Real-time analysis and fault early warning method and system based on digital electric port box transformer data

Abstract

The invention discloses a data real-time analysis and fault early warning method and system based on a digital electric port box transformer substation, and belongs to the technical field of power equipment monitoring and control systems. The method comprises the following steps: acquiring environmental parameters of each component of the box transformer substation in real time through a monitoring network, and performing data smooth denoising to obtain a parameter sequence; constructing a dynamic storage framework based on the sequence in combination with space-time attributes; extracting historical records of the target component to perform multi-dimensional sorting and change rate calculation, and identifying potential anomalies; when the change rate exceeds the limit and the distribution boundary breaks through, generating an alarm sequence and determining a preliminary abnormal mark; comparing the mark with the historical record, fusing the influence weight analysis verification accuracy, and obtaining a final early warning grading result; and parameter summary information is pushed to a control system according to a grading result, response feedback is obtained, and closed-loop control of the operation state is formed. According to the invention, the problem of control lag caused by data noise and delay is solved, and the timeliness, reliability and automation level of box transformer substation control are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power equipment monitoring and control systems, and in particular to a method and system for real-time analysis and fault early warning of digital electric port box transformer data. BACKGROUND

[0002] In the field of operation and maintenance of power systems, box-type transformers (referred to as "box transformers") as key nodes of power distribution are of vital importance to the security and stability of regional power supply. With the popularization of Internet of Things and digital technology, real-time state monitoring based on sensor data has become the mainstream means to improve the efficiency of power equipment management.

[0003] At present, the existing box transformer monitoring schemes mostly realize data collection and basic display functions of environmental parameters (such as temperature, humidity, etc.). However, from the perspective of the control system, these existing technologies have significant deficiencies in realizing forward-looking fault early warning and forming an automated control closed loop, the core problems of which mainly lie in the following aspects: The data processing capability of existing schemes is weak, resulting in poor input signal quality of control decisions. Box transformer environmental parameter data is easily affected by various electromagnetic noise and field interference during real-time collection, and existing technologies lack effective dynamic filtering and denoising mechanisms, so that the data transmitted to the control system contains a large amount of invalid information. The control system makes judgments based on such impure data, which is prone to false alarms or missed alarms, seriously affecting the accuracy of control execution. The data storage and analysis model of existing technologies is rigid, resulting in a serious lag in control response. The current scheme mostly uses a static threshold alarm mechanism, and its data storage structure fails to fully consider the dynamic correlation of parameters in the time and space dimensions. When a device shows early signs of abnormality (such as a slow rise in the temperature of a certain component), the system cannot issue an early warning due to the inability to identify minor deviations through coordinated comparison with historical trends and the status of adjacent components. Only when the parameter value exceeds the fixed safety threshold will an alarm be triggered, and by this time the device may have developed from a sub-healthy state to an irreversible fault, and the control system has missed the best early intervention opportunity.

[0004] The existing technical architecture essentially separates "monitoring" and "control", failing to form an intelligent closed loop of perception-analysis-decision-execution. The existing monitoring system mostly only plays the role of a data dashboard, and the output alarm information often lacks risk classification and specific trend guidance, failing to provide sufficient accurate and forward-looking decision basis for the upper-level control system (such as the load scheduling system, cooling control system). This leads to control behavior always being in a passive mode of "after-the-fact remediation", rather than an active mode of "pre-emptive prevention", greatly restricting the automation and intelligent level of power system operation and maintenance.

[0005] In summary, the existing box transformer monitoring scheme cannot realize the proactive intervention of faults due to slow control response caused by poor data quality and analysis lag. SUMMARY

[0006] The application discloses a digital electric port box transformer data real-time analysis and fault early warning method and system, specifically a method and system for automatically generating hierarchical early warning signals based on dynamic anomaly identification results to drive the control system to respond by real-time acquisition and processing of box transformer multi-component environmental parameters, which can effectively solve the technical problems of slow control response caused by poor data quality and analysis lag in the existing box transformer monitoring scheme, and cannot realize the proactive intervention of faults.

[0007] To solve the above technical problems, the technical scheme adopted by the application is: The digital electric port box transformer data real-time analysis and fault early warning method specifically includes the following steps: Step 1: Real-time acquisition of environmental parameter data from each component of the power equipment through a network of monitoring devices, preliminary screening of parameter deviation indicators and distribution balance states, denoising processing of the collected data using a data smoothing method to obtain a smoothed parameter sequence; Step 2: According to the smoothed parameter sequence, analyze the deviation duration characteristics and fluctuation dynamic indicators in combination with the space-time attribute information, construct a data organization structure, group the environmental parameter data according to the frequency distribution and abnormal proportion indicators, and determine the initial storage container; Step 3: Obtain the parameter data in the initial storage container, calculate the spatial correlation indicators and consistency indicators, if the trend change exceeds the preset condition or the time correlation is lower than the standard, trigger the dynamic adjustment of the initial storage container, and obtain the adjusted storage framework; Step 4: Extract the historical parameter records of the target component from the adjusted storage framework, perform multi-dimensional sorting for peak value comparison and benchmark comparison, and judge the potential abnormal position and time information; Step 5: For the potential abnormal position, calculate the change rate indicator and evaluate it in combination with the trigger condition and risk assessment factors, if the change rate is higher than the condition and the distribution boundary is broken, generate an alarm output sequence and determine the preliminary abnormality mark; Step 6: Through comparison of the preliminary abnormality mark and the historical parameter record, fuse the influence weight analysis and abnormal proportion evaluation, verify the accuracy of the alarm output, and obtain the final early warning classification result; Step 7: According to the final early warning classification result, push the parameter summary information containing the trend change and spatial consistency to the control system, and obtain the response feedback record.

[0008] Step 1 is as follows: The real-time data of the environmental parameters of each component of the power equipment is obtained through the monitoring device network, continuously recorded at a preset collection frequency, and an initial environmental parameter data set is obtained. The data is denoised by using a data smoothing method for the initial environmental parameter data set, and noise interference is eliminated to obtain a smoothed parameter sequence. According to the smoothed parameter sequence, the specific situation of parameter deviation is analyzed, and if a parameter value exceeds a preset threshold range, it is marked as an abnormal parameter point to obtain a marked abnormal data set. Through the marked abnormal data set, feature information of a balanced distribution state is extracted to determine whether there is an unbalanced distribution phenomenon, and an evaluation result of the distribution state is obtained. According to the evaluation result of the distribution state, if an unbalanced distribution phenomenon is found, the weight of the abnormal parameter point is adjusted, and an adjusted parameter distribution set is determined. The adjusted parameter distribution set is obtained, and the change trend of the parameter sequence is continuously monitored in combination with the update frequency of the real-time data to determine whether there is a potential abnormal fluctuation, and a final monitoring analysis result is obtained. Through the final monitoring analysis result, a corresponding parameter adjustment strategy is generated, the running state of the power equipment is dynamically optimized, and an optimized running parameter configuration is obtained.

[0009] Step 2 is specifically as follows: Through the smoothed parameter sequence, in combination with the space-time attribute information, the correlation mode of deviation persistence and fluctuation dynamics is analyzed, a preset classification rule is used to preliminarily group the data, and a classified data set is obtained. According to the classified data set, a logical framework of data organization is constructed for the characteristics of frequency distribution and abnormal proportion, a preset threshold is used to screen the abnormal proportion, and an abnormal data subset is determined. The abnormal data subset is obtained, the fluctuation rule of the parameter sequence in different time periods is analyzed in combination with the dynamic change trend, a statistical tool is used to quantitatively process the fluctuation dynamics, and a fluctuation feature set is obtained. Through the fluctuation feature set, a corresponding storage unit allocation scheme is constructed for the distribution mode of frequency distribution, and if the abnormal proportion of a certain distribution mode exceeds a preset threshold, the data under the mode is marked with a priority, and a marked data unit is determined. According to the marked data unit, in combination with the mapping logic of attribute information and space-time attributes, the rationality of data classification is analyzed, the classification result is optimized and adjusted by using the data organization framework, and an optimized classification structure is obtained. The optimized classification structure is obtained, the distribution balance of the parameter sequence under different classifications is analyzed for the allocation of the storage unit, and if the distribution balance under a certain classification is lower than a preset standard, the data under the classification is re-grouped, and a final storage allocation scheme is determined. Through the final storage allocation scheme, in combination with the fluctuation feature set and the dynamic change trend, the update state of the parameter sequence is continuously tracked, and an automatic tool is used to dynamically adjust the data classification and the storage unit to obtain an adjusted data management framework.

[0010] Step 3 is specifically as follows: Through the parameter data in the initial storage container, the calculation results of the spatial correlation index and the consistency index, the fluctuation of the trend change is analyzed, if the trend change exceeds the preset threshold, the data in the storage container is prioritized, and a sorted data set is obtained; According to the sorted data set, the time-related evaluation result is obtained, if the time correlation is lower than the predetermined standard, the data set is processed in layers, and the data group after layering is determined; The data group after layering is obtained, the dynamic adjustment allocation logic is analyzed according to the capacity limit of the storage container, if the data volume of a certain group exceeds the upper limit of the container, the group is split, and the split data unit is obtained; Through the split data unit, combined with the distribution mode of spatial correlation, the update rule of the storage framework is constructed, the data unit is repositioned by using the preset allocation strategy, and the updated storage layout is determined; The updated storage layout is obtained, the balance state of the data unit in the storage framework is analyzed according to the stability requirement of the consistency index, if the balance state does not reach the predetermined standard, the secondary adjustment mechanism is triggered, and the optimized storage structure is obtained; According to the optimized storage structure, combined with the execution record of dynamic adjustment, the subsequent influence of trend change is continuously tracked, the storage framework is monitored in real time by using an automatic tool, and the final storage configuration is determined; Through the final storage configuration, a long-term management scheme of data storage is constructed according to the persistence of the adjustment result, a log recording tool is used to track the changes of the storage framework, and a complete change archive is obtained.

[0011] Step 4 is as follows: Through the adjusted storage framework, the related historical parameter data of the target component is obtained, the data is preliminarily classified by using the preset screening rule, and the classified parameter set is obtained; According to the classified parameter set, the peak difference and the reference value are compared, if the peak difference exceeds the preset threshold, the data whose peak difference exceeds the threshold is marked, and the abnormal data set after marking is determined; Starting from the marked abnormal data set, the data is prioritized by combining the multi-dimensional sorting logic, and the arranged data sequence is obtained; For the arranged data sequence, the specific distribution of the abnormal position is analyzed, the abnormal position is accurately identified by using the preset positioning tool, and the specific abnormal coordinate point is determined; Through the abnormal coordinate point, combined with the record data of the time node, the time correlation information in the historical parameter is traced back, and the corresponding time period distribution is obtained; According to the time period distribution, the corresponding relationship between the abnormal position and the time node is constructed for the result of the difference analysis, and the final abnormal distribution mode is determined; The final abnormal distribution mode is obtained, the adjustment structure of the storage framework is combined, the parameter optimization strategy for the target component is generated, and the direction of subsequent processing is determined.

[0012] Step 5 is specifically as follows: For the monitoring of the change rate, the rate change data of the target object is obtained from the historical data record, and a preliminary comparison is made in combination with the preset threshold value; If the rate change exceeds the threshold range, an initial abnormal signal is generated, and a preliminary focus point is determined; According to the initial abnormal signal, data extraction is performed on the abnormal position, and the position information is refined using a preset positioning tool to obtain specific abnormal coordinate distribution; Through the abnormal coordinate distribution, the boundary data of the distribution range are combined to analyze whether there is a significant deviation; If the deviation phenomenon is detected, the corresponding deviation identifier is generated, and the distribution characteristics of the abnormality are determined; For the deviation identifier, relevant trigger condition data is obtained, and a comprehensive treatment is performed in combination with the preset rules of risk assessment to obtain a classification result of risk level; According to the classification result of the risk level, a corresponding alarm sequence is generated, the alarms are sorted using a preset priority rule, and the output order of the alarms is determined; Through the output order of the alarm, the abnormal position is confirmed again in combination with the record data of the preliminary marking to obtain a final abnormal marking set; For the final abnormal marking set, the relevance of the abnormal marking is analyzed using a logistic regression model to determine the potential relationship between the abnormal markings and determine the key direction of subsequent processing.

[0013] Step 6 is specifically as follows: Through the comparison of the preliminary marking and the historical parameters, the difference information between the marking data and the parameter record is obtained, a preliminary screening is performed using a preset threshold range, and a difference significant marking set is obtained; According to the difference significant marking set, data extraction is performed on the abnormal proportion, and a weighted calculation is performed in combination with the preset rules of influence weight to determine the weighted value of the abnormal proportion; Through the weighted value of the abnormal proportion, data records related to the alarm output are obtained, if the weighted value exceeds the preset threshold range, a corresponding alarm signal is generated, and the priority of the alarm signal is determined; According to the priority of the alarm signal, the signal data is processed in layers in combination with the logical rules of proportion evaluation, and a layered signal classification is obtained; Through the layered signal classification, aiming at the data demand of early warning grading, a logistic regression model is used to map the classified data to determine the preliminary result of early warning grading; According to the preliminary result of early warning grading, combined with the processing rules of accurate verification, the graded data is compared again to obtain the final output result; Through the final output result, aiming at the business logic of grading determination, the result data is matched with the preset grading standard to judge the final early warning grading state.

[0014] Step 7 is as follows: Through the final early warning grading result, parameter summary data containing trend changes and spatial consistency are sorted out, and the data is sent to the control system to obtain the interactive log after sending; According to the interactive log after sending, the response feedback data returned by the control system is extracted, and the data is analyzed by using the preset field rule to determine the analyzed feedback content; Through the analyzed feedback content, the key fields in the feedback record are classified and processed to obtain the classified data group, and the integrity of the grouped data is judged; If the integrity of the grouped data meets the preset threshold range, the classified data group is prioritized to obtain the sorted feedback sequence; According to the sorted feedback sequence, data filtering is performed according to the demand of result analysis, and a logistic regression model is used to predict the trend of the filtered data to determine the predicted trend direction; Through the predicted trend direction, combined with the spatial consistency parameter summary data, the matching result after comparison is obtained, and the correlation degree of the matching result is judged; If the correlation degree of the matching result reaches the preset standard, the matching result and the response feedback data are integrated and stored to obtain the final business data archive.

[0015] In addition, the application also discloses a real-time analysis and fault early warning system based on digital electric port box variable data, comprising: A data acquisition module is used to acquire environmental parameter data from each component of the power equipment in real time through a monitoring device network; A data processing module is used to perform smoothing and denoising processing on the collected data to obtain a smoothed parameter sequence; A storage management module is used to construct a dynamic storage framework according to the smoothed parameter sequence combined with space-time attribute information; An abnormality detection module is used to extract the historical parameter record of the target component from the adjusted storage framework, perform multidimensional sorting and change rate calculation, and identify potential abnormalities; The early warning generation module is used for generating an alarm sequence based on the change rate index and the distribution boundary breakthrough, verifying the accuracy by comparison with historical parameter records, and obtaining early warning grading results; The control interface module is used for pushing parameter summary information to a control system according to the early warning grading results and obtaining response feedback.

[0016] Further, the data acquisition module comprises a plurality of sensor nodes arranged in each component of the box transformer for collecting temperature and humidity environment parameters, and the sampling frequency is once per minute, and the data is transmitted through a ZigBee wireless network.

[0017] Compared with the prior art, the present application has the following beneficial effects: The present application decomposes the real-time analysis of box transformer data and the fault early warning process into basic data processing units, solving the technical problems of data noise interference, low storage efficiency and abnormal detection delay in the prior art.

[0018] The present application adopts a data smoothing method to perform denoising processing on the real-time collected environment parameter data, obtains a smoothed parameter sequence, effectively eliminates random noise and interference signals, improves data quality and reliability, and provides an accurate basis for subsequent analysis; secondly, a dynamically adjusted storage framework is constructed in combination with space-time attribute information, the data is grouped and priority marked through frequency distribution and abnormal proportion indexes, efficient classification and storage of environment parameter data are realized, the problem of low retrieval efficiency caused by disorganized data in the prior art is solved, so that the historical parameter records of the target component can be quickly located, and continuous trend tracking and abnormal tracing are supported; The present application accurately identifies potential abnormal position and time information through multi-dimensional sorting and change rate index evaluation, generates an alarm output sequence in combination with trigger conditions and risk evaluation factors, and ensures the timeliness and accuracy of abnormal detection.

[0019] The present application verifies the accuracy of the alarm output by comparison of the preliminary abnormal marking with the historical parameter records, fusion of influence weight analysis and abnormal proportion evaluation, obtains the final early warning grading results, pushes parameter summary information containing trend change and spatial consistency to the control system, and obtains response feedback records, so that real-time monitoring and dynamic optimization of the operation state of the power equipment are realized, the response speed, accuracy and operation and maintenance efficiency of the fault early warning are significantly improved, and the safe and stable operation of the power system is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments. It should be understood that the following drawings only show some embodiments of the present application, and therefore should not be considered as limiting the scope, and other related drawings can also be obtained by those skilled in the art without creative labor.

[0021] Figure 1 The overall flowchart of the present application.

[0022] Figure 2 The detailed flowchart of step 3 of the present application. DETAILED DESCRIPTION

[0023] In the following, only some exemplary embodiments are simply described. As those skilled in the art can recognize, the described embodiments can be modified in various different ways without departing from the spirit or scope of the embodiments of the present application. Therefore, the drawings and the description are considered to be exemplary in nature rather than limiting.

[0024] The embodiments of the present application will be described in detail below with reference to the drawings.

[0025] Embodiment 1: Referring to Figure 1 and Figure 2 , the present embodiment discloses a real-time analysis and fault early warning method based on digital electric port box variable data, specifically comprising the following steps: Step 1: Real-time acquisition of environmental parameter data from each component of the power equipment through a monitoring device network, preliminary screening of parameter deviation indicators and distribution balance states, denoising processing of the collected data using a data smoothing method, and obtaining of a smoothed parameter sequence; Step 2: According to the smoothed parameter sequence, combined with space-time attribute information, analyzing deviation persistence characteristics and fluctuation dynamic indicators, constructing a data organization structure, grouping the environmental parameter data according to frequency distribution and abnormal proportion indicators, and determining an initial storage container; Step 3: Obtaining parameter data in the initial storage container, calculating spatial correlation indicators and consistency indicators, if the trend change exceeds the preset condition or the time correlation is lower than the standard, triggering dynamic adjustment of the initial storage container, and obtaining an adjusted storage framework; Step 4: Extracting historical parameter records of the target component from the adjusted storage framework, multi-dimensional sorting according to peak value comparison and benchmark comparison, and judging potential abnormal position and time information; Step 5: For the potential abnormal position, calculating the change rate indicator and combining the trigger condition and risk assessment factors for evaluation, if the change rate is higher than the condition and the distribution boundary is broken, generating an alarm output sequence and determining a preliminary abnormality mark; Step 6: Verify the accuracy of the alarm output by comparing the preliminary anomaly label with the historical parameter record, fusing the impact weight analysis and the anomaly proportion evaluation, and obtaining the final early warning classification result; Step 7: According to the final early warning classification result, push the parameter summary information containing trend change and spatial consistency to the control system to obtain the response feedback record.

[0026] In step 1, the following steps are taken: Through the monitoring device network, real-time data of environmental parameters of each component of the power equipment is obtained, and the initial environmental parameter data set is obtained by continuously recording at a preset collection frequency. For the initial environmental parameter data set, a data smoothing method is used to denoise the data and eliminate noise interference to obtain a smoothed parameter sequence. According to the smoothed parameter sequence, the specific situation of parameter deviation is analyzed. If a parameter value exceeds the preset threshold range, it is marked as an abnormal parameter point to obtain a marked abnormal data set. Through the marked abnormal data set, feature information of balanced distribution state is extracted to determine whether there is an unbalanced distribution phenomenon to obtain an evaluation result of distribution state. According to the evaluation result of distribution state, if an unbalanced distribution phenomenon is found, the weight of the abnormal parameter point is adjusted to determine an adjusted parameter distribution set. The adjusted parameter distribution set is obtained, and the change trend of the parameter sequence is continuously monitored in combination with the update frequency of the real-time data to determine whether there is a potential abnormal fluctuation to obtain a final monitoring analysis result. Through the final monitoring analysis result, a corresponding parameter adjustment strategy is generated to dynamically optimize the running state of the power equipment to obtain an optimized running parameter configuration.

[0027] In specific implementation: For example, in the scenario of power equipment environmental parameter monitoring, suppose that the running state of a transformer is collected and analyzed in real time, and key parameters such as temperature and humidity are focused on.

[0028] For real-time data acquisition of environmental parameters, the temperature and humidity data of each component of the transformer are continuously collected at a frequency of once per minute through a sensor network to form an initial data set.

[0029] Suppose that the temperature data sequence collected in a certain period of time is 38.2, 39.1, 45.3, 38.9, and 40.2 degrees Celsius, and there is obvious abnormal fluctuation.

[0030] In one possible implementation, for data denoising, a moving average method is used to smooth the initial data set.

[0031] For example, taking the average of three data points as the smoothing result, the smoothed sequence is 38.8, 41.1, and 41.5 degrees Celsius, significantly reducing the sudden value caused by noise interference and improving the stability of the data. The advantage of this is that it can more accurately reflect the true trend in subsequent analysis and avoid misjudgment.

[0032] For example, in the parameter deviation analysis phase, assuming that the preset threshold range of temperature is 35.0 to 42.0 degrees Celsius, the smoothed 41.5 has exceeded the threshold, and is therefore marked as an abnormal parameter point, forming an abnormal data set. This step helps to quickly locate potential risk points and provides a basis for subsequent processing.

[0033] In one possible implementation, for the distribution balance state analysis of the abnormal data set, the distribution characteristics of the abnormal points are extracted, and it is found that the temperature abnormal points are concentrated in a certain time period, showing an uneven distribution phenomenon. Through statistics, it is found that abnormal points often occur during high-load operation of the device, reflecting potential heat dissipation problems. Such evaluation results help to reveal the deep problems of device operation.

[0034] For example, in the weight adjustment link, for the abnormal points of uneven distribution, the weight of the abnormal points during the high-load period is increased, for example, from 1.0 to 1.5, to highlight its impact on the overall operation state, forming an adjusted parameter distribution set. This can more accurately reflect the device risk and improve the pertinence of monitoring.

[0035] In one possible implementation, the adjusted parameter distribution set and the real-time data update frequency are combined to continuously monitor the change trend of the temperature sequence. Assuming that the temperature continues to fluctuate between 41.0 and 42.5 in the subsequent period, indicating a potential abnormal fluctuation risk, the final monitoring and analysis result shows that attention should be paid to the performance of the heat dissipation system. This provides data support for subsequent optimization.

[0036] For example, in generating the parameter adjustment strategy, based on the monitoring result, it is suggested to increase the operating frequency of the transformer cooling fan from 30 minutes per hour to 45 minutes, while reducing the upper limit of the load, dynamically optimizing the operation parameter configuration. This not only effectively reduces the temperature abnormal risk, but also prolongs the service life of the device and ensures the stable operation of the power system.

[0037] Further, step 2 is as follows: through the smoothed parameter sequence, combined with the spatiotemporal attribute information, analyze the correlation pattern of deviation persistence and fluctuation dynamics, use the preset classification rule to preliminarily group the data, and obtain the classified data set; according to the classified data set, aiming at the characteristics of frequency distribution and abnormal proportion, construct the logical framework of data organization, use the preset threshold to filter the abnormal proportion, and determine the abnormal data subset; obtain the abnormal data subset, combined with the dynamic change trend, analyze the fluctuation law of the parameter sequence in different time periods, use statistical tools to quantitatively process the fluctuation dynamics, and obtain the fluctuation feature set; through the fluctuation feature set, aiming at the distribution mode of frequency distribution, construct the corresponding storage unit allocation scheme, if the abnormal proportion of a certain distribution mode exceeds the preset threshold, mark the data under this mode with priority, and determine the marked data unit; according to the marked data unit, combined with the mapping logic of attribute information and spatiotemporal attribute, analyze the rationality of data classification, use the data organization framework to optimize and adjust the classification result, and obtain the optimized classification structure; obtain the optimized classification structure, aiming at the allocation of storage units, analyze the distribution balance of the parameter sequence under different classifications, if the distribution balance under a certain classification is lower than the preset standard, re-group the data under this classification, and determine the final storage allocation scheme; through the final storage allocation scheme, combined with the fluctuation feature set and the dynamic change trend, continuously track the update state of the parameter sequence, use the automatic tool to dynamically adjust the data classification and storage unit, and obtain the adjusted data management framework.

[0038] In specific implementation, for example, in the scene of monitoring the environmental parameters of power equipment, the operation data of the transformer is analyzed in depth, combined with the data collection and processing background in the historical information, and the smoothed parameter sequence and the subsequent analysis process are focused on.

[0039] For the smoothed parameter sequence combined with the spatiotemporal attribute information to analyze the correlation pattern of deviation persistence and fluctuation dynamics, the change law of temperature data can be identified through the dual dimensions of time period and space position. Assuming that the temperature data of a transformer in different time periods within a day presents periodic fluctuation, combined with the heat dissipation condition difference of geographical position, the persistence characteristics of temperature deviation in a certain time and component position are analyzed.

[0040] For example, for the preliminary grouping of data using the preset classification rule, the data can be divided into normal, high and abnormal three categories according to the temperature range. Assuming that the normal range is 35.0 to 40.0 degrees Celsius, the high is 40.0 to 42.0 degrees Celsius, and the abnormal is above 42.0 degrees Celsius, through this rule, the data in a day is classified to obtain the classified data set.

[0041] A data organization logical framework is constructed for frequency distribution and abnormal proportion characteristics. It is assumed that a time period with an abnormal proportion exceeding 10% is marked as a high-risk time period. A preset threshold is used to screen an abnormal data subset. For example, if the abnormal proportion of a certain time period reaches 15%, it is classified as a key focus object.

[0042] For example, when analyzing the fluctuation rules of the abnormal data subset, the fluctuation dynamics can be quantified by statistical tools. It is assumed that the temperature data in a certain time period frequently fluctuates between 41.0 and 43.0 degrees Celsius. After quantification, it is found that the fluctuation frequency is 5 times per hour, with 2 times exceeding the normal range, and a fluctuation feature set is obtained. A storage unit allocation scheme is constructed for the frequency distribution pattern. If the abnormal proportion of a certain pattern exceeds the threshold of 15%, the data of the pattern is marked as a priority. For example, the data unit of the high-load time period is marked as priority 1 to ensure that it is focused on in subsequent analysis.

[0043] For example, combined with the labeled data unit and the spatiotemporal attribute mapping logic optimization classification structure, the classification weight can be adjusted according to the location of the device component. It is assumed that the component data near the weak heat dissipation area is given higher attention, and the optimized classification structure is more consistent with the actual operating environment.

[0044] The distribution balance of the storage unit allocation is analyzed. If the data distribution under a certain category is uneven, for example, the data volume of a certain category accounts for only 5%, which is lower than the preset standard of 10%, the data is re-grouped to ensure that the storage scheme is reasonable. For example, by continuously tracking the parameter sequence update state through the final storage allocation scheme, it is assumed that the temperature data is continuously high during the high-load period. An automatic tool is used to dynamically adjust the classification and storage unit, for example, the data of the period is re-allocated to the high-risk category to form an adjusted data management framework. This process ensures the real-time and pertinence of data analysis, providing a reliable basis for subsequent device maintenance.

[0045] Further, step 3 is as follows: Based on the parameter data in the initial storage container, the calculation results of the spatial correlation index and the consistency index, the fluctuation of the trend change is analyzed. If the trend change exceeds the preset threshold, the data in the storage container is prioritized to obtain a sorted data set; According to the sorted data set, the time-related evaluation result is obtained. If the time correlation is lower than the predetermined standard, the data set is processed in layers to determine the data grouping after layering; The data grouping after layering is obtained. The dynamic adjustment allocation logic is analyzed for the capacity limit of the storage container. If the data volume of a certain grouping exceeds the upper limit of the container, the grouping is split to obtain a split data unit; Through the split data units, combined with the spatial correlation distribution pattern, the update rule of the storage framework is constructed, the data units are repositioned using the preset allocation strategy, and the updated storage layout is determined; The updated storage layout is obtained, the balance state of the data units in the storage framework is analyzed according to the stability requirement of the consistency indicator, if the balance state does not reach the predetermined standard, a secondary adjustment mechanism is triggered, and an optimized storage structure is obtained; According to the optimized storage structure, combined with the execution record of dynamic adjustment, the subsequent influence of trend change is continuously tracked, and the storage framework is monitored in real time using an automatic tool to determine the final storage configuration; Through the final storage configuration, a long-term management scheme for data storage is constructed according to the persistence of the adjustment result, a log recording tool is used to track the changes of the storage framework, and a complete change archive is obtained.

[0046] In the scenario of box transformer environmental parameter monitoring, for the processing and analysis of transformer temperature data, the system first extracts parameter data from the initial storage container, for example, obtains the temperature data of a certain box transformer in the past 24 hours, the data points are collected every minute, a total of 1440 data points, and the temperature range is between 40.0 and 85.0 degrees Celsius.

[0047] Then, the spatial correlation index is calculated by algorithm, the Pearson correlation coefficient method is used to analyze the correlation between temperature data of different components at different positions, assuming that the calculation result shows that the correlation coefficient of component A and component B is 0.85, which is higher than the preset threshold value 0.7, indicating that there is strong spatial correlation between the two. At the same time, the consistency index is calculated, the standard deviation is used to analyze the stability of the temperature data, if the standard deviation is 2.5°C, which is lower than the preset standard 3.0°C, it is considered that the data consistency is high. Then, the system detects the trend change, uses the linear regression algorithm to predict the temperature change rate, if the prediction result shows that it increases by 6.0°C per hour, which exceeds the preset condition of 4.0°C per hour, it is determined that the trend is abnormal. In addition, the time correlation is calculated, based on the autocorrelation function to analyze the time sequence dependence between data points, if the correlation coefficient is 0.4, which is lower than the standard value 0.6, it indicates that the time correlation is insufficient.

[0048] Under the condition of triggering the above, the system automatically starts the dynamic adjustment mechanism of the storage container, reassigns the abnormal trend data to the high priority storage unit through the preset rule, for example, marks the period when the proportion of data points whose temperature exceeds 75.0°C reaches 20% as an abnormal period, stores it separately, and adjusts the storage framework. The data of this period is associated with the historical similar abnormal data to form a new storage structure. In order to ensure the logical integrity, the box transformer load rate data is also used as an auxiliary business association, if the load rate exceeds 80% of the rated value, the storage priority of the data in this period is further improved to ensure the comprehensiveness of subsequent analysis.

[0049] Further, step 4 is specifically as follows: Through the adjusted storage framework, the relevant historical parameter data of the target component is obtained, the data is preliminarily classified by using a preset screening rule, and a classified parameter set is obtained; According to the classified parameter set, the peak value difference and the reference value are compared, if the peak value difference exceeds the preset threshold, the part of data whose peak value difference exceeds the threshold is marked, and an abnormal data set after marking is determined; Starting from the abnormal data set after marking, the data is prioritized by combining the logic of multi-dimensional sorting, and a data sequence after arrangement is obtained; For the data sequence after arrangement, the specific distribution of the abnormal position is analyzed, the abnormal position is accurately identified by using a preset positioning tool, and the specific abnormal coordinate point is determined; Through the abnormal coordinate point, the time correlation information in the historical parameters is traced by combining the record data of the time node, and the corresponding time period distribution is obtained; According to the time period distribution, the corresponding relationship between the abnormal position and the time node is constructed according to the result of difference analysis, and the final abnormal distribution mode is determined; The final abnormal distribution mode is obtained, the parameter optimization strategy for the target component is generated by combining the adjustment structure of the storage framework, and the direction of subsequent processing is determined.

[0050] In the scene of power equipment environmental parameter monitoring, for the operation state analysis of the key components of the generator, first, the historical parameter records of the target component are extracted from the adjusted storage framework, for example, the temperature data of the main shaft of a certain generator in the past 7 days is obtained, which is collected once an hour, a total of 168 data points, and the temperature range is between 40.5 and 75.8 degrees Celsius.

[0051] For these data, the peak value comparison is carried out, the maximum value detection algorithm is used, the temperature peak value data is selected, it is assumed that the highest temperature is 75.8 degrees Celsius, which appears at 5 days and 12 hours, and it is compared with the historical peak value record, if the historical average peak value is 70.0 degrees Celsius, the current peak value exceeds the normal range by 5.8 degrees Celsius, and it is marked as a potential abnormal point.

[0052] The benchmark comparison analysis is carried out, the mean plus standard deviation method is used to set the benchmark line, the calculation result is the benchmark temperature 55.0 degrees Celsius plus the standard deviation 8.0 degrees Celsius, the upper limit is 63.0 degrees Celsius, and there are 30 data points in the current data that exceed this upper limit, accounting for about 17.9%, indicating that there is abnormal fluctuation.

[0053] Through the multi-dimensional sorting algorithm, the temperature data is sorted in three dimensions of time, numerical size and deviation from the benchmark. The weighted scoring method is adopted, the time dimension weight is 0.3, the numerical dimension weight is 0.5, and the deviation dimension weight is 0.2. The calculation shows that the abnormal time period is from the 4th day to the 5th day, and the comprehensive score of the 12th hour of the 5th day is 0.92, which is much higher than the threshold value 0.6, and is determined as a key abnormal time point.

[0054] In combination with the generator speed data as a business association, if the speed of this period exceeds the rated value of 90%, for example, reaches 3200 revolutions per minute, and the rated value is 3000 revolutions per minute, the abnormal importance of this time point is further confirmed.

[0055] Based on spatial distribution analysis, the potential abnormal position is determined. Through historical data comparison, it is found that the proportion of temperature sensor data of the front end of the main shaft is 25%, which is higher than that of other positions by 10%. It is inferred that the abnormality may be concentrated in the front end of the main shaft, forming a complete analysis chain.

[0056] Further, step 5 is specifically as follows: For the monitoring of the change rate, the rate change data of the target object is obtained from the historical data record, and a preliminary comparison is made in combination with the preset threshold value; If the rate change exceeds the threshold range, an initial abnormal signal is generated, and a preliminary focus point is determined; According to the initial abnormal signal, data extraction is performed for the abnormal position, and a preset positioning tool is used to refine the position information to obtain specific abnormal coordinate distribution; Through the abnormal coordinate distribution, in combination with the boundary data of the distribution range, it is analyzed whether there is a significant deviation; If the deviation phenomenon is detected, the corresponding deviation identifier is generated, and the distribution characteristics of the abnormality are determined; For the deviation identifier, relevant trigger condition data is obtained, and a comprehensive treatment is performed in combination with the preset rules of risk assessment to obtain the classification result of the risk level; According to the classification result of the risk level, a corresponding alarm sequence is generated, and a preset priority rule is used to sort the alarms to determine the output order of the alarms; Through the output order of the alarm, in combination with the record data of the preliminary mark, the abnormal position is confirmed again to obtain the final abnormal mark set; For the final abnormal mark set, a logistic regression model is used to analyze the relevance of the abnormal marks to determine the potential relationship between the abnormal marks and determine the key direction of subsequent processing.

[0057] In the scenario of power equipment environmental parameter monitoring, in-depth analysis is conducted on the potential abnormal position of the key components of the box transformer. First, focus on the abnormal area of the transformer winding that has been identified, calculate the temperature change rate index. The specific method is to collect temperature data every minute in the past 5 days, a total of 7200 data points, the temperature range is between 40.0 and 85.0 degrees Celsius. Calculate the average change rate per hour through the sliding window algorithm, the window size is set to 60 minutes. It is found that the change rate of the 15th hour of the 3rd day is 0.35 degrees Celsius per hour, which is significantly higher than the threshold condition of the historical average change rate of 0.15 degrees Celsius per hour.

[0058] Combined with the evaluation of the trigger condition, the change rate threshold is set to 0.2 degrees Celsius per hour, and whether the distribution boundary is broken is analyzed. The 95% quantile of the temperature is calculated as 75.0 degrees Celsius using the statistical quantile method, and 12% of the data points in the current data exceed this boundary, indicating that the distribution is abnormal.

[0059] Further introduce risk assessment factors, call the historical fault database, and combine the correlation model of temperature and equipment aging to calculate the risk index corresponding to the current temperature change rate as 0.78, which is higher than the safety threshold of 0.5. At the same time, associate business data such as cooling fan speed, and find that the speed value in this period drops to 1200 revolutions per minute, which is lower than the normal range of 1500 revolutions per minute, which strengthens the reliability of risk assessment.

[0060] If the change rate is higher than the condition and the distribution boundary is broken, an alarm output sequence is automatically generated, including timestamp, 3rd day 15th hour, temperature peak 85.0 degrees Celsius, and risk index 0.78, etc. information, and through the priority sorting algorithm, the alarm is marked as high priority, the preliminary abnormal mark is determined, and pushed to the subsequent analysis module.

[0061] Further, step 6 is as follows: Through the comparison of the preliminary mark with the historical parameters, the difference information between the mark data and the parameter record is obtained, and the preset threshold range is used for preliminary screening to obtain the mark set with significant differences; According to the mark set with significant differences, data extraction is performed on the abnormal proportion, and weighted calculation is performed combined with the preset rules of influence weight to determine the weighted value of the abnormal proportion; Through the weighted value of the abnormal proportion, the data record related to the alarm output is obtained, if the weighted value exceeds the preset threshold range, the corresponding alarm signal is generated, and the priority of the alarm signal is judged; According to the priority of the alarm signal, combined with the logical rules of proportion evaluation, the signal data is processed in layers to obtain the signal classification after layering; Through the classification of the layered signals, in view of the data requirements of the early warning grading, a logistic regression model is used to map the classified data, and the preliminary result of the early warning grading is determined; According to the preliminary result of the early warning grading, combined with the processing rules of accurate verification, the graded data is compared again to obtain the final output result; Through the final output result, in view of the business logic of the grading determination, the result data is matched with the preset grading standard to determine the final early warning grading state.

[0062] In the scene of power equipment environmental parameter monitoring, for the abnormal detection of key parts of power generation machines, first compare the preliminary abnormal mark with the historical parameter record. The specific method is to extract the temperature data record of the same type of equipment in the past 30 days, a total of about 43200 data points, the temperature range is between 45.2 and 78.9 degrees Celsius. Through the time series comparison algorithm, the deviation of the current abnormal mark corresponding temperature fluctuation value and historical mean value is calculated. It is found that the temperature peak value of the current mark period is 85.0 degrees Celsius, which is 15.0 units higher than the historical mean value of 70.0 degrees Celsius, and the deviation rate is 21.4%, which is higher than the set deviation threshold of 15%. Higher than the set deviation threshold of 8%.

[0063] Fusion influence weight analysis, using a multi-factor weighted model, taking temperature fluctuation, running time and load rate as the main influencing factors, respectively assigning weights of 0.5, 0.3 and 0.2, and calculating the comprehensive influence index as 0.82, which is higher than the safety threshold of 0.6, indicating that the current abnormality has a high degree of influence.

[0064] Abnormal proportion evaluation, statistics of the occurrence frequency of similar temperature abnormalities in the past 7 days, found that the abnormal proportion is 9.5%, higher than the normal range of 5%, and combined with the correlation model of abnormal proportion and failure rate in historical data, the probability of current abnormal development into failure is 0.67, which is higher than the warning value of 0.4.

[0065] In order to further verify the accuracy of the alarm output, the cooling liquid flow record in the business data is associated, it is found that the current period flow value is 3.1 liters per minute, which is lower than the standard value of 4.0 liters per minute, which proves the potential risk of temperature abnormality.

[0066] Based on the above analysis results, the system classifies the early warning by a hierarchical algorithm, sets the early warning classification result as a secondary early warning, generates an early warning report containing a timestamp, a temperature peak of 76.3 degrees Celsius, a comprehensive influence index of 0.82, and a failure probability of 0.67, and automatically pushes the early warning report to the monitoring platform to form a complete logical chain from comparison, analysis to classification. The logistic regression model is trained using historical operation and maintenance data, and the feature engineering includes temperature deviation (difference between current value and historical average), change rate (sliding window calculation), load rate (current load to rated value), and environmental temperature compensation. The label is the actual failure state (0-normal, 1-abnormal, 2-failure). The training sample is based on 2000 historical operation records, covering normal, abnormal and failure states, and the model accuracy is ensured to be not less than 80% through 5-fold cross-validation. The model is retrained every quarter using the latest data to maintain its effectiveness.

[0067] Further, step 7 is specifically as follows: Through the final early warning classification result, parameter summary data containing trend changes and spatial consistency are sorted out, data is sent to the control system, and an interaction log after sending is obtained; According to the interaction log after sending, response feedback data returned by the control system is extracted, data is analyzed using a preset field rule, and the analyzed feedback content is determined; Through the analyzed feedback content, the key fields in the feedback record are classified and processed, the classified data groups are obtained, and the integrity of the grouped data is determined; If the integrity of the grouped data meets the preset threshold range, the classified data groups are prioritized, and a feedback sequence after sorting is obtained; According to the feedback sequence after sorting, data is filtered according to the demand of result analysis, a logistic regression model is used to predict the trend of the filtered data, and the predicted trend direction is determined; Through the predicted trend direction, the parameter summary data of spatial consistency are combined and processed, the matching result after comparison is obtained, and the correlation degree of the matching result is determined; If the correlation degree of the matching result meets the preset standard, the matching result and the response feedback data are integrated and stored to obtain the final business data archive.

[0068] The early warning classification threshold is determined based on the box variable operation safety standard and historical failure statistical analysis: Primary early warning (low risk): trend change rate 20%-35%, spatial consistency 0.6-0.8; Secondary early warning (medium risk): trend change rate 35%-50%, spatial consistency 0.4-0.6; Tertiary early warning (high risk): trend change rate >50%, spatial consistency <0.4; The threshold value is determined by 500 historical fault backtracking tests, with a false positive rate controlled within 5%.

[0069] In the process of realizing the early warning grading result pushing and feedback recording, the data analysis system extracts parameters from the early warning grading result. Assuming that the current early warning grading is level three, the trend change parameter shows that the risk index rises from 2.5 to 3.8 within the last 24 hours, with an increase of 52%, and the calculation formula is (3.8-2.5) / 2.5*100%. At the same time, the spatial consistency parameter shows that the coverage rate of the affected area expands from 30% to 45%, and the consistency index calculated by the spatial analysis algorithm is 0.75, indicating that the regional risk distribution is relatively concentrated.

[0070] The above parameters are summarized as structured information, including trend change rate 52% and spatial consistency index 0.75. The automatic generation of push messages is sent to the control system in JSON format through the API interface. The message example is {“level”:3,“trend”:52, “spatial”:0.75}. The push process records the timestamp and sending state to ensure traceability.

[0071] After receiving the message, the parameters are automatically parsed based on the preset rules. If the trend change rate is greater than 50% and the consistency index is greater than 0.7, the secondary response mechanism is triggered to generate a response instruction, such as adjusting the monitoring frequency to once an hour, and storing the response record in the database, including the response level, trigger time and execution state. Assuming that the record ID is 20231001001, the time is 2023-10-01 14:30:00.

[0072] The feedback record is queried once every 30 minutes by a timing task to analyze the response execution rate. Assuming that the current execution rate is 95%, if it is lower than 98%, an optimization suggestion is automatically generated, such as increasing the resource allocation ratio to 20%, and the analysis result is pushed to the management platform to form a closed-loop management logic to ensure the efficient connection of early warning and response.

[0073] Embodiment 2 discloses a real-time analysis and fault early warning system based on digital electric port box transformer data, including a data acquisition module, a data processing module, a storage management module, an anomaly detection module, an early warning generation module and a control interface module.

[0074] In specific implementation: The data acquisition module is composed of temperature sensors and humidity sensors arranged on each component of the box transformer. It collects environmental parameter data in real time through a ZigBee wireless network, with a sampling frequency of once per minute. For example, in the transformer box, the sensor node collects temperature data ranging from -10°C to 100°C with an accuracy of ±0.5°C, and humidity data ranging from 0% to 100% with an accuracy of ±2%. The sensor network covers key components of the box transformer such as transformer windings, switch cabinets, and cable joints, ensuring comprehensive data.

[0075] The data processing module is embedded in the edge computing gateway and uses a moving average algorithm to smooth the collected raw data. The moving average window size is set to 5 data points (corresponding to 5 minutes) based on the thermal time constant of the box transformer, ensuring that both random noise (such as electromagnetic interference) is effectively filtered out and the true temperature trend changes are preserved. For example, for the temperature data sequence [38.2, 39.1, 45.3, 38.9, 40.2] °C, taking a 5-point moving average gives the smoothed sequence [38.8, 41.1, 41.5] °C, effectively eliminating sudden disturbances. The smoothed parameter sequence is uploaded to the cloud platform through the gateway, reducing network transmission load.

[0076] The storage management module is based on the cloud platform and uses the time series database InfluxDB to store the smoothed parameter sequence. It automatically partitions based on spatiotemporal attributes (such as device location coordinates and data timestamps) and sets an abnormality proportion threshold (such as 10%). When the abnormality proportion exceeds the threshold, the storage partition strategy is automatically adjusted, storing abnormal data in high-priority partitions to improve retrieval efficiency.

[0077] The anomaly detection module runs on the server side, extracts the historical parameter records of the target component from the storage management module, and uses a multi-dimensional sorting algorithm (such as a weighted sorting based on peak value, baseline deviation, and time) to identify potential abnormal positions. For example, for temperature data, the rate of change is calculated, and if the rate of change exceeds 0.2°C / min and the distribution boundary (such as the historical 95% quantile) is broken, it is marked as a preliminary anomaly.

[0078] The early warning generation module receives the output of the anomaly detection module, verifies and grades the anomaly labels using a pre-trained logistic regression model. The model input features include: temperature deviation value, rate of change, load rate, spatial consistency index; the output is a three-level warning probability. Decision rule: trigger level 3 warning when high risk probability > 0.7, trigger level 2 warning when medium risk probability > 0.6, trigger level 1 warning when low risk probability > 0.5. The model is automatically updated every quarter, retrained using the last 3 months of operation data. For example, calculate the comprehensive risk index according to the influence weight (such as temperature deviation weight 0.6, humidity deviation weight 0.4), divide the warning into three levels: level 1 (low risk), level 2 (medium risk), level 3 (high risk). The alarm sequence is sorted by priority and pushed to the control interface module.

[0079] The control interface module integrates with the upper control system (such as power distribution automation system) through RESTful API. The pushed parameter summary information includes trend change rate, spatial consistency index and warning level, in JSON format. For example, the push message: {"warning_level": 2, "trend_change": 15%, "spatial_consistency": 0.8}. The control system returns a response feedback, such as adjusting the cooling system operating parameters, and the system records the feedback log for closed-loop optimization.

[0080] The embodiment realizes real-time analysis of box transformer data, fault early warning and automatic control response through modular design. The data flow between system modules is as follows: the data acquisition module sends the original environmental parameters to the edge computing gateway (data processing module) through the ZigBee network; the data processing module uploads the parameter sequence to the cloud platform storage management module after smoothing and denoising; the storage management module dynamically partitions the data and provides historical records to the anomaly detection module; the anomaly detection module outputs preliminary anomaly labels to the early warning generation module; the early warning generation module verifies and grades, and pushes the early warning results to the control system through the control interface module. The feedback returned by the control system (such as adjusting the cooling system parameters) is parsed by the control interface module and stored in the business database for optimizing subsequent early warning strategies, solving the control lag problem caused by data noise and delay, and improving the timeliness, reliability and automation level of box transformer control.

[0081] Although the preferred embodiments of the present application have been described, those skilled in the art can make further changes and modifications to these embodiments once they know the basic inventive concept. Therefore, the appended claims are intended to be interpreted as including all changes and modifications falling within the scope of the present application.

[0082] The above merely describes preferred embodiments of the present application and is not used to limit the present application, and it should be pointed out that any modification, equivalent replacement and improvement within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for real-time analysis and fault warning based on digital electric port box variable data, characterized in that, Specifically comprising the following steps: Step 1: Collecting environmental parameter data from each component of the power equipment in real time through the monitoring device network, preliminarily screening for parameter deviation indicators and distribution balance states, using data smoothing methods to denoise the collected data, and obtaining smoothed parameter sequences; Step 2: According to the smoothed parameter sequences, combined with the spatio-temporal attribute information, analyze the deviation persistence characteristics and fluctuation dynamic indicators, construct the data organization structure, group the environmental parameter data according to the frequency distribution and abnormal proportion indicators, and determine the initial storage container; Step 3: Obtain the parameter data in the initial storage container, calculate the spatial correlation indicators and consistency indicators, if the trend change exceeds the preset condition or the time correlation is lower than the standard, trigger the dynamic adjustment of the initial storage container, and obtain the adjusted storage framework; Step 4: Extract the historical parameter records of the target component from the adjusted storage framework, perform multi-dimensional sorting for peak value comparison and benchmark comparison, and judge the potential abnormal position and time information; Step 5: For the potential abnormal position, calculate the change rate indicator and combine the trigger condition and risk evaluation factors for evaluation, if the change rate is higher than the condition and the distribution boundary is broken, generate an alarm output sequence, and determine the preliminary abnormality mark; Step 6: Through the comparison of the preliminary abnormality mark and the historical parameter record, fuse the influence weight analysis and the abnormal proportion evaluation, verify the accuracy of the alarm output, and obtain the final warning grading result; Step 7: According to the final warning grading result, push the parameter summary information containing trend change and spatial consistency to the control system, and obtain the response feedback record.

2. The method according to claim 1, wherein, Step 1 is as follows: Obtain real-time data of environmental parameters from each component of the power equipment through the monitoring device network, continuously record using a preset collection frequency, obtain an initial environmental parameter data set; for the initial environmental parameter data set, use a data smoothing method to denoise the data, eliminate noise interference, and obtain smoothed parameter sequences; according to the smoothed parameter sequences, analyze the specific situation of parameter deviation, if a parameter value exceeds the preset threshold range, mark it as an abnormal parameter point, obtain a marked abnormal data set; through the marked abnormal data set, extract feature information of the distribution balance state, judge whether there is an uneven distribution phenomenon, obtain an evaluation result of the distribution state; according to the evaluation result of the distribution state, if an uneven distribution phenomenon is found, adjust the weight of the abnormal parameter point, determine an adjusted parameter distribution set; obtain the adjusted parameter distribution set, combine the update frequency of real-time data, continuously monitor the change trend of the parameter sequence, judge whether there is a potential abnormal fluctuation, and obtain the final monitoring analysis result; Through the final monitoring analysis result, generate a corresponding parameter adjustment strategy, dynamically optimize the running state of the power equipment, and obtain the optimized running parameter configuration.

3. The method according to claim 1, wherein, Step 2 is as follows: through the smoothed parameter sequence, combined with the spatiotemporal attribute information, the correlation pattern of the deviation duration and fluctuation dynamics is analyzed, the data is preliminarily grouped using the preset classification rule, and the classified data set is obtained; according to the classified data set, the logical framework of data organization is constructed according to the characteristics of frequency distribution and abnormal proportion, the preset threshold is used to screen the abnormal proportion, and the abnormal data subset is determined; the abnormal data subset is obtained, combined with the dynamic change trend, the fluctuation law of the parameter sequence in different time periods is analyzed, the fluctuation dynamics is quantitatively processed using statistical tools, and the fluctuation feature set is obtained; through the fluctuation feature set, the corresponding storage unit allocation scheme is constructed according to the distribution mode of the frequency distribution, if the abnormal proportion of a certain distribution mode exceeds the preset threshold, the data under the mode is marked with priority, and the marked data unit is determined; according to the marked data unit, combined with the mapping logic of attribute information and spatiotemporal attributes, the rationality of data classification is analyzed, the classification result is optimized and adjusted using the data organization framework, and the optimized classification structure is obtained; the optimized classification structure is obtained, and the distribution balance of the parameter sequence under different classifications is analyzed according to the allocation of the storage unit, if the distribution balance under a certain classification is lower than the preset standard, the data under the classification is re-grouped, and the final storage allocation scheme is determined; Through the final storage allocation scheme, combined with the fluctuation feature set and the dynamic change trend, the update state of the parameter sequence is continuously tracked, the data classification and storage unit are dynamically adjusted using automatic tools, and the adjusted data management framework is obtained.

4. The method according to claim 1, wherein, Step 3 is as follows: Through the parameter data in the initial storage container, the fluctuation of the trend change is analyzed according to the calculation results of the spatial correlation index and the consistency index, if the trend change exceeds the preset threshold, the data in the storage container is sorted by priority, and the sorted data set is obtained; According to the sorted data set, the time-related evaluation result is obtained, if the time correlation is lower than the predetermined standard, the data set is processed in layers, and the layered data grouping is determined; The layered data grouping is obtained, the allocation logic of dynamic adjustment is analyzed according to the capacity limit of the storage container, if the data volume of a certain grouping exceeds the upper limit of the container, the grouping is split, and the split data unit is obtained; Through the split data unit, combined with the distribution mode of the spatial correlation, the update rule of the storage framework is constructed, the data unit is repositioned using the preset allocation strategy, and the updated storage layout is determined; The updated storage layout is obtained, the balance state of the data unit in the storage framework is analyzed according to the stability requirement of the consistency index, if the balance state does not reach the predetermined standard, the secondary adjustment mechanism is triggered, and the optimized storage structure is obtained; According to the optimized storage structure, combined with the execution record of dynamic adjustment, the subsequent influence of the trend change is continuously tracked, the storage framework is monitored in real time using automatic tools, and the final storage configuration is determined; Through the final storage configuration, a long-term management scheme of the data storage is constructed for the persistence of the adjustment result, a log recording tool is used to track the changes of the storage framework, and a complete change archive is obtained.

5. The method according to claim 1, wherein, Step 4 is specifically as follows: Through the adjusted storage framework, relevant historical parameter data of the target component is obtained, a preset screening rule is used to preliminarily classify the data, and a classified parameter set is obtained; According to the classified parameter set, the peak difference and the reference value are compared, if the peak difference exceeds the preset threshold, the part of data whose peak difference exceeds the threshold is marked, and a marked abnormal data set is determined; Starting from the marked abnormal data set, combined with the logic of multi-dimensional sorting, the data is prioritized to obtain a sorted data sequence; For the sorted data sequence, the specific distribution of the abnormal position is analyzed, a preset positioning tool is used to accurately identify the abnormal position, and the specific abnormal coordinate point is determined; Through the abnormal coordinate point, combined with the record data of the time node, the time correlation information in the historical parameter is traced back, and the corresponding time period distribution is obtained; According to the time period distribution, the corresponding relationship between the abnormal position and the time node is constructed according to the result of difference analysis, and the final abnormal distribution mode is determined; Obtain the final abnormal distribution mode, combine the adjustment structure of the storage framework, generate the parameter optimization strategy for the target component, and determine the direction of subsequent processing.

6. The method for real-time analysis and fault early warning based on digital electric port box variable data according to claim 1, characterized in that, Step 5 is specifically as follows: For the monitoring of the change rate, the rate change data of the target object is obtained from the historical data record, and a preliminary comparison is made combined with the preset threshold; If the rate change exceeds the threshold range, an initial abnormal signal is generated, and a preliminary focus point is determined; According to the initial abnormal signal, data extraction is performed on the abnormal position, and a preset positioning tool is used to refine the position information to obtain a specific abnormal coordinate distribution; Through the abnormal coordinate distribution, combined with the boundary data of the distribution range, whether there is a significant deviation is analyzed; If the deviation phenomenon is detected, a corresponding deviation identifier is generated, and the distribution characteristics of the abnormality are determined; For the deviation identifier, the related trigger condition data is obtained, and a preset rule of risk assessment is used for comprehensive processing to obtain a classification result of risk level; According to the classification result of risk level, a corresponding alarm sequence is generated, a preset priority rule is used to sort the alarm, and the output order of the alarm is determined; Through the output order of the alarm, combined with the preliminary marked record data, the abnormal position is confirmed again to obtain a final abnormal marker set; For the final abnormal marker set, a logistic regression model is used to analyze the relevance of the abnormal markers, the potential relationship between the abnormal markers is determined, and the key direction of subsequent processing is determined.

7. The method according to claim 1, wherein, Step 6 is specifically as follows: Through the comparison of the preliminary marker and the historical parameter, the difference information between the marker data and the parameter record is obtained, a preset threshold range is used for preliminary screening, and a difference significant marker set is obtained; According to the difference significant marker set, data extraction is performed on the abnormal proportion, and a preset rule of influence weight is used for weighted calculation to determine the weighted value of the abnormal proportion; The data records related to the alarm output are obtained through the abnormally proportioned weighted values, and if the weighted values exceed a preset threshold range, corresponding alarm signals are generated to determine the priority of the alarm signals; According to the priority of the alarm signals, combined with the logical rules of proportion evaluation, the signal data is processed in layers to obtain the signal classification after layering; Through the signal classification after layering, a logic regression model is used to map and process the classified data according to the data requirements of early warning classification to determine the preliminary results of early warning classification; According to the preliminary results of early warning classification, combined with the processing rules of accurate verification, the classified data is compared twice to obtain the final output results; Through the final output results, the result data is matched with the preset classification standard according to the business logic of classification judgment to determine the final early warning classification state.

8. The method for real-time analysis and fault early warning based on digital electric port box variable data according to claim 1, characterized in that, Step 7 is as follows: Through the final early warning classification results, parameter summary data containing trend changes and spatial consistency are sorted out, data is sent to the control system to obtain the interactive log after sending; According to the interactive log after sending, the response feedback data returned by the control system is extracted, and the feedback content after data analysis is determined by using the preset field rules; Through the feedback content after analysis, the key fields in the feedback record are classified and processed to obtain the data grouping after classification, and the integrity of the grouped data is determined; If the integrity of the grouped data meets the preset threshold range, the data grouping after classification is prioritized to obtain the feedback sequence after sorting; According to the feedback sequence after sorting, data filtering is performed according to the requirements of result analysis, and a logic regression model is used to predict the trend of the filtered data to determine the trend after prediction; Through the trend after prediction, combined with the parameter summary data of spatial consistency, the matching results after comparison are obtained, and the correlation degree of the matching results is determined; If the correlation degree of the matching results meets the preset standard, the matching results and the response feedback data are integrated and stored to obtain the final business data archive.

9. A system for real-time analysis and fault warning based on digital electric port box substation data, used for performing the method for real-time analysis and fault warning based on digital electric port box substation data according to any one of claims 1-8; characterized in that, It includes: A data acquisition module for acquiring environmental parameter data from each component of the power equipment in real time through a monitoring device network; A data processing module for smoothing and denoising the collected data to obtain a smoothed parameter sequence; A storage management module for constructing a dynamic storage framework based on the smoothed parameter sequence and spatiotemporal attribute information; An anomaly detection module for extracting historical parameter records of the target component from the adjusted storage framework, performing multi-dimensional sorting and change rate calculation, and identifying potential anomalies; An early warning generation module for generating an alarm sequence based on the change rate index and distribution boundary breakthrough, and verifying the accuracy by comparing with the historical parameter records to obtain early warning classification results; A control interface module for pushing parameter summary information to the control system according to the early warning classification results and obtaining response feedback.

10. The system according to claim 9, wherein the system further comprises a data analysis module. The data acquisition module includes a plurality of sensor nodes arranged in each component of the box transformer for collecting temperature and humidity environmental parameters with a sampling frequency of once per minute, and the data is transmitted through a ZigBee wireless network.

Images