Oil-immersed transformer operation risk assessment method and system

Abstract

The invention relates to the field of oil-immersed transformers, in particular to an oil-immersed transformer operation risk assessment method and system, and the method comprises the steps: obtaining real-time operation parameters and external influence factors of an oil-immersed transformer; the real-time operation parameters and the external influence factors are preprocessed; based on the preprocessed real-time operation parameters and the external influence factors, outputting a comprehensive risk assessment score; calculating a transformer health index based on the preprocessed real-time operation parameters and the external influence factors, and calculating a risk occurrence probability according to the transformer health index and the comprehensive risk assessment score; and outputting a risk level report based on the comprehensive risk assessment score, the transformer health index and the risk occurrence probability, and generating an early warning signal. According to the method, the aging factor is introduced, and the calculation weights of the comprehensive risk score, the health index and the risk probability are adjusted in combination with the service life of the transformer, so that the evaluation model can dynamically adapt along with the aging degree of the equipment, and the evaluation precision of the transformer is improved.

Description

Technical Field

[0001] This invention relates to the field of oil-immersed transformers, and more particularly to a method and system for assessing the operational risks of oil-immersed transformers. Background Technology

[0002] To ensure the safe and reliable operation of oil-immersed transformers, real-time monitoring of their operating parameters is typically required, including key indicators such as oil temperature, oil level, current, and voltage. Continuous collection of equipment operating status data allows for timely understanding of equipment health conditions, triggering early warning mechanisms when parameters are abnormal. This provides maintenance personnel with technical support for rapid response and precise control, significantly improving the reliability of transformer operation. In power systems, main transformers, high-voltage station service transformers, and standby transformers in large generating units often adopt oil-immersed structures. In addition to routine monitoring via control panel instruments, their daily maintenance should include periodic special inspections of the transformer body and cooling system, along with systematic data collection and in-depth analysis. Based on long-term accumulated operating data and equipment status information, risk assessment models and fault prediction systems for oil-immersed transformers can be constructed, providing a scientific basis for the full lifecycle management of equipment. Oil-immersed transformers face risks such as oil temperature rise, insulation aging, and partial discharge during operation. Existing assessment methods often rely on single parameter thresholds or simple statistical models, lacking dynamic analysis of the synergistic effects of multiple factors, resulting in low risk assessment accuracy and delayed response.

[0003] Therefore, it is necessary to design a method and system for assessing the operational risks of oil-immersed transformers. Summary of the Invention

[0004] To address the technical deficiencies in the background technology, this invention proposes a method and system for assessing the operational risks of oil-immersed transformers, which solves the aforementioned technical problems and meets practical needs. The specific technical solution is as follows: A method and system for assessing the operational risks of oil-immersed transformers, comprising the following steps: Step S1: Obtain the real-time operating parameters and external influencing factors of the oil-immersed transformer; Step S2: Preprocess the real-time operating parameters and external influencing factors; Step S3: Based on the preprocessed real-time operating parameters and external influencing factors, output a comprehensive risk assessment score through a risk assessment formula. The comprehensive risk assessment score integrates the internal operating status and the impact of the external environment. Step S4: Based on the preprocessed real-time operating parameters and external influencing factors, calculate the transformer health index, and calculate the probability of risk occurrence based on the transformer health index and the comprehensive risk assessment score; Step S5: Output a risk level report based on the comprehensive risk assessment score, transformer health index and risk occurrence probability, and generate an early warning signal.

[0005] Furthermore, the real-time operating parameters include oil temperature, load current, ambient temperature, partial discharge quantity, and dissolved gas concentration in the oil. The external influencing factors include ambient humidity, grid voltage fluctuation rate, and the equipment's historical operating years. The equipment's historical operating years are calculated based on the transformer's commissioning date and maintenance records to obtain an aging coefficient, which is used to adjust the weight allocation of the risk assessment formula.

[0006] Furthermore, the preprocessing includes data cleaning, normalization, and missing value imputation. Data cleaning removes noise using filters, normalization employs a minimum-maximum scaling method, and missing value imputation is performed based on time series prediction. The data cleaning uses a low-pass Butterworth filter, and the normalization scaling of each running parameter to the range of 0 to 1 is calculated as follows: original value minus the minimum parameter value, then divided by the maximum parameter value minus the minimum value. The missing value imputation uses an ARIMA time series model with a prediction window of 5 minutes.

[0007] Furthermore, it also includes: Step S6: Based on the risk level report, execute automatic control actions, including adjusting the power of the transformer cooling system or reducing the load current. The automatic control actions are implemented through a PLC controller.

[0008] Furthermore, the specific risk assessment formula is as follows:

[0009] in, This represents the overall risk assessment score, with a dimensionless unit; the higher the value, the higher the risk. This represents the basic risk scaling coefficient, derived from regression analysis of historical accident data; This represents the weight of the i-th running parameter, which is optimized by the support vector machine algorithm during the training phase; This represents the i-th preprocessed running parameter, where i ranges from 1 to n, and n is the total number of parameters; This represents the gain coefficient due to temperature change, reflecting the sensitivity of oil temperature to ambient temperature. This indicates the temperature effect index, calibrated through laboratory temperature rise tests. This indicates the difference between oil temperature and ambient temperature, expressed in degrees Celsius. This represents the partial discharge reduction factor, which is set based on the properties of the insulating material. This represents the normalized partial discharge quantity, calculated by dividing the original partial discharge quantity by a preset threshold.

[0010] Furthermore, the transformer health index is calculated using the following formula:

[0011] in, This represents the transformer health index, with dimensionless units and a value range of 0 to 1, where 0 indicates complete failure and 1 indicates healthy. This represents the load contribution coefficient, determined through load curve analysis. To represent the hyperbolic tangent function, a nonlinear smoothing load impact effect is introduced; This represents the load gain constant, calculated based on the transformer's rated capacity. This represents the normalized load current, calculated by dividing the load current by the rated current, and is dimensionless. The weighting of dissolved gas influence is calibrated from gas chromatography data. It represents the ratio of dissolved gas concentrations in oil, calculated by dividing the hydrogen concentration by the methane concentration, and is dimensionless. This represents the aging acceleration factor, calculated from the equipment's historical operating years. This represents the aging factor, which is calculated by multiplying the number of years of operation by the average annual load rate.

[0012] Furthermore, the probability of risk occurrence is calculated using the following formula:

[0013] in, It represents the probability of risk occurring, with a dimensionless unit and a value range of 0 to 1. Values ​​higher than 0.7 are defined as high risk. The health index coefficient is used to fit the fault data through a logistic regression model. Indicates health index; This represents the comprehensive risk assessment score coefficient, optimized through Monte Carlo simulation; represents the comprehensive risk assessment score; k represents the probability offset constant, which is set based on power grid operation statistics. This represents the natural exponential function.

[0014] Furthermore, the risk level report includes three levels of classification: high risk, medium risk, and low risk. Based on the risk level report's classification, different warning signals are generated for each level and sent to the monitoring platform via a serial communication protocol. The specific grading rules for the risk level classification are as follows: A first threshold and a second threshold are preset. When the probability of a risk occurring is lower than the first threshold, it is judged as low risk; when the probability of a risk occurring is greater than or equal to the first threshold and less than the second threshold, it is judged as medium risk; and when the probability of a risk occurring is greater than or equal to the second threshold, it is judged as high risk.

[0015] An oil-immersed transformer operation risk assessment system includes: The data acquisition module is equipped with a sensor group for collecting internal operating data and external influencing factor data of oil-immersed transformers. The data processing module, connected to the acquisition module, is used to receive the data collected by the acquisition module and perform preprocessing. The calculation module, connected to the data processing module, is used to receive preprocessed data and calculate the comprehensive risk assessment score, transformer health index, and risk occurrence probability based on the preprocessed data. The output module, connected to the calculation module, is used to output a risk level report consisting of a comprehensive risk assessment score, a transformer health index, and a risk occurrence probability, and to generate corresponding early warning signals.

[0016] Furthermore, it also includes a cloud server and an edge computing unit, wherein the cloud server stores historical databases and runs machine learning models, and the edge computing unit is deployed locally at the substation.

[0017] Compared with existing technologies, the oil-immersed transformer operation risk assessment method and system provided by this invention have the following beneficial effects: This invention comprehensively covers the dynamic analysis of the synergistic effects of multiple factors by collecting real-time operating parameters of the transformer body and external environmental factors. By calculating the comprehensive risk score, transformer health index, and risk occurrence probability, it quantifies the operating risk of oil-immersed transformers and presents the operating risk of oil-immersed transformers more intuitively. By introducing an aging factor and adjusting the calculation weights of the comprehensive risk score, health index, and risk probability in combination with the transformer's service life, the assessment model can dynamically adapt to the degree of equipment aging, reducing the false alarm rate of transformer operating risk and improving the accuracy of transformer assessment. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a method for assessing the operational risks of an oil-immersed transformer according to the present invention.

[0019] Figure 2 This is a schematic diagram of a module of an oil-immersed transformer operation risk assessment system according to the present invention. Detailed Implementation

[0020] In the description of this invention, it should be understood that the terms "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "middle," and "inner," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, it should be noted that unless otherwise explicitly specified and limited, the terms "installed," "connected," and "joined" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention through specific circumstances.

[0021] The embodiments of the present invention will be described below with reference to the accompanying drawings and related examples. The embodiments of the present invention are not limited to the following examples, and the present invention relates to the relevant necessary components in this technical field, which should be regarded as well-known technology in this technical field and can be known and mastered by those skilled in this technical field.

[0022] See Figure 1 A method and system for assessing the operational risks of oil-immersed transformers, comprising the following steps: Step S1: Obtain real-time operating parameters and external influencing factors of the oil-immersed transformer; simultaneously obtain internal operating parameters and external influencing factors to cover the main causes of transformer failures. Compared with the single-dimensional assessment of existing technologies that only monitor oil temperature or load current, this method avoids misjudgments caused by the one-sidedness of parameters through multi-parameter collaborative analysis.

[0023] Step S2: Preprocess the real-time operating parameters and external influencing factors; Step S3: Based on the preprocessed real-time operating parameters and external influencing factors, output a comprehensive risk assessment score through a risk assessment formula. The comprehensive risk assessment score integrates the internal operating status and the impact of the external environment. Step S4: Based on the preprocessed real-time operating parameters and external influencing factors, calculate the transformer health index, and calculate the probability of risk occurrence based on the transformer health index and the comprehensive risk assessment score; Step S5: Output a risk level report based on the comprehensive risk assessment score, transformer health index and risk occurrence probability, and generate an early warning signal.

[0024] It should be noted that this invention constructs a full-chain risk assessment system, which achieves risk assessment through multi-source data acquisition, three-layer data processing, dynamic assessment mechanism, and graded response. Multi-source data acquisition corresponds to step S1, covering transformer intrinsic parameters, external environmental factors, and grid status. Three-layer data processing corresponds to step S2, including noise filtering, normalization, and missing value imputation. The dynamic assessment mechanism corresponds to steps S3 and S4, sequentially calculating the comprehensive risk assessment score, transformer health index, and risk occurrence probability. Graded response corresponds to step S5, rating the risk level report based on the risk occurrence probability and generating corresponding early warning signals based on the risk level. The risk level report includes the comprehensive risk assessment score, transformer health index, and risk occurrence probability.

[0025] Existing technologies rely on single parameter thresholds for judgment, failing to integrate external factors such as environmental humidity and voltage fluctuations, leading to biased risk assessments. Furthermore, the lack of dynamic data preprocessing mechanisms results in high misjudgment rates due to noise in the raw data. This invention, by integrating grid volatility and environmental humidity, covers most fault causes. When assessing transformer risks, it improves the accuracy of risk assessments, avoids bias, and enhances data integrity and signal-to-noise ratio through a three-level data processing mechanism. This reduces errors in subsequent calculations of the comprehensive risk assessment score, transformer health index, and risk occurrence probability, thereby improving the accuracy of risk assessments.

[0026] Steps S3 to S5 achieve an upgrade from state quantification to risk prediction through progressive calculations of comprehensive risk score, health index, and risk probability. The comprehensive risk score integrates the real-time impact of internal state and external environment, while the health index characterizes the equipment's health status from three dimensions: load, gas analysis, and aging. Finally, the risk probability combines these two factors to form a quantitative prediction of the likelihood of a fault. This multi-level assessment mechanism, compared to traditional threshold alarms, can more accurately capture the gradual development process of faults. For example, in the early stages of insulation aging, the concentration of dissolved gases in the oil may slowly increase, while traditional methods cannot provide early warnings because a single threshold is not reached. This method, however, identifies faults in advance through gas fault factors in the health index.

[0027] In one embodiment of the present invention, the real-time operating parameters include oil temperature, load current, ambient temperature, partial discharge quantity, and dissolved gas concentration in the oil. The external influencing factors include ambient humidity, grid voltage fluctuation rate, and the equipment's historical operating years. The equipment's historical operating years are calculated based on the transformer's commissioning date and maintenance records to obtain an aging coefficient. The aging coefficient is used to adjust the weight allocation of the risk assessment formula.

[0028] It should be noted that the oil temperature is obtained by directly monitoring the top layer oil temperature of the transformer, reflecting the effectiveness of the cooling system and overload status. The difference between the oil temperature and the ambient temperature directly reflects the transformer's heat dissipation efficiency; an abnormally high difference can pre-indicate a cooling system fault or overload. The load current is collected through the current transformer (CT) and used to calculate the normalized load rate. The ambient temperature and humidity are collected in real time by temperature and humidity sensors installed near the transformer to correct for oil temperature measurement errors. When the ambient humidity exceeds the set value, the moisture content of the transformer insulation paper increases, leading to a decrease in breakdown voltage. This invention incorporates ambient humidity as an external factor into the risk assessment formula, using a weighted allocation system... The aging factor is used to accelerate insulation aging; partial discharge refers to discharge pulses >100pC captured by a high-frequency current sensor, used to diagnose insulation defects; dissolved gas concentration in oil can be obtained by monitoring the concentration of eight gases using an online chromatograph; grid voltage fluctuation rate is obtained from the power grid system and used to calculate the rate of change of the effective voltage value; the historical operating years of equipment are used to calculate the aging coefficient, and the weighting of the aging coefficient is adjusted according to the aging stage. The longer the service life of the transformer, the larger the aging coefficient: the aging coefficient of new equipment is less than 5, the aging coefficient of middle-aged equipment is 5-15, and the aging coefficient of old equipment is greater than 15. The aging coefficient can be used to calculate the aging factor. The actual aging coefficient is calculated based on the service life, and the aging weight in the risk assessment formula is calculated using the aging coefficient.

[0029] This invention introduces the equipment's historical operating years as an aging factor, dynamically adjusting the assessment weights based on the commissioning date and maintenance records, conforming to the aging patterns of transformers. External influencing factors cover environmental humidity and grid voltage fluctuations, improving the assessment's adaptability to complex operating conditions. The introduction of the aging coefficient allows the risk assessment to be dynamically adjusted along with the equipment's lifespan, reducing false alarm rates.

[0030] In one embodiment of the present invention, the preprocessing includes data cleaning, normalization, and missing value imputation. Data cleaning removes noise using filters, normalization employs a minimum-maximum scaling method, and missing value imputation is based on time series prediction. The data cleaning uses a low-pass Butterworth filter, which has a maximum flat amplitude response and an attenuation slope of -20dB / decimal at the cutoff frequency, effectively filtering out high-frequency noise in transformer operating data. The cutoff frequency and order of the filter can be adjusted according to the transformer type. The normalization scaling of each operating parameter to the range of 0 to 1 is performed by subtracting the minimum parameter value from the original value and then dividing by the maximum parameter value minus the minimum. The missing value imputation uses an ARIMA time series model with a prediction window of 5 minutes. When parameters approach extreme values, traditional normalization may lead to computational overflow. This invention avoids division by zero by adding a minimum value to the denominator and uses clipping for outliers exceeding historical maximum values ​​to prevent extreme data from impacting the model.

[0031] The ARIMA time series model captures the temporal correlation of data through the autoregressive order p, the differencing order d, and the moving average order q. Taking load current as an example, when data is missing at a certain moment, the model establishes a prediction equation based on historical data from the previous 60 minutes (12 five-minute windows). When the transformer load current sequence exhibits periodic fluctuations (high during the day and low at night), the ARIMA model can predict the missing value using the values ​​from the previous two periods, the first-order differencing, and the one-period moving average, with the average absolute error controlled within 5%. The five-minute prediction window is the optimal solution in engineering practice—a window that is too short is easily affected by random noise, while a window that is too long leads to prediction lag. After using five-minute ARIMA interpolation, data integrity increased from 85% to 99.2%, providing a reliable data foundation for subsequent risk calculations.

[0032] After data cleaning, normalization, and missing value imputation, the SNR of oil temperature data in this invention increased from 12dB to 35dB, and the SNR of partial discharge increased from 8dB to 28dB, providing high-quality input for subsequent risk assessment formulas. Missing value imputation improved data integrity from 75% in traditional methods to 99.5%, avoiding assessment model failure due to missing data. The preprocessed data more clearly reflects the characteristics of progressive faults. In the early stages of insulation aging, the concentration of dissolved gases in the oil rises slowly. After normalization and noise filtering, the rate of change of gas fault factors can be accurately captured, allowing the Health Index (HI) to begin declining in the early stages of faults, issuing an early warning 3-6 months earlier than traditional threshold alarms.

[0033] In one embodiment of the present invention, it further includes: Step S6: Based on the risk level report, execute automatic control actions. These actions include adjusting the transformer cooling system power or reducing the load current. The automatic control actions are implemented through a PLC controller. Control is forcibly initiated when the probability of risk occurrence is ≥0.7; control is initiated according to a preset strategy when the probability of risk occurrence is ≤0.5 <0.7. Even if the risk level does not reach the threshold, control will still be forcibly initiated if key parameters exceed safety boundaries. 0.5 and 0.7 are preset thresholds that can be autonomously adjusted by the system.

[0034] It should be noted that the automatic control actions are designed based on risk priority ranking: When the comprehensive risk score Rs exceeds the threshold and the oil temperature ΔT > 30℃, the PLC controller will increase the cooling fan power from 50% to 100% for forced heat dissipation. When the probability of risk occurrence is greater than 0.7 and the load rate is greater than 110% of the rated value, the PLC sends a load reduction command through the substation automation system (SCADA) to reduce the load current to 90% of the rated value. This active control can prevent the winding overheating fault caused by long-term overload of the transformer.

[0035] The PLC controller adopts a dual-CPU hot standby architecture, using an industrial-grade PLC such as Siemens S7-400H or ABBAC800M. The main CPU and the standby CPU mirror data in real time through a synchronous bus, with a switching time of <50ms.

[0036] In one embodiment of the present invention, the risk assessment formula is as follows:

[0037] in, This represents the overall risk assessment score, with a dimensionless unit; the higher the value, the higher the risk. This represents the basic risk scaling coefficient, derived from regression analysis of historical accident data; This represents the weight of the i-th running parameter, which is optimized by the support vector machine algorithm during the training phase; This represents the i-th preprocessed running parameter, where i ranges from 1 to n, and n is the total number of parameters; This represents the gain coefficient due to temperature change, reflecting the sensitivity of oil temperature to ambient temperature. This indicates the temperature effect index, calibrated through laboratory temperature rise tests. This indicates the difference between oil temperature and ambient temperature, expressed in degrees Celsius. This represents the partial discharge reduction factor, which is set based on the properties of the insulating material. This represents the normalized partial discharge quantity, calculated by dividing the original partial discharge quantity by a preset threshold.

[0038] As the parameter weight aggregation term, the support vector machine optimizes the weights by maximizing the classification margin during the training phase. This makes the model sensitive to key parameters. When training an SVM on a specific type of transformer, it was found that the weights of partial discharge parameters... =0.35, oil temperature of =0.25, load current =0.15, reflecting that this model of equipment is most sensitive to PD. The training data comes from historical fault cases, such as insulation breakdown accidents caused by excessive partial discharge, which account for 42%, so the model automatically assigns it a high weight.

[0039] This is the temperature effect index, which simulates the nonlinear effects of the difference between oil temperature and ambient temperature. Laboratory temperature rise tests show that... For every 10°C increase, the oxidation rate of the insulating oil increases by 1.8 times; therefore, γ is calibrated to 0.1 / °C through experiments, and β is taken as 0.2 (derived from regression analysis of historical accident data). When When the temperature rises from 20℃ to 40℃, the exponential term changes from... The value of (0.1×20)=0.2×7.389≈1.478 increases to 0.2×54.598≈10.92, causing Rs to rise significantly and accurately reflecting the accelerating effect of temperature risk.

[0040] For partial discharge reduction, based on the theory of insulation material discharge aging, the logarithmic function mapping of PD conforms to the Pareto principle (80% of insulation damage is caused by 20% high-amplitude discharge). δ is set to 0.5 (based on the PDIV (partial discharge initiation voltage) characteristics of oil-paper insulation). When the value increases from 0.1 (normal) to 1.0 (exceeding the limit), the reduction term changes from -0.5 × (1.1)≈−0.048 increased to −0.5× (2)≈−0.347, that is, Rs decreases by 0.347, reflecting the suppressive effect of PD on risk (the higher the PD, the more severe the risk, and the higher Rs should be, but here it is a reduction term, so pay attention to the logic of the formula symbols: The higher, The smaller the value, the more is deducted from Rs, and the higher the actual Rs, which aligns with the logic that higher PD means higher risk.

[0041] In one embodiment of the present invention, the transformer health index is calculated using the following formula:

[0042] in, This represents the transformer health index, with dimensionless units and a value range of 0 to 1, where 0 indicates complete failure and 1 indicates healthy. This represents the load contribution coefficient, determined by load curve analysis, and is used to adjust the weight of the load's impact and the saturation inflection point. To represent the hyperbolic tangent function, a nonlinear smoothing load impact effect is introduced; This represents the load gain constant, calculated based on the transformer's rated capacity, and is used to adjust the weight of the load's influence and the saturation inflection point. It represents the normalized load current, reflecting the ratio of the transformer's current load to its rated capacity. It is calculated by dividing the load current by the rated current and is dimensionless. This indicates the weighting of dissolved gas influence, calibrated from gas chromatography data, and is used to adjust the degree of influence of gas ratios on the health index. It represents the ratio of dissolved gas concentration in oil, calculated by dividing the hydrogen concentration by the methane concentration, and is a transformer fault diagnosis indicator recommended by the International Electrotechnical Commission. This represents the aging acceleration factor, calculated from the equipment's historical operating years, and is used to adjust for differences in the aging rate of different transformer models. This represents the aging factor, calculated by multiplying the number of years of operation by the average annual load rate, reflecting the actual wear and tear of the equipment.

[0043] The load saturation effect term is a nonlinear load modeling module in the transformer health index calculation formula. Its core is to transform the physical influence of load current into a quantitative indicator of insulation life loss through the hyperbolic tangent function. The function is a non-linear mathematical function whose graph exhibits an S-shape. Its characteristic is that it grows approximately linearly when the input value is small, but as the input value increases to a certain extent, the output value gradually approaches a fixed upper limit. Similarly, in a transformer under low load, the core magnetic flux density and winding losses are approximately proportional to the load rate. The function is approximately linear, and the impact of load on health increases linearly; however, when the load rate approaches or exceeds the rated value, the core enters a state of magnetic saturation, and winding losses surge nonlinearly due to the skin effect. The saturation characteristic of the function limits the rate of decline in the health index, preventing the model from outputting extreme results of a "health value crash" due to overload, which is more consistent with the actual withstand capability of the transformer. If the load impact is calculated directly using a linear function, the load's contribution to health will double when the transformer is overloaded by 50%. However, in practice, transformers are designed with a certain overload margin, and the rate of health degradation will not increase proportionally. The saturation characteristic of the function limits the load contribution to near its maximum value, preventing the health index HI from unreasonably dropping to 0 due to overload, and ensuring that the evaluation results conform to engineering reality. Through the calculation of the load saturation effect term, maintenance personnel can intuitively understand the degree of impact of different load rates on transformer health: when... When =0.8, the load contribution is approximately 76% of the maximum value; when When the value is 1.0, the contribution rises to 96%; while when When the coefficient of performance is 1.2, the contribution only increases to 99.5%. This means that the transformer deteriorates fastest when operating near its rated load, while the deterioration rate slows down under overload conditions, providing data support for an operating strategy that avoids long-term full load and allows for short-term overload.

[0044] The gas fault factor is a core module in the transformer health index calculation formula, used to quantify the correlation between insulating oil decomposition gases and equipment faults. Essentially, it identifies latent internal defects by using the concentration ratio of dissolved gases in the oil. When overheating or discharge occurs inside the transformer, the insulating oil and paper decompose to produce characteristic gases. H2 is mainly produced by high-temperature decomposition, while CH4 is a characteristic gas of low-temperature overheating. The ratio of these two gases can effectively distinguish fault types. The gas fault factor transforms abstract fault types into calculable values ​​by linearly amplifying the relationship between the ratio and health degradation. Through the design of characteristic gas ratios and weighting coefficients, the gas fault factor converts latent faults inside the transformer into quantifiable health impact values. It is a core technology module for achieving early fault identification, precise location, and trend tracking, providing a scientific quantitative tool for condition-based maintenance of oil-immersed transformers.

[0045] The aging correction term, a quantitative module for equipment lifecycle within the transformer health index calculation formula, transforms accumulated aging damage into a calculable health degradation value by correlating operating years with historical load intensity. This correction term addresses the significant shortcoming of traditional models that neglect the equipment's service history. Through a two-dimensional calculation of time and intensity, the aging correction term upgrades the transformer aging process from a qualitative assessment to a quantitative evaluation, considering both the time factor of natural aging and the accelerating effect of load operation. This allows the health index to dynamically reflect the health status of the equipment throughout its entire lifecycle. This design provides a scientific quantitative basis for transformer life management, condition-based maintenance, and decommissioning decisions, and is one of the core technologies for achieving precise equipment operation and maintenance.

[0046] In one embodiment of the present invention, the probability of risk occurrence is calculated using the following formula:

[0047] in, It represents the probability of risk occurring, with a dimensionless unit and a value range of 0 to 1. Values ​​higher than 0.7 are defined as high risk. The health index coefficient is used to fit the fault data through a logistic regression model. Indicates health index; This represents the comprehensive risk assessment score coefficient, optimized through Monte Carlo simulation; This represents the comprehensive risk assessment score; k represents the probability offset constant, which is set based on power grid operation statistics and is used to adjust the offset of the overall probability distribution. This represents the natural exponential function. The formula dynamically integrates the health index with the comprehensive risk assessment score using a logistic regression model, outputting a risk occurrence probability of 0-1, thus achieving a paradigm shift in risk assessment from scalar to probabilistic.

[0048] In one embodiment of the present invention, the risk level report includes three levels: high risk, medium risk, and low risk. Based on the risk level report's classification, different warning signals are generated for each level, and these warning signals are sent to the monitoring platform via a serial communication protocol. The specific classification rules are as follows: A first and second threshold are pre-set. When the probability of a risk occurring is lower than the first threshold, it is considered low risk, indicating stable equipment operation with occasional fluctuations within a safe range. When the probability of a risk occurring is greater than or equal to the first threshold but less than the second threshold, it is considered medium risk, indicating potential defects or abnormal parameters requiring close monitoring. When the probability of a risk occurring is greater than or equal to the second threshold, it is considered high risk, indicating an imminent risk of equipment failure requiring emergency handling. The first and second thresholds are set based on power grid operation statistics. The settings of the first and second thresholds are adjusted according to transformer type and operating scenario. A database of first and second thresholds corresponding to different types of transformers and operating scenarios is pre-established. When it is necessary to set the first and second thresholds, parameter data and operating data of oil-immersed transformers are collected to obtain transformer type data and operating scenario data. The first and second thresholds are then directly retrieved from the database by comparison.

[0049] It should be noted that the warning signal includes a risk level identifier, key parameter anomalies, and suggested handling measures. Among the risk level identifiers, low risk is indicated by green, medium risk by yellow, and high risk by red. The suggested handling measures include routine inspection measures for low risk, precise diagnostic measures for medium risk, and immediate load reduction measures for high risk. The corresponding handling measures are selected from these and sent to the monitoring platform. Warning signals of the same level will not be sent repeatedly within a set time to avoid overloading maintenance personnel.

[0050] See Figure 2 An oil-immersed transformer operation risk assessment system, comprising: The data acquisition module is equipped with a sensor group for collecting internal operating data and external influencing factor data of the oil-immersed transformer. The sensor group includes internal operating parameter sensors and external influencing factor sensors, which are used to collect internal operating data and external influencing factor data, respectively. The internal operating parameter sensors include an oil temperature sensor for collecting oil temperature, a partial discharge sensor for collecting partial discharge, and a dissolved gas sensor for collecting dissolved gases in the oil. The external influencing factor sensors include an ambient temperature and humidity sensor for collecting temperature and humidity, and a grid voltage fluctuation rate sensor for collecting grid voltage fluctuation data. The sensor cables are double-shielded, with a grounding resistance of <1Ω, and have passed the IEC 61000-4-3 electromagnetic radiation immunity test. Sensors for key parameters employ a three-sensor redundancy configuration, and outliers are eliminated using a majority voting algorithm to improve data reliability.

[0051] The data processing module, connected to the acquisition module, is used to receive and preprocess the data collected by the acquisition module; it is equipped with a filtering unit, a normalization unit, and an imputation unit, which are used to perform data cleaning, normalization, and missing value imputation operations on the data, respectively.

[0052] The calculation module, connected to the data processing module, is used to receive preprocessed data and calculate the comprehensive risk assessment score, transformer health index, and risk occurrence probability based on the preprocessed data. It has built-in risk assessment model unit, health index model unit, and risk probability model unit, which are used to calculate the comprehensive risk assessment score, transformer health index, and risk occurrence probability, respectively.

[0053] The output module, connected to the calculation module, is used to output a risk level report consisting of a comprehensive risk assessment score, a transformer health index, and a risk occurrence probability, and to generate corresponding early warning signals.

[0054] In one embodiment of the present invention, a cloud server and an edge computing unit are further included. The cloud server stores a historical database and runs a machine learning model, while the edge computing unit is deployed locally at the substation. The historical database uses a distributed database to store transformer lifecycle data, including real-time operating parameters, a fault case library, and maintenance records and equipment files. The machine learning model runs a deep learning framework, iteratively optimizing the risk assessment formula weights, the calibration coefficients of the health index, and the logistic regression parameters of the risk probability model based on historical data. The edge computing unit uses an industrial-grade edge server, deployed in the substation control room, to meet the requirements of real-time data processing, local storage, and electromagnetic interference resistance. It can perform calculations of comprehensive risk assessment scores, transformer health indices, and risk occurrence probabilities, communicate with the PLC controller to achieve automatic control, and has data preprocessing and edge analysis functions, uploading only key feature values ​​to the cloud.

[0055] This invention constructs a comprehensive assessment system encompassing data acquisition, preprocessing, multi-dimensional risk calculation, and tiered response. It integrates operating parameters such as oil temperature, load, and discharge quantity with environmental humidity, voltage fluctuations, and equipment aging factors. By utilizing Butterworth filtering and ARIMA interpolation techniques to improve data quality, and combining a comprehensive risk formula with optimized weights using support vector machines, a health index model based on gas ratios and aging factors, and logistic regression probability calculation, it achieves accurate assessment from state quantification to risk prediction. The accompanying PLC automatic control, three-level early warning mechanism, and cloud-edge collaborative architecture form an intelligent closed loop of "monitoring-assessment-intervention-optimization," significantly improving the early identification capability of transformer faults, reducing false alarm and missed alarm rates, and providing dynamic and intelligent technical support for the full lifecycle management of equipment. This effectively improves the reliability of power grid operation and optimizes maintenance efficiency.

[0056] The above description is only a preferred embodiment of the present invention. It should be noted that those skilled in the art can make several improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method and system for assessing the operational risks of an oil-immersed transformer, characterized in that, Includes the following steps: Step S1: Obtain the real-time operating parameters and external influencing factors of the oil-immersed transformer; Step S2: Preprocess the real-time operating parameters and external influencing factors; Step S3: Based on the preprocessed real-time operating parameters and external influencing factors, output a comprehensive risk assessment score through a risk assessment formula. The comprehensive risk assessment score integrates the internal operating status and the impact of the external environment. Step S4: Based on the preprocessed real-time operating parameters and external influencing factors, calculate the transformer health index, and calculate the probability of risk occurrence based on the transformer health index and the comprehensive risk assessment score; Step S5: Output a risk level report based on the comprehensive risk assessment score, transformer health index and risk occurrence probability, and generate an early warning signal.

2. The method and system for assessing the operational risk of an oil-immersed transformer according to claim 1, characterized in that, The real-time operating parameters include oil temperature, load current, ambient temperature, partial discharge quantity, and dissolved gas concentration in the oil. The external influencing factors include ambient humidity, grid voltage fluctuation rate, and the equipment's historical operating years. The equipment's historical operating years are calculated based on the transformer's commissioning date and maintenance records to obtain an aging coefficient, which is used to adjust the weight allocation of the risk assessment formula.

3. The method and system for assessing the operational risk of an oil-immersed transformer according to claim 1, characterized in that, The preprocessing includes data cleaning, normalization, and missing value imputation. Data cleaning removes noise using filters, normalization employs a minimum-maximum scaling method, and missing value imputation is based on time series prediction. Data cleaning uses a low-pass Butterworth filter, and normalization scales each running parameter to the range of 0 to 1 by subtracting the minimum parameter value from the original value and then dividing by the maximum parameter value minus the minimum value. Missing value imputation uses an ARIMA time series model with a prediction window of 5 minutes.

4. The method and system for assessing the operational risk of an oil-immersed transformer according to claim 1, characterized in that, Also includes: Step S6: Based on the risk level report, execute automatic control actions, including adjusting the power of the transformer cooling system or reducing the load current. The automatic control actions are implemented through a PLC controller.

5. The method and system for assessing the operational risk of an oil-immersed transformer according to claim 1, characterized in that, The specific risk assessment formula is as follows: in, This represents the overall risk assessment score, with a dimensionless unit; the higher the value, the higher the risk. This represents the basic risk scaling coefficient, derived from regression analysis of historical accident data; This represents the weight of the i-th running parameter, which is optimized by the support vector machine algorithm during the training phase; This represents the i-th preprocessed running parameter, where i ranges from 1 to n, and n is the total number of parameters; This represents the gain coefficient due to temperature change, reflecting the sensitivity of oil temperature to ambient temperature. This indicates the temperature effect index, calibrated through laboratory temperature rise tests. This indicates the difference between oil temperature and ambient temperature, expressed in degrees Celsius. This represents the partial discharge reduction factor, which is set based on the properties of the insulating material. This represents the normalized partial discharge quantity, calculated by dividing the original partial discharge quantity by a preset threshold.

6. The method and system for assessing the operational risk of an oil-immersed transformer according to claim 5, characterized in that, The transformer health index is calculated using the following formula: in, This represents the transformer health index, with dimensionless units and a value range of 0 to 1, where 0 indicates complete failure and 1 indicates healthy. This represents the load contribution coefficient, determined through load curve analysis. To represent the hyperbolic tangent function, a nonlinear smoothing load impact effect is introduced; This represents the load gain constant, calculated based on the transformer's rated capacity. This represents the normalized load current, calculated by dividing the load current by the rated current, and is dimensionless. The weighting of dissolved gas influence is calibrated from gas chromatography data. It represents the ratio of dissolved gas concentrations in oil, calculated by dividing the hydrogen concentration by the methane concentration, and is dimensionless. This represents the aging acceleration factor, calculated from the equipment's historical operating years. This represents the aging factor, which is calculated by multiplying the number of years of operation by the average annual load rate.

7. The method and system for assessing the operational risk of an oil-immersed transformer according to claim 1, characterized in that, The probability of risk occurrence is calculated using the following formula: in, It represents the probability of risk occurring, with a dimensionless unit and a value range of 0 to 1. Values ​​higher than 0.7 are defined as high risk. The health index coefficient is used to fit the fault data through a logistic regression model. Indicates health index; This represents the comprehensive risk assessment score coefficient, optimized through Monte Carlo simulation; represents the comprehensive risk assessment score; k represents the probability offset constant, which is set based on power grid operation statistics. This represents the natural exponential function.

8. The method and system for assessing the operational risk of an oil-immersed transformer according to claim 1, characterized in that, The risk level report includes three levels: high risk, medium risk, and low risk. Based on the risk level report's classification, different warning signals are generated for each level and sent to the monitoring platform via a serial communication protocol. The specific classification rules are as follows: A first threshold and a second threshold are preset. When the probability of a risk occurring is lower than the first threshold, it is judged as low risk; when the probability of a risk occurring is greater than or equal to the first threshold and less than the second threshold, it is judged as medium risk. When the probability of a risk occurring is greater than or equal to the second threshold, it is judged as high risk.

9. A risk assessment system for the operation of an oil-immersed transformer, characterized in that, include: The data acquisition module is equipped with a sensor group for collecting internal operating data and external influencing factor data of oil-immersed transformers. The data processing module, connected to the acquisition module, is used to receive the data collected by the acquisition module and perform preprocessing. The calculation module, connected to the data processing module, is used to receive preprocessed data and calculate the comprehensive risk assessment score, transformer health index, and risk occurrence probability based on the preprocessed data. The output module, connected to the calculation module, is used to output a risk level report consisting of a comprehensive risk assessment score, a transformer health index, and a risk occurrence probability, and to generate corresponding early warning signals.

10. The oil-immersed transformer operation risk assessment system according to claim 9, characterized in that, It also includes a cloud server and an edge computing unit, wherein the cloud server stores a historical database and runs machine learning models, and the edge computing unit is deployed locally at the substation.

Images