Transformer data abnormity monitoring method and system based on data processing

Abstract

The invention relates to the technical field of data processing, in particular to a transformer data anomaly monitoring method and system based on data processing. The method comprises the following steps: dividing state data of a target transformer into key state data and expanded state data, determining a target state matrix according to the key state data, performing state prediction through the target state matrix, performing anomaly classification and identification to obtain an anomaly probability vector, and judging the anomaly probability vector to obtain an abnormal state. And whether the target state matrix is updated in combination with the extended state data is determined according to the abnormal probability vector discrimination result so as to recalculate, and through a stage processing mode, the monitoring efficiency can be effectively improved, and the monitoring accuracy can be ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly relates to a method and system for monitoring abnormal transformer data based on data processing. Background Art

[0002] In the operation of a power system, as a core device, the stability of the operation state of a transformer is directly related to the reliable power supply of the power grid. With the advancement of the construction of a smart grid, various monitoring devices are usually deployed on the transformer, such as sensors, on-line monitoring devices, etc., which can collect a large amount of real-time status data, covering multi-dimensional information such as electrical parameters, oil service data, and mechanical vibration.

[0003] However, the increase in the information dimension also means an increase in the computational complexity. When the prior art uses a prediction model to predict transformer data and judge faults or anomalies based on the predicted data, it is difficult to meet the requirement of the timeliness of abnormal monitoring.

[0004] Therefore, how to improve the efficiency of abnormal monitoring of transformer data on the premise of ensuring the accuracy of abnormal monitoring of transformer data has become an urgent problem to be solved. Summary of the Invention

[0005] In order to solve the problem that the abnormal monitoring of transformer data in the prior art needs to process multi-dimensional data, resulting in a high computational complexity and thus a low efficiency of abnormal monitoring of transformer data, the present invention provides a method and system for monitoring abnormal transformer data based on data processing.

[0006] In a first aspect, the present invention provides a method for monitoring abnormal transformer data based on data processing, adopting the following technical solution: A transformer data anomaly monitoring method based on data processing, comprising: obtaining key state data and extended state data of a target transformer at M preset time points respectively, where M is a positive integer; determining a target state matrix according to the M key state data; inputting the target state matrix into a trained state prediction model to obtain predicted state data corresponding to the (M + 1)-th preset time point to the (M + N)-th preset time point respectively, where N is a positive integer; inputting the M first state data and the N predicted state data into a trained anomaly classification model to obtain an anomaly probability vector formed by predicted anomaly probabilities corresponding to K preset anomaly categories respectively, where K is a positive integer; inputting the anomaly probability vector into a trained discrimination model to obtain a discrimination result; if the discrimination result does not meet a first preset condition, determining transformer anomaly information according to the anomaly probability vector; if the discrimination result meets the first preset condition, updating the target state matrix according to the M key state data and the M extended state data, and returning to execute the step of inputting the target state matrix into the trained state prediction model until a new anomaly probability vector is obtained, and determining the transformer anomaly information according to the new anomaly probability vector.

[0007] The beneficial effects are as follows: The state data of the target transformer is divided into key state data and extended state data. The target state matrix is determined according to the key state data, state prediction is carried out through the target state matrix, then anomaly classification and recognition are carried out to obtain an anomaly probability vector, the anomaly probability vector is discriminated, and whether to update the target state matrix by combining the extended state data for recalculation is determined according to the discrimination result of the anomaly probability vector. Through the staged processing method, the monitoring efficiency can be effectively improved and the monitoring accuracy can be guaranteed.

[0008] Further, the target transformer corresponds to a plurality of preset attributes, and the preset attributes have uniquely corresponding priority levels. The preset attributes at least include voltage, current, power, winding DC resistance, oil temperature, oil level, winding temperature, and dissolved gas concentration; The key state data includes state values corresponding to a plurality of preset attributes at the first priority level respectively; The extended state data includes state values corresponding to a plurality of preset attributes from the second priority level to the R-th priority level respectively, where R is a positive integer.

[0009] Further, the process of determining the priority level corresponding to the preset attribute includes the following steps: Obtaining attribute sample data corresponding to each preset attribute respectively; Using the principal component analysis method to perform dimensionality reduction processing on the attribute sample data corresponding to each preset attribute to obtain S reference dimension data, where S is a positive integer; For any attribute sample data, according to the cosine distance between the attribute sample data and each reference dimension data, determine the minimum value among the cosine distances as the associated evaluation value, and determine the reference dimension data corresponding to the associated evaluation value as the associated dimension data of the attribute sample data; Perform clustering processing on the associated evaluation values corresponding to each attribute sample data to obtain a number of clustering sets; For any attribute sample data, according to the clustering set to which the attribute sample data belongs, determine the priority level of the preset attribute corresponding to the attribute sample data.

[0010] The beneficial effects are as follows: Through dimensionality reduction processing by principal component analysis, reference dimension data is obtained, and clustering analysis is performed according to the cosine distance between the attribute sample data and the reference dimension data that is closest, so that the priority level of the preset attribute can be determined according to the characterization ability of the attribute sample data corresponding to the preset attribute, enabling the efficiency of transformer data anomaly monitoring to be improved when using key state data for prediction, and also ensuring the accuracy of transformer data anomaly monitoring in most scenarios.

[0011] Further, the step of "for any attribute sample data, according to the clustering set to which the attribute sample data belongs, determine the priority level of the preset attribute corresponding to the attribute sample data" includes: According to the clustering center values corresponding to each clustering set, sort each clustering set in a preset order to obtain a clustering set sequence; For any attribute sample data, if there is no other attribute sample data in the clustering set to which the attribute sample data belongs that has the same associated dimension data as the attribute sample data, then according to the position of the clustering set to which the attribute sample data belongs in the clustering set sequence, determine the priority level of the preset attribute corresponding to the attribute sample data; If there is another attribute sample data in the clustering set to which the attribute sample data belongs that has the same associated dimension data as the attribute sample data, and the priority level of the preset attribute corresponding to the other attribute sample data has not been determined, then according to the position of the clustering set to which the attribute sample data belongs in the clustering set sequence, determine the priority level of the preset attribute corresponding to the attribute sample data; If there is another attribute sample data in the clustering set to which the attribute sample data belongs that has the same associated dimension data as the attribute sample data, and the priority level of the preset attribute corresponding to the other attribute sample data has been determined, then increase the position of the clustering set to which the attribute sample data belongs in the clustering set sequence by one, and use the increased result as the priority level of the preset attribute corresponding to the attribute sample data.

[0012] The beneficial effects are as follows: By analyzing the clustering set, more priority levels are divided for the preset attributes, further improving the flexibility of transformer data anomaly monitoring. At the same time, the dimension of the key status data is further streamlined according to the correlation between the preset attributes, thereby achieving a further improvement in the efficiency of transformer data anomaly monitoring with a relatively small loss in the accuracy of transformer data anomaly monitoring.

[0013] Further, the step of inputting the anomaly probability vector into the trained discrimination model to obtain a discrimination result includes: Inputting the anomaly probability vector into the trained discrimination model to obtain the discrimination result and its corresponding discrimination probability; Correspondingly, the step of, if the discrimination result meets the first preset condition, updating the target status matrix according to the M key status data and M extended status data, and returning to execute the step of inputting the target status matrix into the trained status prediction model includes: If the discrimination result meets the first preset condition, determining a reference level according to the discrimination probability corresponding to the discrimination result; For any preset time point from the 1st preset time point to the Mth preset time point, determining a target status vector according to the status values of the preset attributes corresponding to the priority levels less than or equal to the reference level in the extended status data at this preset time point and the key status data corresponding to this preset time point; Traversing all the preset time points from the 1st preset time point to the Mth preset time point, obtaining the target status vectors corresponding to each preset time point respectively, forming the target status matrix from the target status vectors corresponding to each preset time point respectively, and returning to execute the step of inputting the target status matrix into the trained status prediction model.

[0014] Further, the step of determining a reference level according to the discrimination probability corresponding to the discrimination result includes: If the discrimination probability corresponding to the discrimination result is greater than a preset probability threshold, determining the maximum priority level as the reference level; Otherwise, mapping the discrimination probability using a preset mapping function, and determining the reference level according to the mapping result.

[0015] Further, the discrimination result is a probability normal type or a probability abnormal type; The first preset condition is that the discrimination result is the probability abnormal type.

[0016] Further, the step of determining transformer anomaly information according to the anomaly probability vector includes: Determining a reference probability threshold according to the K preset anomaly probabilities in the anomaly probability vector; The abnormal information of the transformer is formed by preset abnormal categories respectively corresponding to preset abnormal probabilities greater than the reference probability threshold.

[0017] In a second aspect, the present invention provides a transformer data anomaly monitoring system based on data processing, adopting the following technical solutions: A transformer data anomaly monitoring system based on data processing includes: a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned transformer data anomaly monitoring method based on data processing is implemented.

[0018] By adopting the above technical solutions, the above-mentioned transformer data anomaly monitoring method based on data processing is generated into a computer program and stored in the memory to be loaded and executed by the processor, so as to manufacture a terminal device according to the memory and the processor, which is convenient to use.

[0019] The present invention has the following technical effects: The state data of the target transformer is divided into key state data and extended state data. The target state matrix is determined according to the key state data, and state prediction is carried out through the target state matrix. Then, anomaly classification and recognition are carried out to obtain an anomaly probability vector. The anomaly probability vector is discriminated, and it is determined whether to update the target state matrix by combining the extended state data to recalculate according to the discrimination result of the anomaly probability vector. Through the stage processing method, the monitoring efficiency can be effectively improved and the monitoring accuracy can be guaranteed. Description of the Drawings

[0020] By reading the following detailed description with reference to the drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become easy to understand. In the drawings, several embodiments of the present invention are shown in an exemplary rather than restrictive manner, and the same or corresponding reference numerals are for the same or corresponding parts.

[0021] Figure 1 It is a method flow chart in a transformer data anomaly monitoring method provided by an embodiment of the present invention. Detailed Embodiments

[0022] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0023] It should be understood that when terms such as "first", "second", etc. are used in the claims, the description and the drawings of the present invention, they are only used to distinguish different objects and not to describe a specific order. The terms "comprising" and "including" used in the description and claims of the present invention indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or their combinations.

[0024] An embodiment of the present invention discloses a method for monitoring abnormal transformer data based on data processing. Refer to Figure 1 , which includes steps S1 - S7: S1: Obtain the key status data and extended status data of the target transformer at M preset time points respectively.

[0025] Wherein, M is a positive integer, and the time interval between adjacent preset time points can be a fixed value, such as 5 seconds, 10 seconds, etc. The target transformer may refer to the transformer for which data anomaly monitoring is required.

[0026] Specifically, the target transformer corresponds to a number of preset attributes, and each preset attribute has a uniquely corresponding priority level. The preset attributes at least include voltage, current, power, winding DC resistance, oil temperature, oil level, winding temperature, dissolved gas concentration; The key status data includes the status values corresponding to a number of preset attributes at the first priority level respectively; The extended status data includes the status values corresponding to a number of preset attributes from the second priority level to the Rth priority level respectively.

[0027] Wherein, R is a positive integer, R is less than or equal to the number of preset attributes. The preset attributes may also include vibration signals, noise, etc. Power can be divided into reactive power and active power, and the dissolved gas concentration can be divided into gas concentration, gas concentration, gas concentration, etc.

[0028] The priority level can represent the key degree of the corresponding preset attribute for subsequent prediction and anomaly classification. The lower the priority level, the higher the priority of the corresponding preset attribute.

[0029] Specifically, the process of determining the priority level corresponding to the preset attribute includes the following steps: Obtain the attribute sample data corresponding to each preset attribute respectively; Use the principal component analysis method to perform dimensionality reduction processing on the attribute sample data corresponding to each preset attribute to obtain S reference dimension data, where S is a positive integer; For any attribute sample data, according to the cosine distance between the attribute sample data and each reference dimension data, determine the minimum value among the cosine distances as the associated evaluation value, and determine the reference dimension data corresponding to the associated evaluation value as the associated dimension data of the attribute sample data; Perform clustering processing on the associated evaluation values corresponding to each attribute sample data to obtain several clustering sets; For any attribute sample data, according to the clustering set to which the attribute sample data belongs, determine the priority level of the preset attribute corresponding to the attribute sample data.

[0030] Among them, the principal component analysis method and the method of determining the number S of reference dimension data by means of the threshold method, scree plot method, etc. when using the principal component analysis method are all prior arts and will not be elaborated here.

[0031] The cosine distance can be used to characterize the proximity degree between the corresponding attribute sample data and the corresponding reference dimension data in the high-dimensional space.

[0032] The associated evaluation value can characterize the proximity degree between the corresponding attribute sample data and the closest reference dimension data in the high-dimensional space, and the associated dimension data can refer to the reference dimension data closest to the corresponding attribute sample data in the high-dimensional space.

[0033] The clustering processing can adopt the DBSCAN clustering algorithm, without the implementer setting the number of clusters, thus affecting the clustering effect.

[0034] Specifically, for any attribute sample data, according to the clustering set to which the attribute sample data belongs, determining the priority level of the preset attribute corresponding to the attribute sample data includes: According to the clustering center values corresponding to each clustering set, sort each clustering set in a preset order to obtain a clustering set sequence; For any attribute sample data, if there is no other attribute sample data in the clustering set to which the attribute sample data belongs that has the same associated dimension data as the attribute sample data, then according to the position of the clustering set to which the attribute sample data belongs in the clustering set sequence, determine the priority level of the preset attribute corresponding to the attribute sample data; If there is another attribute sample data in the clustering set to which the attribute sample data belongs that has the same associated dimension data as the attribute sample data, and the priority level of the preset attribute corresponding to the other attribute sample data has not been determined, then according to the position of the clustering set to which the attribute sample data belongs in the clustering set sequence, determine the priority level of the preset attribute corresponding to the attribute sample data; If there is another attribute sample data in the clustering set to which the attribute sample data belongs, and the associated dimension data corresponding to the two attribute sample data is the same, and the priority level of the preset attribute corresponding to the other attribute sample data has been determined, then increment the position of the clustering set to which the attribute sample data belongs in the clustering set sequence by one, and use the incremented result as the priority level of the preset attribute corresponding to the attribute sample data.

[0035] Among them, the clustering center value can be determined according to the mean of the associated evaluation values corresponding to all attribute sample data in the corresponding clustering set, and the preset order is the order of the clustering center values from small to large. Correspondingly, in the clustering set sequence, the clustering set containing attribute sample data with stronger representation ability is more forward.

[0036] When determining the priority level, when there is only one attribute sample data in the clustering set that uses a certain reference dimension data as the associated dimension data, it can be determined that the attribute sample data best represents the reference dimension data. When there are multiple attribute sample data in the clustering set that use a certain reference dimension data as the associated dimension data, only one of them is selected to represent the reference dimension data, so as to avoid setting strongly correlated preset attributes to high priorities, further streamlining the number of preset attributes with high priorities, and thus improving the efficiency of subsequent model inference.

[0037] S2: Determine the target state matrix according to the M key state data.

[0038] Among them, assume that the key state data is P×1 size data, P is the number of preset attributes of the first priority level, and the target state matrix can be Q×M size data, Q is the number of all preset attributes.

[0039] Initialize all elements in the target state matrix to 0. For any key state data, expand the key state data into intermediate data of Q×1 size. Each row of the intermediate data uniquely corresponds to a preset attribute. For any row in the key state data, assign the element value of that row to the corresponding row element of the intermediate data according to the preset attribute corresponding to it. The elements in the intermediate data that are not assigned values are 0. Then assign values to the corresponding column of the target state matrix according to the key state data. Traverse all key state data to obtain the target state matrix.

[0040] S3: Input the target state matrix into the trained state prediction model to obtain the predicted state data corresponding to the (M + 1)-th to (M + N)-th preset time points respectively.

[0041] Among them, N is a positive integer, and N can be set by the implementer himself. The state prediction model can adopt a time-domain convolutional model, a long short-term memory network model, a recurrent network model, etc.

[0042] In this embodiment, a time-domain convolution model is taken as an example for description. The target state matrix determined according to the key state data corresponding to the first preset time point to the Mth preset time point is input into the trained state prediction model, and the predicted state data corresponding to the (M + 1)th preset time point is predicted. Then, according to the key state data corresponding to the second preset time point to the Mth preset time point and the predicted state data corresponding to the (M + 1)th preset time point, the target state matrix is updated and re-input into the trained state prediction model to predict the predicted state data corresponding to the (M + 2)th preset time point, and so on, so as to obtain the predicted state data corresponding to the (M + 1)th preset time point to the (M + N)th preset time point respectively.

[0043] It should be noted that the size of the predicted state data is also Q×1. Therefore, when updating the target state matrix during the prediction process, it can be directly implemented through the splicing operation.

[0044] The training of the time-domain convolution model can adopt the conventional training method, which will not be elaborated here.

[0045] Since there are only P rows of valid data in the target state matrix, and usually P is much smaller than Q, the model inference efficiency based on the target state matrix is relatively high, which can meet the timeliness of transformer data anomaly monitoring.

[0046] S4: Input the M first state data and the N predicted state data into the trained anomaly classification model to obtain an anomaly probability vector formed by the predicted anomaly probabilities corresponding to K preset anomaly categories respectively.

[0047] Among them, K is a positive integer. The preset anomaly categories may include winding faults, core faults, insulation faults, tap changer faults, cooling system faults, etc. The anomaly classification model may include a convolutional layer and a fully connected layer. The input data size of the convolutional layer may be Q×(M + N), and the output size is 1×K.

[0048] The training process of the anomaly classification model can adopt the conventional training method based on class labels and cross-entropy loss function, which will not be elaborated here.

[0049] In one implementation, the implementer can set a sliding window of size Q×L to slide and extract M + N - L input matrices from the matrix formed by the M first state data and the N predicted state data, and then input them into the anomaly classification model respectively. In this case, the input data size of the anomaly classification model is Q×L. Then, according to the output probability vectors corresponding to each input matrix, the anomaly probability vector is determined by means such as mean and weighting. This method aims to reduce the number of parameters of the anomaly classification model, so as to improve the training effect and efficiency of the anomaly classification model and the inference efficiency of the anomaly classification model, but it will increase the number of inferences. The implementer can design the anomaly classification model according to the actual situation.

[0050] S5: Input the abnormal probability vector into the trained discrimination model to obtain the discrimination result.

[0051] Among them, the discrimination model can be implemented by a classification model. The implementer can process the real data of the transformer through the above steps to obtain the sample probability vector, and then compare the abnormal category indicated by the sample probability vector with the abnormal category of the transformer in the real scenario, manually label the discrimination result label, and then use the cross-entropy loss function to train the discrimination model.

[0052] It should be noted that both the abnormal classification model and the discrimination model are classification models, with relatively fast inference speeds and little impact on the efficiency of abnormal monitoring of transformer data, enabling the time-consuming of abnormal monitoring of transformer data to be within an acceptable range.

[0053] Specifically, inputting the abnormal probability vector into the trained discrimination model to obtain the discrimination result includes: Input the abnormal probability vector into the trained discrimination model to obtain the discrimination result and its corresponding discrimination probability; Correspondingly, if the discrimination result meets the first preset condition, update the target state matrix according to the M key state data and M extended state data, and return to execute the step of inputting the target state matrix into the trained state prediction model, including: If the discrimination result meets the first preset condition, determine the reference level according to the discrimination probability corresponding to the discrimination result; For any preset time point from the 1st preset time point to the Mth preset time point, determine the target state vector according to the state values of the preset attributes corresponding to the priority levels less than or equal to the reference level in the extended state data of this preset time point and the key state data corresponding to this preset time point; Traverse all the preset time points from the 1st preset time point to the Mth preset time point to obtain the target state vectors corresponding to each preset time point respectively, form a target state matrix from the target state vectors corresponding to each preset time point respectively, and return to execute the step of inputting the target state matrix into the trained state prediction model.

[0054] Among them, the first preset condition can be used to judge whether the abnormal probability vector conforms to the norm. The discrimination result meeting the first preset condition indicates that the abnormal probability vector does not conform to the norm, and the target state matrix needs to be updated according to the extended state data and the key state data. The specific update method still uses the assignment method, which will not be elaborated here.

[0055] To ensure the efficiency of data anomaly monitoring, a reference level is determined according to the discrimination probability, and the status values of the preset attributes corresponding to the priority levels less than or equal to the reference level in the extended status data are assigned respectively, so as to reduce the inference calculation pressure of the status prediction model as much as possible on the premise of ensuring the accuracy of data anomaly monitoring, and improve the efficiency of data anomaly monitoring.

[0056] Specifically, determining the reference level according to the discrimination probability corresponding to the discrimination result includes: If the discrimination probability corresponding to the discrimination result is greater than the preset probability threshold, the maximum priority level is determined as the reference level; Otherwise, a preset mapping function is used to map the discrimination probability, and the reference level is determined according to the mapping result.

[0057] Among them, when the discrimination probability corresponding to the discrimination result is greater than the preset probability threshold, it indicates that the unreasonable degree of the abnormal probability vector is relatively serious, and the complete status data is required to support accurate judgment.

[0058] The preset probability threshold can be set to 0.8, and the preset mapping function can be y = (2a - 4)×x / (2b - 1)+(4b - a - 4) / (2b - 1), where x is the discrimination probability, y is the mapping result, a is the preset probability threshold, and b is the number of levels of the priority level.

[0059] After obtaining the mapping result, the mapping result is rounded up, and the rounded-up result is used as the reference level.

[0060] S6: If the discrimination result does not meet the first preset condition, the transformer abnormal information is determined according to the abnormal probability vector.

[0061] Among them, the abnormal information can include the target abnormal category of the target transformer, and the abnormal information can be used to send to the manager to play the effect of early warning and indicating verification.

[0062] Specifically, the discrimination result is a probability normal type or a probability abnormal type; The first preset condition is that the discrimination result is a probability abnormal type.

[0063] Optionally, determining the transformer abnormal information according to the abnormal probability vector includes: Determining a reference probability threshold according to K preset abnormal probabilities in the abnormal probability vector; The transformer abnormal information is formed by the preset abnormal categories corresponding to the preset abnormal probabilities greater than the reference probability threshold.

[0064] Among them, the method for determining the reference probability threshold can adopt threshold determination methods such as the mean method and the Otsu threshold method, which will not be elaborated here.

[0065] The preset abnormal categories corresponding to the preset abnormal probabilities greater than the reference probability threshold are all target abnormal categories, forming transformer abnormal information. It can be known that in this embodiment, multiple target abnormal categories can be output simultaneously, which is more convenient for managers to perform abnormal analysis and traceability, and improves the completeness of data abnormal monitoring.

[0066] S7: If the discrimination result meets the first preset condition, update the target state matrix according to the M key state data and the M extended state data, and return to execute the step of inputting the target state matrix into the trained state prediction model until a new abnormal probability vector is obtained, and determine the transformer abnormal information according to the new abnormal probability vector.

[0067] An embodiment of the present invention also discloses a transformer data abnormal monitoring system based on data processing, including a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the transformer data abnormal monitoring method based on the present invention is implemented.

[0068] The above system also includes other components well-known to those skilled in the art such as a communication bus and a communication interface, and their settings and functions are known in the art, so they will not be described in detail here.

[0069] In the present invention, the foregoing memory can be any tangible medium that contains or stores a program, and the program can be used by or in combination with an instruction execution system, apparatus, or device. For example, the computer-readable storage medium can be any suitable magnetic storage medium or magneto-optical storage medium, or any other medium that can be used to store the required information and can be accessed by an application program, module, or both. Any such computer storage medium can be part of the device or accessible or connectable to the device.

[0070] Although this specification has shown and described multiple embodiments of the present invention, it is obvious to those skilled in the art that such embodiments are provided only by way of example. Those skilled in the art will think of many changes, alterations, and alternative ways without departing from the spirit and scope of the present invention. It should be understood that various alternative solutions to the embodiments of the present invention described herein can be adopted in the process of practicing the present invention.

[0071] The above are all preferred embodiments of the present invention. Without restricting the protection scope of the present invention accordingly, therefore: Any equivalent changes made according to the structure, shape, and principle of the present invention shall be covered within the protection scope of the present invention.

Claims

1. A method for monitoring abnormal transformer data based on data processing, characterized in that, Including: Obtain the key state data and extended state data of the target transformer at M preset time points respectively, where M is a positive integer; Determine the target state matrix according to the M key state data; Input the target state matrix into the trained state prediction model to obtain the predicted state data corresponding to the (M + 1)-th preset time point to the (M + N)-th preset time point respectively, where N is a positive integer; Input the M first state data and the N predicted state data into the trained anomaly classification model to obtain an anomaly probability vector formed by the predicted anomaly probabilities corresponding to K preset anomaly categories respectively, where K is a positive integer; Input the anomaly probability vector into the trained discrimination model to obtain a discrimination result; If the discrimination result does not meet the first preset condition, determine the transformer anomaly information according to the anomaly probability vector; If the discrimination result meets the first preset condition, update the target state matrix according to the M key state data and the M extended state data, and return to execute the step of inputting the target state matrix into the trained state prediction model until a new anomaly probability vector is obtained, and determine the transformer anomaly information according to the new anomaly probability vector.

2. The transformer data anomaly monitoring method based on data processing according to claim 1, wherein The target transformer corresponds to a number of preset attributes, and the preset attributes have unique corresponding priority levels. The preset attributes at least include voltage, current, power, winding DC resistance, oil temperature, oil level, winding temperature, dissolved gas concentration; The key state data includes the state values corresponding to a number of preset attributes at the first priority level respectively; The extended state data includes the state values corresponding to a number of preset attributes from the second priority level to the R-th priority level respectively, where R is a positive integer.

3. The transformer data anomaly monitoring method based on data processing according to claim 2, characterized in that, The process of determining the priority level corresponding to the preset attribute includes the following steps: Obtain the attribute sample data corresponding to each preset attribute respectively; Use the principal component analysis method to perform dimensionality reduction processing on the attribute sample data corresponding to each preset attribute to obtain S reference dimension data, where S is a positive integer; For any attribute sample data, according to the cosine distance between the attribute sample data and each reference dimension data, determine the minimum value among the cosine distances as the correlation evaluation value, and determine the reference dimension data corresponding to the correlation evaluation value as the correlation dimension data of the attribute sample data; Perform clustering processing on the correlation evaluation values corresponding to each attribute sample data to obtain a number of clustering sets; For any attribute sample data, determine the priority level of the preset attribute corresponding to the attribute sample data according to the clustering set to which the attribute sample data belongs.

4. The transformer data anomaly monitoring method based on data processing according to claim 3, wherein For any attribute sample data, determining the priority level of the preset attribute corresponding to the attribute sample data according to the clustering set to which the attribute sample data belongs includes: According to the clustering center values corresponding to each clustering set, sort each clustering set in a preset order to obtain a clustering set sequence; For any attribute sample data, if there is no other attribute sample data in the clustering set to which the attribute sample data belongs that has the same associated dimension data as the attribute sample data respectively, then determine the priority level of the preset attribute corresponding to the attribute sample data according to the position of the clustering set to which the attribute sample data belongs in the clustering set sequence; If there is other attribute sample data in the clustering set to which the attribute sample data belongs that has the same associated dimension data as the attribute sample data respectively, and the priority level of the preset attribute corresponding to the other attribute sample data has not been determined, then determine the priority level of the preset attribute corresponding to the attribute sample data according to the position of the clustering set to which the attribute sample data belongs in the clustering set sequence; If there is other attribute sample data in the clustering set to which the attribute sample data belongs that has the same associated dimension data as the attribute sample data respectively, and the priority level of the preset attribute corresponding to the other attribute sample data has been determined, then increment the position of the clustering set to which the attribute sample data belongs in the clustering set sequence by one, and use the incremented result as the priority level of the preset attribute corresponding to the attribute sample data.

5. The method for monitoring abnormal transformer data based on data processing according to claim 4, characterized in that The step of inputting the anomaly probability vector into the trained discrimination model to obtain a discrimination result includes: Input the anomaly probability vector into the trained discrimination model to obtain the discrimination result and its corresponding discrimination probability; Correspondingly, the step of, if the discrimination result meets the first preset condition, updating the target state matrix according to the M key state data and M extended state data, and returning to execute the step of inputting the target state matrix into the trained state prediction model includes: If the discrimination result meets the first preset condition, determine a reference level according to the discrimination probability corresponding to the discrimination result; For any preset time point from the 1st preset time point to the Mth preset time point, determine a target state vector according to the state values of the preset attributes corresponding to the priority levels less than or equal to the reference level in the extended state data of the preset time point and the key state data corresponding to the preset time point; Traverse all preset time points from the 1st preset time point to the Mth preset time point to obtain the target state vectors corresponding to each preset time point respectively, form the target state matrix from the target state vectors corresponding to each preset time point respectively, and return to execute the step of inputting the target state matrix into the trained state prediction model.

6. The transformer data anomaly monitoring method based on data processing according to claim 5, wherein, The step of determining a reference level according to the discrimination probability corresponding to the discrimination result includes: If the discrimination probability corresponding to the discrimination result is greater than a preset probability threshold, determine the maximum priority level as the reference level; Otherwise, map the discrimination probability using a preset mapping function, and determine the reference level according to the mapping result.

7. The transformer data anomaly monitoring method based on data processing according to claim 1, wherein The discrimination result is of a normal probability type or an abnormal probability type; The first preset condition is that the discrimination result is of the abnormal probability type.

8. The method for monitoring abnormal transformer data based on data processing according to claim 1, wherein The step of determining transformer anomaly information according to the anomaly probability vector includes: Determine a reference probability threshold according to K preset abnormal probabilities in the abnormal probability vector; The transformer abnormal information is formed by preset abnormal categories respectively corresponding to the preset abnormal probabilities greater than the reference probability threshold.

9. A transformer data anomaly monitoring system based on data processing, characterized in that, It includes: A processor and a memory, where the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the transformer data abnormal monitoring method based on data processing according to any one of claims 1-8 is implemented.

Images