embodiments of the present disclosure can also be made, as long as it does not deviate from the idea of the present disclosure, and it should also be considered as the disclosed content of the present disclosure.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit the present application; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can still be modified, or some or all of the technical features can be replaced by equivalent replacements; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application, and they should be covered in the scope of the claims and the description of the present application.

Claims

1. A method for predicting the safe operating state of power equipment based on monitoring data, characterized in that, The method comprises: S1: obtaining two consecutive periods before the current time as a first monitoring period and a second monitoring period, obtaining a plurality of state data in the operation of the power equipment in the monitoring period, and obtaining a to-be-predicted time period; S2: obtaining a first data sequence of various state data in the first monitoring period, obtaining a first fluctuation trend index corresponding to the various state data according to the first data sequence of the various state data, and obtaining a first correlation coefficient between any two state data according to the first fluctuation trend index of the various state data; S3: obtaining a second data sequence of various state data in the second monitoring period, obtaining a second fluctuation trend index corresponding to the various state data according to the second data sequence of the various state data, and obtaining a second correlation coefficient between any two state data according to the second fluctuation trend index of the various state data; S4: screening out corresponding two state data with a difference between the first correlation coefficient and the second correlation coefficient less than a preset change threshold as to-be-predicted data, and obtaining a data change of the to-be-predicted data in the prediction time period according to the second data sequence of the to-be-predicted data; According to the first data sequence of the various state data, the first fluctuation trend index corresponding to the various state data specifically comprises: According to the first data sequence of the state data, the difference between adjacent two monitoring values in the first data sequence is obtained as a first relative change, the theoretical maximum value and the theoretical minimum value corresponding to the state data are obtained, and the first absolute change corresponding to the first relative change is obtained according to the first relative change, the theoretical maximum value and the theoretical minimum value; the first relative change / (theoretical maximum value-theoretical minimum value)=first absolute change; The absolute change corresponding to each first absolute change in the first data sequence of each state data is accumulated as the first fluctuation trend index of the corresponding state data.

2. The method of claim 1, wherein the method further comprises: The first correlation coefficient is expressed as: ; wherein, is the first correlation coefficient between the i-th and j-th state data, is the first fluctuation trend indicator of the i-th state data, is the first fluctuation trend indicator of the j-th state data, is a scaling coefficient.

3. The method of claim 1, wherein the method further comprises: According to the second data sequence of the various state data, the second fluctuation trend index corresponding to the various state data specifically comprises: According to the second data sequence of the state data, the difference between adjacent two monitoring values in the second data sequence is obtained as a second relative change, the theoretical maximum value and the theoretical minimum value corresponding to the state data are obtained, and the second absolute change corresponding to the second relative change is obtained according to the second relative change, the theoretical maximum value and the theoretical minimum value; the second relative change / (theoretical maximum value-theoretical minimum value)=second absolute change; The absolute change corresponding to each second absolute change in the second data sequence of each state data is accumulated as the second fluctuation trend index of the corresponding state data.

4. The method of claim 3, wherein the method further comprises: The second correlation coefficient is expressed as: ; wherein is a second correlation coefficient between the i-th and j-th state data, is a second volatility trend indicator of the i-th state data, is a second volatility trend indicator of the j-th state data, is a scaling coefficient.

5. The method of claim 1, wherein the method further comprises: According to the second data sequence of the to-be-predicted data, the data change of the to-be-predicted data in the prediction time period specifically comprises: According to the second data sequence of the to-be-predicted data, the change rate of the to-be-predicted data in the prediction time period is obtained; According to the monitoring value of the to-be-predicted data at the current time and the change rate in the prediction time period, the predicted value of the to-be-predicted data at a plurality of time points in the prediction time period is generated.

6. A power equipment operation safety state prediction system based on monitoring data, characterized by, The system comprises: The acquisition module is configured to acquire two continuous periods before the current time as a first monitoring period and a second monitoring period, acquire a plurality of state data in the operation of the power equipment in the monitoring periods, and acquire a to-be-predicted time period; The first data processing module is configured to acquire a first data sequence of the various state data in the first monitoring period, acquire a first fluctuation trend index corresponding to the various state data according to the first data sequence of the various state data, and acquire a first correlation coefficient between any two state data according to the first fluctuation trend index of the various state data; The second data processing module is configured to acquire a second data sequence of the various state data in the second monitoring period, acquire a second fluctuation trend index corresponding to the various state data according to the second data sequence of the various state data, and acquire a second correlation coefficient between any two state data according to the second fluctuation trend index of the various state data; The data prediction module is configured to screen out corresponding two state data with a difference between the first correlation coefficient and the second correlation coefficient less than a preset change threshold as to-be-predicted data, and acquire a data change of the to-be-predicted data in the prediction time period according to the second data sequence of the to-be-predicted data. The first data processing module is configured to acquire, according to the first data sequence of the state data, a difference between two adjacent monitoring values in the first data sequence as a first relative change, acquire a theoretical maximum value and a theoretical minimum value corresponding to the state data, and acquire a first absolute change corresponding to the first relative change according to the first relative change, the theoretical maximum value, and the theoretical minimum value; accumulate absolute changes corresponding to each first absolute change in the first data sequence of the state data as the first fluctuation trend index; the first relative change / (theoretical maximum value-theoretical minimum value)=the first absolute change.

7. A power equipment operation safety state prediction device based on monitoring data, characterized by, comprising: a memory having stored thereon a computer program; a processor configured to execute the computer program in the memory to implement the power equipment operation safety state prediction method based on monitoring data according to any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The program is executed by the processor to implement the power equipment operation safety state prediction method based on monitoring data according to any one of claims 1 to 5.

Images