A transformer insulation paper aging evaluation method based on furfural and white whale algorithm

Abstract

This invention discloses a method for assessing the aging of transformer insulation paper based on furfural and the beluga algorithm, comprising the following steps: (1) non-uniform accelerated thermal aging experiment; (2) multi-dimensional feature data acquisition; (3) constructing an IBWO-MLP prediction model to predict the degree of polymerization of insulation paper in hot spot areas; (4) establishing a field-based DP estimation model to predict the degree of polymerization of insulation paper in hot spot areas; (5) aging status assessment and lifetime prediction: outputting the remaining lifetime prediction results of insulation paper at transformer hot spots. This invention's method can accurately assess the aging degree of insulation paper in transformer hot spot areas under uninterrupted power supply conditions, with high accuracy and strong applicability, which helps in the early diagnosis of insulation faults.

Description

Technical Field

[0001] This invention relates to the field of power equipment condition assessment and fault prediction technology, and in particular to a method for assessing the aging of transformer insulation paper based on furfural and the beluga algorithm. Background Technology

[0002] Power transformers are core equipment in power grid operation, and their operational reliability directly affects the safety and stability of the power supply system. In actual operation, a significant proportion of transformer failures are related to the aging of insulation materials. The hot-spot areas of the windings have the highest temperatures, and the accelerated rates of local thermal stress and chemical reactions make the insulation paper in these areas the weakest link in the entire insulation system. Severe degradation in these areas can lead to insulation breakdown and power outages. Therefore, timely and accurate assessment of the aging degree of the paper insulation in hot-spot areas is crucial for developing maintenance and replacement strategies and preventing sudden failures.

[0003] Currently, the most intuitive parameters for characterizing the aging state of paper insulation are the degree of polymerization (DP) or tensile strength (TS). However, measuring these parameters requires disassembling the equipment and taking samples, which can cause transformer power outages and even damage to the insulation structure, making them unsuitable for use in online operation. Therefore, this research extensively employs non-destructive testing methods based on chemical product analysis. Furfural (2-Furfural) is a recognized major chemical marker of thermal degradation in paper insulation, with its production amount correlated with the degree of degradation of the insulating paper and unaffected by significant interference from other components in the oil. Under current conditions, furfural detection mainly relies on methods such as high-performance liquid chromatography (HPLC). HPLC detection technology is relatively mature and can accurately detect extremely low concentrations of furfural, with a fast measurement speed. Furthermore, the accompanying software system has powerful data processing and analysis capabilities, capable of generating test reports, providing strong support for the scientific assessment of transformer operating conditions.

[0004] As can be seen from the above, timely and reliable assessment of the remaining life of paper insulation in hotspot areas of UHV or large oil-immersed transformers under non-uniform thermal field conditions is of significant guiding importance and theoretical value for equipment operation, maintenance, repair, and component replacement. In existing technologies, the furfural-polymerization degree empirical model established based on uniform thermal field experiments is insufficiently applicable under complex field operating conditions, especially in hotspot areas, where the prediction results are prone to significant errors. Furthermore, because the formation, release, and degradation processes of furfural are coupled with multiple factors such as temperature gradients and humidity changes, the relationship between furfural and the degree of polymerization exhibits significant nonlinearity and multivariate coupling characteristics. A single chemical index cannot comprehensively and reliably reflect the actual degree of degradation of the insulating paper in hotspot areas. Summary of the Invention

[0005] To address the above shortcomings, this invention provides a method for assessing the aging of transformer insulation paper based on furfural and the White Whale algorithm. It combines an improved White Whale optimization algorithm with an optimized multilayer perceptron (IBWO-MLP) algorithm to assess the condition of paper insulation in hotspot areas. This method can accurately assess the aging degree of insulation paper in transformer hotspot areas under uninterrupted power supply conditions, exhibiting high accuracy and strong applicability, which helps in the early diagnosis of insulation faults. It can effectively improve the accuracy and stability of predictions under non-uniform thermal field conditions, providing a theoretical basis and technical support for transformer insulation condition assessment and operation and maintenance strategy formulation.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A method for evaluating the aging of transformer insulation paper based on furfural and the beluga algorithm includes the following steps: (1) Non-uniform accelerated thermal aging test: The non-uniform thermal aging device was used to accelerate the thermal aging of the oil paper insulation sample, which included the insulation paper and the insulation oil that had been dried and vacuum oil impregnated. (2) Multidimensional feature data acquisition: During the accelerated thermal aging process, the winding temperature at different axial heights is monitored in real time by temperature sensors, and insulating oil samples and insulating paper samples are extracted at preset time intervals to measure furfural concentration, humidity and degree of polymerization of insulating paper at hot spots. (3) Constructing the IBWO-MLP prediction model: The aging time, furfural concentration in oil, moisture content of insulating paper and winding temperature are used as input features to construct a dataset. The improved white whale optimization algorithm is used to optimize the multilayer perceptron prediction model to obtain the IBWO-MLP prediction model, which is used to predict the degree of polymerization of insulating paper in hot spots. (4) Establish a DP estimation model for on-site use: Based on the predicted degree of polymerization of insulating paper, and combined with the improved first-order kinetic aging equation, establish a life estimation model for insulating paper in hot spot areas; (5) Aging status assessment and life prediction: prediction results of the remaining life of the insulating paper at the hot spot of the output transformer.

[0007] Preferably, in the above-mentioned method for evaluating the aging of insulation paper in hot spots of transformers, the non-uniform thermal aging device includes a test tank, winding components, vertical heating pipes, an oil circulation cooling system, and a temperature acquisition controller; the test tank is provided with multiple temperature sensors and oil drain valves along the axial direction from top to bottom, with the top layer temperature being higher than the bottom layer temperature, forming an adjustable non-uniform temperature gradient.

[0008] Preferably, in the above-mentioned method for evaluating the aging of insulating paper in transformer hot spots, the drying treatment is set as follows: drying at 105°C under nitrogen atmosphere for 48 hours; vacuum oil impregnation parameters: impregnation at 60°C under nitrogen atmosphere for 48 hours. This ensures that the initial state of the samples is consistent and the moisture content is extremely low.

[0009] Preferably, in the above-mentioned transformer hot spot area insulation paper aging assessment method, in step (2), the furfural concentration in the oil is measured by high performance liquid chromatography; the humidity is measured by coulometric micro-water analyzer; and the degree of polymerization of the insulation paper is measured by automatic viscometer according to IEC60450 standard.

[0010] Preferably, based on the above-mentioned method for assessing the aging of insulating paper in transformer hot spots, the dataset uses aging time, furfural concentration in oil, moisture content of insulating paper, and winding temperature as input features of the model. The degree of polymerization of insulating paper in hot spots is used as the output target of the IBWO-MLP prediction model. The data is normalized and divided into training and test sets in an 8:2 ratio. The IBWO-MLP model is trained using the training set data to optimize model parameters, and the prediction accuracy of the model is verified using the test set data.

[0011] Preferably, based on the above-mentioned method for assessing the aging of insulation paper in transformer hot spots, an improved white whale optimization algorithm is constructed, specifically including the following steps: (1) Use the reverse learning initialization strategy to initialize the BWO random population; after reverse learning, generate the reverse population X' based on the initial population X. This expression is written as: (1); Where lu and lb represent the upper and lower bounds of the solution, respectively; (2) During the exploration phase, location updates were determined by the swimming behavior of paired beluga whales, as described below: (2); in Let Pi represent the new position of the i-th beluga whale in dimension j, and Pj, where j = 1,2,...,d, represent a random number along dimension 1 to D; Let represent the position of the i-th beluga whale in dimension Pj, and and Let r represent the current positions of the i-th and r-th beluga whales, respectively; r represents a randomly selected beluga whale, and r1 and r2 are random numbers between (0, 1); even and odd numbers represent even and odd numbers, respectively. (3) During the development phase, the Levy flight strategy is implemented to improve the convergence of the algorithm. The mathematical expression of this strategy is as follows: (3); in and These are the positions of the random beluga whale and the optimal beluga whale, respectively. LF represents a randomly generated number that conforms to the Levy distribution; r3 and r4 are random numbers between (0, 1), and c1 is the random jump intensity that quantifies the intensity of the Levy flight. (4) During the whale's fall, this expression can be written as: (4); Where r8-r10 are random numbers generated in the range (0,1); XStep is the step size of the whale's fall; δ is the step factor.

[0012] Preferably, in the above-mentioned method for assessing the aging of insulation paper in transformer hot spots, the execution steps of the IBWO-MLP prediction model are as follows: (1) Set the population size, maximum number of iterations, and search boundary of the hyperparameters for the IBWO algorithm; (2) Use the MLP prediction error as the fitness function of the IBWO algorithm; (3) The beluga whale's position is updated iteratively through the exploration and development of the IBWO algorithm and the whale's fall segment; (4) When the iteration termination condition is met, output the optimal number of neurons and learning rate.

[0013] Preferably, in the above-mentioned method for assessing the aging of insulating paper in transformer hot spots, the improved first-order kinetic aging equation includes a temperature factor and a furfural concentration correction term, the expression of which is: (5).

[0014] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: 1. Compared with existing technologies, the beneficial effects of this invention are primarily reflected in the significant improvement in evaluation accuracy. By constructing a non-uniform thermal aging experimental platform, this invention can simulate the actual temperature gradient distribution inside the transformer in the field, obtaining more representative aging data. Based on this, the improved White Whale Optimization Algorithm (IBWO) is used to optimize the Multilayer Perceptron (MLP), effectively solving the problem of traditional models dealing with the highly nonlinear coupling relationship between multiple variables such as furfural concentration, temperature, and moisture and the degree of polymerization (DP) of the insulating paper. This enables high-precision prediction of the DP value in the most critical hotspot area of ​​the transformer.

[0015] 2. This invention combines an artificial intelligence prediction model with an improved first-order dynamic life equation, enabling the rapid calculation of the remaining life of the insulating paper based on the furfural concentration detected in the oil sample under uninterrupted power conditions. This avoids the drawbacks of destructive sampling of on-site samples by traditional methods, and significantly improves the scientificity, timeliness, and economy of transformer condition assessment and operation and maintenance decisions.

[0016] Compared with the prior art, the present invention has the following significant advantages: 1. Compared with existing technologies, the beneficial effects of this invention are primarily reflected in the significant improvement in evaluation accuracy. By constructing a non-uniform thermal aging experimental platform, this invention can simulate the actual temperature gradient distribution inside the transformer in the field, obtaining more representative aging data. Based on this, the improved White Whale Optimization Algorithm (IBWO) is used to optimize the Multilayer Perceptron (MLP), effectively solving the problem of traditional models dealing with the highly nonlinear coupling relationship between multiple variables such as furfural concentration, temperature, and moisture and the degree of polymerization (DP) of the insulating paper. This enables high-precision prediction of the DP value in the most critical hotspot area of ​​the transformer.

[0017] 2. This invention combines an artificial intelligence prediction model with an improved first-order dynamic life equation, enabling the rapid calculation of the remaining life of the insulating paper based on the furfural concentration detected in the oil sample under uninterrupted power conditions. This avoids the drawbacks of destructive sampling of on-site samples by traditional methods, and significantly improves the scientificity, timeliness, and economy of transformer condition assessment and operation and maintenance decisions. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below.

[0019] Figure 1 This is a flowchart of the power transformer insulation paper life assessment method based on furfural and beluga algorithm, which is an embodiment of the present invention. Figure 2 This is a schematic diagram of oil paper sampling for a transformer insulation paper aging assessment method based on furfural and beluga algorithm according to Embodiment 1 of the present invention; Figure 3 This is a flowchart illustrating the implementation of hotspot area DP prediction based on the IBWO-MLP model in a transformer insulation paper aging assessment method based on furfural and beluga algorithm, as described in Embodiment 1 of the present invention.

[0020] Figure 4 This is a fitness curve diagram of different algorithms for a transformer insulation paper aging assessment method based on furfural and beluga algorithms, according to Embodiment 2 of the present invention. Figure 5 This is a hotspot area DP prediction result diagram based on different prediction models for a transformer insulation paper aging assessment method based on furfural and beluga algorithm according to Embodiment 2 of the present invention. Figure 6 This is a prediction error diagram of different prediction models for a transformer insulation paper aging assessment method based on furfural and beluga algorithm in Embodiment 2 of the present invention. Figure 7This is a verification diagram of the hotspot area DP estimation value of a transformer insulation paper aging assessment method based on furfural and beluga algorithm in Embodiment 2 of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0022] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," 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," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0023] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" 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 mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. Furthermore, the technical features involved in the different embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0024] Example 1 like Figure 1 As shown, a method for assessing the aging status of paper insulation in hot spots of power transformers based on furfural concentration and an improved white whale optimization algorithm is presented. The assessment process is as follows: Figure 1 As shown, it includes the following steps: (1) First, the oil-paper insulation material was dried for 48 hours under vacuum conditions of 105°C / 50 Pa to significantly reduce its internal moisture. Then, the dried insulation paper was immersed in dried oil under vacuum conditions of 60°C / 50 Pa for 48 hours. Subsequently, the insulation paper was uniformly wrapped around the copper coil in a non-uniform aging device, and insulating oil was prepared to completely immerse the insulation paper. In addition, excess air in the aging device was purged and nitrogen was injected.

[0025] (2) The non-uniform aging device in this experiment can provide different temperature gradients, and five temperature sensors are installed at different heights. The winding temperatures monitored from top to bottom are 130℃, 112℃, 92℃, 78℃ and 60℃, respectively. In order to make the device better simulate the non-uniform thermal field of the transformer, the temperature gradient is set with the temperature gradient law of the transformer winding in mind.

[0026] (3) Samples of the oil-paper insulation material are taken every 24 hours. Figure 2 A diagram depicting the sampling process is provided. Samples of oil and oil-impregnated paper were collected at five different heights (h1-h5) along the winding axis. To ensure the accuracy of the results, the insulating paper was sampled six times at the same height for DP and humidity measurements. Additionally, three samples of insulating oil were collected to measure furfural concentration, and the results were averaged.

[0027] (4) The furfural concentration (unit: mg / L) was determined by a Waters Alliance E2695 high performance liquid chromatograph according to IEC 61198 standard; the moisture content of the impregnated paper (unit: %) was analyzed by a Metrohm Karl Fischer coulometer according to IEC 60814 standard; and the degree of polymerization (DP) of the impregnated paper was measured by a Shanghai SRDNCY-6 automatic viscometer according to IEC 60450 standard.

[0028] (5) The IBWO algorithm is applied to optimize the hyperparameters of the MLP model. The IBWO-MLP prediction model process is as follows: Figure 3 The execution steps are as follows: Step 1: Initialize the initial parameters of the IBWO-MLP prediction model, such as the beluga whale population size, number of iterations, and the hyperparameter boundaries lb and ub to be optimized. The sample data is divided into training and testing datasets at a ratio of 80%-20%. Step 2: Standardize the data in the datasets, and (1) and (2) are used to calculate the fitness of individual beluga whales.

[0029] (1); (2); Step 3: Use (1) to calculate the individuals of the reverse population and rank the individual fitness according to (2) to obtain the reverse population. Step 4: Calculate the balance factor Bf according to (3). When Bf > 0.5, the algorithm executes the exploration phase to update the positions of individual beluga whales; when Bf < 0.5, the algorithm switches to the exploitation phase to update the positions of individual beluga whales. Step 5: Calculate the probability of whale fall Wf. If Bf < Wf, enter the whale fall phase to update the positions of individuals; if Bf > Wf, update the fitness F(X). Step 6: Obtain the best individual from the beluga whale population, determine the best number of neurons and learning rate, and replace them into the MLP for training. Then use the test dataset to test the model. Step 7: Determine whether the number of iterations and accuracy meet the requirements. If so, output the DP result; otherwise, go to Steps 4 and 5 to continue the calculation loop.

[0030] (6) Construction of the DP estimation model. This model is developed by using IBWO-MLP to expand the sample data and combining it with the improved first-order kinetic model. First, the traditional first-order kinetic model of cellulose aging is shown as follows: (3); Previous studies considered the value of A as a constant; it can be corrected by the following function. Since the moisture inside the transformer fluctuates, if the moisture in this paper is calculated by detecting the moisture in the oil, errors affecting the estimation accuracy will be introduced. Therefore, the following model is constructed without considering humidity. Therefore, (3) can be written as (4); The DP estimation model can be expressed as: (5); Example 2 The difference between this example and Example 1 is that, to prove the effectiveness of the IBWO optimization algorithm, it is compared with BWO-MLP, particle swarm optimization (PSO), grey wolf optimization (GWO), whale optimization algorithm (WOA), and bat optimization algorithm (BA). The maximum number of iterations for each algorithm is set to 50, and the population size is 20. Each of the above algorithms is tested 15 times, and the fitness results are as Figure 4 shown.

[0031] Based on the IBWO-MLP model, the DP prediction results of the insulating paper in the hot spot area are obtained. At the same time, five other prediction models (BWO-MLP, LSTM, PSO-MLP, SVM, and XGBoost) are also introduced for accuracy comparison. The prediction results are as Figure 5 shown. In addition, the relative errors (%) of the results of different prediction models have also been calculated and shown in Figure 6 .

[0032] like Figure 7 The figure shown is a validation plot of the DP estimates for hotspot areas.

[0033] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for evaluating the aging of transformer insulation paper based on furfural and the beluga algorithm, characterized in that, Includes the following steps: (1) Non-uniform accelerated thermal aging test: The non-uniform thermal aging device was used to accelerate the thermal aging of the oil paper insulation sample, which included the insulation paper and the insulation oil that had been dried and vacuum oil impregnated. (2) Multidimensional feature data acquisition: During the accelerated thermal aging process, the winding temperature at different axial heights is monitored in real time by temperature sensors, and insulating oil samples and insulating paper samples are extracted at preset time intervals to measure furfural concentration, humidity and degree of polymerization of insulating paper at hot spots. (3) Constructing the IBWO-MLP prediction model: The aging time, furfural concentration in oil, moisture content of insulating paper and winding temperature are used as input features to construct a dataset. The improved white whale optimization algorithm is used to optimize the multilayer perceptron prediction model to obtain the IBWO-MLP prediction model, which is used to predict the degree of polymerization of insulating paper in hot spots. (4) Establish a DP estimation model for on-site use: Based on the predicted degree of polymerization of insulating paper, and combined with the improved first-order kinetic aging equation, establish a life estimation model for insulating paper in hot spot areas; (5) Aging status assessment and life prediction: prediction results of the remaining life of the insulating paper at the hot spot of the output transformer.

2. The method for assessing the aging of insulating paper in transformer hot spots according to claim 1, characterized in that: Based on the above-mentioned method for evaluating the aging of insulation paper in transformer hot spots, the non-uniform thermal aging device includes a test tank, winding components, vertical heating pipes, an oil circulation cooling system, and a temperature acquisition controller. The test tank is equipped with multiple temperature sensors and oil drain valves along the axial direction from top to bottom, with the top layer temperature being higher than the bottom layer temperature, forming an adjustable non-uniform temperature gradient.

3. The method for assessing the aging of insulating paper in transformer hot spots according to claim 1, characterized in that: The drying process was set as follows: drying at 105°C for 48 hours under nitrogen atmosphere; vacuum oil immersion parameters: immersion at 60°C to 90°C for 48 hours under nitrogen atmosphere to ensure that the initial state of the sample was consistent and the moisture content was extremely low.

4. The method for assessing the aging of insulating paper in transformer hot spots according to claim 1, characterized in that: In the above-mentioned method for evaluating the aging of insulating paper in transformer hot spots, in step (2), the furfural concentration in the oil is measured by high performance liquid chromatography; the humidity is measured by coulometric micro-water analyzer; and the degree of polymerization of the insulating paper is measured by automatic viscometer according to IEC60450 standard.

5. The method for assessing the aging of insulating paper in hot-spot areas of transformers according to claim 1, characterized in that: The dataset uses aging time, furfural concentration in oil, moisture content of insulating paper, and winding temperature as input features of the model. The degree of polymerization of insulating paper in hot spots is used as the output target of the IBWO-MLP prediction model. The data is normalized and divided into training and test sets in an 8:2 ratio. The IBWO-MLP model is trained using the training set data to optimize the model parameters, and the prediction accuracy of the model is verified using the test set data.

6. The method for assessing the aging of insulating paper in hot-spot areas of transformers according to claim 1, characterized in that: The improved beluga optimization algorithm is constructed by following these steps: (1) Use the reverse learning initialization strategy to initialize the BWO random population; after reverse learning, generate the reverse population X' based on the initial population X. This expression is written as: （1）； Where lu and lb represent the upper and lower bounds of the solution, respectively; (2) During the exploration phase, location updates were determined by the swimming behavior of paired beluga whales, as described below: （2）； in Let Pi represent the new position of the i-th beluga whale in dimension j, and Pj, where j = 1,2,...,d, represent a random number along dimension 1 to D; Let represent the position of the i-th beluga whale in dimension Pj, and and Let r represent the current positions of the i-th and r-th beluga whales, respectively; r represents a randomly selected beluga whale, and r1 and r2 are random numbers between (0, 1); even and odd numbers represent even and odd numbers, respectively. (3) During the development phase, the Levy flight strategy is implemented to improve the convergence of the algorithm. The mathematical expression of this strategy is as follows: （3）； in and These are the positions of the random beluga whale and the optimal beluga whale, respectively. LF represents a randomly generated number that conforms to the Levy distribution; r3 and r4 are random numbers between (0, 1), and c1 is the random jump intensity that quantifies the intensity of the Levy flight. (4) During the whale's fall, this expression can be written as: （4）； Where r8-r10 are random numbers generated in the range (0,1); XStep is the step size of the whale's fall; δ is the step factor.

7. The method for assessing the aging of insulating paper in transformer hot spots according to claim 1, characterized in that: Based on the above-mentioned method for assessing the aging of insulation paper in transformer hot spots, the execution steps of the IBWO-MLP prediction model are as follows: (1) Set the population size, maximum number of iterations, and search boundary of the hyperparameters for the IBWO algorithm; (2) Use the MLP prediction error as the fitness function of the IBWO algorithm; (3) The beluga whale's position is updated iteratively through the exploration and development of the IBWO algorithm and the whale's fall segment; (4) When the iteration termination condition is met, output the optimal number of neurons and learning rate.

8. The method for assessing the aging of insulating paper in hot-spot areas of transformers according to claim 1, characterized in that: Based on the above-mentioned method for assessing the aging of insulating paper in transformer hot spots, the improved first-order kinetic aging equation includes a temperature factor and a furfural concentration correction term, and its expression is as follows: （5）。

Images