Method and system for determining fault maintenance time of power transmission and transformation equipment

Abstract

The invention discloses a method and system for determining the fault maintenance time of power transmission and transformation equipment, and the method comprises the steps: fitting the Weibull distribution of the equipment fault rate according to the historical statistical data of the equipment fault rate, forming a basic fault rate model, and then correcting the basic fault rate model obtained through fitting from a long-term evolution factor and a short-term sudden change factor, thereby obtaining the fault maintenance time of the power transmission and transformation equipment. According to the method, an actual failure rate model is formed, familial defects, individual defects and maintenance behaviors are considered in long-term evolution factors, finally, an equipment failure cumulative probability predicted value is calculated based on the actual failure rate model of the equipment, and an optimal maintenance period is determined by solving a constraint equation that the equipment failure cumulative probability predicted value is equal to a failure cumulative probability threshold value. According to the method, the maintenance period can be determined more reasonably, and over-maintenance or under-maintenance is avoided, so that more accurate and effective decision support is provided for making a medium-and-long-term maintenance plan of equipment.

Description

Technical Field

[0001] The invention belongs to the technical field of power system maintenance, and in particular relates to a method and a system for determining the fault maintenance time of power transmission and transformation equipment. Background Art

[0002] The purpose of condition maintenance of power transmission and transformation equipment is to formulate optimized maintenance strategies in a targeted manner by comprehensively considering the current operating status of the equipment, the development trend of potential failures, and the possible loss consequences of equipment failures, so as to effectively reduce the equipment failure rate, reduce maintenance costs, and improve the safe operation level of the power grid. The condition maintenance of power transmission and transformation equipment is divided into short-term maintenance decisions and medium- and long-term maintenance decisions. Medium- and long-term maintenance mainly adopts regular maintenance methods, that is, based on the statistical analysis results of the equipment failure rate, the maintenance cycle is determined with the failure rate not higher than the empirical value. The main disadvantage of the current method for determining the maintenance cycle is that it does not consider the influence of long-term sudden change factors such as familial defects, individual defects, maintenance behavior, and short-term sudden change factors on the equipment failure rate curve, which leads to over-maintenance or under-maintenance. Summary of the invention

[0003] The purpose of the present invention is to provide a method and system for determining the maintenance time of power transmission and transformation equipment faults, which can reasonably determine the maintenance cycle and avoid over-maintenance or under-maintenance, in view of the above problems existing in the prior art.

[0004] To achieve the above objectives, the technical solution of the present invention is as follows:

[0005] In a first aspect, the present invention provides a method for determining a fault repair time of a power transmission and transformation equipment, the method comprising:

[0006] S1. Use the fault history data of power transmission and transformation equipment to perform segmented fitting on the Weibull distribution function of the equipment during the accidental failure period and the loss failure period, calculate the boundary years between the accidental failure period and the loss failure period, and construct a basic failure rate model for power transmission and transformation equipment;

[0007] S2. Modify the fitted basic failure rate model according to the following formula to obtain the actual failure rate model:

[0008] λ(t)=η×λ0(t-ΔT1-ΔT2);

[0009]

[0010]

[0011] ΔT3=t k×s -t k×(s-1) ;

[0012] In the above formula, λ(t) represents the actual failure rate model finally obtained; λ0(t) represents the equipment failure rate obtained by the basic failure rate model at the time t-ΔT1-ΔT2; η represents the correction coefficient of family defects and individual defects to the Weibull distribution function; ΔT1 is the first time interval, which is used to represent the regression effect of maintenance behavior on the equipment operation time; ΔT2 is the second time interval, which is used to represent the regression effect of short-term sudden change factors on the equipment operation time; ΔT3 represents the time interval between the two most recent health status evaluations of the equipment; t k×s represents the actual operating time of the equipment after the sth health status evaluation after the kth maintenance; t k×(s-1) represents the actual operating time of the equipment after the s-1th health status evaluation after the kth maintenance; JK represents the current health status evaluation value of the equipment; JK0 represents the health status evaluation critical value between normal operation of the equipment and equipment operation failure; 100 represents the equipment operating in the best state; MTBF1 represents the mean time between failures of similar equipment produced by other manufacturers; t 1_i represents the actual operating life of the i-th device among similar devices produced by other manufacturers; n1 represents the total number of failures of similar devices produced by other manufacturers during operation; MTBF2 represents the mean time between failures of the individual device; t2 represents the actual operating life of the individual device; n2 represents the number of failures of the individual device during operation; t k represents the operating time of the equipment individual before the kth maintenance; It represents the equivalent operation time of the equipment after the kth maintenance;

[0013] S3. Calculate the cumulative probability of equipment failure from the completion of the last maintenance to the next moment based on the actual failure rate model, and determine the moment when the cumulative probability of equipment failure reaches the cumulative probability threshold of equipment failure. This moment is the best time for the next maintenance.

[0014] The expression of the fitted basic failure rate model is:

[0015]

[0016] In the above formula, λ0(t) represents the equipment failure rate at time t; α1 and β1 are the shape parameter and scale parameter of the accidental failure period, respectively; α2 and β2 are the shape parameter and scale parameter of the loss failure period, respectively; l represents the dividing line between the accidental failure period and the loss failure period.

[0017] In S1, the BP neural network improved by the LM algorithm is used to fit the parameters α and β of the Weibull distribution function; the specific fitting steps are:

[0018] A1, initialization parameters α, β;

[0019] A2. Calculate the partial derivatives of parameters α and β, and use the LM algorithm to obtain the parameter estimate α after the first iteration. 1 , β 1 and the residual sum of squares;

[0020] A3, based on the parameter estimate α after the first iteration 1 , β 1 Perform the second iteration calculation and obtain the parameter estimation value α after the second iteration through the LM algorithm 2 , β 2 and the residual sum of squares for the second iteration;

[0021] A4. Compare the residual sum of squares of the first iteration and the second iteration. If the residual sum of squares of the second iteration is less than that of the first iteration, the second iteration ends and the process goes to A5. Otherwise, increase the damping coefficient in the LM algorithm in the second iteration and return to A3.

[0022] A5. Return to step A2 to continue iterative calculation until the absolute value of the difference between the parameter estimation value obtained after a certain iteration and the parameter estimation value obtained before a certain iteration is less than the preset system allowable error. The entire iterative process ends and the parameter estimation value obtained in the last iteration is output.

[0023] In S1, the boundary years between the accidental failure period and the loss failure period are obtained by solving the objective function; the calculation formula of the objective function is:

[0024]

[0025] In the above formula, S l represents the objective function; I represents the dividing year between the accidental failure period and the loss failure period; n represents the last year of the loss failure period; λ0(t) 偶 represents the part of the fitted Weibull distribution function corresponding to the wear failure period; λ0(t) 损 represents the part of the fitted Weibull distribution function corresponding to the wear failure period, λ 实_t Indicates the actual failure rate of the equipment at time t.

[0026] The calculation formula for the cumulative probability value of equipment failure is:

[0027]

[0028] In the above formula, Px represents the cumulative probability of equipment failure from the completion of the kth maintenance to the next maintenance; t′ represents the equivalent service life of the equipment at the current time, t x Indicates the next maintenance time of the equipment;

[0029] In step S3, the calculation formula for the cumulative probability threshold of equipment failure is:

[0030]

[0031] In the above formula, P max Indicates the cumulative probability threshold of equipment failure; T O Indicates the design life of the equipment; T max Indicates the maximum interval between scheduled inspections of equipment.

[0032] In a second aspect, the present invention provides a system for determining the fault maintenance time of power transmission and transformation equipment, the system comprising a basic failure rate model fitting module, an actual failure rate model building module, and an optimal maintenance time calculation module:

[0033] The basic failure rate model fitting module is used to use the fault history data of the power transmission and transformation equipment to perform piecewise fitting on the Weibull distribution function of the equipment during the accidental failure period and the loss failure period, calculate the boundary years between the accidental failure period and the loss failure period, and construct the basic failure rate model of the power transmission and transformation equipment;

[0034] The actual failure rate model building module is used to correct the fitted basic failure rate model according to the following formula to obtain the actual failure rate model:

[0035] λ(t)=η×λ0(t-ΔT1-ΔT2);

[0036]

[0037] ΔT3=t k×s -t k×(s-1) ;

[0038] In the above formula, λ(t) represents the actual failure rate model finally obtained; λ0(t) represents the equipment failure rate obtained by the basic failure rate model at the time t-ΔT1-ΔT2; η represents the correction coefficient of family defects and individual defects to the Weibull distribution function; ΔT1 is the first time interval, which is used to represent the regression effect of maintenance behavior on the equipment operation time; ΔT2 is the second time interval, which is used to represent the regression effect of short-term sudden change factors on the equipment operation time; ΔT3 represents the time interval between the two most recent health status evaluations of the equipment; t k×s represents the actual operating time of the equipment after the sth health status evaluation after the kth maintenance; t k×(s-1) represents the actual operating time of the equipment after the s-1th health status evaluation after the kth maintenance; JK represents the current health status evaluation value of the equipment; JK0 represents the health status evaluation critical value between normal operation of the equipment and equipment operation failure; 100 represents the equipment operating in the best state; MTBF1 represents the mean time between failures of similar equipment produced by other manufacturers; t1_ix represents the actual operating life of the i-th device among similar devices produced by other manufacturers; n1 represents the total number of failures of similar devices produced by other manufacturers during operation; MTBF2 represents the mean time between failures of the individual device; t2 represents the actual operating life of the individual device; n2 represents the number of failures of the individual device during operation; t k represents the operating time of the equipment individual before the kth maintenance; It represents the equivalent operation time of the equipment after the kth maintenance;

[0039] The optimal maintenance time calculation module is used to calculate the cumulative probability value of equipment failure from the completion of the last maintenance to the next time based on the actual failure rate model, and determine the time when the cumulative probability value of equipment failure reaches the cumulative probability threshold value of equipment failure. This time is the optimal time for the next maintenance.

[0040] The expression of the fitted basic failure rate model is:

[0041]

[0042] In the above formula, λ0(t) represents the equipment failure rate at time t; α1 and β1 are the shape parameter and scale parameter of the accidental failure period, respectively; α2 and β2 are the shape parameter and scale parameter of the loss failure period, respectively; l represents the dividing line between the accidental failure period and the loss failure period.

[0043] The basic failure rate model fitting module uses the Macquart method or the BP neural network improved by the LM algorithm to fit the parameters α and β of the Weibull distribution function; the specific fitting steps are:

[0044] A1, initialization parameters α, β;

[0045] A2. Calculate the partial derivatives of parameters α and β, and use the LM algorithm to obtain the parameter estimate α after the first iteration. 1 , β 1 and the residual sum of squares;

[0046] A3, based on the parameter estimate α after the first iteration 1 , β 2 Perform the second iteration calculation and obtain the parameter estimation value α after the second iteration through the LM algorithm 2 , β 2 and the residual sum of squares for the second iteration;

[0047] A4. Compare the residual sum of squares of the first iteration and the second iteration. If the residual sum of squares of the second iteration is less than that of the first iteration, the second iteration ends and the process goes to A5. Otherwise, increase the damping coefficient in the LM algorithm in the second iteration and return to A3.

[0048] A5. Return to step A2 to continue iterative calculation until the absolute value of the difference between the parameter estimation value obtained after a certain iteration and the parameter estimation value obtained before a certain iteration is less than the preset system allowable error. The entire iterative process ends and the parameter estimation value obtained in the last iteration is output.

[0049] The basic failure rate model fitting module obtains the boundary years between the accidental failure period and the loss failure period by solving the objective function; the calculation formula of the objective function is:

[0050]

[0051] In the above formula, S l represents the objective function; l represents the dividing year between the accidental failure period and the loss failure period; n represents the last year of the loss failure period; λ0(t) 偶 represents the part of the fitted Weibull distribution function corresponding to the wear failure period; λ0(t) 损 represents the part of the fitted Weibull distribution function corresponding to the wear failure period, λ 实_t Indicates the actual failure rate of the equipment at time t.

[0052] The optimal maintenance time calculation module calculates the cumulative probability value of equipment failure according to the following formula:

[0053]

[0054] In the above formula, P1 represents the cumulative probability of equipment failure from the completion of the kth maintenance to the next maintenance; t′ represents the equivalent service life of the equipment at the current time, t x Indicates the next maintenance time of the equipment;

[0055] The cumulative probability threshold of equipment failure is calculated according to the following formula:

[0056]

[0057] In the above formula, P max represents the cumulative probability threshold of equipment failure; T0 represents the design life of the equipment; T max Indicates the maximum interval between scheduled inspections of equipment.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] The method for determining the fault maintenance time of power transmission and transformation equipment described in the present invention first forms a basic failure rate model by fitting the Weibull distribution of the equipment failure rate based on the historical statistical data of the equipment failure rate, and then corrects the fitted basic failure rate model from two aspects of long-term evolution factors and short-term sudden change factors to form an actual failure rate model, and the long-term evolution factors take into account family defects, individual defects, and maintenance behavior. Finally, the cumulative probability prediction value of equipment failure is calculated based on the actual failure rate model of the equipment, and the optimal maintenance cycle is determined by solving the constraint equation that the cumulative probability prediction value of the equipment failure is equal to the cumulative probability threshold value of the failure, thereby providing more accurate and effective decision support for the formulation of medium- and long-term maintenance plans for the equipment, and avoiding over-maintenance or under-maintenance. Therefore, the present invention can more reasonably determine the maintenance cycle and avoid over-maintenance or under-maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 The figure is a flow chart of the maintenance method of the present invention.

[0061] Figure 2 The structural diagram of the maintenance system of the present invention is DETAILED DESCRIPTION

[0062] The present invention is further described in detail below in conjunction with specific implementations and drawings.

[0063] Embodiment 1:

[0064] See also Figure 1 A method for determining the time for troubleshooting of power transmission and transformation equipment is provided, which is performed in the following steps:

[0065] S1. Assume that the power transmission and transformation equipment to be analyzed works in the accidental failure period and the loss failure period. Since the change law of the equipment accidental failure period and the loss failure period conforms to the bathtub curve, the Weibull distribution function is used to characterize the characteristics of the bathtub curve in different periods. The Weibull distribution function is widely used in reliability analysis, especially for the distribution form of aging cumulative failure of electromechanical products; the expression of the basic failure rate model designed by the Weibull distribution function is:

[0066]

[0067] In the above formula, λ0(t) represents the equipment failure rate at time t; α1 and β1 are the shape parameter and scale parameter of the accidental failure period respectively; α2 and β2 are the shape parameter and scale parameter of the loss failure period respectively; l represents the dividing year between the accidental failure period and the loss failure period;

[0068] (1) The fault history statistical data of the power transmission and transformation equipment are used to perform segmented fitting of the Weibull distribution function of the equipment during the accidental failure period and the loss failure period. Specifically, the BP neural network improved by the LM algorithm is used to fit the parameters α and β of the Weibull distribution function. The specific fitting process is as follows:

[0069] A1, initialization parameters α, β;

[0070] A2. Calculate the partial derivatives of parameters α and β, and use the LM algorithm to obtain the parameter estimate α after the first iteration. 1 , β 1 and the residual sum of squares;

[0071] A3, based on the parameter estimate α after the first iteration 1 , β 1 Perform the second iteration calculation and obtain the parameter estimation value α after the second iteration through the LM algorithm 2 , β 2 and the residual sum of squares for the second iteration;

[0072] A4. Compare the residual sum of squares of the first iteration and the second iteration. If the residual sum of squares of the second iteration is less than that of the first iteration, the second iteration ends and the process goes to A5. Otherwise, increase the damping coefficient μ in the LM algorithm during the second iteration and return to A3.

[0073] A5, return to step A2 to continue iterative calculation until the absolute value of the difference between the parameter estimation value obtained after a certain iteration and the parameter estimation value obtained before a certain iteration is less than the preset error tolerance value ε, the entire iterative process ends, and the parameter estimation value obtained in the last iteration is output;

[0074] BP neural network includes input layer, which is used to input data; hidden layer, which generally contains one or more layers of hidden units; output layer, which is used to output the shape parameter α and scale parameter β of Weibull distribution; mean square error (MSE) is usually selected as loss function; input data is forward propagated and error is back propagated to obtain output value (i.e. estimated shape parameter and scale parameter). Forward propagation means that sample data is input from input layer, processed by each hidden layer layer by layer and transmitted to output. If the output result of output layer does not reach the expected value, it turns to error back propagation; error back propagation is to propagate the output error through hidden layer layer by layer, find the minimum value of error function to continuously correct the weight and bias of network, the process of continuous adjustment of weight and bias is the learning and training process of network, until the error reaches the expected range.

[0075] Assuming that the output layer target vector of the BP network is Y and the activation function is f(), C can be used to represent the corresponding deviation function W and input weight P, that is, C = f(p, w), and r is introduced.t (W) = y t -f(p t , W) form a vector r(W), so E(W) can be expressed as:

[0076]

[0077] In the above formula, J is the Jacobian matrix of r;

[0078] Based on the Jacobian matrix, when the second-order derivative of r(W) is not considered, the Hessian matrix is ​​constructed:

[0079]

[0080] The Hessian matrix form can be approximately transformed into H = J T J, and use this to complete the transformation of the Gauss-Newton formula. After the above calculation transformation, the second-order derivative of the objective function E(W) can be transformed into the solution of the first-order derivative, and the calculation workload is greatly reduced; the expression of the Gauss-Newton formula is as follows:

[0081]

[0082] The Gauss-Newton formula is improved to obtain the LM method iteration formula:

[0083]

[0084] In the above formula, i is the unit matrix; μ is the damping coefficient;

[0085] The learning and training methods of BP neural network can be selected as Gauss-Newton method (μ→0) and standard gradient method (μ→∞), and the weight variables and deviations are continuously adjusted to finally meet the training requirements. The training formula is as follows:

[0086]

[0087] The traditional BP neural network algorithm uses the gradient descent method to update the weights and biases of the network, while the LM algorithm adopted in the present invention adds a more complex second-order optimization process on this basis, which makes the convergence faster. The LM algorithm, also known as the Levenberg-Marquardt algorithm, is an optimization algorithm for solving nonlinear least squares problems, combining the advantages of the gradient descent method and the Newton method. In the training process of the LM algorithm, the training error allowable value ε, the damping coefficient μ, and the constant β are introduced; the threshold and connection weights are initialized, and the damping coefficient μ=μ0 and the number of iterations k=0 are set; the objective function E(W), the network output and J are calculated, and the calculation of ΔW is completed on this basis; if E(W k)<ε, the network stops computing; otherwise, it continues computing E(W k+1 ) According to the threshold vector and weight W k+1 =W k +ΔW is calculated, if E(W k+1 )<E(W k ), and meets k=k+1, μ=μ / β, then the network output and objective function can be returned for recalculation; conversely, if μ=μβ, ΔW can be directly calculated;

[0088] (2) By solving the objective function, the boundary years between the accidental failure period and the loss failure period are obtained to form a basic failure rate model of the power transmission and transformation equipment; the calculation formula of the objective function is:

[0089]

[0090] In the above formula, S l represents the objective function; l represents the dividing year between the accidental failure period and the loss failure period; n represents the last year of the loss failure period; λ0(t) 偶 represents the part of the fitted Weibull distribution function corresponding to the wear failure period; λ0(t) 损 represents the part of the fitted Weibull distribution function corresponding to the wear failure period; 实_t represents the actual failure rate of the equipment at time t;

[0091] S2. Since factors such as familial defects, individual defects, and maintenance behaviors will affect the basic failure rate curve of equipment over a longer time scale, in order to reflect the impact of the above long-term evolution factors, the basic failure rate model needs to be modified accordingly: For familial defects, the actual mean failure-free time of equipment produced by the same manufacturer is compared with the overall mean failure-free time of equipment of the same type, and the ratio of the two is used to perform a linear correction on the basic failure rate of similar products; for individual defects, similar to familial defects, in order to reflect the individual differences between different equipment produced by the same manufacturer, the failure rate curve of equipment produced by the same manufacturer needs to be linearly corrected according to the actual operation of the equipment; maintenance behaviors can improve Improve the performance of the equipment, reduce the equipment failure rate, and extend the service life of the equipment, which is reflected in the equipment failure rate curve. Maintenance behavior has a regression effect on the operating life of the equipment; maintenance behavior cannot restore the equipment to its original state, because the equipment itself is affected by material aging, and maintenance only repairs the deteriorated state of the equipment to a certain extent. Therefore, the equivalent service age of the equipment after the current maintenance cannot be less than the equivalent service age after the previous maintenance; for short-term sudden change factors, the actual equivalent operating time of the equipment at the current moment is calculated in combination with the equipment health status evaluation score, so as to obtain the equipment failure rate after considering the short-term sudden change factors, so as to realize the correction of the basic failure rate model of the equipment by the short-term sudden change factors;

[0092] Specifically, the fitted basic failure rate model is corrected according to the following formula to obtain the actual failure rate model:

[0093] λ(t)=η×λ0(t-ΔT1-ΔT2);

[0094]

[0095] ΔT3=t k×s -t k×(s-1) ;

[0096] In the above formula, λ(t) represents the actual failure rate model finally obtained; λ0(t) represents the equipment failure rate obtained by the basic failure rate model at the time t-ΔT1-ΔT2; η represents the correction coefficient of family defects and individual defects to the Weibull distribution function; ΔT1 is the first time interval, which is used to represent the regression effect of maintenance behavior on the equipment operation time; ΔT2 is the second time interval, which is used to represent the regression effect of short-term sudden change factors on the equipment operation time; ΔT3 represents the time interval between the two most recent health status evaluations of the equipment, which is generally a fixed value; t k×s represents the actual operating time of the equipment after the Sth health status evaluation after the kth maintenance; t k×(s-1) represents the actual operating time of the equipment after the s-1th health status evaluation after the kth maintenance; JK represents the current health status evaluation value of the equipment; JK0 represents the critical value of the health status evaluation between normal operation of the equipment and equipment operation failure, which is generally 80 points, and 100 points represents that the equipment is operating in the best state; MTBF1 represents the average failure-free time of similar equipment produced by other manufacturers; t 1_i represents the actual operating life of the i-th device among similar devices produced by other manufacturers; n1 represents the total number of failures of similar devices produced by other manufacturers during operation; MTBF2 represents the mean time between failures of the individual device; t2 represents the actual operating life of the individual device; n2 represents the number of failures of the individual device during operation; t k represents the operating time of the equipment individual before the kth maintenance; It represents the equivalent operation time of the equipment after the kth maintenance;

[0097] S3. Calculate the cumulative probability value P1 of equipment failure from the last maintenance completion to the next moment based on the actual failure rate model, and calculate the cumulative probability threshold value Px of equipment failure ax Finally, determine the time when the cumulative probability of equipment failure reaches the cumulative probability threshold of equipment failure, which is the best time for the next maintenance t best ; The calculation formula for the cumulative probability value of equipment failure is:

[0098]

[0099] In the above formula, P1 represents the cumulative probability of equipment failure from the completion of the kth maintenance to the next maintenance; t′ represents the equivalent service life of the equipment at the current time, t x Indicates the next maintenance time of the equipment;

[0100] Considering that the time of the equipment in the early failure period and the wear and tear failure period is relatively short compared to the entire operating life of the equipment, the midpoint of the equipment's accidental failure period is replaced by the midpoint of the equipment's design life, and the calculation formula for the cumulative probability threshold of equipment failure is obtained as follows:

[0101]

[0102] In the above formula, P max Indicates the cumulative probability threshold of equipment failure; T O Indicates the design life of the equipment; T max Indicates the maximum interval between scheduled inspections of equipment.

[0103] In order to verify the effectiveness of the maintenance method proposed in the present invention, a 220kV transformer of a certain type in a certain power grid is taken as an example, and a simulation analysis is performed on it to obtain the historical statistical data of the failure rate of the transformer in different service time periods as shown in Table 1. The MTBF1 is 6.32 years and the MTBF2 is 7.6 years. The transformer has been in operation for 8 years, with a design life of 20 years and a maximum maintenance interval of 8 years. The basic failure rate fitting curve is obtained as shown in Table 1. Figure 2 As shown in the figure, the optimal demarcation period l is 15 years, and the optimal time for the next maintenance is finally determined as t best is 10.5 years; the basic failure rate model is as follows:

[0104]

[0105] Table 1 Historical statistics of failure rate

[0106]

[0107] Compared with the traditional equipment condition maintenance scheme, the present invention can effectively avoid the reduction of equipment effective utilization rate caused by "over-maintenance", as well as the waste of human and material resources caused by it, and reduce the impact on the reliability of power supply of the power grid. When the equipment condition evaluation score JK is close to the critical score value, the timing of its next maintenance should be slightly extended on the basis of the maximum maintenance interval.

[0108] Embodiment 2:

[0109] See also Figure 2A system for determining the fault maintenance time of power transmission and transformation equipment includes a basic failure rate model fitting module, an actual failure rate model construction module, and an optimal maintenance time calculation module; the basic failure rate model fitting module is used to use the failure history data of the power transmission and transformation equipment to perform segmented fitting on the Weibull distribution function of the equipment in the accidental failure period and the loss failure period, calculate the boundary years between the accidental failure period and the loss failure period, and construct the basic failure rate model of the power transmission and transformation equipment; specifically, the BP neural network improved by the LM algorithm is used to fit the parameters α and β of the Weibull distribution; the specific fitting steps are:

[0110] A1, initialization parameters α, β;

[0111] A2. Calculate the partial derivatives of parameters α and β, and use the LM algorithm to obtain the parameter estimate α after the first iteration. 1 , β 1 and the residual sum of squares;

[0112] A3, based on the parameter estimate α after the first iteration 1 , β 1 Perform the second iteration calculation and obtain the parameter estimation value α after the second iteration through the LM algorithm 2 , β 1 and the residual sum of squares for the second iteration;

[0113] A4. Compare the residual sum of squares of the first iteration and the second iteration. If the residual sum of squares of the second iteration is less than that of the first iteration, the second iteration ends and the process goes to A5. Otherwise, increase the damping coefficient in the LM algorithm in the second iteration and return to A3.

[0114] A5, return to step A2 to continue iterative calculation until the absolute value of the difference between the parameter estimation value obtained after a certain iteration and the parameter estimation value obtained before a certain iteration is less than the preset system allowable error, the entire iterative process ends, and the parameter estimation value obtained in the last iteration is output;

[0115] Specifically, the boundary years between the accidental failure period and the loss failure period are obtained by solving the objective function. The calculation formula of the objective function is:

[0116]

[0117] In the above formula, S l represents the objective function; l represents the dividing year between the accidental failure period and the loss failure period; n represents the last year of the loss failure period; λ0(t) 偶 represents the part of the fitted Weibull distribution function corresponding to the wear failure period; λ0(t) 损 represents the part of the fitted Weibull distribution function corresponding to the wear failure period;

[0118] The expression of the basic failure rate model is:

[0119]

[0120] In the above formula, λ0(t) represents the equipment failure rate at time t; α1 and β1 are the shape parameter and scale parameter of the accidental failure period respectively; α2 and β2 are the shape parameter and scale parameter of the loss failure period respectively; l represents the dividing year between the accidental failure period and the loss failure period;

[0121] The actual failure rate model building module is used to correct the fitted basic failure rate model according to the following formula to obtain the actual failure rate model:

[0122] λ(t)=η×λ0(t-ΔT1-ΔT2);

[0123]

[0124] ΔT3=t k×s -t k×(s-1) ;

[0125] In the above formula, λ(t) represents the actual failure rate model finally obtained; λ0(t) represents the fitted Weibull distribution function; η represents the correction coefficient of family defects and individual defects on the Weibull distribution function; ΔT1 represents the regression effect of maintenance behavior on the equipment operation time; ΔT2 represents the regression effect of short-term sudden change factors on the equipment operation time; ΔT3 represents the time interval between the two most recent health status evaluations of the equipment; t k×s represents the equivalent operating time of the equipment after the sth health status evaluation after the kth maintenance; tk ×(s-1) represents the equivalent operating time of the equipment after the S-1th health status evaluation after the kth maintenance; JK represents the current health status evaluation value of the equipment; JK0 represents the health status evaluation critical value between normal operation of the equipment and equipment operation failure; 100 represents the equipment operating in the best state; MTBF1 represents the mean time between failures of similar equipment produced by other manufacturers; T 1_i represents the actual operating life of the i-th device among similar devices produced by other manufacturers; n1 represents the total number of failures of similar devices produced by other manufacturers during operation; MTBF2 represents the mean time between failures of the individual device; t2 represents the actual operating life of the individual device; n2 represents the number of failures of the individual device during operation; t k represents the operating time of the equipment individual before the kth maintenance; It represents the equivalent operation time of the equipment after the kth maintenance;

[0126] The optimal maintenance time calculation module is used to calculate the cumulative probability value of equipment failure from the completion of the last maintenance to the next time based on the actual failure rate model, and determine the time when the cumulative probability value of equipment failure reaches the cumulative probability threshold value of equipment failure. This time is the optimal time for the next maintenance, wherein the calculation formula of the cumulative probability value of equipment failure is:

[0127]

[0128] In the above formula, P1 represents the cumulative probability of equipment failure from the completion of the kth maintenance to the next maintenance; t′ represents the equivalent service life of the equipment at the current time, t x Indicates the next maintenance time of the equipment;

[0129] The calculation formula for the cumulative probability threshold of equipment failure is:

[0130]

[0131] In the above formula, P max Indicates the cumulative probability threshold of equipment failure; T O Indicates the design life of the equipment; T max Indicates the maximum interval between scheduled inspections of equipment.

Claims

1. A method for determining the fault repair time of power transmission and transformation equipment, characterized in that: The method comprises: S1. Use the fault history data of power transmission and transformation equipment to perform segmented fitting on the Weibull distribution function of the equipment during the accidental failure period and the loss failure period, calculate the boundary years between the accidental failure period and the loss failure period, and construct a basic failure rate model for power transmission and transformation equipment; S2. Modify the fitted basic failure rate model according to the following formula to obtain the actual failure rate model: λ(t)=η×λ0(t-ΔT1-ΔT2); λT3=t k×s -t k×(s-1) ; In the above formula, λ(t) represents the actual failure rate model finally obtained; λ0(t) represents the equipment failure rate obtained by the basic failure rate model at the time t-ΔT1-ΔT2; η represents the correction coefficient of family defects and individual defects to the Weibull distribution function; ΔT1 is the first time interval, which is used to represent the regression effect of maintenance behavior on the equipment operation time; ΔT2 is the second time interval, which is used to represent the regression effect of short-term sudden change factors on the equipment operation time; ΔT3 represents the time interval between the two most recent health status evaluations of the equipment; t k×s represents the actual operating time of the equipment after the sth health status evaluation after the kth maintenance; t k ×(s-1) represents the actual operating time of the equipment after the S-1th health status evaluation after the kth maintenance; JK represents the current health status evaluation value of the equipment; JK0 represents the health status evaluation critical value between normal operation of the equipment and equipment operation failure; 100 represents the equipment operating in the best state; MTBF1 represents the mean time between failures of similar equipment produced by other manufacturers; f 1_i represents the actual operating life of the i-th device among similar devices produced by other manufacturers; n1 represents the total number of failures of similar devices produced by other manufacturers during operation; MTBF2 represents the mean time between failures of the individual device; t2 represents the actual operating life of the individual device; n2 represents the number of failures of the individual device during operation; t k represents the operating time of the equipment individual before the kth maintenance; It represents the equivalent operation time of the equipment after the kth maintenance; S3. Calculate the cumulative probability of equipment failure from the completion of the last maintenance to the next moment based on the actual failure rate model, and determine the moment when the cumulative probability of equipment failure reaches the cumulative probability threshold of equipment failure. This moment is the best time for the next maintenance.

2. A method for determining the fault repair time of power transmission and transformation equipment according to claim 1, characterized in that: The expression of the fitted basic failure rate model is: In the above formula, λ0(t) represents the equipment failure rate at time t; α1 and β1 are the shape parameter and scale parameter of the accidental failure period, respectively; α2 and β2 are the shape parameter and scale parameter of the loss failure period, respectively; l represents the dividing line between the accidental failure period and the loss failure period.

3. A method for determining the fault repair time of power transmission and transformation equipment according to claim 2, characterized in that: In S1, the BP neural network improved by the LM algorithm is used to fit the parameters α and β of the Weibull distribution function. The specific fitting process is: A1, initialization parameters α, β; A2. Calculate the partial derivatives of parameters α and β, and use the LM algorithm to obtain the parameter estimate α after the first iteration. 1 , β 1 And the residual sum of squares of the BP neural network; A3, based on the parameter estimate α after the first iteration 1 , β 1 Perform the second iteration calculation and obtain the parameter estimation value α after the second iteration through the LM algorithm 2 , β 2 and the residual sum of squares for the second iteration; A4. Compare the residual sum of squares of the first iteration and the second iteration. If the residual sum of squares of the second iteration is less than that of the first iteration, the second iteration ends and the process goes to A5. Otherwise, increase the damping coefficient in the LM algorithm in the second iteration and return to A3. A5. Return to step A2 to continue iterative calculation until the absolute value of the difference between the parameter estimation value obtained after a certain iteration and the parameter estimation value obtained before a certain iteration is less than the preset system allowable error. The entire iterative process ends and the parameter estimation value obtained in the last iteration is output.

4. A method for determining the fault repair time of power transmission and transformation equipment according to claim 2, characterized in that: In S1, the boundary years between the accidental failure period and the loss failure period are obtained by solving the objective function; the calculation formula of the objective function is: In the above formula, S l represents the objective function; l represents the dividing year between the accidental failure period and the loss failure period; n represents the last year of the loss failure period; λ0(t) 偶 represents the part of the fitted Weibull distribution function corresponding to the wear failure period; λ0(t) 损 represents the part of the fitted Weibull distribution function corresponding to the wear failure period; 实_t Indicates the actual failure rate of the equipment at time t.

5. A method for determining the fault repair time of power transmission and transformation equipment according to claim 1 or 2, characterized in that: The calculation formula for the cumulative probability value of equipment failure is: In the above formula, P1 represents the cumulative probability value of equipment failure from the completion of the kth maintenance to the next maintenance; t′ represents the equivalent service life of the equipment at the current time, t x Indicates the next maintenance time of the equipment; In step S3, the calculation formula for the cumulative probability threshold of equipment failure is: In the above formula, P max represents the cumulative probability threshold of equipment failure; T0 represents the design life of the equipment; T max Indicates the maximum interval between scheduled inspections of equipment.

6. A system for determining the time for troubleshooting of power transmission and transformation equipment, characterized in that: The system includes a basic failure rate model fitting module, an actual failure rate model building module, and an optimal maintenance time calculation module: The basic failure rate model fitting module is used to use the fault history data of the power transmission and transformation equipment to perform piecewise fitting on the Weibull distribution function of the equipment during the accidental failure period and the loss failure period, calculate the boundary years between the accidental failure period and the loss failure period, and construct the basic failure rate model of the power transmission and transformation equipment; The actual failure rate model building module is used to correct the fitted basic failure rate model according to the following formula to obtain the actual failure rate model: λ(t)=η×λ0(t-ΔT1-ΔT2); ΔT3=t k×s -t k×(s-1) ; In the above formula, λ(t) represents the actual failure rate model finally obtained; λ0(t) represents the equipment failure rate obtained by the basic failure rate model at the time t-ΔT1-ΔT2; η represents the correction coefficient of family defects and individual defects on the Weibull distribution function; ΔT1 is the first time interval, which is used to represent the regression effect of maintenance behavior on the equipment operation time; ΔT2 is the second time interval, which is used to represent the regression effect of short-term sudden change factors on the equipment operation time; ΔT3 represents the time interval between the two most recent health status evaluations of the equipment; t k×s It represents the actual operation time of the equipment after the sth health status evaluation after the kth maintenance; t k ×(s-1) represents the actual operation time of the equipment after the S-1th health status evaluation after the kth maintenance; JK represents the current health status evaluation value of the equipment; JK0 represents the health status evaluation critical value between normal operation of the equipment and equipment operation failure; 100 means the equipment is running in the best condition; MTBF1 means the average time between failures of similar equipment produced by other manufacturers; t 1_i represents the actual operating life of the i-th device among similar devices produced by other manufacturers; n1 represents the total number of failures of similar devices produced by other manufacturers during operation; MTBF2 represents the mean time between failures of the individual device; t2 represents the actual operating life of the equipment; n2 represents the number of failures that occurred during the operation of the equipment; t k represents the operating time of the equipment individual before the kth maintenance; It represents the equivalent operation time of the equipment after the kth maintenance; The optimal maintenance time calculation module is used to calculate the cumulative probability value of equipment failure from the completion of the last maintenance to the next time based on the actual failure rate model, and determine the time when the cumulative probability value of equipment failure reaches the cumulative probability threshold value of equipment failure. This time is the optimal time for the next maintenance.

7. A system for determining fault repair time of power transmission and transformation equipment according to claim 6, characterized in that: The expression of the fitted basic failure rate model is: In the above formula, λ0(t) represents the equipment failure rate at time t; α1 and β1 are the shape parameter and scale parameter of the accidental failure period, respectively; α2 and β2 are the shape parameter and scale parameter of the loss failure period, respectively; l represents the dividing line between the accidental failure period and the loss failure period.

8. A system for determining the fault repair time of power transmission and transformation equipment according to claim 7, characterized in that: The basic failure rate model fitting module uses the BP neural network improved by the LM algorithm to fit the parameters α and β of the Weibull distribution function. The specific fitting steps are: A1, initialization parameters α, β; A2. Calculate the partial derivatives of parameters α and β, and use the LM algorithm to obtain the parameter estimate α after the first iteration. 1 , β 1 and the residual sum of squares; A3, based on the parameter estimate α after the first iteration 1 , β 1 Perform the second iteration calculation and obtain the parameter estimation value α after the second iteration through the LM algorithm 2 , β 2 and the residual sum of squares for the second iteration; A4. Compare the residual sum of squares of the first iteration and the second iteration. If the residual sum of squares of the second iteration is less than that of the first iteration, the second iteration ends and the process goes to A5. Otherwise, increase the damping coefficient in the LM algorithm in the second iteration and return to A3. A5. Return to step A2 to continue iterative calculation until the absolute value of the difference between the parameter estimation value obtained after a certain iteration and the parameter estimation value obtained before a certain iteration is less than the preset system allowable error. The entire iterative process ends and the parameter estimation value obtained in the last iteration is output.

9. A system for determining the fault repair time of power transmission and transformation equipment according to claim 7, characterized in that: The basic failure rate model fitting module obtains the boundary years between the accidental failure period and the loss failure period by solving the objective function; the calculation formula of the objective function is: In the above formula, S l represents the objective function; l represents the dividing year between the accidental failure period and the loss failure period; n represents the last year of the loss failure period; λ0(t) 偶 represents the part of the fitted Weibull distribution function corresponding to the wear failure period; λ0(t) 损 represents the part of the fitted Weibull distribution function corresponding to the wear failure period, λ 实_t Indicates the actual failure rate of the equipment at time t.

10. A system for determining the fault repair time of power transmission and transformation equipment according to claim 6 or 7, characterized in that: The optimal maintenance time calculation module calculates the cumulative probability value of equipment failure according to the following formula: In the above formula, P1 represents the cumulative probability value of equipment failure from the completion of the kth maintenance to the next maintenance; t′ represents the equivalent service life of the equipment at the current time, t x Indicates the next maintenance time of the equipment; The cumulative probability threshold of equipment failure is calculated according to the following formula: In the above formula, P max represents the cumulative probability threshold of equipment failure; T0 represents the design life of the equipment; T max Indicates the maximum interval between scheduled inspections of equipment.

Images