Intelligent power utilization safety monitoring and management system

Abstract

The invention belongs to the technical field of power utilization safety monitoring and management, and particularly discloses and provides an intelligent power utilization safety monitoring and management system which comprises a data sensing module, a risk analysis module, a scheme generation module, a scheme correction module and a scheme feedback terminal. According to the method, the average change rate and the fluctuation stability are calculated based on the current time sequence data, and then the trend and the dispersion degree of parameter change are quantified, so that the problem that slow deterioration cannot be captured by a traditional fixed threshold value method is solved, the capacity of sensing hidden risks is improved, and the method is suitable for popularization and application. The sampling frequency of the sensor is dynamically adjusted based on the risk evolution probability, effective balance between monitoring efficiency and resource consumption is achieved, meanwhile, whether prediction adjustment reaches the standard or not is evaluated based on actual early warning data in combination with historical data, when prediction adjustment does not reach the standard, reverse adjustment is conducted on a prediction adjustment scheme, and the prediction adjustment efficiency is improved. The quantile of the reference normal interval is dynamically corrected, and continuous optimization of the early warning strategy of the system is ensured.

Description

Technical Field

[0001] This invention belongs to the field of electricity safety monitoring and management technology, and relates to an intelligent electricity safety monitoring and management system. Background Technology

[0002] With increasing societal demands for electrical safety, smart electricity systems have become a key technological means to prevent electrical fires and ensure the stable operation of power facilities. Traditional electricity safety monitoring systems typically rely on preset fixed thresholds for anomaly detection. Their early warning strategies lack dynamic adaptability and struggle to adapt to the complex operating conditions in power distribution systems caused by load fluctuations, changes in equipment operating status, and environmental factors. This severely restricts the accuracy and reliability of the monitoring system.

[0003] To overcome the aforementioned shortcomings, several data analysis-based improvement schemes have emerged in the existing technology. For example, Chinese invention patent CN114862293A discloses a smart electricity safety management method and system. This scheme determines standard parameters by obtaining enterprise registration information, collects current data from power supply nodes, compares it with the standard parameters, calculates the electricity risk rate, and finally triggers an alert based on a preset risk rate threshold. This method, by incorporating personalized enterprise information, improves the relevance of risk assessment to a certain extent.

[0004] The existing technologies described above have the following shortcomings: 1. The current system uses a fixed sampling frequency for data collection, which cannot dynamically adjust the monitoring density according to the evolving risk situation. In high-risk scenarios, a fixed sampling frequency may lead to the loss of key data features, affecting the timeliness of risk identification. In low-risk scenarios, it results in the unnecessary consumption of monitoring resources.

[0005] 2. The current system's early warning judgment relies on preset, static risk rate thresholds and lacks a closed-loop feedback mechanism that can self-optimize based on historical early warning results. Consequently, when dynamic risk trends change in the power system, the fixed thresholds cannot adapt to them, leading to a gradual decline in the accuracy of risk assessment and making it difficult to achieve continuous, reliable, and accurate early warnings. Summary of the Invention

[0006] In view of this, in order to solve the problems mentioned in the background technology, a smart electricity safety monitoring and management system is proposed.

[0007] The objective of this invention can be achieved through the following technical solution: This invention provides a smart electricity safety monitoring and management system, including: a data sensing module, which collects current time-series data of various time-series parameters through sensors deployed at key monitoring points in the power distribution system.

[0008] The risk analysis module calculates the average rate of change and volatility stability of each time series parameter based on the current time series data, and performs trend comparison analysis in conjunction with historical data to determine the probability of risk evolution at each key monitoring point.

[0009] The scheme generation module dynamically adjusts the sampling frequency of the sensor based on the risk evolution probability, and dynamically calculates and adjusts the early warning threshold according to the parameter change trend to generate a predictive adjustment scheme.

[0010] The scheme correction module collects actual early warning data during the effect evaluation period after the power distribution system implements the predicted adjustment scheme to calculate the actual early warning effect index, and evaluates whether the predicted adjustment meets the standard by combining historical data. When the predicted adjustment does not meet the standard, the predicted adjustment scheme is reversed to generate a corrected predicted adjustment scheme.

[0011] The solution feedback terminal feeds back the revised prediction and adjustment plan to the power safety monitoring and management platform.

[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention calculates the average rate of change and fluctuation stability of each time series parameter based on the current time series data, thereby quantifying the trend and dispersion of parameter changes, thus solving the problem of slow deterioration or intermittent anomalies that cannot be captured by the traditional fixed threshold method, and improving the ability to perceive hidden risks.

[0013] (2) This invention dynamically adjusts the sensor sampling frequency based on the risk evolution probability, assigns a higher sampling frequency level to the risk probability range with a higher failure rate, thereby achieving more complete data features during high-risk periods and reasonably saving monitoring resources during low-risk periods, thus achieving an effective balance between monitoring efficiency and resource consumption.

[0014] (3) This invention dynamically calculates and adjusts the early warning threshold by analyzing the trend of parameter changes, and establishes a negative feedback mechanism between the threshold and the trend of parameter changes by using the coupled calculation of the trend slope and the early warning threshold. This enables the system to automatically lower the threshold to provide early warning when the parameters deteriorate, and to raise the threshold to reduce false alarms when the parameters improve, thereby dynamically optimizing the reliability of the early warning.

[0015] (4) This invention evaluates whether the prediction adjustment meets the standard based on actual early warning data and historical data, and when the prediction adjustment does not meet the standard, it reverses the prediction adjustment scheme and dynamically corrects the quantile of the benchmark normal interval, thus ensuring the continuous optimization of the system's early warning strategy.

[0016] (5) By using the quantile statistical method to construct the baseline normal interval of each time series parameter, the present invention effectively filters out random fluctuations and noise interference in historical data, thereby enhancing the anti-interference capability of the system and providing a reliable basis for subsequent anomaly identification. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram showing the connections of the various modules in the system of the present invention.

[0019] Figure 2 This is a schematic diagram showing the connection steps for determining the probability of risk evolution in this invention.

[0020] Figure 3 This is a schematic diagram showing the connection steps of the prediction and adjustment scheme of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see Figure 1 As shown, the present invention provides a smart electricity safety monitoring and management system, which includes: a data sensing module, a risk analysis module, a scheme generation module, a scheme correction module, and a scheme feedback terminal.

[0023] In the above, the data perception module is connected to the risk analysis module and the solution generation module, the solution generation module is connected to the risk analysis module and the solution correction module, and the solution correction module is also connected to the solution feedback terminal.

[0024] The data sensing module collects current time-series data of various time-series parameters through sensors deployed at key monitoring points in the power distribution system. These time-series parameters include, but are not limited to, current, residual current, and temperature parameters.

[0025] The risk analysis module calculates the average rate of change and volatility stability of each time series parameter based on the current time series data, and performs trend comparison analysis in conjunction with historical data to determine the probability of risk evolution at each key monitoring point.

[0026] For example, the calculation of the average rate of change of each time series parameter includes: obtaining the values ​​of each time series parameter at each moment in the current monitoring period from the current time series data, and constructing the value change curve of each time series parameter.

[0027] The slope of the change curve is extracted as the average rate of change of each time series parameter.

[0028] For example, the calculation of the fluctuation stability of each time series parameter includes: calculating the standard deviation and mean of the numerical sequence based on the values ​​of each time series parameter at each time point within the current monitoring period.

[0029] The ratio of the standard deviation to the mean is used as the volatility stability of each time series parameter.

[0030] Please see Figure 2 As shown, for example, determining the risk evolution probability of each key monitoring point includes: Q1, constructing a baseline normal range for each time series parameter based on the historical data.

[0031] Furthermore, the construction of the baseline normal interval for each time series parameter includes: Q1-1, selecting historical monitoring periods that are adjacent to the current monitoring period and marked as having no time series parameter deviation from historical data as the baseline monitoring period.

[0032] It should be noted that "proximity" refers to the M consecutive historical monitoring cycles preceding the current monitoring cycle, where M is a configurable parameter, for example, 5-20 cycles. To achieve a balance between the timeliness and statistical stability of the benchmark data, the specific value of M is dynamically correlated with the monitoring frequency of the power distribution system. Its adaptive adjustment rule is that the value of M should cover a typical system operating cycle. As an example, if the system monitoring frequency is once per hour, then M is 24 to cover one day; if the monitoring frequency is once every 15 minutes, then M is 96 to cover one day; if the monitoring frequency is once per day, then M is 7 to cover one week. Simultaneously, to ensure the continuous operation of the system, if the number of cycles meeting the conditions within the last M cycles is insufficient, the system has a fault-tolerant mechanism that automatically expands the selection range to historical monitoring cycles within the last 30 days to ensure the availability of the benchmark construction function. It should be understood that if, after expanding to the last 30 days, there are still no historical monitoring cycles meeting the condition of no time-series parameter deviation, the system uses preset industry standard parameters to construct the initial benchmark normal range.

[0033] It should be noted that the historical monitoring period without time-series parameter deviation refers to a period in which the system did not trigger any warnings or record any faults, and where the average rate of change and fluctuation stability of all time-series parameters were within the corresponding benchmark range. This ensures that the selected benchmark monitoring period can represent the safe operating state of the system. By using such data as the source for constructing the benchmark, the reference value of the constructed benchmark normal range is effectively guaranteed, thereby providing a reliable and accurate judgment benchmark for subsequent risk identification.

[0034] Q1-2. Obtain the average rate of change of each time series parameter at each time point from the historical data of the benchmark monitoring period, and construct a set of average rate of change values ​​for each time series parameter.

[0035] Q1-3. For the set of average change rate values ​​of each time series parameter within the benchmark monitoring period, take the P1 quantile as the lower limit and the P2 quantile as the upper limit to obtain the benchmark normal range of the average change rate of each time series parameter.

[0036] It should be noted that the P1 quantile is used to determine the lower boundary of the benchmark normal range, representing the lower critical point of the normal fluctuation range of each time series parameter within the benchmark monitoring period. The P2 quantile is used to determine the upper boundary of the benchmark normal range, representing the upper critical point of the normal fluctuation range of each time series parameter within the benchmark monitoring period.

[0037] The average rate of change characterizes the trend and speed of parameter change. During normal operation, the system is expected to change gradually; drastic, rapid changes are more likely to indicate potential failures or shocks. To more sensitively capture such abnormal trends, a relatively narrow tolerance range is set. For example, P1=10 and P2=90 can be set to focus on 80% of the main data, maintaining greater vigilance for faster trends at both ends of the range.

[0038] Q1-4. Obtain the fluctuation stability of each time series parameter at each time point from the historical data of the benchmark monitoring period, and construct a set of fluctuation stability values ​​for each time series parameter.

[0039] Q1-5. For the set of fluctuation stability values, take the P3 quantile as the lower limit and the P4 quantile as the upper limit to obtain the normal range of fluctuation stability for each time series parameter.

[0040] It should be noted that the P3 quantile is used to determine the lower boundary of the baseline normal range, representing the lower critical point of the normal fluctuation range of each time series parameter within the baseline monitoring period. The P4 quantile is used to determine the upper boundary of the baseline normal range, representing the upper critical point of the normal fluctuation range of each time series parameter within the baseline monitoring period.

[0041] The principle behind fluctuation stability is that it characterizes the degree of dispersion of parameters around their mean. A healthy system typically exhibits stable fluctuations. Therefore, a stable benchmark is desired that covers the vast majority of normal fluctuations and filters out accidental noise. Based on this, a relatively wide tolerance range can be set. Preferably, P3=25 and P4=75, i.e., using the interquartile range, can be set to reflect the fluctuation level of the middle 50% of the most stable data, thereby avoiding misjudging normal or slightly larger fluctuations as abnormal.

[0042] Q2. Compare the average rate of change and fluctuation stability of each time series parameter with its baseline normal range. If the time series parameter is not within its baseline normal range, then the time series parameter is determined to be an abnormal time series parameter.

[0043] Q3. For each key monitoring point, count the number of time series parameters that are identified as abnormal.

[0044] Q4. The ratio of the number of abnormal time series parameters to the total number of time series parameters at key monitoring points is used as the risk evolution probability of each key monitoring point.

[0045] The scheme generation module dynamically adjusts the sampling frequency of the sensor based on the risk evolution probability, and dynamically calculates and adjusts the early warning threshold according to the parameter change trend to generate a prediction and adjustment scheme.

[0046] For example, the dynamic adjustment of the sensor sampling frequency includes: calculating the risk evolution probability of each key monitoring point in each historical monitoring period based on the historical data, and obtaining records of whether a fault occurred during the same period.

[0047] It should be noted that the calculation method for the risk evolution probability of each key monitoring point in each historical monitoring period is the same as the method for the risk evolution probability in the current monitoring period, and will not be repeated here.

[0048] The risk evolution probability of each key monitoring point in each historical monitoring period is sorted from low to high according to its value, and the sorted risk evolution probability is divided into multiple continuous risk probability intervals according to a preset ratio.

[0049] It should be noted that the preset ratio is used to divide the ranking sequence of historical risk evolution probabilities into multiple probability intervals with different risk levels. This ratio is set by equally dividing the entire ranking sequence of historical risk evolution probabilities to form 3 to 5 consecutive risk probability intervals. Preferably, the system defaults to dividing into 3 intervals: a low-risk interval (0% to 33% after ranking), a medium-risk interval (33% to 67% after ranking), and a high-risk interval (67% to 100% after ranking). The number of intervals can be configured and adjusted by the power safety monitoring and management platform according to the actual risk level classification requirements of the power distribution system.

[0050] Calculate the proportion of the number of periods in which failures occurred in all historical monitoring periods within each risk probability interval to the total number of periods, and use this as the failure rate for each risk probability interval.

[0051] The risk probability intervals are sorted from low to high according to their failure rates.

[0052] The highest sampling frequency level is assigned to the risk probability interval with the highest failure rate, the second highest sampling frequency level is assigned to the risk probability interval with the second highest failure rate, and so on, to obtain the risk probability interval corresponding to each sampling frequency.

[0053] The sampling frequency of each key monitoring point is obtained by matching the risk evolution probability of each key monitoring point with the risk probability interval corresponding to each sampling frequency.

[0054] For example, the generation of the prediction adjustment scheme includes: for each time series parameter in each key monitoring point, obtaining the average rate of change of the current monitoring period, and the continuous rate of change of the historical data. The average rate of change for each monitoring period is used to construct trend curves for each time series parameter in chronological order.

[0055] It should be noted that the continuous Each monitoring period is used to construct a time series with statistical significance and trend analysis value. This represents the size of the historical data window used to analyze the trend of parameter changes. The value of is dynamically related to the probability of risk evolution. In a specific embodiment, this is achieved through the formula... In the formula For the number of monitoring cycles, and These are the preset maximum and minimum backtracking periods, for example... , , This is the floor function. This represents the probability of risk evolution.

[0056] This formula ensures that the system can automatically reduce the risk as the probability of risk evolution increases. This value is used to enhance sensitivity in detecting accelerating risk evolution trends. Conversely, when the probability of risk evolution is low, the value can be increased. Values ​​to obtain a smoother, more robust view of long-term trends.

[0057] The trend change curve is linearly fitted, and the slope is obtained as the trend slope of each time series parameter.

[0058] The trend slope is coupled with the early warning threshold to obtain the adjustment early warning threshold for each time series parameter at each key monitoring point.

[0059] For example, the formula for calculating the adjusted warning threshold is: In the formula The adjusted warning threshold is calculated as follows: This is the preset baseline warning threshold for this time series parameter at this key monitoring point. The slope is the trend obtained through linear fitting. A predefined, positive-zero scaling factor used to control the strength of the influence of the trend slope on the threshold adjustment.

[0060] The purpose of this design is to establish a linear negative feedback relationship between the trend slope and the threshold adjustment amount. When the trend slope... hour, , making This means lowering the warning threshold. This makes the system more sensitive when parameters show an upward or deteriorating trend, enabling it to issue warnings earlier, with the aim of proactively preventing risks and reducing the false negative rate.

[0061] When the trend slope hour, , making This means raising the warning threshold. This helps avoid overreacting to positive improvement trends, reducing false alarm rates, and making the system run more smoothly with a lower trend slope. The larger the absolute value, the stronger the trend, and the greater the adjustment range of the threshold.

[0062] Wherein, scaling factor It is determined by linear regression analysis of the trend slope and the optimal early warning threshold in historical data. The larger the value, the greater the threshold adjustment caused by the same trend slope, and the more sensitive the system is to trend changes. The smaller the value, the smoother the system's response to trends, and the more conservative the threshold adjustment. Preferably, The initial value can be determined by analyzing the relationship between trend changes and the optimal early warning threshold in historical data, and can be optimized based on the feedback of scheme correction during system operation.

[0063] To ensure the effectiveness of the warning threshold, a safety range constraint must be imposed on the adjusted warning threshold. Specifically, the warning threshold value is limited to between 10% and 150% of the rated value of the physical quantity corresponding to the time-series parameter. If the calculated warning threshold exceeds this range, the system automatically sets it as the upper or lower limit of the safety range.

[0064] The adjusted sampling frequency and the adjusted early warning threshold of each time series parameter at each key monitoring point are encapsulated to generate a prediction adjustment scheme.

[0065] The scheme correction module collects actual early warning data during the effect evaluation period after the power distribution system implements the prediction adjustment scheme to calculate the actual early warning effect index, and evaluates whether the prediction adjustment meets the standard by combining historical data. When the prediction adjustment does not meet the standard, the prediction adjustment scheme is reversed to generate a corrected prediction adjustment scheme.

[0066] It should be noted that the duration of the effectiveness evaluation period is dynamically linked to the system's monitoring frequency to ensure the collection of statistically significant operational data. Specifically, the effectiveness evaluation period comprises W complete monitoring cycles. As an example, if the system monitoring frequency is once per hour, the effectiveness evaluation period is set to 24 monitoring cycles; if the monitoring frequency is once per day, the effectiveness evaluation period is set to 7 monitoring cycles. The value of W typically ranges from 1 to 7 and can be configured according to the timeliness requirements of the evaluation in actual applications.

[0067] To avoid the correction logic from failing, if no fault occurs during the effect evaluation period, the false alarm rate is 0, and the system will only make corrections based on the false alarm rate; if no warning is triggered at the same time, the correction will be skipped and the current solution will be retained.

[0068] For example, the calculation of actual early warning effectiveness indicators includes: obtaining the number of correct early warnings, the number of false early warnings, and the number of missed faults from actual early warning data.

[0069] The total number of actual faults is obtained by summing the number of correct warnings and the number of missed faults. The ratio of the number of missed faults to the total number of actual faults is then used as the actual missed fault rate.

[0070] The total number of warnings triggered by the system is obtained by summing the number of correct warnings and the number of false alarms. The ratio of the number of false alarms to the total number of warnings triggered by the system is then used as the actual false alarm rate.

[0071] The actual underreporting rate and the actual false alarm rate are used as indicators of the actual early warning effectiveness.

[0072] For example, the assessment of whether the prediction adjustment meets the standard includes: obtaining the historical actual underreporting rate and historical actual false alarm rate from historical data within the historical monitoring period immediately adjacent to the current effect assessment period.

[0073] The actual false alarm rate and the actual false alarm rate are compared with the corresponding historical actual false alarm rate and historical actual false alarm rate, respectively.

[0074] When the actual false negative rate is less than the historical actual false negative rate and the actual false positive rate is less than the historical actual false positive rate, it indicates that the current forecast adjustment plan has improved the system's risk capture capability and judgment accuracy, and the overall early warning performance of the system has been comprehensively optimized. The forecast adjustment plan is deemed to meet the standard. Otherwise, the forecast adjustment plan is deemed to fail to meet the standard.

[0075] When the actual false negative rate is greater than or equal to the historical false negative rate, it indicates that the current scheme may have a reduced ability to identify real risks due to excessively high warning thresholds or insufficient monitoring sensitivity, thus posing a security risk.

[0076] When the actual false alarm rate is greater than or equal to the historical actual false alarm rate, it indicates that the current solution may be due to the warning threshold being set too low or the judgment conditions being too sensitive, resulting in the system generating too many invalid warnings, which reduces monitoring efficiency and user experience.

[0077] When either the false alarm rate or the false alarm rate increases, it indicates that the prediction adjustment scheme has failed to achieve the expected optimization effect in actual application, or even leads to the deterioration of system performance. At this time, it is judged as unqualified and the scheme correction mechanism is triggered.

[0078] Please see Figure 3 As shown, for example, the generation of the corrected prediction adjustment scheme includes: subtracting the actual missed rate and the actual false alarm rate from the corresponding historical actual missed rate and the historical actual false alarm rate, and then selecting the difference with the larger absolute value as the key deviation.

[0079] The key deviation is multiplied by a fixed proportional constant to obtain the dynamic adjustment step size. When the main deviation is an increase in the false negative rate, the difference between the quantile and the dynamic adjustment step size is calculated to obtain the adjusted quantile.

[0080] It should be added that the fixed proportional constant is a proportional coefficient dynamically calculated based on the system deviation state, used to convert the absolute value of the critical deviation, which reflects the performance gap of the system, into a specific adjustment step size for the quantile.

[0081] A fixed scaling constant is determined using a predefined dynamic scaling function. This function is configured such that the larger the critical deviation, the larger the calculated scaling factor.

[0082] In one specific embodiment, the linear mapping formula can be used for calculation: In the formula and These are the preset upper and lower limits of the proportional coefficient, for example. , . This is the critical deviation amount calculated so far. The preset deviation normalization benchmark can be the largest historical critical deviation observed since the system began operating, or a fixed value set empirically, such as 0.5, which represents 50% of the deviation. During the initial operation phase of the system, the largest historical critical deviation has not yet been observed. Use the default value of 0.5. Wait until the system has accumulated more than 90 days of operation. It will be automatically updated to the maximum critical deviation actually observed during this period.

[0083] When the system performance deviation is significant, the above formula will automatically calculate a larger scaling factor, thereby generating a larger adjustment step size. This allows the system to quickly perform coarse adjustments, rapidly narrowing the performance gap and improving convergence efficiency.

[0084] When the system performance approaches its optimal state, the formula calculates a smaller proportionality coefficient, resulting in a smaller adjustment step size. This allows the system to smoothly fine-tune, precisely approximating the optimal state and avoiding repeated oscillations around the optimal value. By dynamically mapping key deviations to proportionality coefficients through a predefined mathematical function, the system achieves the goal of adaptively adjusting the correction step size based on real-time performance differences, thus balancing convergence speed and system stability.

[0085] When the main deviation is an increase in the false alarm rate, the quantile is summed with the dynamic adjustment step size to obtain the adjusted quantile.

[0086] Apply safety range constraints to the adjusted quantiles, and reconstruct the revised prediction adjustment scheme based on the quantiles with safety range constraints.

[0087] It should be added that the specific process of applying a safety range constraint to the adjusted quantiles is as follows: Define a lower safety limit and a higher safety limit for the quantiles, for example, a lower safety limit of 0 and a higher safety limit of 100. Compare the adjusted quantiles with the lower and higher safety limits; if the adjusted quantiles are less than the lower safety limit, set them as the lower safety limit; if the adjusted quantiles are greater than the higher safety limit, set them as the higher safety limit; if the adjusted quantiles are within the safety range, maintain the adjusted quantiles.

[0088] It should be noted that the reconstructed and corrected prediction adjustment scheme will use the new quantile parameters, after being processed by the safety range constraint, as the latest input for constructing the baseline normal interval, and will completely re-execute the above prediction adjustment scheme generation process to generate the corrected prediction adjustment scheme.

[0089] Specifically, the reconstruction process of the revised forecast adjustment scheme is as follows: First, based on the new quantile parameters, the average rate of change benchmark normal range and the volatility stability benchmark normal range of each time series parameter are recalculated according to the above method for constructing the benchmark normal range of each time series parameter.

[0090] Then, using the updated baseline normal range, the risk evolution probability of each key monitoring point is recalculated according to the method described above for determining the risk evolution probability of each key monitoring point.

[0091] Finally, the updated risk evolution probability is substituted into the above method of dynamically adjusting the sensor sampling frequency and generating a prediction adjustment scheme to generate a corrected prediction adjustment scheme, which includes the corrected sampling frequency and the corrected warning threshold.

[0092] The proposed solution feedback terminal will send the revised prediction and adjustment plan back to the electricity safety monitoring and management platform.

[0093] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0094] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0095] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0096] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0097] Finally, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart power utilization safety monitoring management system, characterized in that: The system comprises: a data sensing module, which collects current time series data of each time series parameter through sensors deployed at each key monitoring point in the power distribution system; a risk analysis module, which calculates the average change rate and fluctuation stability of each time series parameter based on the current time series data, and performs trend comparison analysis combined with historical data to determine the risk evolution probability of each key monitoring point; a scheme generation module, which dynamically adjusts the sampling frequency of the sensors based on the risk evolution probability, and dynamically calculates and adjusts the early warning threshold according to the parameter change trend to generate a prediction adjustment scheme; a scheme correction module, which collects actual early warning data to calculate actual early warning effect indicators during the effect evaluation period after the power distribution system executes the prediction adjustment scheme, evaluates whether the prediction adjustment meets the standard combined with historical data, and generates a corrected prediction adjustment scheme by reverse adjustment when the prediction adjustment does not meet the standard; a scheme feedback terminal, which feeds back the corrected prediction adjustment scheme to the power utilization safety monitoring and management platform.

2. The intelligent power utilization safety monitoring and management system according to claim 1, characterized in that: The calculation of the average change rate of each time series parameter comprises: obtaining the values of each time series parameter at each time in the current monitoring period from the current time series data to construct a value change curve of each time series parameter; extracting the slope from the change curve as the average change rate of each time series parameter.

3. The intelligent power utilization safety monitoring and management system of claim 1, wherein: The calculation of the fluctuation stability of each time series parameter comprises: calculating the standard deviation and mean value of the value sequence respectively based on the values of each time series parameter at each time in the current monitoring period; taking the ratio of the standard deviation to the mean value as the fluctuation stability of each time series parameter.

4. The intelligent power utilization safety monitoring and management system of claim 1, wherein: The determination of the risk evolution probability of each key monitoring point comprises: Q1, constructing a baseline normal interval of each time series parameter based on the historical data; Q2, comparing the average change rate and fluctuation stability of each time series parameter with its baseline normal interval, if the time series parameter is not located in its baseline normal interval, it is determined that the time series parameter is an abnormal time series parameter; Q3, for each key monitoring point, the number of abnormal time series parameters determined is counted; Q4, taking the ratio of the number of abnormal time series parameters to the total number of time series parameters of the key monitoring point as the risk evolution probability of the key monitoring point.

5. The intelligent power utilization safety monitoring management system according to claim 4, characterized in that: The construction of the baseline normal interval of each time series parameter comprises: selecting a historical monitoring period adjacent to the current monitoring period and marked as no time series parameter deviation as a baseline monitoring period from the historical data; obtaining the average change rate of each time series parameter at each time from the historical data of the baseline monitoring period to construct an average change rate value set of each time series parameter; for the average change rate value set of each time series parameter in the baseline monitoring period, taking the P1 quantile as the lower limit and the P2 quantile as the upper limit to obtain the average change rate baseline normal interval of each time series parameter; obtaining the fluctuation stability of each time series parameter at each time from the historical data of the baseline monitoring period to construct a fluctuation stability value set of each time series parameter; for the fluctuation stability value set, taking the P3 quantile as the lower limit and the P4 quantile as the upper limit to obtain the fluctuation stability baseline normal interval of each time series parameter.

6. The intelligent power utilization safety monitoring and management system of claim 1, wherein: The dynamic adjustment of the sampling frequency of the sensors comprises: Based on the historical data, the risk evolution probability of each key monitoring point in each historical monitoring period is calculated, and the record of whether a fault occurs in the same period is obtained; The risk evolution probability of each key monitoring point in each historical monitoring period is sorted from low to high according to its value, and the sorted risk evolution probability is divided into multiple continuous risk probability intervals according to a preset proportion; The proportion of the number of periods in which a fault occurs in all historical monitoring periods in each risk probability interval to the total number of periods is calculated, and is taken as the fault rate of each risk probability interval; Each risk probability interval is sorted from low to high according to its fault rate; The highest sampling frequency level is assigned to the risk probability interval with the highest fault rate, the second highest sampling frequency level is assigned to the risk probability interval with the second highest fault rate, and so on, to obtain the risk probability interval corresponding to each sampling frequency; The risk evolution probability of each key monitoring point is matched with the risk probability interval corresponding to each sampling frequency, to obtain the sampling frequency of each key monitoring point.

7. The intelligent power utilization safety monitoring management system according to claim 1, characterized in that: The generation of the predicted adjustment scheme includes: For each time series parameter in each key monitoring point, the average change rate of the current monitoring period and the average change rate of the continuous monitoring periods in the historical data are obtained, and a trend change curve of each time series parameter is constructed in time sequence. Linear fitting is performed on the trend change curve, and then the slope is obtained therefrom as the trend slope of each time sequence parameter; The trend slope is coupled with the early warning threshold to obtain the adjustment early warning threshold of each time sequence parameter in each key monitoring point; The adjusted sampling frequency is encapsulated with the adjusted early warning threshold of each time sequence parameter in each key monitoring point to generate a predicted adjustment scheme. 8.The smart power utilization safety monitoring and management system of claim 1, wherein: The calculation of the actual early warning effect index includes: The number of correct early warnings, the number of false positive early warnings, and the number of missed fault reports are obtained from the actual early warning data; The number of correct early warnings and the number of missed fault reports are summed to obtain the total number of actual faults, and then the ratio of the number of missed fault reports to the total number of actual faults is taken as the actual missed fault rate; The number of correct early warnings and the number of false positive early warnings are summed to obtain the total number of early warnings triggered by the system, and then the ratio of the number of false positive early warnings to the total number of early warnings triggered by the system is taken as the actual false positive rate; The actual missed fault rate and the actual false positive rate are taken as the actual early warning effect index. 9.The smart power utilization safety monitoring and management system of claim 1, wherein: The evaluation of whether the predicted adjustment meets the standard includes: The historical actual missed fault rate and the historical actual false positive rate in the historical monitoring period adjacent to the current effect evaluation period are obtained from the historical data; The actual missed fault rate and the actual false positive rate are compared with the corresponding historical actual missed fault rate and historical actual false positive rate, respectively; When the actual missed fault rate is less than the historical actual missed fault rate and the actual false positive rate is less than the historical actual false positive rate, it is determined that the predicted adjustment scheme meets the standard, otherwise, it is determined that the predicted adjustment scheme does not meet the standard.

10. The intelligent power safety monitoring and management system of claim 5, wherein: The generation of the corrected predicted adjustment scheme includes: The actual missed fault rate and the actual false positive rate are subtracted from the corresponding historical actual missed fault rate and historical actual false positive rate, respectively, and then the difference with a larger absolute value is selected as the key deviation; The key deviation is multiplied by a fixed proportion constant to obtain a dynamic adjustment step, when the main deviation is the increase of the missed fault rate, the quantile number is subtracted from the dynamic adjustment step to obtain the adjusted quantile number; When the main deviation is the increase of the false positive rate, the quantile number is summed with the dynamic adjustment step to obtain the adjusted quantile number; The safety range constraint is imposed on the adjusted quantile, and a revised prediction adjustment scheme is reconstructed based on the quantile after the safety range constraint is imposed.

Images