Transformer bank operation state evaluation and optimization method considering cooling system transformation

Abstract

The invention relates to the field of transformer bank operation optimization methods, in particular to a transformer bank operation state evaluation and optimization method considering cooling system transformation, which comprises the following steps of: S1, determining a mathematical mapping relationship among the number of cooling fans, the rotating speed type of the fans, the load rate and an operation temperature drop value of an oil-immersed self-cooling transformer; establishing a transformer cooling system transformation efficacy model; s2, determining unknown parameters in the transformation efficacy model of the transformer cooling system according to the field operation data; s3, generating load rate data of the transformers in a future planning period according to the historical load rate data of the transformers; establishing a mathematical relationship between the operating temperature and the aging factor of the transformer to obtain a transformer operating state evaluation method considering the transformation influence of the cooling system; and S4, on the basis of S1, S2 and S3, forming a transformer bank cooling system transformation scheme optimization decision-making method considering the cost constraint so as to determine a cooling system transformation scheme of each transformer. According to the invention, an accurate optimization scheme can be obtained.

Description

Technical Field

[0001] The present invention relates to the field of transformer group operation optimization methods, and in particular to a transformer group operation status evaluation and optimization method considering cooling system transformation. Background Art

[0002] Power transformers are the core equipment in power generation systems, playing the role of power transfer and voltage conversion. Oil-immersed self-cooling transformers are widely used in power generation systems due to their simple structure, high efficiency, and low maintenance. Many regional substations directly connected to power plants are equipped with oil-immersed self-cooling transformers. This type of transformer mainly uses the natural convection of oil to drive the heat generated by the transformer to the surface of the oil tank wall and the location of the heat dissipation pipe, and then dissipates the heat under the action of air convection and air heat conduction.

[0003] In recent years, with the changes in the national economy and the operation mode of the power grid, the power sector requires the load capacity of transformers in specific areas to be increased to ensure sufficient power supply. However, the increase in the transformer load rate will lead to an increase in load loss and heat generation. At this time, if the cooling and heat dissipation capacity of the transformer is insufficient, it may cause the transformer to overheat, which will not only accelerate its aging process, but also affect its stable operation and even shorten its expected service life. Taking some areas as an example, due to the influence of the manufacturing process and the external operating environment, some oil-immersed self-cooling transformers have an operating temperature exceeding the warning line under high temperature and high load conditions, which exposes the problem of insufficient heat dissipation capacity of the cooling system of the oil-immersed self-cooling transformer. To this end, the project site personnel usually modify the cooling system of the oil-immersed self-cooling transformer by installing a high-speed cooling fan to reduce the operating temperature of the transformer under high load conditions and improve its operating status.

[0004] However, the current transformer cooling system modification (i.e., cooling fan installation) mainly relies on the empirical methods of on-site operators, and lacks accurate and effective mathematical models to analyze and evaluate the impact of cooling fans. This deficiency makes it impossible to objectively quantify the improvement effect of installing cooling fans on the operating status of transformers. Especially when a certain transformer group needs to undergo centralized cooling system modification, the existing methods cannot maximize the operating status of the entire transformer group within a limited cost, and it is difficult to achieve cost-effective optimization.

[0005] At present, in terms of transformer cooling system transformation efficiency modeling, existing methods usually rely on the expert experience of on-site operators to determine the number of cooling fans and speed settings that need to be installed. However, due to the lack of a mathematical model that quantitatively describes the relationship between the degree of transformer operating temperature drop and the number of cooling fans and fan speed types, it is impossible to deeply analyze the impact of different cooling fan numbers and speed selections on the improvement of transformer operating temperature.

[0006] In terms of transformer operating status assessment considering cooling system modification, the operating status and performance aging of the transformer are usually determined by its operating temperature. However, due to the lack of a mathematical model that quantitatively describes the relationship between the degree of transformer operating temperature drop and the number of cooling fans and fan speed types, existing methods cannot consider the impact of cooling system modification in transformer operating status and aging assessment.

[0007] In terms of optimizing the operating status of transformers considering the transformation of cooling systems, existing methods usually analyze individual transformers and determine the transformation plan for each transformer one by one. However, this method is difficult to determine the overall transformation plan for the transformer group under limited investment costs, and may not achieve the optimal improvement of the operating status of the entire transformer group. Summary of the invention

[0008] The present invention aims to provide a method for evaluating and optimizing the operating status of a transformer group taking into account the transformation of a cooling system, so as to solve the problem that the existing method cannot determine the impact of the transformation of the cooling system on the transformer.

[0009] The transformer group operation status evaluation and optimization method considering the cooling system transformation in this scheme includes the following steps:

[0010] S1, based on the multi-parameter Sigmoid function, determine the mathematical mapping relationship between the number of cooling fans, fan speed type, load rate and the operating temperature drop value of the oil-immersed self-cooling transformer, and establish a data-driven transformer cooling system transformation efficiency model;

[0011] S2, based on the field operation data, the residual minimization method with the introduction of the Elastic Net regularization method is used to determine the unknown parameters in the transformer cooling system transformation efficacy model established in S1;

[0012] S3, based on the historical load rate data of each transformer, the acceptance-rejection sampling method is used to generate the load rate data of the transformer in the future planning period; then the precise mathematical relationship between the transformer operating temperature and its aging factor is established, and a transformer operating status evaluation method considering the impact of cooling system transformation is proposed;

[0013] S4, based on the transformer cooling system transformation efficacy model established in S1, the transformer load rate generation method in S2, and the transformer operating status evaluation method, forms an optimization decision-making method for the transformer group cooling system transformation plan considering cost constraints to determine the cooling system transformation plan for each transformer.

[0014] The beneficial effects of this program are:

[0015] The multiple fan-related parameters in the transformer cooling system are mathematically mapped to the temperature of the transformer, and the corresponding model is established to quantitatively evaluate the impact of adding different numbers of cooling fans and selecting different fan speed types on the transformer operating temperature. Then, the operating status evaluation and optimization decision of the transformer group after the cooling system transformation can be carried out, which can achieve effective quantification of the cooling system transformation and meet the actual needs of the project for effective evaluation and efficient optimization of the transformer group operating status. The transformer aging factor is used as the characteristic quantity to describe its operating status, and the precise mathematical relationship between the transformer operating temperature and its aging factor is established. The number of cooling fans installed on each transformer in the transformer group and the selected fan speed type are used as decision variables to obtain the specific transformation plan for each transformer in the transformer group, and the impact of the cooling system on the transformer under different transformation conditions is accurately evaluated, making the transformer cooling system transformation more practical.

[0016] Furthermore, the establishment process of the transformer cooling system transformation efficacy model in S1 includes:

[0017] S1.1, obtain the field operation data of the oil-immersed self-cooling transformer that has undergone cooling system transformation as data input. The field operation data includes: the number of cooling fans N, the rated speed V of each cooling fan, and the transformer load data P op , Transformer operating temperature T op ;

[0018] S1.2, based on the transformer load data, the transformer load rate data is calculated and the calculation formula is expressed as:

[0019]

[0020] Where P op is the actual load of the transformer at hour h, P rate is the rated capacity of the transformer;

[0021] S1.3, based on the transformer load rate data, calculate the maximum operating temperature T of the transformer without cooling system modification hst , the maximum operating temperature T in the hth hour hst,h The calculation formula is:

[0022] T hst,h =Θ E,h +ΔΘ O,h +ΔΘ Y,h ;

[0023] Where, T hst,h is the hot spot temperature of the transformer at the hth hour; ΔΘ E,h is the ambient temperature at hour h; ΔΘ O,h and ΔΘ Y,hThey are the temperature rise of the transformer top oil temperature to the ambient temperature and the temperature rise of the winding hot spot temperature to the top oil temperature;

[0024] S1.4, based on the actual operating temperature of the transformer and the virtual maximum operating temperature of the winding, calculate the drop in the transformer operating temperature after the cooling system is modified. The calculation formula is:

[0025] T dec,h =T hst,h -T op,h ;

[0026] Where, T hst,h is the virtual maximum operating temperature of the transformer at the hth hour, T op,h is the actual operating temperature of the transformer; T dec,h is the drop in transformer operating temperature caused by the transformation of the cooling system;

[0027] S1.5, based on the Sigmoid function, establish the number of transformer cooling fans N, fan speed type V, and load rate L op The operating temperature drop value T of oil-immersed self-cooling transformer dec The mathematical equation between is expressed as:

[0028]

[0029] In the formula, W is the sensitivity coefficient of the adjustment temperature drop to be estimated; h is the time-varying coupling factor after considering the interaction of the number of cooling fans, the rated speed of the fans, and the load rate; are parameters to be estimated.

[0030] The beneficial effect is that the various parameters of the cooling system modified in the transformer are associated, and the influence on the transformer under the premise of designing different parameters can be quantified.

[0031] Further, the temperature rise Δθ of the transformer top oil temperature to the ambient temperature in S1.3 is O,h and ΔΘ Y,h The calculation formulas are:

[0032]

[0033]

[0034] In the formula, the symbols init and ulti represent the start and end time of the hth hour respectively; ΔΘ O,R and ΔΘ Y,R They are the temperature rise of the transformer top oil temperature relative to the ambient temperature and the winding virtual maximum operating temperature relative to the top oil temperature under rated load; are the temperature rise of the top oil temperature of the transformer relative to the ambient temperature at the beginning and end of the hth hour respectively; are the temperature rise of the virtual maximum operating temperature of the transformer winding at the beginning and end of the hth hour relative to the top oil temperature of the transformer; are the load rates of the transformer at the beginning and end of the hth hour, υ TO and H are the transformer oil and winding temperature rise time constants respectively; R is the ratio of the transformer rated load loss to the no-load loss; σ1 and σ2 are constant parameters determined by the cooling method of the oil-immersed self-cooling transformer, and their values ​​are 0.8 and 0.5 respectively.

[0035] The beneficial effect is that the temperature rise calculation of the transformer oil temperature to the ambient temperature takes time and load into consideration, thereby improving the accuracy of the calculation result.

[0036] Further, the estimation of unknown parameters in S2 includes:

[0037] S2.1, based on the mathematical model established in S1.5, the Huber loss function is used as the model parameter to estimate the loss function, which is calculated as:

[0038]

[0039]

[0040] Where: Loss(θ) is the sum of all residual losses; δ is the error control threshold; g(ω i ) is the residual loss of the i-th group of data; θ is the estimated value of the unknown parameters in the model; is the predicted value based on the model parameters θ;

[0041] S2.2, introduce the Elastic Net regularization method and rewrite the sum of residual losses Loss(θ) in S2.1 as:

[0042]

[0043] Where: j is the number of unknown parameters; is the regularization term; λ1 and λ2 are regularization parameters that control the weight of the regularization term;

[0044] S2.3, using the optimization method based on gradient descent to iteratively adjust the parameters, after initializing the parameters θ, the gradient of the loss function Loss(θ) for each parameter is calculated as:

[0045]

[0046] S2.4, update the parameter θ according to the gradient, and the update process is expressed as:

[0047]

[0048] Where: α is the learning rate, which controls the magnitude of each update;

[0049] S2.5: Repeat S2.3 and S2.4 until the loss function stops decreasing and the optimal parameter estimate of the model is determined.

[0050] The beneficial effect is that the method used in the estimation process of each unknown parameter can improve the accuracy of the estimated result.

[0051] Further, the S3 includes:

[0052] S3.1, using clustering algorithm to extract N l The typical daily average load rate curve of the transformer is shown as follows:

[0053]

[0054] Where: They represent the average load factor values ​​from the 1st to the 24th hour in the daily average load factor curve of Article 1 respectively;

[0055] S3.2, based on the historical growth trend of transformer load rate and combined with possible future load growth factors, predict the growth rate of transformer load rate in the next few years

[0056]

[0057] Where: W It represents the cumulative annual load growth rate of the transformer in the future W years;

[0058] S3.3, from N l Extract a typical curve from the typical daily average load rate curve of the transformer. Assume that the curve extracted is the lth curve, and the average value of 24*W is Standard deviation is SD%*K rate The log-normal distribution of w=1,2,...,W,i=1,2,...,24,where K rate is the rated load rate of the transformer, and the log-normal distribution is expressed as:

[0059]

[0060] In the formula, represents the lognormal distribution corresponding to the i-th hour in the w-th year; μ i,w and σ i,w denote the mean and standard deviation parameters of the lognormal distribution respectively;

[0061] S3.4, according to the rejection-acceptance sampling principle, select the standard normal distribution G(x) as the proposed distribution;

[0062] S3.5, calculation of constant c normal , so that the target distribution function is magnified by the proposed distribution function c normal The complete envelope of . That is, for any x, the following conditions must be met:

[0063]

[0064] S3.6, Generate a random sample Y from the proposed distribution G(x) sam , and generate a random number U that follows a uniform distribution between [0,1] sam , calculate the candidate value of the load factor at the i-th hour in the w-th year And calculate the acceptance probability, expressed as:

[0065]

[0066] S3.7, Candidate value judgment, if A accept ≤U sam , then accept as a random sample of the load rate at hour i in year w, otherwise the sample is rejected and sampling is restarted;

[0067] S3.8, repeat S3.3 to S3.7 in chronological order until enough load rate curves of the transformer for the next W years are generated;

[0068] S3.9, after completing S3.8, the virtual maximum operating temperature of the transformer and the operating temperature drop value after considering the specific cooling system modification can be calculated according to the formula in step 1; thereby, the operating temperature time series of the transformer in the next W years after considering the cooling system modification is calculated, and the transformer operating temperature after considering the cooling system modification is calculated by the following formula:

[0069] T op,i,w =T hst,i,w -T dec,i,w ;

[0070] Where: T hst,i,w is the operating temperature of the transformer at the i-th hour in the w-th year without considering the impact of the cooling system modification; T dec,i,w is the drop in operating temperature of the transformer at the i-th hour in the w-th year after the cooling system modification;

[0071] S3.10, using the transformer accelerated aging factor as the characteristic quantity of the transformer's future operating state, based on the transformer's operating temperature time series calculated in S3.9 for the next W years, evaluate the transformer's operating state in the next W years:

[0072]

[0073] In the formula, is the hot spot temperature reference value, B is an empirical constant with a value of 15000, Z is the Zth future load rate curve of the transformer generated by S3.8, T op,w,k The operating temperature of the transformer after the cooling system modification is considered in the future w and k years.

[0074] The beneficial effect is that the quantitative evaluation of the transformer operating status considering the impact of cooling system transformation during the planning period can fully capture the volatility and random uncertainty of the transformer's future load rate and improve the accuracy of the evaluation results.

[0075] Furthermore, in S3.3, the mean and standard deviation parameters of the lognormal distribution are calculated using the load mean The standard deviation is SD%*Krate. The two parameters are calculated by the following formula:

[0076]

[0077] The beneficial effect is that the calculation of mean and standard deviation parameters can improve the accuracy of the calculation results.

[0078] Furthermore, the step of determining the cooling system modification plan for each transformer in S4 includes:

[0079] S4.1, the goal of the optimization decision model for the transformer group cooling system transformation plan considering cost constraints is to optimize the overall operating status of the transformer group during the planning period. The objective function of the optimization model is:

[0080]

[0081] Where: FA is the overall operating status evaluation value of the transformer group during the planning period, F aging , s is the operating status evaluation value of the sth transformer in the transformer group during the planning period;

[0082] S4.2, the constraints that the transformer group cooling system transformation plan optimization decision model needs to meet include the cooling system transformation investment cost constraint during the planning period and the number of cooling fans installed on each transformer;

[0083] S4.3, input the electrical parameters, operating parameters, and historical load curve parameters of the transformer group, introduce the simulated degradation algorithm into the iterative process of the conventional particle swarm, and after each update of the particle speed and position, judge whether the particle swarm algorithm accepts the new global optimal solution according to the simulated annealing criterion.

[0084] The beneficial effect is that the transformation plan formed under various parameters and constraints can meet the needs of actual operation and reduce economic costs.

[0085] Furthermore, the constraints in S4.2 are respectively expressed as:

[0086] The cooling system renovation investment cost constraint means that the cooling system renovation investment cost during the planning period should be less than the predetermined investment cost, which can be expressed as:

[0087] C cost ≤C set ;

[0088] In the formula, C set is the acceptable upper limit of investment cost during the planning period; Ccost is the total investment cost during the planning period, and the total investment cost is expressed as:

[0089]

[0090] Where V is the total number of cooling fans with different speeds; J v is the investment cost of installing a single type v cooling fan; num s,v is the number of the vth type cooling fans installed on the sth transformer in the transformer group;

[0091] The number of cooling fans is limited. During the planning period, the number of cooling fans that can be installed on each transformer should be less than the upper limit of the predetermined number of cooling fans, which can be expressed as:

[0092]

[0093] Where: num s,se t It is the upper limit of the cooling fans that can be installed on the sth transformer in the transformer group.

[0094] The beneficial effect is: taking the total investment cost as the key constraint, an optimization decision model for the transformer group cooling system transformation plan considering the cost constraint is established, and the improved particle swarm algorithm is used to solve it, which can obtain the optimal cooling system transformation plan under the optimal cost conditions, making the operation of the overall equipment more energy-saving and economical.

[0095] Further, the simulated annealing criterion in S4.3 is expressed as:

[0096] If f p(r)-f g (r)≤0, then accept the new solution;

[0097] If f p (r)-f g (r)>0, with acceptance probability P act Accept the new solution with probability P ac The calculation formula is:

[0098]

[0099] Where: f p (r) and f g (r) represent the fitness values ​​of the individual optimal position and the global optimal position at the rth iteration, that is, the overall operating status evaluation value of the transformer group during the planning period; TP r represents the simulated annealing temperature at the rth iteration, and the simulated annealing temperature is calculated by the following formula:

[0100]

[0101] Where: ζ represents the simulated temperature attenuation coefficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0102] Figure 1 A schematic block diagram of an embodiment of a transformer group operation status evaluation and optimization method considering cooling system modification according to the present invention;

[0103] Figure 2 The optimization decision model solving process in the embodiment of the transformer group operation status evaluation and optimization method considering the cooling system transformation of the present invention;

[0104] Figure 3 A schematic diagram of the application of the cooling system transformation efficacy model in the embodiment of the transformer group operation status evaluation and optimization method considering the cooling system transformation of the present invention;

[0105] Figure 4 A diagram showing the number of cooling fans required to be installed on different transformers in an embodiment of a transformer group operation status evaluation and optimization method considering cooling system modification of the present invention;

[0106] Figure 5 These are the transformer group operating status evaluation results under two schemes in the transformer group operating status evaluation and optimization method embodiment considering cooling system modification of the present invention. DETAILED DESCRIPTION

[0107] The following is further explained in detail through specific implementation methods.

[0108] Example

[0109] Transformer group operation status evaluation and optimization method considering cooling system modification, such as Figure 1 As shown, the following steps are included:

[0110] S1, based on the multi-parameter Sigmoid function, determine the mathematical mapping relationship between the number of cooling fans, fan speed type, load rate and the operating temperature drop value of the oil-immersed self-cooling transformer, and establish a data-driven transformer cooling system transformation efficiency model. The establishment process of the transformer cooling system transformation efficiency model includes:

[0111] S1.1, obtain the field operation data of the oil-immersed self-cooling transformer that has undergone cooling system transformation as data input. The field operation data includes: the number of cooling fans N installed on each transformer recorded in the equipment ledger, the rated speed V of each cooling fan, and the transformer load data P recorded in the equipment operation data. op , obtain the transformer operating temperature T through the online infrared monitoring system op .

[0112] S1.2, based on the transformer load data, the transformer load rate data is calculated and the calculation formula is expressed as:

[0113]

[0114] Where P op is the actual load of the transformer at hour h, P rate is the rated capacity of the transformer. To facilitate subsequent calculations, the load rate expressed in percentage can be converted to a constant, for example, 80% can be converted to 0.8.

[0115] S1.3, based on the transformer load rate data, calculate the maximum operating temperature T of the transformer without cooling system modification hst , the highest operating temperature is the hot spot temperature, the highest operating temperature T in the hth hour hst,h The calculation formula is:

[0116] T hst,h =Θ E,h +ΔΘ O,h +ΔΘ Y,h ;

[0117] Where, T hst,h is the hot spot temperature of the transformer at the hth hour; ΔΘ E,h is the ambient temperature at hour h; ΔΘ O,h and ΔΘ Y,h They are respectively the temperature rise of the transformer top oil temperature to the ambient temperature and the temperature rise of the winding hot spot temperature to the top oil temperature.

[0118] The temperature rise of the transformer top oil temperature to the ambient temperature is Δθ O,h and ΔΘY,h The calculation formulas are:

[0119]

[0120]

[0121] In the formula, the symbols init and ulti represent the start and end time of the hth hour respectively; ΔΘ O,R and ΔΘ Y,R They are the temperature rise of the transformer top oil temperature relative to the ambient temperature and the winding virtual maximum operating temperature relative to the top oil temperature under rated load; are the temperature rise of the top oil temperature of the transformer relative to the ambient temperature at the beginning and end of the hth hour respectively; are the temperature rise of the virtual maximum operating temperature of the transformer winding at the beginning and end of the hth hour relative to the top oil temperature of the transformer; are the load rates of the transformer at the start and end of the hth hour, respectively, which can be calculated by step S1.2, υ TO and H are the transformer oil and winding temperature rise time constants respectively; R is the ratio of the transformer rated load loss to the no-load loss; σ1 and σ2 are constant parameters determined by the cooling method of the oil-immersed self-cooling transformer, and their values ​​are 0.8 and 0.5 respectively.

[0122] S1.4, based on the actual operating temperature of the transformer and the virtual maximum operating temperature of the winding, calculate the drop in the transformer operating temperature after the cooling system is modified. The calculation formula is:

[0123] T dec,h =T hst,h -T op,h ;

[0124] Where, T hst,h is the virtual maximum operating temperature of the transformer at the hth hour, T op,h is the actual operating temperature of the transformer; T dec,h is the drop in transformer operating temperature caused by the transformation of the cooling system.

[0125] S1.5, generally speaking, the increase in the number of cooling fans and the increase in the rated speed of the fans will improve the cooling effect on the transformer. Therefore, it can be considered that the number of cooling fans and the rated speed of the fans are positively correlated with the drop in transformer operating temperature. However, according to expert experience, under the premise of determining the number of cooling fans and the speed type, the higher the load rate of the transformer, the higher its operating temperature will be, which will impose certain restrictions on the cooling effect. Therefore, the storage capacity is negatively correlated with the load rate and the temperature drop value. Therefore, based on the Sigmoid function, the linear and nonlinear effects of the three influencing factors are combined to describe the influence of the number of cooling fans, the fan speed type and the load rate on the transformer temperature, and the number of transformer cooling fans N, the fan speed type V, and the load rate L are established. op The operating temperature drop value T of oil-immersed self-cooling transformer dec The mathematical equation between is expressed as:

[0126]

[0127]

[0128] In the formula, W is the sensitivity coefficient of the adjustment temperature drop to be estimated; h is the time-varying coupling factor after considering the interaction of the number of cooling fans, the rated speed of the fans, and the load rate; All are parameters to be estimated. It should be noted that, in order to ensure the fitting effect of the subsequent model, normalization and other methods can be used in advance to standardize the data such as the number of transformer cooling fans, the rated speed of the fans, and the load rate, which will not be repeated here.

[0129] S2, based on the field operation data, uses the residual minimization method with the introduction of the Elastic Net regularization method to determine the unknown parameters in the transformer cooling system transformation efficacy model established in S1. The determination of the unknown parameters includes:

[0130] S2.1, based on the mathematical model established in S1.5, the Huber loss function is used as the model parameter estimation loss function. The Huber loss function combines the advantages of L1 and L2 norms. For small residuals, the Huber loss is the same as the square loss in the traditional minimum residual method; for large residuals, the Huber loss uses absolute values ​​to reduce the influence of outliers. The calculation process is:

[0131]

[0132] Where: Loss(θ) is the sum of all residual losses; δ is the error control threshold; g(ω i ) is the residual loss of the i-th group of data; θ is the estimated value of the unknown parameters in the model; is the predicted value based on the model parameters θ.

[0133] S2.2, in the process of minimizing the residual, the model may overfit the training data. Therefore, the Elastic Net regularization method is introduced to penalize the parameter size while estimating the parameters, so that some parameters tend to zero, thereby achieving feature selection. The sum of the residual loss Loss(θ) in S2.1 is rewritten as:

[0134]

[0135] Where: j is the number of unknown parameters; is the regularization term; λ1 and λ2 are regularization parameters that control the weight of the regularization term. In practical applications, it should be noted that larger regularization parameter values ​​will force more model parameters to become zero, thereby simplifying the model. Smaller regularization parameter values ​​allow the model to fit the training data more, but are prone to overfitting.

[0136] S2.3, the optimization method based on gradient descent is used to iteratively adjust the parameters so that the value of the loss function Loss(θ) gradually decreases until the optimal parameter estimate is found. After initializing the parameter θ, the gradient of the loss function Loss(θ) for each parameter is calculated as:

[0137]

[0138] S2.4, update the parameter θ according to the gradient, and the update process is expressed as:

[0139]

[0140] Where: α is the learning rate, which controls the magnitude of each update.

[0141] S2.5: Repeat S2.3 and S2.4 until the loss function stops decreasing and the optimal parameter estimate of the model is determined.

[0142] S3, based on the historical load rate data of each transformer, the acceptance-rejection sampling method is used to generate the load rate data of the transformer in the future planning period; then the precise mathematical relationship between the transformer operating temperature and its aging factor is established, and a transformer operating status evaluation method considering the impact of cooling system transformation is proposed, which is used to quantitatively evaluate the transformer operating status considering the impact of cooling system transformation during the planning period, and provide a model basis for the subsequent optimization of the transformer group operating status. Specifically include:

[0143] S3.1, as described in S1 and S2, the transformer load factor is the key input data for calculating its operating temperature. In order to evaluate the operating status of the transformer in the future planning period, this step aims to simulate the transformer load factor data in the future planning period based on the transformer historical load factor data. The clustering algorithm is used to extract N l The typical daily average load rate curve of the transformer is shown in Figure 1. Clustering algorithms such as K-medoids, etc. The lth daily average load rate curve is expressed as:

[0144]

[0145] Where: They respectively represent the average load rate values ​​from the 1st to the 24th hour in the 1st daily average load rate curve.

[0146] S3.2, based on the historical growth trend of transformer load rate, combined with possible future load growth factors, such as new loads, policy changes, etc., predict the growth rate of transformer load rate in the next few years

[0147]

[0148] Where: ψW represents the cumulative annual load growth rate of the transformer in the future W years.

[0149] S3.3, from N l Extract a typical curve from the typical daily average load rate curve of the transformer. Assume that the curve extracted is the lth curve, and the average value of 24*W is Standard deviation is SD%*K rate The log-normal distribution of w=1,2,...,W,i=1,2,...,24,where K rate is the rated load rate of the transformer, and the normal distribution is expressed as:

[0150]

[0151] In the formula, represents the lognormal distribution corresponding to the i-th hour in the w-th year; μ i,w and σ i,w are the mean and standard deviation parameters of the lognormal distribution, respectively.

[0152] The mean and standard deviation parameters of the lognormal distribution, using the mean of the load The standard deviation is SD%*Krate. The two parameters are calculated by the following formula:

[0153]

[0154] S3.4, because the lognormal distribution is obtained by exponential transformation of the normal distribution, according to the rejection-acceptance sampling principle, the standard normal distribution G(x) is selected as the recommended distribution, and x is the variable;

[0155] S3.5, calculation of constant c normal , so that the target distribution function is magnified by the proposed distribution function c normal The complete envelope of . That is, for any x, the following conditions must be met:

[0156]

[0157] S3.6, Generate a random sample Y from the proposed distribution G(x) sam , and generate a random number U that follows a uniform distribution between [0,1] sam , calculate the candidate value of the load factor at the i-th hour in the w-th year And calculate the acceptance probability, expressed as:

[0158]

[0159] S3.7, Candidate value judgment, if A accept ≤U sam , then accept as a random sample of the load rate at hour i in year w, otherwise the sample is rejected and sampling is restarted;

[0160] S3.8, repeat S3.3 to S3.7 in chronological order until enough load rate curves of the transformer for the next W years are generated;

[0161] S3.9, after completing S3.8, the virtual maximum operating temperature of the transformer and the operating temperature drop value after considering the specific cooling system modification can be calculated according to the formula in step 1; thereby, the operating temperature time series of the transformer in the next W years after considering the cooling system modification is calculated, and the transformer operating temperature after considering the cooling system modification is calculated by the following formula:

[0162] T op,i,w =T hst,i,w -T dec,i,w ;

[0163] Where: T hst,i,w is the operating temperature of the transformer at the i-th hour in the w-th year without considering the impact of the cooling system modification, calculated by S1.3; T dec,i,w is the drop in operating temperature of the transformer at the i-th hour in the w-th year after the cooling system transformation is performed, which is calculated by the cooling system transformation efficacy model established by S1-S2.

[0164] S3.10 adopts the transformer aging factor calculation formula provided in the IEEE C57.91-2011 standard, takes the transformer accelerated aging factor as the characteristic quantity of the transformer's future operating state, and evaluates the transformer's operating state in the next W years based on the transformer's operating temperature time series in the next W years calculated in S3.9:

[0165]

[0166] In the formula, is the hot spot temperature reference value, the IEEE recommended value is 110℃, B is the empirical constant, the value is 15000, Z is the Zth future load rate curve of the transformer generated by S3.8, T op,w,k is the operating temperature of the transformer in the future w and k years after the cooling system is modified. aging >1 indicates that the transformer is in an accelerated aging stage and exceeds its normal operating capacity. aging <1 indicates that the transformer is underutilized, which may lead to asset waste. Therefore, in engineering, it is usually hoped that the transformer operating state quantity F aging Keep at 1.

[0167] S4, based on the transformer cooling system transformation efficacy model established in S1, the transformer load rate generation method in S2, and the transformer operation status evaluation method, forms an optimization decision method for transformer group cooling system transformation plan considering cost constraints to determine the cooling system transformation plan for each transformer. The steps for determining the cooling system transformation plan for each transformer include:

[0168] S4.1, the goal of the optimization decision model for the transformer group cooling system transformation plan considering cost constraints is to optimize the overall operating status of the transformer group during the planning period. The objective function of the optimization model is:

[0169]

[0170] Where: FA is the overall operating status evaluation value of the transformer group during the planning period, F aging , s It is the operating status evaluation value of the sth transformer in the transformer group during the planning period.

[0171] S4.2, the constraints that the transformer group cooling system transformation optimization decision model needs to meet include the cooling system transformation investment cost constraint during the planning period and the number of cooling fans installed on each transformer. The constraints are expressed as follows:

[0172] The cooling system renovation investment cost constraint means that the cooling system renovation investment cost during the planning period should be less than the predetermined investment cost, which can be expressed as:

[0173] Ccost ≤C set ;

[0174] In the formula, C set is the acceptable upper limit of investment cost during the planning period; Ccost is the total investment cost during the planning period, and the total investment cost is expressed as:

[0175]

[0176] Where V is the total number of cooling fans with different speeds; J v is the investment cost of installing a single type v cooling fan; num s,v is the number of the vth type cooling fans installed on the sth transformer in the transformer group;

[0177] The number of cooling fans is limited. Due to the size of the transformer, that is, the space limitation, the number of cooling fans that can be installed on each transformer during the planning period should be less than the predetermined upper limit of the number of cooling fans, expressed as:

[0178]

[0179] Where: num s,se t is the upper limit of the number of cooling fans that can be installed on the sth transformer in the transformer group. By optimizing the above model, the specific number of cooling fans of each type that need to be installed on each transformer in the transformer group can be determined.

[0180] S4.3, input the electrical parameters, operating parameters, and historical load curve parameters of the transformer group, and use conventional particle swarm and other intelligent optimization algorithms to complete the model solution; in order to ensure engineering practicability and improve solution efficiency, the simulated degradation algorithm is introduced into the iterative process of the conventional particle swarm. After each update of the particle speed and position, the simulated annealing criterion is used to determine whether the particle swarm algorithm accepts the new global optimal solution. The simulated annealing criterion is expressed as:

[0181] If f p (r)-f g (r)≤0, then accept the new solution;

[0182] If f p (r)-f g (r)>0, with acceptance probability P act Accept the new solution with probability P ac The calculation formula is:

[0183]

[0184] Where: f p (r) and f g(r) represent the fitness values ​​of the individual optimal position and the global optimal position at the rth iteration, that is, the overall operating status evaluation value of the transformer group during the planning period; TP r represents the simulated annealing temperature at the rth iteration, and the simulated annealing temperature is calculated by the following formula:

[0185]

[0186] Where: Represents the simulated temperature attenuation coefficient.

[0187] like Figure 2 As shown in the figure, the solution process of the optimization decision model of the transformer group cooling system transformation plan considering cost constraints includes:

[0188] S4.3.1, input basic parameters such as electrical parameters, operating parameters, historical load factor data, and particle swarm algorithm parameters of the transformer group;

[0189] S4.3.2, let the algorithm iteration number r = 1, randomly generate the initial generation particle population that meets the conditions, the initial generation particle population is the number of each type of cooling fans installed on each transformer;

[0190] S4.3.3, for each particle, one particle is a transformer group cooling system transformation plan, and the transformer group operating status index value under the cooling system transformation plan is calculated based on S1-S4;

[0191] S4.3.4, update the individual optimal position and global optimal position obtained in the rth iteration, and update the speed and position of each particle;

[0192] S4.3.5, judging whether to accept the new global optimal solution according to the simulated annealing criterion, and if accepted, updating the annealing temperature synchronously;

[0193] S4.3.6, determine whether r is greater than or equal to R set If yes, update r to r+1 and return to S4.3.3. If no, the iteration ends and the optimal solution is output.

[0194] Taking the historical load rate and other operating data collected from a transformer group as an example, according to the above transformer cooling system transformation efficiency modeling method, the relationship between the number of cooling fans, fan speed type, load rate and transformer operating temperature drop value is established as follows: Figure 3 As shown, the load rate is taken as 0.6 as an example. Figure 3It can be seen that, under the condition of fixed load rate, the greater the number of cooling fans installed or the greater the rated speed of the cooling fans, the more obvious the improvement effect on the transformer operating temperature, which conforms to the actual engineering law. It can also be concluded that the proposed transformer cooling system transformation efficiency modeling method can effectively describe the mapping relationship between "the number of cooling fans, fan speed type, load rate" and "transformer operating temperature".

[0195] In order to verify the effectiveness of the proposed transformer group operation status evaluation and optimization method, it is assumed that the cooling fan speed of all transformers is 900 rpm and the planning period is 1 year. At the same time, according to cost constraints, 6 transformers with an operating age of more than 15 years in a certain area are selected as the analysis object. The transformer group can be installed with up to 12 cooling fans. According to the method of the patented invention, Figure 4 and Figure 5 The number of cooling fans required for different transformers and the corresponding transformer group operating status evaluation values ​​are shown. Figure 4 and Figure 5 The operating status evaluation values ​​of the transformer group after 12 cooling fans are evenly distributed to each transformer (i.e., the comparison scheme) are also listed.

[0196] Depend on Figure 3 It can be seen that the number of cooling fans required for transformer No. 2 is 5, while transformer No. 5 does not need to be installed with cooling fans. This is mainly because according to the on-site operation data, the overall load rate of transformer No. 2 is significantly higher than that of transformer No. 5, resulting in a correspondingly higher operating temperature. Therefore, in order to ensure the overall operating status of the transformer group during the planning period, it is necessary to install more cooling fans for transformer No. 2. Figure 5 It can be seen that compared with the method of evenly distributing cooling fans, the proposed method can further improve the operating status of the transformer group in the future planning period (the optimal operating status is set to 1). Through the above example analysis, it can be concluded that the proposed transformer group operating status evaluation and optimization method considering the cooling system transformation provides a more scientific and reasonable transformer group transformation plan for on-site operation and maintenance personnel.

[0197] Compared with the existing methods, this embodiment firstly establishes a mathematical mapping model between "the number of cooling fans, the type of fan speed, the load rate" and the "transformer operating temperature drop value" based on historical statistical data and simulation data, using a multi-parameter Sigmoid function; then, using the transformer aging factor as the characteristic quantity to describe its operating state, an accurate mathematical relationship between the "transformer operating temperature" and its "aging factor" is established, and a transformer operating state evaluation method based on the aging factor is proposed; based on the above steps, an optimization decision model for the transformer group cooling system transformation plan considering cost constraints is established, and an improved particle swarm algorithm is used to solve it, so that the specific transformation plan of each transformer in the transformer group can be obtained, and the optimization and improvement of the overall operating state of the transformer group is achieved.

[0198] The above is only an embodiment of the present invention, and the common knowledge such as the known specific structure and characteristics in the scheme is not described in detail here. It should be pointed out that for those skilled in the art, several deformations and improvements can be made without departing from the structure of the present invention, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A method for evaluating and optimizing the operation status of a transformer group considering the transformation of the cooling system, characterized in that: The following steps are involved: S1, based on the multi-parameter Sigmoid function, determine the mathematical mapping relationship between the number of cooling fans, fan speed type, load rate and the operating temperature drop value of the oil-immersed self-cooling transformer, and establish a data-driven transformer cooling system transformation efficiency model; S2, based on the field operation data, the residual minimization method with the introduction of the Elastic Net regularization method is used to determine the unknown parameters in the transformer cooling system transformation efficacy model established in S1; S3, based on the historical load rate data of each transformer, the load rate data of the transformer in the future planning period is generated by using the acceptance-rejection sampling method; Then, an accurate mathematical relationship between transformer operating temperature and its aging factor is established, and a transformer operating status assessment method considering the impact of cooling system modification is proposed; S4, based on the transformer cooling system transformation efficacy model established in S1, the transformer load rate generation method in S2, and the transformer operating status evaluation method, forms an optimization decision-making method for the transformer group cooling system transformation plan considering cost constraints to determine the cooling system transformation plan for each transformer.

2. The transformer group operation status evaluation and optimization method considering cooling system modification according to claim 1 is characterized in that: The process of establishing the transformer cooling system transformation efficiency model in S1 includes: S1.1, obtain the field operation data of the oil-immersed self-cooling transformer that has undergone cooling system transformation as data input. The field operation data includes: the number of cooling fans N, the rated speed V of each cooling fan, and the transformer load data P op , Transformer operating temperature T op ; S1.2, based on the transformer load data, the transformer load rate data is calculated and the calculation formula is expressed as: Where P op is the actual load of the transformer at the hth hour, P rate is the rated capacity of the transformer; S1.3, based on the transformer load rate data, calculate the maximum operating temperature T of the transformer without cooling system modification hst , the maximum operating temperature T in the hth hour hst,h The calculation formula is: T hst,h =Θ E,h +DTH O,h +DTH Y,h ; Where, T hst,h is the hot spot temperature of the transformer at the hth hour; ΔΘ E,h is the ambient temperature at hour h; ΔΘ O,h and ΔΘ Y,h They are the temperature rise of the transformer top oil temperature to the ambient temperature and the temperature rise of the winding hot spot temperature to the top oil temperature; S1.4, based on the actual operating temperature of the transformer and the virtual maximum operating temperature of the winding, calculate the drop in the transformer operating temperature after the cooling system is modified. The calculation formula is: T dec,h =T hst,h -T op,h ; Where, T hst,h is the virtual maximum operating temperature of the transformer at the hth hour, T op,h is the actual operating temperature of the transformer; T dec,h is the drop in transformer operating temperature caused by the transformation of the cooling system; S1.5, based on the Sigmoid function, establish the number of transformer cooling fans N, fan speed type V, and load rate L op The operating temperature drop value T of oil-immersed self-cooling transformer dec The mathematical equation between is expressed as: In the formula, W is the sensitivity coefficient of the adjustment temperature drop to be estimated; h is the time-varying coupling factor after considering the interaction of the number of cooling fans, the rated speed of the fans, and the load rate; are parameters to be estimated.

3. The transformer group operation status evaluation and optimization method considering cooling system modification according to claim 2 is characterized in that: The temperature rise Δθ of the transformer top oil temperature to the ambient temperature in S1.3 O,h and ΔΘ Y,h The calculation formulas are: In the formula, the symbols init and ulti represent the start and end time of the hth hour respectively; ΔΘ O,R and ΔΘ Y,R They are the temperature rise of the transformer top oil temperature relative to the ambient temperature and the winding virtual maximum operating temperature relative to the top oil temperature under rated load; and are the temperature rise of the top oil temperature of the transformer relative to the ambient temperature at the beginning and end of the hth hour respectively; and are the temperature rise of the virtual maximum operating temperature of the transformer winding at the beginning and end of the hth hour relative to the top oil temperature of the transformer; and are the load rates of the transformer at the beginning and end of the hth hour, υ TO and H are the transformer oil and winding temperature rise time constants respectively; R is the ratio of the transformer rated load loss to the no-load loss; σ1 and σ2 are constant parameters determined by the cooling method of the oil-immersed self-cooling transformer, and their values ​​are 0.8 and 0.5 respectively.

4. The transformer group operation status evaluation and optimization method considering cooling system modification according to claim 2 is characterized in that: The estimation of unknown parameters in S2 includes: S2.1, based on the mathematical model established in S1.5, the Huber loss function is used as the model parameter to estimate the loss function, which is calculated as: Where: Loss(θ) is the sum of all residual losses; δ is the error control threshold; g(ω i ) is the residual loss of the i-th group of data; θ is the estimated value of the unknown parameters in the model; is the predicted value based on the model parameters θ; S2.2, introduce the Elastic Net regularization method and rewrite the sum of residual losses Loss(θ) in S2.1 as: Where: j is the number of unknown parameters; is the regularization term; λ1 and λ2 are regularization parameters that control the weight of the regularization term; S2.3, using the optimization method based on gradient descent to iteratively adjust the parameters, after initializing the parameters θ, the gradient of the loss function Loss(θ) for each parameter is calculated as: S2.4, update the parameter θ according to the gradient, and the update process is expressed as: Where: α is the learning rate, which controls the magnitude of each update; S2.5: Repeat S2.3 and S2.4 until the loss function stops decreasing and the optimal parameter estimate of the model is determined.

5. The transformer group operation status evaluation and optimization method considering cooling system modification according to claim 4 is characterized in that: The S3 includes: S3.1, using clustering algorithm to extract N l The typical daily average load rate curve of the transformer is shown as follows: Where: They represent the average load factor values ​​from the 1st to the 24th hour in the daily average load factor curve of Article 1 respectively; S3.2, based on the historical growth trend of transformer load rate and combined with possible future load growth factors, predict the growth rate of transformer load rate in the next few years ψ: ψ=[ψ1,ψ2,L,ψ W-1 ,ψ W ]; Where: W It represents the cumulative annual load growth rate of the transformer in the future W years; S3.3, from N l Extract a typical curve from the typical daily average load rate curve of the transformer. Assume that the curve extracted is the lth curve, and the average value of 24*W is Standard deviation is SD%*K rate The log-normal distribution of w=1,2,...,W,i=1,2,...,24,where K rate is the rated load rate of the transformer, and the normal distribution is expressed as: In the formula, represents the lognormal distribution corresponding to the i-th hour in the w-th year; μ i,w and σ i,w represent the mean and standard deviation parameters of the lognormal distribution respectively; S3.4, according to the rejection-acceptance sampling principle, select the standard normal distribution G(x) as the proposed distribution; S3.5, calculation of constant c normal , so that the target distribution function is magnified by the proposed distribution function c normal The complete envelope of . That is, for any x, the following conditions must be met: S3.6, Generate a random sample Y from the proposed distribution G(x) sam , and generate a random number U that follows a uniform distribution between [0,1] sam , calculate the candidate value of the load factor at the i-th hour in the w-th year And calculate the acceptance probability, expressed as: S3.7, Candidate value judgment, if A accept ≤U sam , then accept as a random sample of the load rate at hour i in year w, otherwise the sample is rejected and sampling is restarted; S3.8, repeat S3.3 to S3.7 in chronological order until enough load rate curves of the transformer for the next W years are generated; S3.9, after completing S3.8, the virtual maximum operating temperature of the transformer and the operating temperature drop value after considering the specific cooling system modification can be calculated according to the formula in step 1; thereby, the operating temperature time series of the transformer in the next W years after considering the cooling system modification is calculated, and the transformer operating temperature after considering the cooling system modification is calculated by the following formula: T op,i,w =T hst,i,w -T dec,i,w ; Where: T hst,i,w is the operating temperature of the transformer at the i-th hour in the w-th year without considering the impact of the cooling system modification; T dec,i,w is the drop in operating temperature of the transformer at the i-th hour in the w-th year after the cooling system modification; S3.10, using the transformer accelerated aging factor as the characteristic quantity of the transformer's future operating state, based on the transformer's operating temperature time series calculated in S3.9 for the next W years, evaluate the transformer's operating state in the next W years: Where, T H is the hot spot temperature reference value, B is an empirical constant with a value of 15000, Z is the Zth future load rate curve of the transformer generated by S3.8, T op,w,k The operating temperature of the transformer after the cooling system modification is considered in the future w and k years.

6. The transformer group operation status evaluation and optimization method considering cooling system modification according to claim 5 is characterized in that: In S3.3, the mean and standard deviation parameters of the lognormal distribution are calculated using the load mean The standard deviation is SD%*Krate. The two parameters are calculated by the following formula:

7. The transformer group operation status evaluation and optimization method considering cooling system modification according to claim 5 is characterized in that: The step of determining the cooling system transformation plan for each transformer in S4 includes: S4.1, the goal of the optimization decision model for the transformer group cooling system transformation plan considering cost constraints is to optimize the overall operating status of the transformer group during the planning period. The objective function of the optimization model is: Where: FA is the overall operating status evaluation value of the transformer group during the planning period, F aging , s is the operating status evaluation value of the sth transformer in the transformer group during the planning period; S4.2, the constraints that the transformer group cooling system transformation plan optimization decision model needs to meet include the cooling system transformation investment cost constraint during the planning period and the number of cooling fans installed on each transformer; S4.3, input the electrical parameters, operating parameters, and historical load curve parameters of the transformer group, introduce the simulated degradation algorithm into the iterative process of the conventional particle swarm, and after each update of the particle speed and position, judge whether the particle swarm algorithm accepts the new global optimal solution according to the simulated annealing criterion.

8. The transformer group operation status evaluation and optimization method considering cooling system modification according to claim 7 is characterized in that: The constraints in S4.2 are expressed as follows: The cooling system renovation investment cost constraint means that the cooling system renovation investment cost during the planning period should be less than the predetermined investment cost, which can be expressed as: C cost ≤C set ; In the formula, C set is the acceptable upper limit of investment cost during the planning period; Ccost is the total investment cost during the planning period, and the total investment cost is expressed as: Where V is the total number of cooling fans with different speeds; J v is the investment cost of installing a single type v cooling fan; num s,v is the number of the vth type cooling fans installed on the sth transformer in the transformer group; The number of cooling fans is limited. During the planning period, the number of cooling fans that can be installed on each transformer should be less than the upper limit of the predetermined number of cooling fans, which can be expressed as: Where: num s,se t It is the upper limit of the cooling fans that can be installed on the sth transformer in the transformer group.

9. The transformer group operation status evaluation and optimization method considering cooling system modification according to claim 7 is characterized in that: The simulated annealing criterion in S4.3 is expressed as: If f p (r)-f g (r)≤0, then accept the new solution; If f p (r)-f g (r)>0, with acceptance probability P act Accept the new solution with probability P ac The calculation formula is: Where: f p (r) and f g (r) represent the fitness values ​​of the individual optimal position and the global optimal position at the rth iteration, that is, the overall operating status evaluation value of the transformer group during the planning period; TP r represents the simulated annealing temperature at the rth iteration, and the simulated annealing temperature is calculated by the following formula: Where: ζ represents the simulated temperature attenuation coefficient.

Images