Transformer state comprehensive monitoring system based on 5G

Abstract

The invention relates to the technical field of transformer state monitoring, and discloses a 5G-based transformer state comprehensive monitoring system, which comprises a monitoring unit used for collecting state comprehensive data of a transformer to generate a plurality of state comprehensive data packets, and sending the state comprehensive data packets to a processing unit through a 5G module; and the processing unit is used for establishing a processing model library, selecting a processing model based on the multi-state comprehensive data packet, generating an adjusting instruction based on the processing model, and sending the adjusting instruction to the monitoring unit through a 5G module. According to the invention, through accurate monitoring and effective processing of the transformer, the occurrence frequency of transformer faults can be reduced, so that the stability of the whole power system is improved; the stable operation state of the transformer is beneficial for guaranteeing the continuity of power supply, and power fluctuation and power failure accidents caused by faults of the transformer are avoided; and the utilization efficiency of operation and maintenance resources is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of transformer condition monitoring, and particularly to a comprehensive transformer condition monitoring system based on 5G. Background Art

[0002] With the development of artificial intelligence and big data technologies, transformer condition monitoring systems are also evolving towards intelligence and networking to improve the accuracy and efficiency of monitoring.

[0003] Transformer fault types are diverse, including insulation aging, partial discharge, overheating, etc. Diagnostic methods and indicators for each fault are different. In addition, the operating environment and load conditions of the transformer also affect fault diagnosis. Therefore, how to accurately diagnose the type and degree of transformer faults is a complex issue. Summary of the Invention

[0004] The objective of the present invention is that the comprehensive transformer condition monitoring system based on 5G realizes comprehensive, accurate, real-time monitoring and early warning of the transformer operating state through advanced sensing technologies, data analysis algorithms, and the advantages of the 5G network.

[0005] To achieve the above objective, the present invention provides a comprehensive transformer condition monitoring system based on 5G, including:

[0006] A monitoring unit, configured to collect comprehensive transformer condition data to generate multiple comprehensive condition data packets, and send them to the processing unit through a 5G module;

[0007] A processing unit, configured to establish a processing model library, select a processing model based on multiple comprehensive condition data packets, generate an adjustment instruction based on the processing model, and send it to the monitoring unit through a 5G module;

[0008] The processing unit includes:

[0009] A first processing module: configured to generate a state sample set according to the comprehensive transformer condition data of historical transformers;

[0010] A second processing module: configured to establish a processing model according to each state sample, and generate a processing model library by combining all the processing models;

[0011] A third processing module: configured to judge the operating state of the current transformer according to multiple comprehensive condition data packets, select a corresponding state sample based on the operating state of the current transformer, and select a corresponding processing model based on the state sample;

[0012] A fourth processing module: configured to generate an adjustment instruction based on the selected corresponding processing model.

[0013] In some embodiments of the present invention, the first processing module is further configured to:

[0014] Divide the state types of the transformer based on the obtained comprehensive historical state data of the transformer to generate multiple state types;

[0015] Obtain the comprehensive historical state data of the transformer corresponding to each state type;

[0016] The comprehensive historical state data of the transformer includes the historical state data of different monitoring points of the transformer, and generates an operating state feature set A corresponding to the current state type based on the historical state data of different monitoring points, A = {a1, a2... ai... an};

[0017] Among them, ai represents the operating data characteristic value of the monitoring point corresponding to the i-th monitoring point under the current state type, and n represents the total number of monitoring points;

[0018] Combine all the state types of the transformer in the historical state data and the operating state feature set A corresponding to each state type to generate a state sample set B, B = {b1, b2... bj... bm};

[0019] Among them, bj represents the j-th state sample, and m represents the number of state samples.

[0020] In some embodiments of the present invention, the second processing module is further configured to:

[0021] Obtain the historical processing data corresponding to the current state sample, including: processing strategy data and the historical state data of the transformer after processing;

[0022] Generate a primary processing strategy for the current state sample based on the processing strategy data;

[0023] Combine the historical state data of the transformer and the current primary processing strategy to generate an evaluation value C of the current primary processing strategy;

[0024] Judge whether to correct the current primary processing strategy by comparing the evaluation value of the current primary processing strategy with a preset value D;

[0025] If C≥D; then correct the current primary processing strategy to generate a secondary processing strategy;

[0026] If C<D; then retain the current primary processing strategy as the secondary processing strategy;

[0027] Combine the current state sample and the secondary processing strategy corresponding to the current state sample to generate a processing model of the current state sample, and combine all the processing models to generate a processing model library.

[0028] In some embodiments of the present invention, when generating the primary processing strategy of the current state sample, it includes:

[0029] For the obtained operating status feature set A, the operating data feature value ai is compared with the preset value of the operating data of the corresponding monitoring point in sequence, and the operating data feature value where ai is less than the preset value of the operating data is marked as an abnormal feature value;

[0030] Based on all the abnormal feature values, determine the current status sample and generate multiple control instructions;

[0031] Based on the historical processing strategy data, sort all the control instructions to generate the primary processing strategy of the current status sample.

[0032] In some embodiments of the present invention, when generating the evaluation value C of the current primary processing strategy, it includes:

[0033] Obtain the execution period t of the primary processing strategy;

[0034] Based on historical data, predict the transformer load change curve during the execution period t;

[0035] Obtain the proportion of the period during which the transformer load change curve exceeds the preset value within the execution period t;

[0036] Based on the proportion of the period, generate the time reference value T of the current primary processing strategy;

[0037] Based on historical data, judge the correlation between the abnormal feature values in the operating status feature set A;

[0038] Based on the correlation between the abnormal feature values, generate the correlation reference value Y of the current primary processing strategy;

[0039] Combine the time reference value T and the correlation reference value Y of the current primary processing strategy to generate the evaluation value C of the current primary processing strategy;

[0040] C = f1 * T + f2 * Y;

[0041] Wherein, f1 is the weight of the time reference value T, f2 is the weight of the correlation reference value Y, f1 + f2 = 1, P ∈ (0, 1), Y ∈ (0, 1).

[0042] In some embodiments of the present invention, when generating the secondary processing strategy, it includes:

[0043] Based on historical data, set the first preset value Z of the optimization reference value Y;

[0044] If Y ≤ Z, execute the primary optimization instruction to generate the secondary processing strategy;

[0045] If Y > Z, execute the secondary optimization instruction to generate the secondary processing strategy.

[0046] In some embodiments of the present invention, the primary optimization instruction includes:

[0047] Predict the reference value of the load capacity of the transformer under the current primary processing strategy;

[0048] And calculate the load difference within the execution period t based on the load change curve of the transformer within the execution period t and the reference value of the load capacity of the transformer under the current primary processing strategy;

[0049] Generate a transformer scheduling instruction based on the load difference within the execution period t;

[0050] Generate a secondary processing strategy by combining the transformer scheduling instruction and the primary processing strategy.

[0051] In some embodiments of the present invention, the secondary optimization instruction includes:

[0052] Obtain the associated abnormal eigenvalue and historical experience data to determine the factors causing the failure;

[0053] Generate one or more corresponding correction instructions based on the factors causing the failure;

[0054] Predict the implementation effect of each correction instruction and generate a corresponding set of implementation effect reference values E, E = {e1, e2... ek... ev};

[0055] Wherein, ek represents the implementation effect reference value of the kth correction instruction, and v represents the total number of correction instructions;

[0056] Select the correction instruction corresponding to the maximum implementation effect reference value to generate a secondary processing strategy.

[0057] In some embodiments of the present invention, the third processing module is further configured to:

[0058] Receive a variety of state comprehensive data packets of the current transformer sent by the monitoring unit based on the 5G module;

[0059] Generate an operating state feature set A1 of the current monitoring time node by combining the variety of state comprehensive data packets and historical data;

[0060] Determine the state sample of the transformer at the current monitoring time node by combining the operating state feature set A1 of the current monitoring time node and the state sample set B;

[0061] Select a corresponding processing model by combining the state sample of the transformer at the current monitoring time node and the processing model library.

[0062] In some embodiments of the present invention, the fourth processing module is further configured to:

[0063] Obtain the adjustment parameter range of the selected processing model in the third processing module;

[0064] Generate an adjustment instruction by combining the adjustment parameter range and the data in the comprehensive data packet of various current states of the transformer.

[0065] Compared with the prior art, the beneficial effects of a 5G-based comprehensive transformer status monitoring system provided by an embodiment of the present invention are as follows:

[0066] The monitoring unit can collect comprehensive status data of different monitoring points of the transformer, covering various aspects of information; such comprehensive data collection helps to more accurately understand the operating status of the transformer.

[0067] Data transmission through the 5G module enables the processing unit to analyze and make decisions based on the latest data, reducing the problem of inaccurate monitoring caused by data transmission delay or loss.

[0068] By analyzing the comprehensive status data of historical transformers, dividing the status types and generating a status sample set, various possible status situations of the transformer can be better covered.

[0069] By evaluating and correcting the primary processing strategy to generate a secondary processing strategy and then constructing a processing model library, the accuracy and adaptability of the model can be improved, enabling it to give appropriate processing suggestions when facing transformers in different operating states.

[0070] Combining the current comprehensive data packet of various states and historical data, generating a set A1 of operating status characteristics, comparing it with the status sample set B to determine the status sample of the current transformer, and then selecting the corresponding processing model can accurately judge the current operating status of the transformer and accurately locate the problem in a complex operating environment.

[0071] In the process of constructing the processing strategy, by analyzing abnormal characteristic values, predicting the load capacity of the transformer, considering various factors such as the load change curve, potential problems can be discovered before a fault occurs.

[0072] By analyzing associated abnormal characteristic values to determine the fault factors, predicting the implementation effect of the correction instruction and selecting the optimal solution, it helps to take effective preventive measures at the incipient stage of the fault and avoid the deterioration of the fault. BRIEF DESCRIPTION OF THE DRAWINGS

[0073] Figure 1 is a structural diagram of a 5G-based comprehensive transformer status monitoring system provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0074] The following further describes in detail the specific embodiments of the present invention with reference to the drawings and embodiments. The following embodiments are used to illustrate the present invention but are not intended to limit the scope of the present invention.

[0075] In the description of the present invention, it should be understood that the orientation or positional relationship indicated by the terms "center", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation to the present invention.

[0076] The terms "first" and "second" are only used for descriptive purposes and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise stated, the meaning of "a plurality" is two or more.

[0077] In the description of the present invention, it should be noted that unless otherwise clearly specified and defined, the terms "installed", "connected", and "coupled" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected or indirectly connected through an intermediate medium, and it may be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0078] Embodiment 1:

[0079] A 5G-based comprehensive transformer status monitoring system provided by an embodiment of the present invention, as Figure 1 shown, includes:

[0080] A monitoring unit, configured to collect comprehensive transformer status data to generate a variety of comprehensive status data packets and send them to the processing unit through a 5G module;

[0081] A processing unit, configured to establish a processing model library, select a processing model based on a variety of comprehensive status data packets, generate an adjustment instruction based on the processing model, and send it to the monitoring unit through a 5G module;

[0082] The processing unit includes:

[0083] A first processing module: configured to generate a status sample set according to the comprehensive status data of historical transformers;

[0084] A second processing module: configured to establish a processing model according to each status sample and generate a processing model library by combining all the processing models;

[0085] The third processing module: used to comprehensively judge the operating state of the current transformer based on various states of the data packet, select the corresponding state sample based on the operating state of the current transformer, and select the corresponding processing model based on the state sample;

[0086] The fourth processing module: used to generate an adjustment instruction based on the selected corresponding processing model.

[0087] The monitoring unit includes a gas monitoring system for transformer oil, which can continuously and online detect gas components such as hydrogen, carbon monoxide, carbon dioxide, acetylene, ethylene, methane, ethane, etc. dissolved in the oil and the micro-water content, and conducts fault warning monitoring for oil-immersed transformers.

[0088] And through methods such as the graphical method and the three-ratio method, it makes fault analysis and diagnosis, and can quickly complete the monitoring and warning of dissolved gases in transformer oil with high precision and high reliability.

[0089] Analyze and judge the current state and the trend of state development of the transformer, provide analysis and diagnosis results for existing or potential faults of the equipment, so as to provide a reliable decision-making basis for formulating the state maintenance plan of the transformer.

[0090] Obtain a monitoring system for the grounding current of the transformer core and clamping parts, monitor the grounding current of the transformer core, has a continuous real-time automatic detection function, has a stable working ability, and a perfect fault alarm function. When the external power supply disappears, the data is not lost.

[0091] Through thermal infrared imaging, real-time infrared thermal imaging is carried out on the transformer bushing, the high- and low-voltage leads of the transformer, and the transformer body. Real-time collect the temperature of the key areas of the transformer, and provide problem data and infrared thermal imaging display.

[0092] Real-time collect environmental temperature, such as parameters such as humidity, temperature, and noise.

[0093] Embodiment 2:

[0094] The first processing module is also used for:

[0095] Based on the historical state comprehensive data of the obtained transformer, divide the state types of the transformer to generate multiple state types;

[0096] Obtain the historical state comprehensive data of the transformer corresponding to each state type;

[0097] The historical state comprehensive data of the transformer includes the historical state data of different monitoring points of the transformer, and based on the historical state data of different monitoring points, generate an operating state feature set A corresponding to the current state type, A = {a1, a2... ai... an};

[0098] Among them, ai represents the characteristic value of the operation data of the monitoring point corresponding to the i-th monitoring point in the current state type, and n represents the total number of monitoring points;

[0099] Combining all the state types of the transformer in the historical state data and the operation state feature set A corresponding to each state type to generate a state sample set B, B = {b1, b2... bj... bm};

[0100] Among them, bj represents the j-th state sample, and m represents the number of state samples.

[0101] In this embodiment, based on the obtained historical state comprehensive data of the transformer, the state types of the transformer are classified to generate multiple state types. These include different state types such as normal operation state, overload state, overheat state, partial discharge state, etc.

[0102] Obtain the historical state comprehensive data of the transformer corresponding to each state type. These data may include the historical state data of different monitoring points such as the oil temperature, voltage, current, oil level height, vibration frequency, etc. of the transformer.

[0103] Based on the historical state data of different monitoring points, generate the operation state feature set A corresponding to the current state type. For example, if there are n monitoring points, then A = {a1, a2... ai... an}, where ai represents the characteristic value of the operation data of the i-th monitoring point corresponding to the current state type.

[0104] Combining all the state types of the transformer in the historical state data and the operation state feature set A corresponding to each state type to generate a state sample set B. For example, B = {b1, b2... bj... bm}, where bj represents the j-th state sample, m represents the number of state samples, and bj = (j, A), that is, each state sample contains a state number j and the corresponding operation state feature set A.

[0105] Embodiment 3:

[0106] The second processing module is further configured to:

[0107] Obtain the historical processing data corresponding to the current state sample, including: processing strategy data and the historical state data of the transformer after processing;

[0108] Generate a primary processing strategy for the current state sample based on the processing strategy data;

[0109] Combine the historical state data of the transformer and the current primary processing strategy to generate an evaluation value C of the current primary processing strategy;

[0110] Judge whether to correct the current primary processing strategy by comparing the evaluation value of the current primary processing strategy with a preset value D;

[0111] If C ≥ D, then modify the current primary processing strategy to generate a secondary processing strategy;

[0112] If C < D, then retain the current primary processing strategy as the secondary processing strategy;

[0113] Combine the current state sample and the corresponding secondary processing strategy of the current state sample to generate a processing model for the current state sample, and combine all processing models to generate a processing model library.

[0114] In this embodiment, historical processing data corresponding to the current state sample is obtained from the historical database of the system. These data cover the processing strategies adopted for previous similar state samples and the historical state data of the transformer after processing.

[0115] For the processing strategy data, various operations are detailed, such as the specific parameters and execution times of operations such as adjusting the tap position of the transformer tap changer, adjusting the power of the cooling system, and changing the load distribution.

[0116] The historical state data of the transformer after processing includes the data changes of monitoring points such as oil temperature, voltage, current, and oil level after implementing the processing strategy, as well as the overall operation stability index of the transformer.

[0117] Generate the primary processing strategy for the current state sample based on the processing strategy data. First, analyze the operation modes and effects in the processing strategy data. For example, if adjusting the tap position of the tap changer had a good effect on stabilizing the voltage in past similar states, then adjusting the tap position of the tap changer will be listed as an important operation option in the current primary processing strategy.

[0118] At the same time, consider the synergistic relationship between different operations. For example, when adjusting the power of the cooling system, it may be necessary to simultaneously pay attention to the change in the oil level to ensure the normal operation of the cooling system, and this synergistic relationship will also be incorporated into the primary processing strategy.

[0119] Judge whether to modify the current primary processing strategy by comparing the evaluation value C of the current primary processing strategy with the preset value D. The preset value D is preset based on various factors such as the operation requirements, safety, and economy of the transformer.

[0120] For example, if D is set to 0.6, when C ≥ 0.6, it indicates that there may be room for improvement in the current primary processing strategy and it needs to be modified to generate a secondary processing strategy; when C < 0.6, it means that the current primary processing strategy is relatively effective under the evaluation of historical data, and the current primary processing strategy is retained as the secondary processing strategy.

[0121] Generate a processing model for the current state sample by combining the current state sample and the corresponding secondary processing strategy. The processing model includes the features of the state sample and the content of the corresponding secondary processing strategy.

[0122] Combine the processing models of each state sample and generate a processing model library by combining all the processing models. This processing model library provides a basis for subsequently selecting a suitable processing model according to the actual operating state of the transformer.

[0123] Example 4:

[0124] When generating the primary processing strategy for the current state sample, it includes:

[0125] Obtain the operating state feature set A, and successively compare the operating data feature value ai with the preset operating data value of the corresponding monitoring point. Mark the operating data feature value ai that is less than the preset operating data value as an abnormal feature value.

[0126] Determine the current state sample based on all the abnormal feature values and generate multiple control instructions.

[0127] Sort all the control instructions based on the historical processing strategy data to generate the primary processing strategy for the current state sample.

[0128] In this embodiment, after obtaining the operating state feature set A, for each operating data feature value ai (i = 1, 2,..., n) therein, compare it with the preset operating data value of the corresponding monitoring point. For example, for the oil temperature monitoring point, its preset operating data value may be a normal operating temperature range (such as 60 - 80 °C). If the current oil temperature feature value ai is lower than 60 °C, then this ai is marked as an abnormal feature value.

[0129] This comparison process is carried out for each monitoring point one by one. For the voltage monitoring point, there is its corresponding rated voltage range as the preset value; for the current monitoring point, there is also a preset value set according to the rated current of the transformer, etc.

[0130] When all the abnormal feature values are found, determine the current state sample based on these abnormal feature values. For example, if abnormal feature values of oil temperature, voltage, and oil level are found, the combination of these abnormal feature values can determine a specific current state sample, which represents a specific abnormal operating state of the transformer at a certain moment.

[0131] Generate multiple control instructions based on the determined abnormal characteristic values. If the oil temperature is an abnormal characteristic value and is lower than the normal range, the possible control instruction generated may be to increase the heating power of the cooling system (assuming that the too low oil temperature affects the performance of the transformer in a low-temperature environment); if the voltage is lower than the preset value, the instruction to adjust the transformer tap changer to increase the voltage may be generated; if the oil level is too low, the instruction to supplement the transformer oil may be generated, etc.

[0132] Sort all the control instructions based on the historical processing strategy data to generate the first-level processing strategy of the current state sample. The historical processing strategy data includes the effects and priority information of the control instructions taken in the past in similar abnormal states.

[0133] For example, historical data shows that when the oil temperature is low and the voltage is low, first adjusting the tap changer to increase the voltage and then increasing the heating power of the cooling system has a better overall recovery effect on the transformer. Then, sort the corresponding control instructions in this order.

[0134] For some instructions that may have a greater impact on the operation of the transformer, such as instructions involving internal structure adjustment of the transformer, they may be ranked relatively later, while some relatively mild instructions with less impact on the overall operation (such as increasing the monitoring frequency) may be ranked relatively earlier to ensure that when dealing with the abnormal state of the transformer, the problem can be solved in a timely manner and the safe and stable operation of the transformer can be guaranteed to the greatest extent.

[0135] Example 5:

[0136] When generating the evaluation value C of the current first-level processing strategy, it includes:

[0137] Obtain the execution period t of the first-level processing strategy;

[0138] Predict the transformer load change curve within the execution period t based on historical data;

[0139] Obtain the proportion of the period during which the transformer load change curve within the execution period t exceeds the preset value;

[0140] Generate the time reference value T of the current first-level processing strategy based on the period proportion;

[0141] Judge the relevance between the abnormal characteristic values in the operation state feature set A based on historical data;

[0142] Generate the association reference value Y of the current first-level processing strategy based on the relevance between the abnormal characteristic values;

[0143] Combine the time reference value T and the association reference value Y of the current first-level processing strategy to generate the evaluation value C of the current first-level processing strategy;

[0144] C = f1 * T + f2 * Y;

[0145] Among them, f1 is the weight of the time reference value T, f2 is the weight of the associated reference value Y, f1 + f2 = 1, P ∈ (0, 1), Y ∈ (0, 1).

[0146] In this embodiment, the execution period t of the primary processing strategy is clarified, and this execution period t is determined according to the operating conditions of the transformer and the characteristics of the processing strategy. For example, for some processing strategies with short-term adjustments, the execution period t may be several hours; while for strategies involving long-term equipment maintenance and status adjustment, the execution period t may be several days or even weeks.

[0147] Predict the transformer load change curve within the execution period t based on historical data. Utilize historical load data and adopt appropriate prediction methods, such as the ARIMA model (Autoregressive Integrated Moving Average model) in time series analysis or regression algorithms in machine learning (such as linear regression, decision tree regression, etc.).

[0148] Taking the ARIMA model as an example, first conduct a stationarity test on the historical load data to determine the differencing order d, then determine the autoregressive order p and the moving average order q according to the autocorrelation function (ACF) and the partial autocorrelation function (PACF), establish the ARIMA(p, d, q) model, input relevant parameters, and predict the transformer load change curve within the execution period t.

[0149] Set a preset value for the transformer load, and this preset value is determined according to factors such as the rated capacity of the transformer and the requirements for safe operation. For example, for a transformer with a rated capacity of 1000 kVA, considering the safe overload capacity, the preset value may be set to 1100 kVA (i.e., 110% of the rated capacity).

[0150] Obtain the proportion of the time period during which the transformer load change curve within the execution period t exceeds the preset value. By analyzing the predicted load change curve, statistically calculate the time length during which the load exceeds the preset value, and then divide it by the total duration of the execution period t to obtain the proportion of the time period.

[0151] Generate the time reference value T of the current primary processing strategy according to the proportion of the time period. For example, if the proportion of the time period is 0.2 (i.e., 20%), it can be converted into the time reference value T through a mapping function (such as a linear mapping). Assume the mapping function is T = 1 - 0.2 = 0.8 (this is just a simple example, and the actual mapping function may be more complex).

[0152] Judge the correlation between abnormal eigenvalue in the operating state feature set A based on historical data. Adopt a correlation analysis method, such as Pearson correlation coefficient analysis. For two abnormal eigenvalues ai and aj in the operating state feature set A, calculate their Pearson correlation coefficient.

[0153] If the correlation coefficient r is close to 1 or -1, it indicates that the two abnormal eigenvalues are highly correlated; if r is close to 0, it indicates a weak correlation.

[0154] Example 6:

[0155] When generating the secondary processing strategy, it includes:

[0156] Set the first preset value Z of the optimization reference value Y based on historical data;

[0157] If Y ≤ Z, execute the primary optimization instruction to generate the secondary processing strategy;

[0158] If Y > Z, execute the secondary optimization instruction to generate the secondary processing strategy.

[0159] In this embodiment, the first preset value Z of the optimization reference value Y is set based on historical data. The determination of this preset value Z is obtained through the analysis of a large amount of historical data. For example, by analyzing the distribution of the associated reference value Y of the transformer in different operating states in the past, a suitable critical value is selected as Z. Suppose after analysis, it is found that when the associated reference value Y is below 0.3, better results can be obtained by using the primary optimization instruction, then Z is set to 0.3.

[0160] If Y ≤ Z, execute the primary optimization instruction to generate the secondary processing strategy. This is because when the associated reference value Y is small, it means that the correlation between abnormal eigenvalues is weak, and using a relatively simple primary optimization instruction may be sufficient to handle the current state of the transformer.

[0161] If Y > Z, execute the secondary optimization instruction to generate the secondary processing strategy. When Y is large, it indicates that the correlation between abnormal eigenvalues is strong, and more complex secondary optimization instructions are needed to handle it.

[0162] Example 7:

[0163] The primary optimization instruction includes:

[0164] Predict the load capacity reference value of the transformer under the current primary processing strategy;

[0165] And calculate the load difference within the execution period t based on the transformer load change curve within the execution period t and the load capacity reference value of the transformer under the current primary processing strategy;

[0166] Generate a transformer scheduling instruction based on the load difference within the execution period t;

[0167] Combine the transformer scheduling instruction and the primary processing strategy to generate the secondary processing strategy.

[0168] In this embodiment, for the primary optimization instruction, first predict the reference value of the load capacity of the transformer under the current primary processing strategy. This can be predicted based on the rated parameters of the transformer, the current operating status (such as oil temperature, voltage, etc.), and historical load capacity data. For example, using a machine learning algorithm (such as support vector regression), taking the oil temperature, voltage, current, etc. of the transformer as input features, training a model to predict the reference value of the load capacity.

[0169] Calculate the load difference within the execution period t based on the load change curve of the transformer within the execution period t and the reference value of the load capacity of the transformer under the current primary processing strategy. Obtain the load value at each time point from the predicted load change curve, then subtract the reference value of the load capacity to get the load difference at each time point, and finally sum up to obtain the load difference within the execution period t.

[0170] Generate a transformer scheduling instruction based on the load difference within the execution period t. If the load difference is positive, that is, the load exceeds the load capacity, a scheduling instruction to reduce the load may be generated, such as adjusting the load distribution, reducing the access of certain devices, etc.; if the load difference is negative, the load may be appropriately increased.

[0171] Combine the transformer scheduling instruction and the primary processing strategy to generate a secondary processing strategy. For example, incorporate the scheduling instruction into the primary processing strategy, supplement or adjust the original control instruction to form a complete secondary processing strategy.

[0172] Embodiment 8:

[0173] The secondary optimization instruction includes:

[0174] Obtain the abnormal eigenvalue with an association and historical experience data to determine the factors causing the failure;

[0175] Generate one or more corresponding correction instructions based on the factors causing the failure;

[0176] Predict the implementation effect of each correction instruction and generate a corresponding set of implementation effect reference values E, E = {e1, e2... ek... ev};

[0177] Among them, ek represents the implementation effect reference value of the kth correction instruction, and v represents the total number of correction instructions;

[0178] Select the correction instruction corresponding to the maximum value of the implementation effect reference value to generate a secondary processing strategy.

[0179] In this embodiment, first, abnormal eigenvalue and historical experience data with associations are obtained. For example, if it is found that there is an association between the abnormal eigenvalue of oil temperature and the abnormal eigenvalue of oil level, and historical experience data is combined at the same time, this data may include information such as the fault conditions, maintenance operations, and equipment performance changes recorded in the past under similar abnormal oil temperature and oil level conditions.

[0180] By comprehensively analyzing these associated abnormal eigenvalues and historical experience data, the factors causing the fault are determined. For example, it may be found that the fault of the cooling system causes the oil temperature to rise and the oil level to drop at the same time. For example, the blockage of the cooling pipeline affects the circulation and heat dissipation of the oil, and this is the determined fault factor.

[0181] One or more corresponding correction instructions are generated based on the factors causing the fault. For the case of blockage of the cooling pipeline, the possible generated correction instructions may include instructions such as cleaning the cooling pipeline, checking whether the oil pump of the cooling system is working properly, and replacing the filter screen in the cooling system.

[0182] Predict the implementation effect of each correction instruction and generate a corresponding set of implementation effect reference values E, E = {e1, e2…ek…ev}. This prediction process may be based on simulation analysis, historical data comparison, or expert experience model. For example, for the correction instruction of cleaning the cooling pipeline, an implementation effect reference value e1 can be determined according to data such as the time when the oil temperature and oil level return to normal after cleaning the cooling pipeline in the past, and the subsequent stable operation duration of the transformer.

[0183] Select the correction instruction corresponding to the maximum value of the implementation effect reference value to generate a secondary processing strategy. Find the correction instruction corresponding to the maximum value in the set E. Assume that e3 is the maximum value, then the 3rd corresponding correction instruction is selected as the secondary processing strategy. This is because this instruction has the best implementation effect in the prediction and is expected to most effectively solve the potential fault of the transformer.

[0184] Embodiment 9:

[0185] The third processing module is further used for:

[0186] Receiving a comprehensive data packet of various states of the current transformer sent by the monitoring unit based on the 5G module;

[0187] Combining the comprehensive data packet of various states and historical data to generate an operating state feature set A1 at the current monitoring time node;

[0188] Combining the operating state feature set A1 at the current monitoring time node and the state sample set B to determine the state sample of the transformer at the current monitoring time node;

[0189] Combining the state sample of the transformer at the current monitoring time node and the processing model library to select the corresponding processing model.

[0190] In this embodiment, by utilizing the high-speed communication capability of the 5G module, a comprehensive data packet of various states of the current transformer sent by the monitoring unit is received. The 5G module can ensure the fast and stable transmission of data. The comprehensive data packet of various states may include various state information such as the oil temperature, voltage, current, oil level, etc. of the transformer, as well as metadata such as the acquisition time of these state information and the acquisition device number.

[0191] Based on the comprehensive data packet of various states and historical data, an operating state feature set A1 at the current monitoring time node is generated. Key state information is extracted from the comprehensive data packet of various states, for example, values such as oil temperature, voltage, current, and oil level are extracted. At the same time, these data are calibrated and supplemented with reference to historical data. For example, if the historical data shows the normal range of the oil temperature under the current ambient temperature, then the rationality of the current oil temperature data can be judged according to this range, and the relevant judgment results are also added to the operating state feature set A1.

[0192] Based on the operating state feature set A1 at the current monitoring time node and the state sample set B, the state sample of the transformer at the current monitoring time node is determined. The state sample set B contains sample data of the transformer in different operating states in the past, and each sample data has a corresponding operating state feature set. By matching and comparing the current operating state feature set A1 with the samples in the state sample set B, the most similar sample is found, so as to determine the state sample of the transformer at the current monitoring time node. For example, a distance metric method such as the Euclidean distance can be used, and the sample with the smallest distance is the current state sample.

[0193] Based on the state sample of the transformer at the current monitoring time node and the processing model library, the corresponding processing model is selected. Different processing models corresponding to different state samples are stored in the processing model library. After determining the state sample of the transformer at the current monitoring time node, the corresponding processing model can be found in the processing model library. For example, if the current state sample is "the oil temperature is too high and the voltage is slightly low", then the processing model for this state is found in the processing model library, and this model may include processing strategies such as adjusting the power of the cooling system and fine-tuning the tap position of the transformer.

[0194] Embodiment 10:

[0195] The fourth processing module is further configured to:

[0196] Obtain the adjustment parameter range of the selected processing model in the third processing module;

[0197] Generate an adjustment instruction by combining the adjustment parameter range and the data in the comprehensive data packet of various states of the current transformer.

[0198] In this embodiment, the fourth processing module first obtains the adjustment parameter range of the selected processing model in the third processing module. The processing model is determined for a specific transformer state and includes a series of adjustment parameters. For example, for a model that processes the problem of excessively high transformer oil temperature, the adjustment parameters may include the power adjustment range of the cooling system, assumed to be [30%, 80%], which is the adjustment parameter range of the cooling system power.

[0199] Generate an adjustment instruction by combining the adjustment parameter range and the data in the comprehensive data packet of various states of the current transformer. Extract the data related to the adjustment parameters from the comprehensive data packet of various states of the current transformer. For example, the current oil temperature is 90°C (assuming the normal range is 60 - 80°C). Combining with the cooling system power adjustment parameter range [30%, 80%], since the oil temperature is too high, an adjustment instruction to increase the cooling system power to 70% may be generated. This adjustment instruction is obtained by comprehensively considering the adjustment parameter range and the current transformer state data, aiming to restore the transformer to the normal operating state.

[0200] Finally, it should be noted that: Obviously, those skilled in the art can make various changes and deformations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and its equivalent technologies, the present invention also intends to include these changes and deformations.

[0201] The above is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention.

Claims

1. A 5G-based transformer status comprehensive monitoring system, characterized in that: Including: A monitoring unit, configured to collect comprehensive status data of a transformer to generate multiple comprehensive status data packets, and send them to a processing unit through a 5G module; A processing unit, configured to establish a processing model library, select a processing model based on multiple comprehensive status data packets, generate an adjustment instruction based on the processing model, and send it to the monitoring unit through a 5G module; The processing unit includes: A first processing module: configured to generate a status sample set according to the comprehensive status data of historical transformers; A second processing module: configured to establish a processing model according to each status sample, and generate a processing model library by combining all the processing models; A third processing module: configured to judge the operating status of the current transformer according to multiple comprehensive status data packets, select a corresponding status sample based on the operating status of the current transformer, and select a corresponding processing model based on the status sample; A fourth processing module: configured to generate an adjustment instruction for the selected corresponding processing model.

2. The 5G-based transformer status comprehensive monitoring system according to claim 1, characterized in that: The first processing module is further configured to: Classify the status types of the transformer based on the obtained comprehensive historical status data of the transformer to generate multiple status types; Obtain the comprehensive historical status data of the transformer corresponding to each status type; The comprehensive historical status data of the transformer includes the historical status data of different monitoring points of the transformer, and generates an operating status feature set A corresponding to the current status type based on the historical status data of different monitoring points, A = {a1, a2... ai... an}; Wherein, ai represents the operating data characteristic value of the monitoring point corresponding to the i-th monitoring point under the current status type, and n represents the total number of monitoring points; Generate a status sample set B by combining the historical status data, all the status types of the transformer, and the operating status feature set A corresponding to each status type, B = {b1, b2... bj... bm}; Wherein, bj represents the j-th status sample, and m represents the number of status samples.

3. The 5G-based transformer status comprehensive monitoring system according to claim 2, characterized in that: The second processing module is further configured to: Obtain the historical processing data corresponding to the current status sample, including: processing strategy data and the historical status data of the transformer after processing; Generate a primary processing strategy for the current status sample based on the processing strategy data; Generate an evaluation value C of the current primary processing strategy by combining the historical status data of the transformer and the current primary processing strategy; Judge whether to correct the current primary processing strategy by comparing the evaluation value of the current primary processing strategy with a preset value D; If C≥D; then correct the current primary processing strategy to generate a secondary processing strategy; If C<D; then retain the current primary processing strategy as the secondary processing strategy; Generate a processing model for the current status sample and the secondary processing strategy corresponding to the current status sample, and generate a processing model library by combining all the processing models.

4. The 5G-based transformer status comprehensive monitoring system according to claim 3, characterized in that: When generating the primary processing strategy for the current status sample, it includes: Obtain the operating status feature set A, compare the operating data characteristic value ai with the preset value of the operating data of the corresponding monitoring point in turn, and mark the operating data characteristic value ai less than the preset value of the operating data as an abnormal characteristic value; Determine the current status sample based on all the abnormal characteristic values and generate multiple control instructions; All control instructions are sorted based on historical processing strategy data to generate a primary processing strategy for the current state sample.

5. The 5G-based transformer status comprehensive monitoring system according to claim 4, characterized in that: The generating of the evaluation value C of the current primary processing strategy includes: Obtain the execution period t of the first-level processing strategy; Predict the transformer load change curve within the execution period t based on historical data; Obtain the proportion of time periods in which the transformer load change curve exceeds a preset value within the execution period t; Generate a time reference value T of the current primary processing strategy based on the time period ratio; Determine the correlation between abnormal feature values ​​in the operating status feature set A based on historical data; Generate a correlation reference value Y of the current primary processing strategy based on the correlation between abnormal feature values; Generate an evaluation value C of the current primary processing strategy by combining the time reference value T and the associated reference value Y of the current primary processing strategy; C = f1*T+f2*Y; Among them, f1 is the weight of the time reference value T, f2 is the weight of the associated reference value Y, f1+f2=1, P∈(0,1), Y∈(0,1).

6. The 5G-based integrated transformer status monitoring system according to claim 5, characterized in that: The generating of the secondary processing strategy includes: Setting a first preset value Z of the optimization reference value Y based on historical data; If Y≤Z, the first-level optimization instruction is executed to generate the second-level processing strategy; If Y>Z, execute the secondary optimization instruction to generate the secondary processing strategy.

7. The 5G-based transformer status comprehensive monitoring system according to claim 6, characterized in that: The first-level optimization instructions include: Predict the reference value of the transformer's load capacity under the current primary processing strategy; and calculating the load difference within the execution period t based on the transformer load change curve within the execution period t and the load capacity reference value of the transformer under the current primary processing strategy; Generate transformer dispatching instructions based on the load difference within the execution period t; The secondary processing strategy is generated by combining the transformer dispatch instructions and the primary processing strategy.

8. The 5G-based transformer status comprehensive monitoring system according to claim 7, characterized in that: The secondary optimization instructions include: Obtain associated abnormal characteristic values ​​and historical experience data to determine the factors that cause the failure; generating one or more corresponding correction instructions based on the factors causing the fault; Predict the implementation effect of each correction instruction and generate a corresponding implementation effect reference value set E, E = {e1, e2…ek…ev}; Wherein, ek represents the reference value of the implementation effect of the kth correction instruction, and v represents the total number of correction instructions; Select the correction instruction corresponding to the maximum value of the implementation effect reference value to generate a secondary processing strategy.

9. The 5G-based transformer status comprehensive monitoring system according to claim 8, characterized in that: The third processing module is also used for: Based on the 5G module, the comprehensive data packet of various states of the current transformer sent by the monitoring unit is received; Combine multiple state comprehensive data packages and historical data to generate the operating state feature set A1 of the current monitoring time node; Determine the state sample of the transformer at the current monitoring time node by combining the operating state feature set A1 and the state sample set B at the current monitoring time node; Select the corresponding processing model based on the transformer status sample at the current monitoring time node and the processing model library.

10. The 5G-based transformer status comprehensive monitoring system according to claim 9, characterized in that: The fourth processing module is further used for: Obtaining an adjustment parameter interval of the processing model selected in the third processing module; The adjustment instruction is generated by combining the adjustment parameter range and the data in the comprehensive data package of the current transformer's various states.

Images