Transformer intelligent monitoring and fault early warning system and method

Abstract

The invention relates to the technical field of transformer monitoring, in particular to an intelligent monitoring and fault early warning system and method for a transformer. The method comprises the following steps: setting acquisition parameters, respectively acquiring working data of different acquisition targets of a plurality of transformers according to the acquisition parameters, preprocessing all the working data to obtain preprocessed data, then calculating the qualified rate of each transformer based on the preprocessed data, and screening out abnormal transformers according to the qualified rates of the transformers. And finally, calculating the work score of each abnormal transformer, and screening out the faulty transformer according to the work scores of the abnormal transformers. The method can detect a plurality of transformers at the same time, and screen out the faulty transformer from the plurality of transformers based on the work scores of the transformers. The work score is calculated based on different detection items of the transformer, and each detection item has a corresponding weight coefficient, so that the fault transformer can be screened out in time according to different detection items.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer monitoring, and in particular to a transformer intelligent monitoring and fault early warning system and method. Background Art

[0002] ‌A transformer is a device that uses the principle of electromagnetic induction to change AC voltage. Its main components are the primary coil, the secondary coil and the iron core (magnetic core). The main functions of a transformer are: voltage conversion, current conversion, impedance conversion, isolation, voltage stabilization (magnetic saturation transformer), etc. Transformers are basic equipment for power transmission and distribution, and are widely used in industry, agriculture, transportation, urban communities and other fields.

[0003] A Chinese patent with announcement number CN115268350B discloses a fault warning method and system for a voltage-stabilizing transformer. The method groups the fault factor set based on the fault phenomenon clustering result, generates the fault factor grouping result, traverses the accessory fault factor set to generate a fault probability list and determines whether the preset fault probability is met, sorts the fault factors that meet the preset fault probability, and matches the associated fault phenomenon clustering result according to the fault factor sorting result for warning. However, the prior art only clusters the fault factors, but does not mention how to conduct detailed inspections on different detection items of the transformer, resulting in the inability to quickly screen out the transformers with the most serious faults. Summary of the invention

[0004] The purpose of the present invention is to solve the problems existing in the background technology and to propose a transformer intelligent monitoring and fault early warning system and method.

[0005] The technical solution of the present invention: On the one hand, the present application provides a transformer intelligent monitoring and fault warning method, comprising: Set acquisition parameters, and respectively acquire working data of different acquisition targets of multiple transformers according to the acquisition parameters; Preprocess all working data to obtain preprocessed data; Calculate the qualified rate of each transformer based on the preprocessed data, and screen out abnormal transformers according to the qualified rate of the transformers; The working score of each abnormal transformer is calculated, and the faulty transformers are screened out according to the working scores of the abnormal transformers.

[0006] Preferably, the acquisition parameters are set, and the working data of different acquisition targets of multiple transformers are respectively acquired according to the acquisition parameters, including: Create a work data table; A transformer is selected, and collection parameters are set for each collection target of the transformer; the collection targets include transformer temperature data, transformer voltage data, and transformer oil quality data, and the collection parameters include collection period and collection frequency; According to the acquisition parameters of each acquisition target of the transformer, the working data of multiple acquisition targets are respectively acquired, thereby obtaining multiple sub-data groups, and the multiple sub-data groups are put into the working data table; Return to select a transformer until all transformers are selected and the working data of each transformer is obtained.

[0007] Preferably, all working data are preprocessed to obtain preprocessed data, including: Randomly select a transformer and all the working data of the transformer from the working data table; Select a sub-data group from all the working data of the transformer, and determine whether each acquisition node of the sub-data group has corresponding acquisition data; If there is no corresponding collected data under the collection node of the sub-data group, the collection node with corresponding collected data will be recorded as a missing node; Copy the collected data corresponding to any collection node adjacent to the missing node to the missing node to fill the missing node; If each acquisition node of the sub-data group has corresponding acquisition data, then return to select a sub-data group from all the working data of the transformer until all the sub-data groups of the transformer are selected to obtain the pre-processed data of the transformer; Return to randomly select a transformer and all the working data of the transformer from the working data table until all transformers are selected to obtain the preprocessing data of all transformers.

[0008] Preferably, after the step of marking the acquisition node with corresponding acquisition data as a missing node if there is an acquisition node in the sub-data group but no corresponding acquisition data, the method further includes: Set missing threshold; Determine whether the number of missing nodes in the sub-data group is greater than or equal to a missing threshold; If the number of missing nodes in the sub-data group is greater than or equal to the missing threshold, the transformer corresponding to the sub-data group is recorded as a faulty transformer.

[0009] Preferably, the qualified rate of each transformer is calculated based on the preprocessed data, and abnormal transformers are screened out according to the qualified rate of the transformers, including: Select a transformer and all preprocessed data of the transformer; Select a sub-data group; Set eligibility thresholds; Determine whether each working data of the sub-data group is greater than or equal to a qualified threshold; If each working data of the sub-data group is greater than or equal to the qualified threshold, the sub-data group is qualified; Return to select a sub-data group until all sub-data groups of the transformer are selected, and calculate the qualified rate of the sub-data groups of the transformer; Return to select a transformer and all pre-processed data of the transformer until all transformers are selected and the qualified rate of each transformer is obtained.

[0010] Preferably, the qualified rate of each transformer is calculated based on the preprocessed data, and abnormal transformers are screened out according to the qualified rate of the transformers, and further includes: Select all transformers; All transformers are sorted from small to large according to the qualified rate of the transformers, and the transformers in the first M are screened out; the first M transformers are recorded as abnormal transformers.

[0011] Preferably, before calculating the working score of each abnormal transformer and screening out the faulty transformer according to the working score of the abnormal transformer, the method includes: Acquire historical data of multiple transformers; each historical data of a transformer includes multiple historical sub-data groups; Select a historical sub-data group from the historical data of multiple transformers, and calculate the average value of the historical sub-data group; Return and select a historical sub-data group from the historical data of multiple transformers until the average value of each historical sub-data group is obtained; The average values ​​of all historical sub-data groups are normalized to obtain a normalized average value, which is recorded as the score of the sub-data group.

[0012] Preferably, calculating the working score of each abnormal transformer and screening out the faulty transformer according to the working score of the abnormal transformer includes: Select an abnormal transformer and all preprocessed data of the abnormal transformer; Setting a corresponding weight coefficient for each sub-data group included in the preprocessed data; The working score of the abnormal transformer is calculated by combining the weight coefficient of the sub-data group through formula 1; Formula 1; in, is the working score of the abnormal transformer, is the score of the ith sub-dataset of the anomaly transformer, is the historical score of the i-th sub-data group of the anomaly transformer, is the weight coefficient of the abnormal i-th sub-data group, is the total number of sub-data groups; Return and select an abnormal transformer and all preprocessed data of the abnormal transformer until the working scores of all abnormal transformers are obtained.

[0013] Preferably, calculating the working score of each abnormal transformer and screening out the faulty transformer according to the working score of the abnormal transformer also includes: All abnormal transformers are sorted in descending order according to the work scores, and the top N abnormal transformers are selected; the top N abnormal transformers are recorded as faulty transformers; Suspend the operation of the faulty transformer and issue an early warning.

[0014] On the other hand, the present application also provides a transformer intelligent monitoring and fault warning system, including: A collection component, through which the working data of the transformer is collected; A processing component, wherein the processing component includes a processing unit and a data unit, processes the working data of the transformer through the data unit, and executes the transformer intelligent monitoring and fault warning method as described in any one of the above on the transformer through all processing units combined with the processing results.

[0015] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects: By setting the acquisition parameters, the working data of different acquisition targets of multiple transformers are collected respectively according to the acquisition parameters, and all the working data are preprocessed to obtain preprocessed data. Then, the qualified rate of each transformer is calculated based on the preprocessed data, and the abnormal transformers are screened out according to the qualified rate of the transformer. Finally, the working score of each abnormal transformer is calculated, and the faulty transformer is screened out according to the working score of the abnormal transformer. The present application can detect multiple transformers at the same time, and screen out the faulty transformer from the multiple transformers based on the working score of the transformer. Since the working score is calculated based on different detection items of the transformer, and each detection item has a corresponding weight coefficient, the present application can focus on different detection items to screen out faulty transformers in time. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of a flow chart of a transformer intelligent monitoring and fault early warning method proposed by the present invention; Figure 2 This is a structural schematic diagram of a transformer intelligent monitoring and fault early warning system proposed by the present invention.

[0017] Reference numerals: 100, acquisition component; 200, processing component; 201, processing unit; 202, data unit. DETAILED DESCRIPTION

[0018] Embodiment 1, as Figure 1 As shown, the present invention proposes a transformer intelligent monitoring and fault early warning method, comprising: S100, setting acquisition parameters, and respectively acquiring working data of different acquisition targets of multiple transformers according to the acquisition parameters; S200, preprocessing all working data to obtain preprocessed data; S300, calculating the qualified rate of each transformer based on the preprocessed data, and screening out abnormal transformers according to the qualified rate of the transformers; S400, calculating the working score of each abnormal transformer, and screening out the faulty transformers according to the working scores of the abnormal transformers.

[0019] In the present invention, by setting acquisition parameters, the working data of different acquisition targets of multiple transformers are respectively acquired according to the acquisition parameters, and all the working data are preprocessed to obtain preprocessed data, and then the qualified rate of each transformer is calculated based on the preprocessed data, and abnormal transformers are screened out according to the qualified rate of the transformer, and finally the working score of each abnormal transformer is calculated, and the faulty transformer is screened out according to the working score of the abnormal transformer. The present application can detect multiple transformers at the same time, and screen out faulty transformers from multiple transformers based on the working score of the transformer. Since the working score is calculated based on different detection items of the transformer, and each detection item has a corresponding weight coefficient, the present application can focus on different detection items to screen out faulty transformers in time.

[0020] In an optional embodiment, the S100 includes: S110, creating a work data table; S120, selecting a transformer, and setting acquisition parameters for each acquisition target of the transformer; the acquisition targets include transformer temperature data, transformer voltage data, and transformer oil quality data group, and the acquisition parameters include acquisition cycle and acquisition frequency; Specifically, the types of transformer collection targets can be adjusted according to the transformer detection items. Each collection target corresponds to a transformer detection item. The more detection items a transformer has, the more transformer collection targets there are. S130, respectively collecting working data of a plurality of collection targets according to the collection parameters of each collection target of the transformer, thereby obtaining a plurality of sub-data groups, and putting the plurality of sub-data groups into a working data table; Specifically, since each acquisition target corresponds to a detection item of a transformer, each sub-data group contains the same detection data, that is, for transformer A, if transformer A contains sub-data group A-1, sub-data group A-2 and sub-data group A-3, then the working data contained in sub-data group A-1 is the transformer temperature data of different acquisition nodes, the working data contained in sub-data group A-2 is the transformer voltage data of different acquisition nodes, and the working data contained in sub-data group A-3 is the transformer oil quality data of different acquisition nodes; S140, return to step S120, until all transformers are selected, and the working data of each transformer is obtained.

[0021] It should be noted that the present application is to monitor multiple transformers on the transmission line at the same time. Therefore, when collecting transformer data, it is necessary to collect the working data of multiple transformers at the same time, so as to facilitate the calculation of the working scores of multiple transformers, and finally screen out the faulty transformers according to the working scores of the transformers.

[0022] In order to ensure that the working data of multiple transformers are sufficiently credible, when setting the acquisition parameters, the acquisition parameters of different transformers for the same acquisition target should be kept as consistent as possible, so as to ensure that the working data of the same acquisition target of different transformers are comparable to each other, thereby facilitating the subsequent calculation of the working scores of different transformers.

[0023] In an optional embodiment, the S200 includes: S210, randomly selecting a transformer and all working data of the transformer from the working data table; S220, selecting a sub-data group from all the working data of the transformer, and determining whether each collection node of the sub-data group has corresponding collection data; S230, if there is a collection node in the sub-data group but no corresponding collection data, the collection node with corresponding collection data is recorded as a missing node; S240, copying the collection data corresponding to any collection node adjacent to the missing node to the missing node to fill the missing node; S250, if each acquisition node of the sub-data group has corresponding acquisition data, return to step S220 until all sub-data groups of the transformer are selected to obtain the pre-processed data of the transformer; S260, return to step S210, until all transformers are selected, and the preprocessing data of all transformers are obtained.

[0024] It should be noted that since there may be omissions in the process of collecting the working data of the transformer, resulting in missing working data, if the missing data cannot be supplemented in time, it may affect the subsequent calculation of the transformer's working score.

[0025] In an optional embodiment, after S230, the method further includes: S231, setting a missing threshold; S232, determining whether the number of missing nodes in the sub-data group is greater than or equal to a missing threshold; S233: If the number of missing nodes in the sub-data group is greater than or equal to the missing threshold, the transformer corresponding to the sub-data group is recorded as a faulty transformer.

[0026] It should be noted that if a sub-data group has multiple missing nodes, it may be that the acquisition component corresponding to the sub-data group is damaged or the transformer corresponding to the sub-data group is faulty. In either case, an early warning needs to be issued for the transformer.

[0027] In an optional embodiment, the S300 includes: S310, selecting a transformer and all preprocessed data of the transformer; S320, selecting a sub-data group and calculating the average value of the sub-data group; S330, setting a qualified range based on the average value of each sub-data group; S340, determining whether each working data of the sub-data group is within a qualified range; S350, if each working data of the sub-data group is within the qualified range, the sub-data group is qualified; S360, returning to select a sub-data group until all sub-data groups of the transformer are selected, and calculating the qualified rate of the sub-data groups of the transformer; S370, returning to select a transformer and all pre-processed data of the transformer until all transformers are selected, and obtaining the qualified rate of each transformer.

[0028] It should be noted that when executing step 330, the average value of the sub-data group can be used as a benchmark, and the deviation of 20% can be used as a deviation threshold to obtain the qualified range of the sub-data group. Since the working data of the transformer is in a relatively stable state when it is working, if there is a sub-data group whose working data is not within the qualified range of the sub-data group, it means that the detection item corresponding to the sub-data group may be abnormal. If a transformer has multiple sub-data groups corresponding to abnormal detection items, it will result in a low qualified rate of the transformer, and it can be preliminarily determined that the transformer is an abnormal transformer.

[0029] In an optional embodiment, the S300 further includes: S380, select all transformers; S390, sorting all transformers from small to large according to the qualified rates of the transformers, screening out the transformers in the first M order; and recording the first M transformers as abnormal transformers.

[0030] It should be noted that the larger the value of M is, the greater the number of abnormal transformers will be. After calculating the qualified rate of each transformer, the first M transformers with lower qualified rates are screened out. Since these M transformers have the lowest qualified rates, the possibility of problems with these M transformers is the greatest, so these transformers are marked as abnormal transformers to facilitate subsequent steps to determine whether the abnormal transformers are actually faulty.

[0031] In an optional embodiment, before S400, the following steps are included: K100, obtains historical data of multiple transformers; each transformer's historical data contains multiple historical sub-data groups; K200, selects a historical sub-data group from the historical data of multiple transformers and calculates the average value of the historical sub-data group; K300, returns to select a historical sub-data group from the historical data of multiple transformers until the average value of each historical sub-data group is obtained; K400, normalize the average values ​​of all historical sub-data groups to obtain the normalized average value, and record the normalized average value as the score of the sub-data group; Specifically, normalization is to scale the data to a specific range, usually to eliminate the impact of data of different magnitudes on model training and prediction. Normalization can improve model performance, speed up model training, and improve model accuracy.

[0032] It should be noted that before calculating the working score of the transformer through formula 1, it is necessary to first determine the score of each collection target of the transformer, and then the working score of the transformer can be calculated by combining the score of the collection target and the weight coefficient of the collection target. The reason why the historical average value of the collection target is selected as the scoring standard for the collection target is to ensure that the working data collected in real time is of reference to the historical working data.

[0033] In an optional embodiment, the S400 includes: S410, selecting an abnormal transformer and all preprocessed data of the abnormal transformer; S420, setting a corresponding weight coefficient for each sub-data group included in the pre-processed data; S430, calculating the working score of the abnormal transformer by combining the weight coefficient of the sub-data group through formula 1; Formula 1; in, is the working score of the abnormal transformer, is the score of the ith sub-dataset of the anomaly transformer, is the historical score of the i-th sub-data group of the anomaly transformer, is the weight coefficient of the abnormal i-th sub-data group, is the total number of sub-data groups; Specifically, when calculating the score of the i-th sub-data group of the abnormal transformer, the average value of the sub-data group is calculated first, and then normalized to obtain the score of the sub-data group; Since the weight coefficient can be adjusted freely, the key detection items of transformer fault detection can be adjusted by adjusting the size of the weight coefficient. The larger the weight coefficient corresponding to the acquisition target, the more emphasis is placed on the detection items corresponding to the acquisition target. S440, returning to step S410, until the working scores of all abnormal transformers are obtained.

[0034] It should be noted that, through the aforementioned embodiment, M abnormal transformers with low pass rates are determined from multiple transformers. In order to further find the faulty transformer from the abnormal transformers, the working score of each abnormal transformer is calculated by formula 1, and the working score is used as the basis for evaluating whether the abnormal transformer is normal.

[0035] The working score is composed of the scores of each sub-data group of the transformer, so the working score can reflect whether the transformer is working normally. The higher the working score, the further the abnormal transformer is from the normal working state of the transformer, that is, the greater the possibility of failure.

[0036] Since a corresponding weight coefficient can be set for each sub-data group, different collection targets are given different weights, so that when checking transformer faults, different detection items can be focused on. For example, the collection targets of the transformer are transformer temperature data, transformer voltage data and transformer oil quality data, and the corresponding weights of each collection target are 1.0, 1.0 and 2.0 respectively. It can be seen that the main target of this transformer fault inspection is the transformer oil quality data collection target. Since the weight coefficient corresponding to the transformer oil quality data is relatively high, as long as there is a little abnormal value in the transformer oil quality data, the transformer's working score will increase rapidly, so that it is easier to find transformers with faults in the transformer oil quality data.

[0037] In an optional embodiment, the S400 further includes: S450, sorting all abnormal transformers in descending order according to the work scores, screening out the top N abnormal transformers; recording the top N abnormal transformers as faulty transformers; S460, suspend the operation of the faulty transformer and issue an early warning.

[0038] It should be noted that the working score of each abnormal transformer is calculated by formula 1. Since the larger the working score, the greater the probability that the abnormal transformer will have problems in this detection, N abnormal transformers with higher working scores are screened out, and these N abnormal transformers are marked as faulty transformers, and the work of the faulty transformer is suspended.

[0039] The present application detects the qualified rate of each transformer first, thereby preliminarily screening out M abnormal transformers with low qualified rates from multiple transformers, and then calculates the working scores of the M abnormal transformers, and uses the working scores as a criterion for judging whether the transformer is working normally, thereby screening out N faulty transformers with high working scores, suspending the operation of the faulty transformer and issuing an early warning after the faulty transformer is detected, and reducing the possibility of missed judgment and misjudgment during transformer detection by performing double detection on the transformer.

[0040] The present application also provides a transformer intelligent monitoring and fault warning system, including a collection component 100 and a processing component 200.

[0041] The working data of the transformer is collected by the collection component 100, and the processing component 200 includes a processing unit 201 and a data unit 202. The working data of the transformer is processed by the data unit 202, and the transformer intelligent monitoring and fault early warning method described in the first embodiment is performed on the transformer by the processing unit 201 in combination with the processing result; Specifically, the present application detects multiple transformers at the same time, so each transformer is provided with at least one corresponding acquisition component 100 .

[0042] It should be noted that the working data of multiple transformers are collected by the collection component 100. Since the collection component 100 is communicatively connected with the processing component, the collected working data can be transmitted to the processing component 200. Then, the processing component 200 combines the working data of multiple transformers to screen out faulty transformers and issue an early warning.

[0043] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge scope of technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. A transformer intelligent monitoring and fault early warning method, characterized in that: include: Set acquisition parameters, and respectively acquire working data of different acquisition targets of multiple transformers according to the acquisition parameters; Preprocess all working data to obtain preprocessed data; Calculate the qualified rate of each transformer based on the preprocessed data, and screen out abnormal transformers according to the qualified rate of the transformers; The working score of each abnormal transformer is calculated, and the faulty transformers are screened out according to the working scores of the abnormal transformers.

2. A transformer intelligent monitoring and fault early warning method according to claim 1, characterized in that: Set the acquisition parameters and collect the working data of different acquisition targets of multiple transformers according to the acquisition parameters, including: Create a work data table; A transformer is selected, and collection parameters are set for each collection target of the transformer; the collection targets include transformer temperature data, transformer voltage data, and transformer oil quality data, and the collection parameters include collection period and collection frequency; According to the acquisition parameters of each acquisition target of the transformer, the working data of multiple acquisition targets are respectively acquired, thereby obtaining multiple sub-data groups, and the multiple sub-data groups are put into the working data table; Return to select a transformer until all transformers are selected and the working data of each transformer is obtained.

3. A transformer intelligent monitoring and fault early warning method according to claim 2, characterized in that: All working data are preprocessed to obtain preprocessed data, including: Randomly select a transformer and all the working data of the transformer from the working data table; Select a sub-data group from all the working data of the transformer, and determine whether each acquisition node of the sub-data group has corresponding acquisition data; If there is no corresponding collected data under the collection node of the sub-data group, the collection node with corresponding collected data will be recorded as a missing node; Copy the collected data corresponding to any collection node adjacent to the missing node to the missing node to fill the missing node; If each acquisition node of the sub-data group has corresponding acquisition data, then return to select a sub-data group from all the working data of the transformer until all the sub-data groups of the transformer are selected to obtain the pre-processed data of the transformer; Return to randomly select a transformer and all the working data of the transformer from the working data table until all transformers are selected to obtain the preprocessing data of all transformers.

4. A transformer intelligent monitoring and fault early warning method according to claim 3, characterized in that: After the step of marking the acquisition node with corresponding acquisition data as a missing node if there is an acquisition node in the sub-data group but no corresponding acquisition data, the method further includes: Set missing threshold; Determine whether the number of missing nodes in the sub-data group is greater than or equal to a missing threshold; If the number of missing nodes in the sub-data group is greater than or equal to the missing threshold, the transformer corresponding to the sub-data group is recorded as a faulty transformer.

5. A transformer intelligent monitoring and fault early warning method according to claim 4, characterized in that: The qualified rate of each transformer is calculated based on the preprocessed data, and abnormal transformers are screened out according to the qualified rate of the transformers, including: Select a transformer and all preprocessed data of the transformer; Select a sub-data group; Set eligibility thresholds; Determine whether each working data of the sub-data group is greater than or equal to a qualified threshold; If each working data of the sub-data group is greater than or equal to the qualified threshold, the sub-data group is qualified; Return to select a sub-data group until all sub-data groups of the transformer are selected, and calculate the qualified rate of the sub-data groups of the transformer; Return to select a transformer and all pre-processed data of the transformer until all transformers are selected and the qualified rate of each transformer is obtained.

6. A transformer intelligent monitoring and fault early warning method according to claim 5, characterized in that: The qualified rate of each transformer is calculated based on the preprocessed data, and abnormal transformers are screened out according to the qualified rate of the transformers. It also includes: Select all transformers; All transformers are sorted from small to large according to the qualified rate of the transformers, and the transformers in the first M are screened out; the first M transformers are recorded as abnormal transformers.

7. A transformer intelligent monitoring and fault early warning method according to claim 6, characterized in that: Before calculating the working score of each abnormal transformer and screening out the faulty transformer according to the working score of the abnormal transformer, the following steps are included: Acquire historical data of multiple transformers; each historical data of a transformer includes multiple historical sub-data groups; Select a historical sub-data group from the historical data of multiple transformers, and calculate the average value of the historical sub-data group; Return and select a historical sub-data group from the historical data of multiple transformers until the average value of each historical sub-data group is obtained; The average values ​​of all historical sub-data groups are normalized to obtain a normalized average value, which is recorded as the score of the sub-data group.

8. A transformer intelligent monitoring and fault early warning method according to claim 7, characterized in that: Calculate the working score of each abnormal transformer, and screen out faulty transformers based on the working scores of abnormal transformers, including: Select an abnormal transformer and all preprocessed data of the abnormal transformer; Setting a corresponding weight coefficient for each sub-data group included in the preprocessed data; The working score of the abnormal transformer is calculated by combining the weight coefficient of the sub-data group through formula 1; Formula 1; in, is the working score of the abnormal transformer, is the score of the ith sub-dataset of the anomaly transformer, is the historical score of the i-th sub-data group of the anomaly transformer, is the weight coefficient of the abnormal i-th sub-data group, is the total number of sub-data groups; Return and select an abnormal transformer and all preprocessed data of the abnormal transformer until the working scores of all abnormal transformers are obtained.

9. A transformer intelligent monitoring and fault early warning method according to claim 8, characterized in that: Calculate the working score of each abnormal transformer, and screen out the faulty transformers according to the working scores of the abnormal transformers, and also include: All abnormal transformers are sorted in descending order according to the work scores, and the top N abnormal transformers are selected; the top N abnormal transformers are recorded as faulty transformers; Suspend the operation of the faulty transformer and issue an early warning.

10. A transformer intelligent monitoring and fault warning system, characterized in that: include: A collection component, through which the working data of the transformer is collected; A processing component, wherein the processing component includes a processing unit and a data unit, the working data of the transformer is processed by the data unit, and the transformer intelligent monitoring and fault warning method as described in any one of claims 1 to 9 is executed on the transformer by the processing unit in combination with the processing result.

Images