A Method for Evaluating the Maintenance Effectiveness of Transformer Cooling Systems Based on Big Data from Maintenance Inspections

Abstract

This invention discloses a method for evaluating the maintenance effect of transformer cooling systems based on maintenance big data, belonging to the field of industrial big data technology. The technical solution is as follows: Acquire comprehensive maintenance operation data; anchor maintenance events at the maintenance completion time and divide observation windows before and after maintenance; perform condition comparability regularization on the temperature control response condition boundary to form equivalent operating condition sample pairs and a set of condition deviation samples; extract cooling execution trajectories from the equivalent operating condition sample pairs, perform cooling execution burden decomposition, obtain cooling control input characterization quantities, and mark cooling execution interference; convert the temperature operating state into a temperature rise response, forming a unit temperature control benefit improvement result and a temperature benchmark interference set; then construct a cooling group role transfer relationship diagram, perform benefit attribution correction, generate a net contribution result for maintenance effect, and output the maintenance effect evaluation result.

Description

Technical Field

[0001] This invention relates to the field of industrial big data technology, specifically to a method for evaluating the maintenance effect of transformer cooling systems based on maintenance big data. Background Technology

[0002] Transformer cooling systems typically consist of fans, oil pumps, radiators, cooler assemblies, temperature acquisition circuits, and cooling control circuits. Their operation directly affects the top-layer oil temperature, winding temperature, and the transformer's long-term load capacity. In actual operation and maintenance scenarios, the cooling system does not operate continuously in a fixed state but automatically engages and disengages based on load levels, ambient temperature, temperature thresholds, and substation control strategies. When high summer temperatures, short-term overloads, or adjustments to the operating mode occur, the control system maintains the temperature within the allowable range by increasing the number of fans, oil pumps, or cooler assemblies. Maintenance work also involves a combination of operations. A single overhaul may involve radiator cleaning, fan bearing treatment, oil pump test run, control cabinet wiring tightening, temperature sensor calibration, and standby cooler group switching inspection. Taking the main transformer summer overhaul as an example, the top oil temperature may remain in the same temperature control range before and after the overhaul, but the number of cooler groups in operation, the frequency of fan start-up and shutdown, the continuous operating time of the oil pump, and the records of manual forced switching have changed after the overhaul. At the same time, the load, environmental heat dissipation conditions, daily cycle operating status, and available cooling resources may also be different before and after the overhaul, making it difficult to directly compare the operating samples before and after the overhaul.

[0003] Current evaluations of transformer cooling system maintenance effectiveness largely rely on closed-loop maintenance work orders, the number of temperature alarms, post-maintenance trial operation results, and comparisons of temperature curves before and after maintenance. While these methods meet general acceptance record requirements, they struggle to identify the true maintenance benefits after automatic control intervention. This is because temperature results are influenced by heat load, environmental heat dissipation conditions, cooling capacity, and control input. When cooling efficiency decreases, the system may still maintain temperature stability by investing more cooling resources, preventing the temperature curve from showing anomalies. Furthermore, the actual benefits after maintenance may manifest as adjustments to cooling workload, changes in standby group participation status, disappearance of forced switching requirements, or restoration of consistent command feedback. The changes in the temperature of the cooler group under maintenance may be manifested in the status of the standby cooler group, the rotating cooler group, or the control loop, rather than directly reflected in the temperature index changes of the cooler group under maintenance itself. If we continue to use the decrease in temperature or the reduction in alarms as the main evaluation criteria, it is easy to mistake the temperature changes caused by the reduction in load or the cooling of the environment as maintenance benefits. It is also easy to deny effective maintenance when the temperature change is not significant, which will affect the subsequent maintenance cycle, spare parts configuration, and cooling system status evaluation. Therefore, there is an urgent need for a maintenance effectiveness evaluation method that can combine maintenance operation data, temperature control response condition boundaries, cooling execution trajectory, and cooling group role transfer relationship to identify the actual maintenance contribution formed by maintenance behavior under automatic switching conditions. Summary of the Invention

[0004] To achieve the above objectives, the present invention provides the following technical solution: The method for evaluating the maintenance effectiveness of transformer cooling systems based on big data from maintenance includes the following steps: S1. Obtain comprehensive maintenance and operation data of the transformer cooling system. Anchor the maintenance event at the time of maintenance completion and divide the observation window before and after maintenance. Perform condition comparability normalization on the temperature control response condition boundary that affects the comparability of temperature response within the observation window. Unify the samples before and after maintenance in terms of heat input, heat dissipation boundary, operating cycle and cooling resource conditions to obtain equivalent operating condition sample pairs. Write the samples that do not enter the equivalent operating condition sample pairs and retain the cooling execution data into the condition deviation sample set. S2 extracts the cooling execution trajectory from the equivalent operating condition sample pair, decomposes the cooling execution burden, converts the cooling resource occupation, cooling action maintenance and control action switching into the same evaluation scale, and obtains the cooling control input characterization quantity through the rated capacity calibration of the cooler group. At the same time, the forced intervention trajectory and feedback deviation trajectory are marked as cooling execution interference. S3 converts the temperature operating state in the equivalent operating condition sample pair into a temperature rise response after deducting the influence of the environmental boundary, performs temperature rise response purification, and obtains the unit temperature control benefit improvement result according to the correspondence between the temperature rise response and the cooling control input characterization quantity. Samples lacking temperature reference are written into the temperature reference interference set. S4 generates a role transfer relationship diagram of the cooling group based on the changes in the operating responsibilities before and after the overhaul of the cooling group. It performs a benefit attribution correction on the unit temperature control benefit improvement results, and combines the cooling execution interference, temperature reference interference set and operating condition deviation sample set to form the net contribution result of maintenance effect, and outputs the maintenance effect evaluation result.

[0005] Furthermore, the comprehensive maintenance and operation data should include at least maintenance event data, temperature operation data, load operation data, environmental boundary data, and cooling execution data; maintenance event data should include at least the maintenance object, maintenance action, maintenance completion time, and maintenance conclusion; temperature operation data should include at least the top oil temperature, winding temperature, and cooler outlet temperature; load operation data should include at least the active load, rated load, and operating mode label; environmental boundary data should include at least the ambient temperature and sampling time; and cooling execution data should include at least the cooler group activation / deactivation status, cooling equipment operation status, start / stop records, forced activation / deactivation records, control command issuance status, and actual feedback status. The comprehensive maintenance and operation data establishes data association relationships under the same maintenance event based on the maintenance object and sampling time.

[0006] Furthermore, the anchoring of maintenance events and the division of observation windows include: marking the maintenance completion time as a time anchor point and marking the maintenance object as an object anchor point; retrieving previous and subsequent maintenance events for the same maintenance object based on the object anchor point; extracting continuous operating records before the time anchor point that do not cross previous maintenance events to obtain the pre-maintenance candidate interval; extracting continuous operating records after the time anchor point that do not cross subsequent maintenance events to obtain the post-maintenance candidate interval; determining the window length based on the historical temperature rise recovery process of similar maintenance events or post-maintenance trial operation records; and extracting operating records corresponding to the window length from the pre-maintenance candidate interval and the post-maintenance candidate interval to obtain the pre-maintenance observation window and the post-maintenance observation window.

[0007] Furthermore, the temperature control response condition boundary includes at least the heat input boundary, heat dissipation boundary, operating cycle boundary, and cooling resource boundary; the heat input boundary is formed by load operation data, the heat dissipation boundary is formed by environmental boundary data, the operating cycle boundary is formed by the daily cycle interval to which the sampling time belongs, and the cooling resource boundary is formed by the status of the cooler group that can participate in commissioning and decommissioning. The comparability normalization of operating conditions includes: performing load scale normalization on the heat input boundary to obtain the heat input description result; performing environmental distribution normalization on the heat dissipation boundary to obtain the heat dissipation boundary description result; performing daily cycle alignment on the operating cycle boundary to obtain the cycle operation description result; performing available resource calibration on the cooling resource boundary to obtain the cooling resource description result; and writing the heat input description result, heat dissipation boundary description result, cycle operation description result, and cooling resource description result into the same temperature control response operating condition boundary record.

[0008] Furthermore, the formation of equivalent operating condition sample pairs and operating condition deviation sample sets includes: forming an operating condition matching benchmark using historical non-maintenance operating samples with continuous operating records and no maintenance events or cooling fault alarms; performing first gating on samples in the pre-maintenance and post-maintenance observation windows based on the operating mode label; identifying the source of cooling resource changes based on the cooling resource description results; and when the cooling resource change is formed by the maintenance object in the maintenance event data changing from an abnormal restricted state, fault lockout state, or maintenance isolation state to a state where it can participate in commissioning / decommissioning, writing the corresponding sample into the inspection... Repair the intrinsic resource recovery sample; when the cooling resource change does not correspond to the maintenance object in the maintenance event data, execute the second gating based on the correspondence between the cooling resource description result and the operating condition matching benchmark; based on the heat input description result, heat dissipation boundary description result and cycle operation description result, perform pairing confirmation on the pre-maintenance sample, post-maintenance sample and maintenance intrinsic resource recovery sample after the second gating; write the pre-maintenance sample and post-maintenance sample that pass the pairing confirmation into the equivalent operating condition sample pair; write the sample that does not pass the pairing confirmation and contains cooling execution data into the operating condition deviation sample set.

[0009] Furthermore, the formation of cooling execution trajectory, cooling execution interference, and cooling execution interference rollback results includes: organizing the cooling execution data into chain segments according to the time sequence between the control command issuance state, the cooling equipment operation state, and the actual feedback state; writing cooling execution chain segments with corresponding control commands, cooling equipment operations, and actual feedback states into the complete cooling execution trajectory; writing cooling execution chain segments containing forced switching records into the forced intervention trajectory; writing cooling execution chain segments lacking actual feedback states or whose actual feedback states do not correspond to the cooling equipment operation states into the feedback deviation trajectory; performing source comparison on the forced intervention trajectory and feedback deviation trajectory in the pre-maintenance and post-maintenance observation windows; when a forced intervention trajectory or feedback deviation trajectory exists before maintenance and the corresponding chain segment transitions to the complete cooling execution trajectory after maintenance, a cooling execution interference rollback result is formed; when a forced intervention trajectory or feedback deviation trajectory still exists after maintenance, cooling execution interference is written.

[0010] Furthermore, the breakdown of cooling execution burden includes: extracting the cooling resource occupation process, cooling action maintenance process, and control action switching process from the complete cooling execution trajectory; performing resource occupation calibration on the cooling resource occupation process; performing operation maintenance calibration on the cooling action maintenance process; performing action switching calibration on the control action switching process; writing the resource occupation calibration results, operation maintenance calibration results, and action switching calibration results into the basic cooling execution burden; calibrating the execution capacity of the basic cooling execution burden based on the rated capacity of the cooler group to form a cooling control input characterization quantity; taking cooling execution interference as the source of interference in the net contribution result of maintenance effect, and taking the cooling execution interference rollback result as the basis for forming the forced intervention rollback attribution path.

[0011] Furthermore, the formation of the temperature rise response, temperature rise response purification, and temperature reference interference set includes: determining the target evaluation temperature according to the maintenance object; when the maintenance object is a fan, oil pump, radiator, or cooler assembly, writing the top oil temperature into the main temperature control response and the winding temperature into the verification response; when the maintenance object is a temperature acquisition loop or cooling control loop, writing the top oil temperature response and winding temperature response into the temperature data verification record; performing environmental boundary subtraction processing on the target evaluation temperature based on the environmental boundary data at the same sampling time to obtain the temperature rise response; when the maintenance event data includes temperature sensor verification or replacement actions, correcting the temperature reference before and after maintenance based on the verification record; and writing samples lacking temperature reference basis into the temperature reference interference set.

[0012] Furthermore, the formation of the unit temperature control benefit improvement result includes: within the equivalent operating condition sample pair, writing the temperature rise response and cooling control input characterization quantity into the input response binding segment; identifying the temperature rise decline state after cooling input from the input response binding segment; identifying the temperature rise suppression state under the equivalent operating condition from the input response binding segment; identifying the thermal stability maintenance state within the target control range from the input response binding segment; the target control range is determined by the transformer operating procedures, equipment manufacturer's temperature control setpoint, or station temperature control strategy record; writing the temperature rise decline state, temperature rise suppression state, and thermal stability maintenance state into the temperature response description quantity; writing the temperature response description quantity and cooling control input characterization quantity into the unit temperature control benefit result; and comparing the unit temperature control benefit results in the pre-maintenance observation window and the post-maintenance observation window according to the same equivalent operating condition sample pair to form the unit temperature control benefit improvement result.

[0013] Furthermore, the formation of the cooling group role transfer relationship diagram and the net contribution result of maintenance effect includes: writing the operating roles undertaken by the cooler group in the pre-maintenance observation window and the post-maintenance observation window into the role node; the operating roles include automatic load role, standby role, rotation participation role, abnormal restricted role, and forced entry role; forming role establishment records based on the duration of the role, the number of times the role appears, and the entry ratio of the cooler group in the corresponding observation window; when the role establishment record reaches the role establishment benchmark, the corresponding operating role is written into the role node; the role establishment benchmark is determined by the duration distribution, the number of times the role appears, and the entry ratio distribution of the corresponding operating role in the historical non-maintenance operation samples; writing the changes in the operating roles of the same cooler group between the pre-maintenance observation window and the post-maintenance observation window into the role transfer edge; when the maintained cooler group changes from the abnormal restricted role to the automatic load role or the rotation participation role, the corresponding role transfer edge is written into the maintenance object recovery attribution path; when the standby role changes from participating in cooling load state to waiting for insurance... When the temperature rise response is within the target control range and the status is maintained, the corresponding role transfer edge is written into the backup load attribution path. When the forced input role is transferred to the automatic load role under automatic control or the rotating participation role, or when the forced intervention trajectory or feedback deviation trajectory is transferred into the complete cooling execution trajectory, the corresponding role transfer edge is written into the forced intervention rollback attribution path. Based on the maintenance object recovery attribution path, the backup load attribution path, and the forced intervention rollback attribution path, the benefit attribution correction is performed on the unit temperature control benefit improvement result. Among them, the benefit attribution correction is written into the corresponding attribution path according to the equivalent operating condition sample pair, the cooler group identifier, and the role transfer edge identifier. When a unit temperature control benefit improvement result corresponds to multiple role transfer edges, corresponding attribution records are formed respectively, and the net contribution result of maintenance effect is written according to the role establishment record corresponding to each attribution record. The unit temperature control benefit improvement result after benefit attribution correction, the cooling execution interference, the cooling execution interference rollback result, the temperature reference interference set, and the operating condition deviation sample set are written into the net contribution result of maintenance effect.

[0014] This invention provides a method for evaluating the maintenance effectiveness of transformer cooling systems based on big data from maintenance inspections, which has the following beneficial effects: 1. This invention anchors maintenance events at the time of maintenance completion and performs comparability normalization of temperature control response operating condition boundaries within the pre-maintenance and post-maintenance observation windows, forming equivalent operating condition sample pairs. Simultaneously, samples that do not meet comparability conditions but retain cooling execution data are written into the operating condition deviation sample set. This ensures that subsequent evaluations are based on samples with unified heat input, heat dissipation boundaries, operating cycles, and cooling resource conditions. This eliminates interference from load reduction, environmental cooling, or changes in operating modes on the assessment of maintenance benefits, improving the comparability of samples before and after maintenance. For changes in cooling resources caused by the recovery of the maintained object's state, this invention includes them in the maintenance-endogenous resource recovery sample, avoiding the misidentification of cooling resource recovery caused by maintenance actions as external operating condition deviations. Furthermore, since operating condition deviation samples are not directly discarded but rather used as sources of interference in the subsequent net contribution results of maintenance effectiveness, they also retain abnormal operating segments and prevent excessive screening of evaluation samples, ensuring that subsequent evaluations have both a comparability basis and interference tracing capabilities.

[0015] 2. Compared to methods that judge maintenance effectiveness solely based on the number of temperature alarms, the closed-loop status of maintenance work orders, or the results of post-maintenance trial runs, this invention extracts the cooling execution trajectory within an equivalent operating condition sample pair and decomposes the cooling execution burden on this trajectory. It converts cooling resource usage, control action disturbances, and command feedback deviations into a single evaluation metric, and then combines this with the rated capacity of the cooler unit to form a characterization quantity for cooling control input. This process allows the execution burden consumed by the fan, oil pump, and cooler unit to maintain the temperature control target under automatic switching conditions to be expressed separately. It can identify maintenance benefits that are masked by the automatic control strategy when the temperature curve remains stable, thus achieving [the desired effect]. This invention reveals the maintenance effect from the perspective of control input; furthermore, it performs a source comparison on the forced intervention trajectory and feedback deviation trajectory before and after maintenance, and writes the changes in the complete cooling execution trajectory after maintenance into the cooling execution interference rollback results, so that the removal of manual forced intervention and the recovery of command feedback can be used as maintenance benefits in subsequent attribution. At the same time, this invention separates the cooling execution interference caused by forced switching, feedback loss, and feedback mismatch from the complete cooling execution trajectory, which also prevents manual forced intervention or feedback abnormalities from being mistakenly counted as normal cooling input, so that a mutual verification relationship is formed between the cooling control input characterization quantity and the cooling execution interference.

[0016] 3. This invention further incorporates the temperature rise response after deducting the environmental boundary effect and the cooling control input characterization quantity into the input response binding segment to form a unit temperature control benefit improvement result. Based on this, a cooling group role transfer relationship diagram is constructed, and benefit attribution correction is performed on the unit temperature control benefit improvement result. Through the maintenance object recovery attribution path, the standby load attribution path, and the forced intervention retreat attribution path, it can identify situations where maintenance benefits are manifested in changes in standby waiting roles, rotation participation roles, or forced input roles, thus extending maintenance benefits from temperature results to changes in the cooling group's operational responsibilities. The benefit attribution correction is based on equivalent operating condition samples. By binding paths to the cooler group identifier and role transfer edge identifier, the benefits of objects that have not actually changed are not mistakenly attributed to maintenance objects. At the same time, since the net contribution result of maintenance effect integrates the unit temperature control benefit improvement result, role transfer correction result, cooling execution interference, temperature benchmark interference set and operating condition deviation sample set, it can still identify the decrease in control input, release of reserve load and forced intervention rollback when the temperature change is not obvious. This makes the final maintenance effect evaluation result have the linkage support of data source, attribution path and interference mark, which enhances the interpretability of maintenance quality review, anomaly tracking and next cycle maintenance arrangement. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is the S4 flowchart of the present invention. Detailed Implementation

[0018] 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 and Figure 2 This embodiment provides a method for evaluating the maintenance effectiveness of transformer cooling systems based on big data from maintenance inspections, including the following specific steps: In one implementation, when executing S1, comprehensive maintenance and operation data of the transformer cooling system is acquired, and equivalent operating condition sample pairs under maintenance event anchor points are constructed. The comprehensive maintenance and operation data includes maintenance event data, temperature operation data, load operation data, environmental boundary data, and cooling execution data. Maintenance event data includes maintenance object, maintenance action, maintenance completion time, and maintenance conclusion. Temperature operation data includes top oil temperature, winding temperature, and cooler outlet temperature. Load operation data includes active load, rated load, and operating mode label. Environmental boundary data includes ambient temperature and sampling time. Cooling execution data includes cooler group activation / deactivation status, cooling equipment operation status, start / stop records, forced activation / deactivation records, control command issuance status, and actual feedback status. The above data are obtained by the maintenance work order system, transformer online monitoring system, dispatch load record, station environmental monitoring device, and cooling control system, respectively. The data acquisition interface, timestamp reading, missing value marking, and abnormal data removal adopt existing industrial data acquisition and preprocessing methods. The processing focus of this implementation is to associate data from different sources with the same maintenance object and the same sampling time under the same maintenance event.

[0020] When the data time base is unified, the sampling time of the temperature operation data is used as the main time base. The load operation data, environmental boundary data, and cooling execution data are mapped to the corresponding sampling time. When the sampling periods of different systems are inconsistent, a common sampling period supported by both the online monitoring system and the cooling control system within the same station is used for alignment. In this embodiment, the common sampling period is 5 minutes. The basis for this value is that a 5-minute sampling period can cover the commonly used recording granularity of the transformer online monitoring system and can retain the temperature response changes after the fans, oil pumps, and cooler groups are put into operation or deactivated. When the original recording periods within the station are different, the common sampling period is jointly determined by the original sampling period of the equipment and the cooling control status recording period. If any data source does not have a corresponding record within the common sampling period, a missing mark is written to the data item at that sampling time, and the data is not replaced by supplementary data.

[0021] When forming maintenance event anchor points, the maintenance completion time is marked as the time anchor point, and the maintenance object is marked as the object anchor point. The maintenance object is used to limit the cooler group, fan, oil pump, radiator, temperature acquisition circuit or cooling control circuit corresponding to a maintenance action. The previous maintenance event of the same maintenance object is retrieved from the object anchor point to the time anchor point, and continuous operation records that do not cross the previous maintenance event are extracted to obtain the pre-maintenance candidate interval. The subsequent maintenance events of the same maintenance object are retrieved from the object anchor point to the time anchor point, and continuous operation records that do not cross the subsequent maintenance events are extracted to obtain the post-maintenance candidate interval. Through this process, the evaluation scope of the current maintenance event will not cover adjacent maintenance events of the same maintenance object.

[0022] The observation window length is determined based on the historical temperature rise recovery process or post-maintenance trial operation records of similar maintenance events. Similar maintenance events refer to historical maintenance events with the same maintenance object category and maintenance action category. The historical temperature rise recovery process refers to the process in which the temperature rise response enters the target control range after maintenance is completed, and there are no forced switching records or feedback deviations during the corresponding time period. The target control range is determined by the transformer operation procedures, the equipment manufacturer's temperature control setpoint, or the station's temperature control strategy records. When the number of historical samples of similar maintenance events reaches the minimum sample size specified in the station's maintenance evaluation rules, the historical temperature rise recovery process is used to determine the observation window length. When the number of historical samples does not reach the minimum sample size specified in the station's maintenance evaluation rules, the post-maintenance trial operation records or the post-maintenance observation period in the operation and maintenance procedures are used to determine the observation window length. The observation window must not cross adjacent maintenance events of the same maintenance object.

[0023] Within the observation windows before and after maintenance, the temperature control response condition boundaries affecting the comparability of temperature responses are determined. These boundaries include heat input boundaries, heat dissipation boundaries, operating cycle boundaries, and cooling resource boundaries. The heat input boundary is formed from load operation data and is used to define the background of the transformer operating load's effect on the temperature response. The heat dissipation boundary is formed from environmental boundary data and is used to define the background of the ambient temperature's effect on the heat dissipation process. The operating cycle boundary is formed from the daily cycle interval to which the sampling time belongs and is used to define the background of the effect of day-night load changes and station operating rhythm on the temperature response. The cooling resource boundary is formed from the status of cooler groups that can participate in commissioning and decommissioning and is used to define the resource conditions for participating in cooling execution before and after maintenance. Cooler groups that are out of service, under maintenance isolation, communication unavailable, or under fault lockout are not included in the status of cooler groups that can participate in commissioning and decommissioning.

[0024] The temperature control response condition boundary is normalized for comparability, the heat input boundary is normalized for load scale, and the relative relationship between active load and rated load is written into the heat input description result. The heat dissipation boundary is normalized for environmental distribution, and the position of the ambient temperature in the historical ambient temperature distribution of the same transformer is written into the heat dissipation boundary description result. The reason for using historical ambient temperature distribution for normalization is that the ambient temperature distribution is different in different seasons and different sites. Directly using the absolute value of ambient temperature as the pairing basis will lead to inconsistencies in the heat dissipation boundary conditions of samples before and after maintenance. The operating cycle boundary is aligned for daily cycle, and the sampling time is written into the corresponding hour interval, peak-valley interval or station operating shift interval to obtain the cycle operation description result. The cooling resource boundary is calibrated for available resources, and the status of the cooler group that can participate in commissioning and decommissioning is written into the cooling resource description result. The heat input description result, heat dissipation boundary description result, cycle operation description result and cooling resource description result are written into the same temperature control response condition boundary record.

[0025] The operating condition matching benchmark is formed from historical non-maintenance operation samples. Historical non-maintenance operation samples refer to historical operation samples that have not experienced maintenance events, have not experienced cooling fault alarms, and have continuous operation records. Historical non-maintenance operation samples are used to characterize the range of natural operating condition differences of the same transformer under non-maintenance conditions. When forming the operating condition matching benchmark, the natural difference distribution of heat input description results, heat dissipation boundary description results, cycle operation description results, and cooling resource description results in the historical non-maintenance operation samples are statistically analyzed, and the boundary of sample pairing before and after maintenance is determined by the natural difference distribution. For transformers with seasonal temperature changes, the historical non-maintenance operation samples are limited to historical samples that are consistent with the ambient temperature quantile interval of the current maintenance event. For transformers with frequent changes in operating mode, the historical non-maintenance operation samples are grouped according to the operating mode label to form the operating condition matching benchmark.

[0026] The formation of equivalent operating condition sample pairs includes sample gating and pairing confirmation. First, based on the operating mode label, the samples in the pre-maintenance observation window and the post-maintenance observation window are subjected to the first gating. Samples with inconsistent operating mode labels are not included in the subsequent pairing confirmation. Then, based on the correspondence between the cooling resource description results and the operating condition matching benchmark, the second gating is performed. Samples whose cooling resource description results exceed the operating condition matching benchmark and are not written into the maintenance endogenous resource recovery sample are not included in the subsequent pairing confirmation. Samples that have been written into the maintenance endogenous resource recovery sample continue to enter the pairing confirmation and are used as candidate sources for the maintenance object recovery attribution path in the subsequent cooling group role transfer relationship diagram. Subsequently, based on the heat input description results, heat dissipation boundary description results, and cycle operation description results, the pre-maintenance samples and post-maintenance samples after the second gating are paired and confirmed. The pre-maintenance samples and post-maintenance samples that pass the pairing confirmation are written into the equivalent operating condition sample pairs.

[0027] Before executing the second gating, the source of the cooling resource change is first identified. If the change in the cooling resource description result corresponds to the maintenance object in the maintenance event data, and the maintenance object changes from an abnormal restricted state, fault-locked state, or maintenance isolation state to a state where it can participate in commissioning or decommissioning, then the corresponding sample is written into the maintenance-endogenous resource recovery sample. The maintenance-endogenous resource recovery sample is not directly excluded as an external operating condition deviation, but is used in the subsequent pairing confirmation along with the heat input description result, heat dissipation boundary description result, and periodic operation description result for judgment. If the change in the cooling resource description result does not correspond to the maintenance object in the maintenance event data, or if the change originates from the adjustment of the scheduling operation mode, temporary equipment shutdown, communication unavailability, or resource status change not caused by this maintenance, then the second gating is executed according to the correspondence between the cooling resource description result and the operating condition matching benchmark.

[0028] Samples that fail pairing confirmation but contain cooling execution data are written into the operating condition deviation sample set. The operating condition deviation sample set is used to record operating segments that deviate from the heat input boundary, heat dissipation boundary, operating cycle boundary, or cooling resource boundary before and after maintenance. This set is not used as a direct evaluation sample for the unit temperature control benefit improvement result, but is used as a source of operating condition interference when the net contribution result of subsequent maintenance effect is formed. For samples that fail pairing confirmation and do not contain cooling execution data, they are written into the invalid sample record and are not included in the subsequent cooling execution trajectory extraction.

[0029] Through this step, the maintenance event anchor point defines the boundary for evaluating the maintenance effect, the temperature control response condition boundary defines the basis for comparing samples before and after maintenance, the equivalent operating condition sample pair provides input for the subsequent cooling control input characterization quantity and the unit temperature control benefit improvement result, the operating condition deviation sample set provides the source of interference for the subsequent net contribution result of maintenance effect, the equivalent operating condition sample pair output by S1 enters S2 and S3, and the operating condition deviation sample set output by S1 enters S4.

[0030] In one implementation, after obtaining the equivalent operating condition sample pair in S1, S2 is executed to extract the cooling execution trajectory from the equivalent operating condition sample pair and perform cooling execution burden decomposition to form a cooling control input characterization quantity. At the same time, cooling execution interference is marked. The cooling control input characterization quantity is used to characterize the cooling execution burden corresponding to maintaining the temperature control target within the equivalent operating condition sample pair. The object processed in S2 is cooling execution data, which includes the cooler group activation / deactivation status, cooling equipment operation status, start / stop records, forced activation / deactivation records, control command issuance status, and actual feedback status. Temperature operation data is not directly written into the cooling control input characterization quantity. Temperature operation data is used in S3 to form the temperature rise response and unit temperature control benefit improvement results.

[0031] Specifically, cooling execution data at the corresponding sampling time is read from equivalent operating condition sample pairs. The cooling execution data is then processed in a chain-like manner according to the time sequence between control command issuance status, cooling equipment operation status, and actual feedback status, forming cooling execution chain segments. Each cooling execution chain segment starts with a cooling control command or a cooling equipment status change and is associated with the cooling equipment operation status and actual feedback status of the same cooler group within the corresponding time range. Control command issuance status includes activation commands, deactivation commands, rotation commands, start / stop commands, and hold commands. Cooling equipment operation status is determined by the cooler group activation / deactivation status, fan operation status, oil pump operation status, and cooler group operation status. Actual feedback status is determined by the remote signaling status, operation hold status, or fault feedback status returned by the cooling control system. The acquisition, timestamp reading, and status coding of control commands, cooling equipment operation status, and actual feedback status adopt the existing recording methods of the substation monitoring system or cooling control system. The key processing point of this implementation is to organize the scattered records into cooling execution chain segments with a sequential relationship.

[0032] When organizing cooling execution chain segments, feedback confirmation time limits are set, determined in the following order: If the cooling control system stores a device action confirmation time limit, that device action confirmation time limit is used; if the cooling control system does not store a device action confirmation time limit and historical normal commissioning / discharge records for the same model of cooling equipment exist, the 95th percentile time from the issuance of the control command to the return of the actual feedback status in the historical normal commissioning / discharge records is used; if the cooling control system does not store a device action confirmation time limit and the historical normal commissioning / discharge records do not reach the minimum sample size specified in the station's maintenance evaluation rules, the equipment manufacturer's action confirmation time limit or the station's... The internal remote signaling confirmation rules and historical normal commissioning / discharge records refer to commissioning / discharge records that simultaneously include control commands, cooling equipment operation status, and actual feedback status, but do not include forced commissioning / discharge records or feedback deviation records. In this implementation, the feedback confirmation time limit is within the range of 30 seconds to 300 seconds. This range is derived from the electrical control actions, mechanical responses, and remote signaling refresh processes of the fan, oil pump, and cooler group. If the equipment manufacturer's action confirmation time limit or the station's internal remote signaling confirmation rules explicitly give a value exceeding this range, then the equipment manufacturer's action confirmation time limit or the station's internal remote signaling confirmation rules shall prevail, and the corresponding basis shall be written into the maintenance evaluation record.

[0033] When a cooling execution chain segment has a correspondence between control commands, cooling equipment actions, and actual feedback states, the cooling execution chain segment is written into the complete cooling execution trajectory. The correspondence includes: the cooler group pointed to by the control command is consistent with the cooler group that took action; the control command type is consistent with the cooling equipment action state; the actual feedback state returns within the feedback confirmation time limit and corresponds to the action state; the input command corresponds to the cooler group input state; the output command corresponds to the cooler group output state; the rotation command corresponds to the change of the operating responsibilities of the rotating cooler group; and the maintenance command corresponds to the cooling equipment operation maintenance state. The complete cooling execution trajectory is used to form the basic cooling execution burden in the future.

[0034] When a cooling execution chain segment contains a forced switching record, the cooling execution chain segment is written into the forced intervention trajectory. The forced switching record includes manual forced switching, manual forced reversal, temporary test input, manual input after temporary maintenance isolation is lifted, and cooling equipment actions triggered by non-automatic strategies. The forced intervention trajectory is not written into the basic cooling execution burden, but into the cooling execution interference. The reason for this is that forced switching will change the cooling resource occupancy status and temperature control results, but this action does not belong to the cooling execution burden formed under the automatic control strategy. If the forced intervention trajectory is directly written into the basic cooling execution burden, the subsequent unit temperature control benefit improvement results will be mixed with the temperature control changes caused by manual intervention.

[0035] When a cooling execution chain segment lacks actual feedback status, or when the actual feedback status does not correspond to the cooling equipment's operating status, the cooling execution chain segment is written into the feedback deviation trajectory. Lack of actual feedback status means that the actual feedback status of the corresponding cooler group is not obtained within the feedback confirmation time limit. A mismatch between the actual feedback status and the cooling equipment's operating status means that the control command or action record shows the cooler group is engaged, but the actual feedback status is withdrawn, faulty, locked, or invalid; or the control command or action record shows the cooler group is withdrawn, but the actual feedback status is still engaged. The feedback deviation trajectory is written into the cooling execution interference, which is used in S4 as a source of interference for the net contribution result of maintenance effectiveness.

[0036] A source comparison is performed on the forced intervention trajectory and feedback deviation trajectory in the pre-maintenance observation window and the post-maintenance observation window. The source comparison is based on the maintenance event anchor point, cooler group identifier, control command type, and cooling execution chain segment number. When a forced intervention trajectory or feedback deviation trajectory exists in the pre-maintenance observation window, and the corresponding cooling execution chain segment in the post-maintenance observation window transitions to a complete cooling execution trajectory, this change is written into the cooling execution interference rollback result. The cooling execution interference rollback result is used to characterize the state change of manual forced intervention being released or command feedback deviation disappearing after maintenance. When a forced intervention trajectory or feedback deviation trajectory still exists in the post-maintenance observation window, the corresponding chain segment is written into the cooling execution interference.

[0037] Subsequently, the cooling execution burden is decomposed on the complete cooling execution trajectory. The cooling execution burden decomposition includes extracting the cooling resource occupation process, cooling action maintenance process, and control action switching process from the complete cooling execution trajectory. The cooling resource occupation process is formed by the cooler group's activation and deactivation status, which is used to characterize the cooling resource status participating in temperature control execution during the sample period. The cooling action maintenance process is formed by the continuous operation status of the fan, oil pump, or cooler group, which is used to characterize the operation maintenance status of the cooling equipment during the sample period. The control action switching process is formed by start-stop records, activation and deactivation change records, and rotation change records, which are used to characterize the process of cooling control action switching during the sample period.

[0038] Resource occupancy calibration is performed during the cooling resource occupancy process. During resource occupancy calibration, the input status, output status, standby status, and rotation status of the cooler group during the sample period are written into a unified resource occupancy record. Cooler groups that are in a fault-locked, maintenance-isolated, or communication-unavailable state are not written into the normal resource occupancy record. Resource occupancy calibration is based on whether the cooler group is in a state where it can participate in input / output and actually assumes cooling responsibilities. This process makes the input status, standby status, and rotation status of the cooler group enter the same cooling resource occupancy process.

[0039] The operation and maintenance calibration is performed during the cooling action maintenance process. During operation and maintenance calibration, the fan operation status, oil pump operation status, and cooler group operation status are written into the operation and maintenance process according to the sample time period. If the fan operation status and oil pump operation status exist simultaneously in the same cooler group, they are recorded separately according to the functional objects of the cooling equipment and then written into the operation and maintenance calibration result of the cooler group. For single-cycle remote signaling changes that do not continuously meet the status confirmation conditions, they are not written into the operation and maintenance calibration result. The status confirmation condition is preferentially adopted by the status de-jitter configuration of the station cooling control system. When the status de-jitter configuration is not set, the same status for three consecutive remote signaling refresh cycles is used as the status confirmation condition. The basis for this value is that the status change within a single remote signaling refresh cycle may be caused by communication jitter or sampling error, and three consecutive remote signaling refresh cycles can form continuous confirmation of the cooling equipment operation status.

[0040] Action switching calibration is performed during the control action switching process. During action switching calibration, the start / stop of the fan, the start / stop of the oil pump, the engagement of the cooler group, the deactivation of the cooler group, and the rotation switching are written into the action switching record. The start / stop record is associated with the corresponding control command, action status, and actual feedback status. For consecutive engagement and deactivation actions of the same cooler group, if there is an anti-frequent start / stop interval in the station's cooling control strategy, the anti-frequent start / stop interval is used as the merging judgment basis; if there is a minimum start / stop interval in the equipment manufacturer's data, the equipment manufacturer's data is used as the merging judgment basis; if there is a stable holding time for the same model of equipment in the maintenance and test records, the maintenance and test records are used as the merging judgment basis. If none of the above criteria exist, the consecutive engagement and deactivation actions are not merged, and the consecutive engagement and deactivation actions are written into the action switching record separately. This process can avoid manually setting the start / stop merging threshold when there is a lack of criteria.

[0041] After completing resource occupancy calibration, operation maintenance calibration, and action switching calibration, the results of these calibrations are written into the basic cooling execution burden. The basic cooling execution burden is derived only from the complete cooling execution trajectory and does not include the forced intervention trajectory or feedback deviation trajectory. The forced intervention trajectory and feedback deviation trajectory are written into the cooling execution interference and participate in the formation of the net contribution result of maintenance effect in S4.

[0042] Furthermore, the basic cooling load is calibrated based on the rated capacity of the cooler group, forming a characteristic quantity of cooling control input. The rated capacity of the cooler group is determined by the equipment manufacturer's nameplate, cooler group design data, station equipment ledger, or maintenance and test records. For air-cooled cooler groups, the rated capacity is based on the rated air volume, rated motor power, or the cooling capacity indicated by the manufacturer. For cooler groups with oil pumps participating in oil circulation, the rated capacity is based on the rated oil flow rate, rated motor power, or the cooling capacity indicated by the manufacturer. Cooler groups of the same specification adopt the same rated capacity calibration rules, while cooler groups of different specifications are calibrated separately according to their respective rated capacities. This process ensures that the basic cooling load of cooler groups of different specifications has a unified evaluation basis when forming a characteristic quantity of cooling control input.

[0043] When forming the characterization of cooling control input, the basic cooling execution burden is unified to the same evaluation scale. The same evaluation scale adopts a dimensionless range of 0 to 1. The resource occupancy calibration result uses the status of cooling resources participating in temperature control execution during the sample period as the scale source; the operation and maintenance calibration result uses the operation and maintenance status of cooling equipment during the sample period as the scale source; the action switching calibration result uses the rated allowable number of equipment start-stops, the station's anti-frequent start-stop strategy, or the action switching record in the maintenance test record as the scale source. 0 indicates that the corresponding cooling execution burden has not occurred, and 1 indicates that the corresponding cooling execution burden has reached the upper limit state under the corresponding scale source. After all calibration results enter the same evaluation scale, they are then calibrated by the rated capacity of the cooler group to form the characterization of cooling control input.

[0044] In a preferred implementation, the resource occupancy calibration result is formed by the number of cooler groups actually undertaking cooling responsibilities during the sample period, the corresponding duration of undertaking responsibilities, and the total number of cooler groups that can participate in commissioning and decommissioning. The operation and maintenance calibration result is formed by the ratio between the effective operation and maintenance duration of the fan, oil pump, or cooler group during the sample period and the length of the sample period. The action switching calibration result is formed by the ratio between the number of start-stop, commissioning and decommissioning, or rotation switching during the sample period and the rated allowable number of start-stops for the corresponding equipment, the upper limit of action switching in the station's anti-frequent start-stop strategy record, or maintenance test record. The resource occupancy calibration result, operation and maintenance calibration result, and action switching calibration result are all initially limited to the range of 0 to 1, and then the capacity is calibrated according to the rated capacity of the cooler group to obtain the cooling control input characterization quantity. Through this processing, different cooler group specifications, operation and maintenance durations, and action switching frequencies can be included in the subsequent unit temperature control benefit improvement results under the same dimensionless evaluation scale.

[0045] When forming the characterization quantity for cooling control input, the resource occupancy calibration results, operation maintenance calibration results, and action switching calibration results are first homogenized so that all three types of calibration results are recorded in the direction that the larger the value, the higher the cooling execution burden; then, the resource occupancy weight, operation maintenance weight, and action switching weight are determined. The resource occupancy weight, operation maintenance weight, and action switching weight are primarily determined by the station's maintenance evaluation rules; if the station's maintenance evaluation rules do not record them, they are determined based on the contribution of the resource occupancy calibration results, operation maintenance calibration results, and action switching calibration results to the fluctuation of the temperature response descriptor quantity in the same type of historical non-maintenance operation samples; if the contribution is insufficient to determine the contribution, an equal weighting method is used.

[0046] When calibrating the capacity of cooler groups with different rated capacities, a rated capacity ratio is formed based on the proportion of the rated capacity of a single cooler group to the total rated capacity of all cooler groups that can be put into operation or decommissioned. This rated capacity ratio is then used to correct the resource utilization calibration results, operation maintenance calibration results, and action switching calibration results of the corresponding cooler groups. The corrected three types of calibration results are then weighted and synthesized according to resource utilization weight, operation maintenance weight, and action switching weight to obtain the cooling control input characterization quantity. Through this process, the cooling control input characterization quantity simultaneously reflects the degree of cooling resource participation, the degree of cooling equipment operation maintenance, and the degree of control action switching, and enables cooler groups of different specifications to participate in the formation of subsequent unit temperature control benefit improvement results under the same dimensionless evaluation scale.

[0047] This step does not use top oil temperature, winding temperature, or cooler outlet temperature to directly form the cooling control input characteristic quantity. Temperature operation data is used in S3 to form the temperature rise response and unit temperature control benefit improvement result. S2 outputs the cooling control input characteristic quantity, cooling execution interference, and cooling execution interference backoff result. The cooling control input characteristic quantity enters S3 to form the input response binding segment with the temperature rise response and participates in the formation of the unit temperature control benefit improvement result. The cooling execution interference enters S4 to form the net contribution result of maintenance effect together with the operating condition deviation sample set and the temperature reference interference set. In this way, the cooling execution burden formed under automatic switching conditions and the interference links caused by forced switching and feedback deviation are recorded separately. Subsequent evaluation can distinguish the cooling control input characteristic quantity and cooling execution interference under the same maintenance event.

[0048] In one implementation, after obtaining the equivalent operating condition sample pair in S1 and the cooling control input characterization quantity in S2, S3 is executed to convert the temperature operating state in the equivalent operating condition sample pair into a temperature rise response after deducting the influence of the environmental boundary. Temperature rise response purification is performed, and the unit temperature control benefit improvement result is obtained according to the correspondence between the temperature rise response and the cooling control input characterization quantity. The objects processed in S3 include the equivalent operating condition sample pair, temperature operating data, environmental boundary data, and cooling control input characterization quantity. The equivalent operating condition sample pair defines the comparable operating conditions of the samples before and after maintenance; the cooling control input characterization quantity characterizes the cooling execution burden corresponding to maintaining the temperature control target; the temperature operating data is used to form the temperature rise response; and the environmental boundary data is used to deduct the influence of external heat dissipation conditions on temperature evaluation.

[0049] The target evaluation temperature is determined according to the maintenance object. When the maintenance object is a fan, oil pump, radiator, or cooler group, the top oil temperature is written into the main temperature control response, and the winding temperature is written into the verification response. Fans, oil pumps, radiators, and cooler groups directly affect the oil circuit heat exchange and overall cooling capacity. The top oil temperature is used to characterize the cooling system's control result on the main transformer oil temperature, and the winding temperature is used to verify the local thermal response under load. When the maintenance object is a temperature acquisition loop or cooling control loop, the top oil temperature response and winding temperature response are written into the temperature data verification record; in the same Within an equivalent operating condition sample pair, when both the top oil temperature response and the winding temperature response have completed the temperature rise response formation, and both types of temperature rise responses have corresponding sampling times and target control interval sources, the temperature data is written into a valid record; samples lacking corresponding sampling times, lacking target control interval sources, or entering the temperature reference interference set are not written into a valid temperature data record, and the cooler outlet temperature is used to identify the local heat transfer status of the cooler group; when the maintenance object involves heat transfer abnormalities in a single cooler group, the cooler outlet temperature is written into the auxiliary source of the temperature data verification record.

[0050] Based on the environmental boundary data at the same sampling time, environmental boundary subtraction processing is performed on the target evaluation temperature to obtain the temperature rise response. Specifically, the processing is as follows: the target evaluation temperature and the ambient temperature at the same sampling time are used to form a relative temperature state, and this relative temperature state is written into the temperature rise response. The reason for using the temperature rise response in subsequent processing is that the original temperature is simultaneously affected by the ambient temperature, load status, and cooling execution. Based on the fact that S1 has already normalized the heat input, heat dissipation boundary, operating cycle, and cooling resource conditions through equivalent operating condition samples, S3 further subtracts the influence of the environmental boundary at the same sampling time, so that the temperature operating state can be included in the input response binding segment.

[0051] Before performing temperature rise response purification, first check whether the maintenance event data includes temperature sensor calibration or replacement actions. If the maintenance event data includes temperature sensor calibration or replacement actions, the temperature baseline before and after maintenance is corrected according to the calibration record. The calibration record includes the calibrated temperature measurement point, calibration time, calibration deviation, and calibration conclusion. When the calibration record gives the temperature measurement point deviation value, the temperature baseline of the corresponding measurement point is adjusted according to the deviation value. When the calibration record only gives a qualified or unqualified conclusion and does not give the deviation value, the corresponding sample is written into the temperature baseline interference set and does not enter the direct formation process of the unit temperature control benefit improvement result. When the maintenance event data does not include temperature sensor calibration or replacement actions, the original acquisition baseline of the same temperature measurement point before and after maintenance is used. Samples lacking temperature baseline basis are written into the temperature baseline interference set and participate in the formation of the net contribution result of maintenance effect in S4.

[0052] Within the equivalent operating condition sample pair, the temperature rise response and cooling control input characterization quantity are written into the input response binding segment. The input response binding segment takes the same equivalent operating condition sample pair as the boundary and records the temperature rise response, cooling control input characterization quantity, target evaluation temperature type, target control interval source, and data validity mark corresponding to the sample before and after maintenance. The temperature rise response and cooling control input characterization quantity should correspond to the same sampling time, the same maintenance event anchor point, and the same equivalent operating condition sample pair number. Records that do not meet the correspondence are not written into the input response binding segment.

[0053] The temperature rise and fall status after cooling activation is identified from the activation response binding segments. The temperature rise and fall status is determined by the change process of the temperature rise response during the response observation period after cooling activation. The response observation period is determined in the following order: if the station temperature control strategy record has specified the response observation period, the station temperature control strategy record is used; if the equipment manufacturer's data provides the cooling action response period, the equipment manufacturer's data is used; if the historical record of the same type of cooling action reaches the minimum sample size specified by the station maintenance evaluation rules, the time required for the temperature rise response to enter the target control range in the historical record is used; if none of the above criteria exist, for the activation of fans, oil pumps, or cooler groups, the response observation period is 30 minutes to 6 hours. This range is applicable to the temperature response observation after the cooling equipment is activated, and is derived from the confirmation of the electrical action of the cooling equipment, changes in oil circulation, and the heat inertia transfer process of the main transformer. If the station temperature control strategy record or the equipment manufacturer's data provides a dedicated time period that exceeds this range, the corresponding dedicated time period is used, and this criterion is written into the maintenance evaluation record.

[0054] The temperature rise suppression state under equivalent operating conditions is identified from the input response binding segments. The temperature rise suppression state is determined by the trend of the temperature rise response within the equivalent operating condition sample as a function of the sampling time. To exclude fluctuations at individual sampling points, the temperature rise suppression state is identified using continuous sampling segments. The length of the continuous sampling segments is determined based on the common sampling period. If the on-site online monitoring system has a temperature trend judgment period set, that temperature trend judgment period is used. If no temperature trend judgment period is set, a continuous sampling segment is formed by no less than 3 continuous sampling points. When the common sampling period is 5 minutes, the continuous sampling segment should be no less than 15 minutes. This value is based on the fact that 1 or 2 sampling points are easily affected by communication jitter, sampling errors, or instantaneous load disturbances, and 3 continuous sampling points can form the smallest continuous judgment unit for the temperature rise change trend.

[0055] The thermal stability maintenance state within the target control range is identified from the input response binding segments. The target control range is determined by the transformer operating procedures, the equipment manufacturer's temperature control setpoint, or the station's temperature control strategy record. When all three sources exist simultaneously, the actual temperature control strategy record in operation within the station is used. When the station's temperature control strategy record is missing, the equipment manufacturer's temperature control setpoint is used. When the equipment manufacturer's temperature control setpoint is missing, the temperature control requirements in the transformer operating procedures are used. The thermal stability maintenance state is formed by the continuous state of the temperature rise response within the target control range. When the target control range is adjusted before and after maintenance, the samples before and after the adjustment are written into different input response binding segments and are not directly compared in the temperature control benefit results of the same unit.

[0056] The temperature rise decline state, temperature rise suppression state, and thermal stability maintenance state are all written into the temperature response descriptor. The temperature response descriptor is used to record the temperature rise decline process after cooling is put into operation, the temperature rise trend under equivalent operating conditions, and the continuous state within the target control range. The temperature response descriptor is recorded using a unified evaluation scale, which is formed by the temperature rise response distribution in historical non-maintenance operation samples. The definition of historical non-maintenance operation samples is consistent with S1, that is, historical operation samples in which no maintenance events have occurred, no cooling fault alarms have occurred, and the operation records are continuous. The temperature rise decline state, temperature rise suppression state, and thermal stability maintenance state are scaled according to the corresponding temperature rise state distribution of historical non-maintenance operation samples and written into the same temperature response descriptor. Through this processing, the temperature response descriptor will not directly mix the Celsius, time length, and state records.

[0057] The temperature response descriptor and the cooling control input characterization are written into the unit temperature control benefit result. The unit temperature control benefit result is the result after binding the temperature response descriptor and the corresponding cooling control input characterization within the same equivalent operating condition sample pair. When forming the unit temperature control benefit result, it is first confirmed that the temperature response descriptor and the cooling control input characterization have the same sampling time, the same equivalent operating condition sample pair number, and the same maintenance event anchor point. Records that meet the correspondence are written into the unit temperature control benefit result, and records that do not meet the correspondence are not included in the unit temperature control benefit result. Based on the reason, they are written into the temperature reference interference set, the operating condition deviation sample set, or the cooling execution interference, respectively.

[0058] When generating the unit temperature control benefit result, the temperature response descriptor is first converted into a temperature control response effective quantity ranging from 0 to 1. The temperature control response effective quantity is recorded in the direction that the larger the value, the more sufficient the temperature rise and fall state, the more stable the temperature rise suppression state, and the more reliable the thermal stability maintenance state. Then, the cooling control input characterization quantity in the same input response binding segment is read. When the cooling control input characterization quantity is greater than the minimum effective cooling input quantity, the unit input response relationship between the temperature control response effective quantity and the cooling control input characterization quantity is written into the unit temperature control benefit result.

[0059] The minimum effective cooling input is primarily determined by the station's maintenance evaluation rules or cooling control strategy records. If neither the station's maintenance evaluation rules nor the cooling control strategy records are recorded, the 5th percentile of the non-zero cooling control input characteristic value in the historical non-maintenance operation samples of the same type is used as the minimum effective cooling input. If the historical non-maintenance operation samples of the same type are insufficient to form the 5th percentile, the minimum value of the non-zero cooling control input characteristic value within the current observation window is used as the minimum effective cooling input, and a sample basis mark is written into the maintenance effect evaluation results. Samples with a minimum effective cooling input are not included in the direct formation of the unit temperature control benefit result, and a data validity mark is written into them. Through this process, the unit temperature control benefit result is used to characterize the temperature control response benefit corresponding to the unit cooling control input under effective cooling input conditions, avoiding the formation of unfounded unit benefit results under conditions of no cooling input or extremely low cooling input.

[0060] Using the same equivalent operating condition sample pair, the unit temperature control benefit results in the pre-maintenance observation window and the post-maintenance observation window are compared to form the unit temperature control benefit improvement result. When performing the comparison, the equivalent operating condition sample pair is used as the smallest comparison unit. The pre-maintenance sample and the post-maintenance sample are normalized in S1 at the heat input boundary, heat dissipation boundary, operating cycle boundary, and cooling resource boundary. The unit temperature control benefit improvement result is used to record the change state between the unit temperature control benefit results before and after maintenance. If the temperature value does not decrease directly after maintenance, but the temperature response description quantity ranks higher than the pre-maintenance sample under the same evaluation scale, and the corresponding cooling control input characterization quantity does not enter the cooling execution interference, the corresponding change is written into the unit temperature control benefit improvement result. If the temperature value decreases after maintenance, but the corresponding input response binding segment has cooling execution interference, temperature reference interference set, or operating condition deviation sample set marker, the marker is transmitted to S4 along with the unit temperature control benefit improvement result and is not directly identified as the net contribution result of maintenance effect in S3.

[0061] This step outputs the unit temperature control benefit improvement result and the temperature reference disturbance set. The unit temperature control benefit improvement result enters S4 and is corrected for benefit attribution by comparing it with the cooling group role transfer relationship diagram. The temperature reference disturbance set enters S4 and, together with the cooling execution disturbance and the operating condition deviation sample set, participates in the formation of the net contribution result of maintenance effect. Through this step, the maintenance effect evaluation is transformed from a direct comparison of temperature values ​​before and after maintenance to an evaluation of the input response correspondence between the temperature rise response and the input characteristic quantity of cooling control under the same equivalent operating condition.

[0062] In one implementation, after obtaining the cooling control input characterization quantity in S2 and the unit temperature control benefit improvement result in S3, S4 is executed. A cooling group role transfer relationship diagram is formed according to the change of operating responsibilities before and after the cooler group maintenance. The benefit attribution of the unit temperature control benefit improvement result is corrected. The net contribution result of maintenance effect is formed by combining the cooling execution interference, the temperature reference interference set and the operating condition deviation sample set. The maintenance effect evaluation result is output. The objects processed in this step include the cooling control input characterization quantity, the unit temperature control benefit improvement result, the cooling execution interference, the temperature reference interference set and the operating condition deviation sample set.

[0063] Specifically, the operating roles of each cooler group are first read from the pre-maintenance and post-maintenance observation windows, and the operating roles undertaken by the cooler group in each observation window are written into the role nodes. The operating roles include automatic carrying role, standby role, rotation participation role, abnormal restriction role, and forced entry role. Automatic carrying role means that the cooler group undertakes the routine cooling task according to the automatic control strategy; standby role means that the cooler group is in a state where it can participate in the start-up and shutdown but is not undertaking the current routine cooling task; rotation participation role means that the cooler group participates in periodic operation according to the preset rotation strategy; abnormal restriction role means that the cooler group has a fault alarm, feedback mismatch, lockout, or cannot be entered according to the control command; forced entry role means that the cooler group is put into operation by manual command, temporary test command, or non-automatic strategy. The operating role is determined by the cooler group's start-up and shutdown status, operating mode label, forced start-up and shutdown record, control command issuance status, and actual feedback status.

[0064] Before forming a role node, a role establishment record is first formed based on the duration of the role, the number of times the role appears, and the percentage of input in the corresponding observation window of the cooler group. When the role establishment record reaches the role establishment benchmark, the corresponding running role is written into the role node. The role establishment benchmark is determined by the distribution of the duration, the number of times the role appears, and the percentage of input in the corresponding running role in the historical non-maintenance operation samples. When the historical non-maintenance operation samples are insufficient, the role establishment benchmark is determined by the station cooling control strategy record or maintenance evaluation rules.

[0065] When multiple operating roles exist within the same observation window for the same cooler group, the operating roles within that observation window are determined according to role determination priority. The role determination priority is as follows: mandatory activation role, abnormal restricted role, automatic undertaking role, rotating participation role, and standby waiting role. The basis for setting this priority is that the mandatory activation role directly reflects manual or temporary strategy intervention, the abnormal restricted role directly reflects an incomplete execution link, the automatic undertaking role and the rotating participation role reflect the operating responsibilities under automatic control, and the standby waiting role reflects the status of being able to be activated but not undertaking regular cooling tasks. If the mandatory activation record covers the automatic activation / deactivation record, the corresponding time period is written as the mandatory activation role; if the control command does not correspond to the actual feedback status, the corresponding time period is written as the abnormal restricted role.

[0066] Subsequently, the changes in the operating roles of the same cooler group between the pre-maintenance observation window and the post-maintenance observation window are written into the role transfer edge. The role transfer edge includes the cooler group identifier, maintenance event identifier, pre-maintenance operating role, post-maintenance operating role, role change window, and corresponding unit temperature control benefit improvement result identifier. If there are multiple role changes in the same direction within the post-maintenance observation window of the same cooler group, the role changes in the same direction are written into the same role transfer edge, and the corresponding time interval is retained. The role transfer relationship diagram of the cooler group is formed by the role nodes and the role transfer edge.

[0067] When a cooler group under maintenance changes from an abnormal restricted role to an automatic bearing role or a rotating participation role, the corresponding role transfer edge is written into the maintenance object recovery attribution path. If the same maintenance event involves multiple maintenance objects, the corresponding cooler groups are matched according to the maintenance objects in the maintenance event data. Cooler groups not listed as maintenance objects are not written into the maintenance object recovery attribution path.

[0068] When a standby role transitions from the "participating in cooling load" state to the "standby holding" state, and the temperature rise response is within the target control range, the corresponding role transfer edge is written into the standby load attribution path. The "participating in cooling load" state indicates that the standby cooler group was put into operation and undertook the cooling task in the observation window before maintenance; the "standby holding" state indicates that the standby cooler group remained in the "available" state in the observation window after maintenance and did not undertake routine cooling tasks. The target control range is determined by S3 and comes from the transformer operation procedures, the equipment manufacturer's temperature control setpoint, or the station's temperature control strategy record. If a standby role enters the "standby holding" state, but the temperature rise response is not within the target control range, the corresponding role transfer edge is not written into the standby load attribution path, and a verification mark is written into the maintenance effect evaluation result.

[0069] When a forced-initiated role is transferred to an automatic-bearing role or a rotating participation role under automatic control, the corresponding role transfer edge is written into the forced intervention rollback attribution path. If a forced switching record still exists after maintenance, or if a forced switching record and a feedback deviation trajectory exist simultaneously, the corresponding role transfer edge is not written into the forced intervention rollback attribution path, but into the cooling execution interference. The cooling execution interference is formed by S2 and is used to record the forced intervention trajectory and the feedback deviation trajectory.

[0070] After obtaining the recovery attribution path, standby load attribution path, and forced intervention rollback attribution path for the maintenance object, a benefit attribution correction is performed on the unit temperature control benefit improvement result. This benefit attribution correction does not use a fixed priority allocation method; instead, it establishes a binding relationship based on equivalent operating condition sample pairs, cooler group identifiers, role transfer edge identifiers, and maintenance event anchor points. When the maintenance object's recovery attribution path and the unit temperature control benefit improvement result point to the same maintenance object and the same maintenance event, the corresponding unit temperature control benefit improvement result is written into the maintenance object's recovery attribution path. Similarly, when the standby load attribution path and the unit temperature control benefit improvement result point to the same... When an equivalent operating condition sample pair and the same standby cooler group are used, the corresponding unit temperature control benefit improvement result is written into the standby load attribution path. When the forced intervention rollback attribution path and the unit temperature control benefit improvement result point to the same post-maintenance observation window, the same cooler group, and the same cooling execution interference rollback result, the corresponding unit temperature control benefit improvement result is written into the forced intervention rollback attribution path. If a unit temperature control benefit improvement result corresponds to multiple role transfer edges, attribution records are formed separately. Each attribution record saves the corresponding cooler group identifier, role transfer edge identifier, equivalent operating condition sample pair number, and maintenance event anchor point.

[0071] Simultaneously, the cooling execution interference formed by S2, the temperature reference interference set formed by S3, and the operating condition deviation sample set formed by S1 are introduced into the process of forming the net contribution result of maintenance effect. The cooling execution interference is used to record the forced intervention trajectory and feedback deviation trajectory; the temperature reference interference set is used to record samples that lack temperature reference basis; the operating condition deviation sample set is used to record samples that have not entered the equivalent operating condition sample pair but retain cooling execution data. If the unit temperature control benefit improvement result corresponding to a maintenance event is marked by cooling execution interference, temperature reference interference set, or operating condition deviation sample set, the corresponding interference is written into the net contribution result of maintenance effect.

[0072] The formation of role transfer correction results includes: generating path establishment records for the maintenance object recovery attribution path, the forced intervention rollback attribution path, and the backup carrier attribution path, respectively. The path establishment record includes the path type, establishment period, corresponding observation window length, and corresponding unit temperature control benefit improvement result identifier; forming the path base quantity based on the ratio between the path establishment period and the corresponding observation window length; when multiple attribution paths exist for the same maintenance event, determining the main attribution path in the order of maintenance object recovery attribution path, forced intervention rollback attribution path, and backup carrier attribution path, and forming the role transfer correction base result based on the path base quantity of the main attribution path, and then converting it into a role transfer correction result of 0 to 1 through the quantile interval of historical samples of the same type of maintenance event.

[0073] The interference deduction result is formed by: separately calculating the interference coverage ratio of cooling execution interference, temperature reference interference set, and operating condition deviation sample set within the observation window corresponding to the maintenance event. The interference coverage ratio of cooling execution interference is formed by the ratio between the number of samples covered by the forced intervention trajectory and feedback deviation trajectory and the total number of samples in the observation window. The interference coverage ratio of the temperature reference interference set is formed by the ratio between the number of samples lacking temperature reference and the total number of samples in the observation window. The interference coverage ratio of the operating condition deviation sample set is formed by the ratio between the number of samples that did not enter the equivalent operating condition sample pair but retained cooling execution data and the total number of samples in the observation window. The maximum value among the above interference coverage ratios is taken as the basic result of interference deduction, and then converted into an interference deduction result of 0 to 1 through the quantile interval of historical samples of the same type of maintenance event.

[0074] In this embodiment, the net contribution of maintenance effect is calculated using the following formula:

[0075] in, This indicates a maintenance event; Indicates maintenance event The corresponding net contribution of maintenance effect; This indicates the improvement in unit temperature control benefits corresponding to the maintenance event, which is formed by comparing the unit temperature control benefits before and after maintenance within the same equivalent operating condition sample in S3. Indicates maintenance event The corresponding role transfer correction results are formed by the maintenance object recovery attribution path, the backup bearer attribution path, the forced intervention rollback attribution path, and the cooling execution interference rollback results. The interference deduction result corresponding to the maintenance event is formed by cooling execution interference, temperature reference interference set, and operating condition deviation sample set; and Both are dimensionless positive contribution weights, used to limit the positive contribution ratio of unit temperature control benefit improvement and role transfer correction results in the net contribution result of maintenance effect, respectively. and The values ​​of are all in the range of 0 to 1, and satisfy . + =1; This is a dimensionless interference deduction weight, used to limit the degree to which the interference deduction result reduces the net contribution of the maintenance effect. The value range is from 0 to 1, and Do not participate and Normalization.

[0076] Before being entered into the formula, the unit temperature control benefit improvement result, role transfer correction result, and interference deduction result were all converted into dimensionless results (0 to 1) using the quantile intervals of historical samples of similar maintenance events, and then homogenized. Specifically, the unit temperature control benefit improvement result and role transfer correction result were recorded in the direction where larger values ​​indicate stronger positive maintenance contributions; the interference deduction result was recorded in the direction where larger values ​​indicate stronger residual interference. The net contribution of maintenance effect is a dimensionless ranking value used to classify the maintenance effect level among similar maintenance events.

[0077] The composite weight is primarily determined by the station's maintenance evaluation rules. If the station's maintenance evaluation rules do not specify composite weights, the positive contribution weights of the unit temperature control benefit improvement results and role transfer correction results are determined based on the consistency between the unit temperature control benefit improvement results and role transfer correction results and subsequent re-inspection conclusions in historical samples of similar maintenance events. Higher consistency results correspond to greater positive contribution weights, and the two consistency results are normalized to form the composite weight. and ,and and For dimensionless values ​​between 0 and 1, satisfying + =1, interference deduction weight The weight is determined independently of the positive contribution weight, and its value ranges from 0 to 1; when the station maintenance evaluation rules record the interference deduction level and its weight mapping relationship, it is determined according to the weight mapping relationship. When the in-station maintenance evaluation rules only record the interference deduction level but not the weight mapping relationship, the interference deduction level is mapped equidistantly to the 0 to 1 interval from low to high, forming... The lowest interference deduction level corresponds to 0, the highest interference deduction level corresponds to 1, and the intermediate interference deduction levels are equidistant from 0 to 1 according to the level order. When the station maintenance evaluation rules do not record the interference deduction level, it is determined based on the consistency between the interference deduction results in historical samples of similar maintenance events and the subsequent re-inspection anomaly conclusions. The higher the consistency, The larger the value, the better.

[0078] When the number of historical samples of the same type of maintenance event does not reach the minimum sample size stipulated in the station's maintenance evaluation rules, or when subsequent re-inspection conclusions are insufficient to form a consistent result, a conservative weighting method is adopted to determine the weighting. and Take 0.5 respectively, and... Set the value to 1, and simultaneously write the sample basis marker into the maintenance effect evaluation results.

[0079] The net contribution of maintenance effect is a dimensionless ordinal value, and its value is not limited to the range of 0 to 1. The maintenance effect level is determined according to the quantile position of the net contribution of maintenance effect in the distribution of historical net contribution results of the same type of maintenance event.

[0080] The reason for adopting this formula is that the unit temperature control benefit improvement result is used to characterize the change in input response between temperature rise response and cooling control input under the same operating conditions; the role transfer correction result is used to characterize the attribution relationship of maintenance benefits in the change of cooler group operation responsibilities; and the interference deduction result is used to characterize the impact of residual forced intervention, feedback deviation, missing temperature benchmark and operating condition deviation on the evaluation result. The three types of results are dimensionless, homogenized and weighted before participating in the formation of the net contribution result of maintenance effect, which can avoid the direct synthesis of results of different natures without scale unification.

[0081] Historical samples of similar maintenance events refer to historical samples with the same maintenance object category and maintenance action category, and with complete pre-maintenance observation windows, post-maintenance observation windows, cooling execution data, and temperature operation data. When the number of historical samples of similar maintenance events reaches the minimum sample size stipulated in the station's maintenance evaluation rules, this type of historical sample is used to form a quantile interval. When the number of historical samples of similar maintenance events does not reach the minimum sample size stipulated in the station's maintenance evaluation rules, historical samples of the same maintenance object category and maintenance actions belonging to the same category of cooling capacity recovery, control loop recovery, or temperature acquisition recovery are used to form a quantile interval, and the sample source mark is written into the maintenance effect evaluation result. When the number of replacement historical samples still does not reach the minimum sample size stipulated in the station's maintenance evaluation rules, the net contribution result of maintenance effect and the insufficient sample mark are output, and quantile level division is not performed.

[0082] The maintenance effectiveness assessment results include the maintenance effectiveness level, the maintenance effectiveness maintenance result, and the maintenance recommendation for the next cycle. The maintenance effectiveness level is determined according to the position of the net contribution of the maintenance effectiveness in the distribution of historical net contribution results of similar maintenance events. When the net contribution of the maintenance effectiveness is above the 75th percentile, the first maintenance effectiveness level is output; when the net contribution of the maintenance effectiveness is between the 25th and 75th percentiles, the second maintenance effectiveness level is output; when the net contribution of the maintenance effectiveness is below the 25th percentile, the third maintenance effectiveness level is output. The 25th and 75th percentiles are derived from the distribution of historical net contribution results of similar maintenance events and are used to form the low, middle, and high ranges. If the maintenance effectiveness assessment results contain an insufficient sample marker, the net contribution of the maintenance effectiveness is output, but the maintenance effectiveness level is not output.

[0083] The maintenance effect retention result is formed based on the change in the net contribution of maintenance effect within the continuous observation window after maintenance. The length of the continuous observation window is consistent with the length of the post-maintenance observation window in S1. When the station maintenance evaluation rule sets a review cycle, the review cycle is used as the continuous observation window. When the net contribution of maintenance effect changes from the first maintenance effect level to the second or third maintenance effect level, or from the second maintenance effect level to the third maintenance effect level, a maintenance effect retention verification prompt is output. The value of the two consecutive observation windows is based on the following: a single observation window may be affected by temporary load changes or environmental boundary changes, and two consecutive observation windows can form the review conditions for the post-maintenance status change. When the station maintenance evaluation rule stipulates a longer review cycle, the station maintenance evaluation rule shall be followed.

[0084] The next maintenance cycle recommendations are formed based on the maintenance effectiveness level, attribution path, and source of interference. When the maintenance effectiveness level is the third level and the recovery attribution path for the repaired object has not been formed, a re-inspection recommendation for the repaired cooler group is output. When the standby load attribution path has not been formed and the standby standby role is still participating in the cooling load state, a standby cooler group load verification recommendation is output. When the forced intervention rollback attribution path has not been formed and there is still a forced intervention trajectory in the cooling execution interference, a control strategy and manual forced switching record verification recommendation is output. When the temperature benchmark interference set persists, a temperature measurement point verification recommendation is output. When the operating condition deviation sample set persists in multiple observation windows, a subsequent evaluation window re-selection recommendation is output.

[0085] This step outputs the net contribution result of maintenance effect and the maintenance effect evaluation result. S4 forms a progressive relationship with S1 to S3: S1 provides equivalent operating condition sample pairs and operating condition deviation sample sets; S2 provides cooling control input characterization quantities and cooling execution interference; S3 provides unit temperature control benefit improvement results and temperature benchmark interference sets; S4 generates the net contribution result of maintenance effect based on the above results.

[0086] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units 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.

[0087] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0088] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes 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.

Claims

1. A method for evaluating the maintenance effectiveness of transformer cooling systems based on big data from maintenance inspections, characterized in that, The steps include: Acquire comprehensive maintenance and operation data of the transformer cooling system, anchor the maintenance event at the maintenance completion time, and divide the observation window before and after maintenance. Perform condition comparability normalization on the temperature control response condition boundary that affects the comparability of temperature response within the observation window, so that the samples before and after maintenance are unified in terms of heat input, heat dissipation boundary, operating cycle and cooling resource conditions, to obtain equivalent operating condition sample pairs, and write the samples that do not enter the equivalent operating condition sample pairs but retain cooling execution data into the operating condition deviation sample set. Cooling execution trajectories are extracted from equivalent operating condition sample pairs, cooling execution burden is decomposed, cooling resource occupation, cooling action maintenance and control action switching are converted into the same evaluation scale, and cooling control input characterization quantity is obtained by calibrating the rated capacity of the cooler group. At the same time, forced intervention trajectory and feedback deviation trajectory are marked as cooling execution interference. The temperature operating state in the equivalent operating condition sample pair is converted into a temperature rise response after deducting the influence of the environmental boundary. Temperature rise response purification is performed, and the unit temperature control benefit improvement result is obtained according to the correspondence between the temperature rise response and the cooling control input characterization quantity. Samples lacking temperature reference are written into the temperature reference interference set. A role transfer relationship diagram of the cooling group is formed based on the changes in the operational responsibilities of the cooling group before and after maintenance. The benefit attribution of the unit temperature control benefit improvement results is adjusted, and the net contribution result of maintenance effect is formed by combining the cooling execution interference, temperature reference interference set and operating condition deviation sample set. The maintenance effect evaluation result is then output.

2. The method for evaluating the maintenance effect of a transformer cooling system based on big data from maintenance, as described in claim 1, is characterized in that... The comprehensive maintenance and operation data should include at least maintenance event data, temperature operation data, load operation data, environmental boundary data, and cooling execution data. Maintenance event data should include at least the maintenance object, maintenance action, maintenance completion time, and maintenance conclusion. Temperature operation data should include at least the top oil temperature, winding temperature, and cooler outlet temperature. Load operation data should include at least the active load, rated load, and operating mode label. Environmental boundary data should include at least the ambient temperature and sampling time. Cooling execution data should include at least the cooler group activation / deactivation status, cooling equipment operation status, start / stop records, forced activation / deactivation records, control command issuance status, and actual feedback status. The comprehensive maintenance and operation data establishes data association relationships under the same maintenance event based on the maintenance object and sampling time.

3. The method for evaluating the maintenance effect of a transformer cooling system based on maintenance big data as described in claim 2, characterized in that, The anchoring of maintenance events and the division of observation windows include: marking the maintenance completion time as the time anchor point and marking the maintenance object as the object anchor point; retrieving the previous and subsequent maintenance events for the same maintenance object based on the object anchor point; extracting continuous operation records before the time anchor point that do not cross the previous maintenance event to obtain the pre-maintenance candidate interval; extracting continuous operation records after the time anchor point that do not cross the subsequent maintenance event to obtain the post-maintenance candidate interval; determining the window length based on the historical temperature rise recovery process of similar maintenance events or the post-maintenance trial operation records; and extracting operation records corresponding to the window length from the pre-maintenance candidate interval and the post-maintenance candidate interval to obtain the pre-maintenance observation window and the post-maintenance observation window.

4. The method for evaluating the maintenance effect of a transformer cooling system based on big data from maintenance, as described in claim 1, is characterized in that... The temperature control response condition boundary includes at least the heat input boundary, heat dissipation boundary, operating cycle boundary, and cooling resource boundary; the heat input boundary is formed by load operation data, the heat dissipation boundary is formed by environmental boundary data, the operating cycle boundary is formed by the daily cycle interval to which the sampling time belongs, and the cooling resource boundary is formed by the status of the cooler group that can participate in commissioning and decommissioning; The comparability normalization of operating conditions includes: performing load scale normalization on the heat input boundary to obtain the heat input description result; performing environmental distribution normalization on the heat dissipation boundary to obtain the heat dissipation boundary description result; performing daily cycle alignment on the operating cycle boundary to obtain the cycle operation description result; performing available resource calibration on the cooling resource boundary to obtain the cooling resource description result; and writing the heat input description result, heat dissipation boundary description result, cycle operation description result, and cooling resource description result into the same temperature control response operating condition boundary record.

5. The method for evaluating the maintenance effect of a transformer cooling system based on maintenance big data as described in claim 4, characterized in that, The formation of equivalent operating condition sample pairs and operating condition deviation sample sets includes: forming an operating condition matching benchmark using historical non-maintenance operating samples with continuous operating records and no maintenance events or cooling fault alarms; performing first gating on samples in the pre-maintenance and post-maintenance observation windows based on the operating mode label; identifying the source of cooling resource changes based on the cooling resource description results; and writing the corresponding sample into the maintenance endogenous data when the cooling resource change is formed by the maintenance object in the maintenance event data changing from an abnormal restricted state, fault lockout state, or maintenance isolation state to a state that can participate in commissioning / decommissioning. Resource recovery samples; when the change in cooling resources does not correspond to the maintenance object in the maintenance event data, the second gating is executed based on the correspondence between the cooling resource description results and the operating condition matching benchmark; based on the heat input description results, heat dissipation boundary description results and cycle operation description results, pairing confirmation is performed on the pre-maintenance samples, post-maintenance samples and maintenance-endogenous resource recovery samples after the second gating; the pre-maintenance samples and post-maintenance samples that pass the pairing confirmation are written into the equivalent operating condition sample pair; the samples that do not pass the pairing confirmation and contain cooling execution data are written into the operating condition deviation sample set.

6. The method for evaluating the maintenance effect of a transformer cooling system based on big data from maintenance, as described in claim 1, is characterized in that... The formation of cooling execution trajectory, cooling execution interference, and cooling execution interference rollback results includes: organizing cooling execution data into chain segments according to the time sequence between control command issuance status, cooling equipment operation status, and actual feedback status; writing cooling execution chain segments with corresponding control commands, cooling equipment operation, and actual feedback status into the complete cooling execution trajectory; writing cooling execution chain segments containing forced switching records into the forced intervention trajectory; writing cooling execution chain segments lacking actual feedback status or whose actual feedback status does not correspond to the cooling equipment operation status into the feedback deviation trajectory; performing source comparison on the forced intervention trajectory and feedback deviation trajectory in the pre-maintenance and post-maintenance observation windows; when a forced intervention trajectory or feedback deviation trajectory exists before maintenance and the corresponding chain segment transitions to the complete cooling execution trajectory after maintenance, a cooling execution interference rollback result is formed; when a forced intervention trajectory or feedback deviation trajectory still exists after maintenance, a cooling execution interference is written.

7. The method for evaluating the maintenance effect of a transformer cooling system based on big data from maintenance, as described in claim 6, is characterized in that... The breakdown of cooling execution burden includes: extracting the cooling resource occupation process, cooling action maintenance process, and control action switching process from the complete cooling execution trajectory; performing resource occupation calibration on the cooling resource occupation process; performing operation maintenance calibration on the cooling action maintenance process; performing action switching calibration on the control action switching process; writing the resource occupation calibration results, operation maintenance calibration results, and action switching calibration results into the basic cooling execution burden; calibrating the execution capacity of the basic cooling execution burden based on the rated capacity of the cooler group to form a cooling control input characterization quantity; taking cooling execution interference as the source of interference in the net contribution result of maintenance effect, and taking the cooling execution interference rollback result as the basis for forming the forced intervention rollback attribution path.

8. The method for evaluating the maintenance effect of a transformer cooling system based on big data from maintenance, as described in claim 1, is characterized in that... The formation of the temperature rise response, temperature rise response purification, and temperature reference interference set includes: determining the target evaluation temperature according to the maintenance object; when the maintenance object is a fan, oil pump, radiator, or cooler group, writing the top oil temperature into the main temperature control response and the winding temperature into the verification response; when the maintenance object is a temperature acquisition loop or cooling control loop, writing the top oil temperature response and winding temperature response into the temperature data verification record; performing environmental boundary subtraction processing on the target evaluation temperature based on the environmental boundary data at the same sampling time to obtain the temperature rise response; when the maintenance event data includes temperature sensor verification or replacement actions, correcting the temperature reference before and after maintenance based on the verification record; and writing samples lacking temperature reference basis into the temperature reference interference set.

9. The method for evaluating the maintenance effect of a transformer cooling system based on big data from maintenance, as described in claim 8, is characterized in that... The formation of the unit temperature control benefit improvement result includes: within the equivalent operating condition sample pair, writing the temperature rise response and cooling control input characterization quantity into the input response binding segment; identifying the temperature rise decline state after cooling input from the input response binding segment; identifying the temperature rise suppression state under the equivalent operating condition from the input response binding segment; identifying the thermal stability maintenance state within the target control range from the input response binding segment; the target control range is determined by the transformer operating procedures, equipment manufacturer's temperature control setpoint, or station temperature control strategy record; writing the temperature rise decline state, temperature rise suppression state, and thermal stability maintenance state into the temperature response description quantity; writing the temperature response description quantity and cooling control input characterization quantity into the unit temperature control benefit result; and comparing the unit temperature control benefit results in the pre-maintenance observation window and the post-maintenance observation window according to the same equivalent operating condition sample pair to form the unit temperature control benefit improvement result.

10. The method for evaluating the maintenance effect of a transformer cooling system based on maintenance big data according to claim 9, characterized in that, The formation of the cooling group role transfer relationship diagram and the net contribution result of maintenance effect includes: writing the operating roles undertaken by the cooler group in the pre-maintenance observation window and the post-maintenance observation window into the role node; the operating roles include automatic load role, standby role, rotation participation role, abnormal restricted role, and forced operation role; forming role establishment records based on the duration, frequency, and operation ratio of the role of the cooler group in the corresponding observation window; when the role establishment record reaches the role establishment benchmark, the corresponding operating role is written into the role node; the role establishment benchmark is determined by the duration distribution, frequency distribution, and operation ratio distribution of the corresponding operating role in the historical non-maintenance operation samples; writing the changes in the operating roles of the same cooler group between the pre-maintenance observation window and the post-maintenance observation window into the role transfer edge; when the maintained cooler group changes from the abnormal restricted role to the automatic load role or the rotation participation role, the corresponding role transfer edge is written into the maintenance object recovery attribution path; when the standby role changes from the participating cooling load state to the standby hold state... When the temperature rise response is within the target control range, the corresponding role transfer edge is written into the backup load attribution path; when the forced input role is transferred to the automatic load role under automatic control or the rotating participation role, or when the forced intervention trajectory or feedback deviation trajectory is transferred into the complete cooling execution trajectory, the corresponding role transfer edge is written into the forced intervention rollback attribution path; based on the maintenance object recovery attribution path, the backup load attribution path, and the forced intervention rollback attribution path, the benefit attribution correction is performed on the unit temperature control benefit improvement result; among which, the benefit attribution correction is written into the corresponding attribution path according to the equivalent operating condition sample pair, the cooler group identifier, and the role transfer edge identifier; when a unit temperature control benefit improvement result corresponds to multiple role transfer edges, corresponding attribution records are formed respectively, and the net contribution result of maintenance effect is written according to the role establishment record corresponding to each attribution record; the unit temperature control benefit improvement result after benefit attribution correction, the cooling execution interference, the cooling execution interference rollback result, the temperature reference interference set, and the operating condition deviation sample set are written into the net contribution result of maintenance effect.

Images