A method, device and storage medium for predicting the remaining life of a high-power thyristor

Abstract

The present invention discloses a method, device and storage medium for predicting the remaining life of a high-power thyristor, wherein the prediction method includes the following steps: firstly, a high-temperature blocking test is performed on the thyristor, and the test time in the high-temperature blocking test of the thyristor is converted into the operating time of the thyristor in actual working conditions by an acceleration factor, a comprehensive health index of the thyristor is constructed according to the measurement data, the weight coefficient of the health index is solved by a sparse Bayesian algorithm, and the hyperparameters of the model are updated by maximizing the marginal likelihood function, and finally, a degradation model of the thyristor is established according to the comprehensive health index of the thyristor to evaluate the remaining life of the thyristor. The comprehensive health index of the thyristor constructed in the present invention can comprehensively reflect the health status of the thyristor, and the thyristor remaining life prediction method based on the index can characterize the influence of the degradation of different characteristic parameters on the remaining life of the thyristor, so that the remaining life prediction of the thyristor is more accurate.

Description

Technical Field

[0001] The present invention belongs to the technical field of high-power thyristor remaining life prediction, and in particular relates to a high-power thyristor remaining life prediction method, device and storage medium. Background Art

[0002] Thyristors are the core components for power conversion in UHV DC transmission converter valves, and their reliability is of great significance to the safe and stable operation of DC transmission systems. Accurate prediction of the remaining life of thyristors can help operation and maintenance personnel replace them in time before they fail, avoiding damage to the DC transmission system caused by thyristor failure, thereby improving the reliability of the DC transmission system.

[0003] In the converter valve, thyristors have different operating states, and their health status is reflected by multiple characteristic parameters, such as on-state voltage drop, leakage current, reverse recovery charge, gate current, etc. At present, the research on thyristor remaining life prediction mainly selects one or two of the above characteristic parameters for prediction, lacks overall consideration of the above typical characteristic parameters of thyristors, and it is difficult to fully and accurately reflect the health status and degradation process of thyristors.

[0004] Therefore, in order to fully reflect the health status of thyristors and accurately measure their remaining service life, it is necessary to make full use of multiple performance indicators of thyristors to describe the health status of thyristors, and to construct a thyristor remaining service life prediction method based on multi-dimensional characteristic parameters to provide more accurate guidance for relevant staff.

[0005] Therefore, the applicant proposes a method, device and storage medium for predicting the remaining life of a high-power thyristor. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention provides a method for predicting the remaining service life of a high-power thyristor, which comprehensively reflects the health status of the thyristor, accurately measures its remaining service life, makes full use of multiple performance indicators of the thyristor to describe the health status of the thyristor, and constructs a thyristor remaining service life prediction method based on multi-dimensional characteristic parameters, providing more accurate guidance for relevant personnel.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] A method for predicting the remaining life of a high-power thyristor, the method comprising the following steps:

[0009] S1. Perform a high temperature blocking test on the thyristor and collect degradation data on the thyristor on-state voltage drop, leakage current, reverse recovery charge and gate current;

[0010] S2. Calculate the acceleration factor, convert the operating time of the thyristor in the test to the operating time under actual working conditions; calculate the degradation rate of each characteristic parameter;

[0011] S3. Construct a comprehensive health index of the thyristor based on the thyristor high temperature blocking test data, and determine the weight coefficient and failure threshold of the comprehensive health index;

[0012] S4. Establish a thyristor degradation model based on comprehensive indicators, estimate the unknown parameters of the degradation model, and predict the remaining life of the thyristor.

[0013] As a further solution of the present invention: the specific method of the thyristor high temperature blocking test described in step S1 is:

[0014] Set the thyristor in blocking state and apply reverse voltage to the sample. The voltage value is 80% of the maximum reverse bias voltage. Set the ambient temperature to 120℃. Detect the on-state voltage drop, leakage current, reverse recovery charge and gate current of the thyristor every 25 hours. Stop the test after 1000 hours.

[0015] As a further solution of the present invention: the specific steps of step S2 are as follows:

[0016] The calculation formula of the acceleration factor is as follows:

[0017]

[0018] in, is the acceleration factor, is the voltage applied to the thyristor during the test, is the average voltage of the thyristor under actual working conditions, is the activation energy, with a value of 0.6, k is the Boltzmann constant, which is , is the temperature of the thyristor in the test, is the temperature of the thyristor during normal operation under actual working conditions;

[0019] The calculation formula for converting the operating time of the thyristor in the test to the operating time under actual working conditions is as follows:

[0020]

[0021] in, is the time under actual operating conditions, The time for the high temperature blocking test to run;

[0022] The calculation formula of the degradation rate of each characteristic parameter is as follows:

[0023]

[0024] in , , , They are the thyristor on-state voltage drop, leakage current, reverse recovery charge and gate current. ( k =1,2,…, K ) time degradation rate, , , , They are the thyristor on-state voltage drop, leakage current, reverse recovery charge and gate current. The degradation data at the moment, , , , They are the rated values ​​of the thyristor's on-state voltage drop, leakage current, reverse recovery charge and gate current respectively.

[0025] As a further solution of the present invention: the calculation formula of the thyristor comprehensive health index described in step S3 is as follows:

[0026]

[0027] in, is the comprehensive health index vector of the thyristor, is the data matrix of the characteristic parameters of the thyristor, in which each characteristic parameter of each sample is tested. Observation, is the weight coefficient vector of the thyristor, , , , are the weight coefficients of thyristor on-state voltage drop, leakage current, reverse recovery charge and gate current, respectively. is the noise vector of the thyristor, where ;

[0028] The calculation of the weight coefficient includes the following steps;

[0029] 3-1 Initialize the distribution of model parameters;

[0030] 3-2 Construct the posterior distribution of the weight coefficient based on the Bayesian formula and estimate the mean and standard deviation of the weight coefficient;

[0031] 3-3 Calculate the hyperparameters required for the sparse Bayesian algorithm by maximizing the marginal likelihood function;

[0032] 3-4 Repeat steps 3-2 and 3-3 until the estimated value of the weight coefficient converges, output the model parameters, and obtain the weight coefficient;

[0033] The weight coefficient calculation method described in step 3-2 is as follows:

[0034] According to the sparse Bayesian algorithm, the weight coefficient is set ( p =1, 2, 3, 4) through hyperparameters Sampling is obtained, ,in , , , They are weight coefficients , , , The corresponding hyperparameters, assuming , we can get the posterior distribution function of the weight coefficient as follows:

[0035]

[0036] in

[0037]

[0038]

[0039] in

[0040] in

[0041] I is the 4th-order identity matrix

[0042] According to the above formula, the posterior distribution of the weight coefficient is:

[0043]

[0044] in ,

[0045] It can be seen that the posterior distribution of the weight coefficient follows a Gaussian distribution, so the mean and standard deviation of the weight coefficient are:

[0046]

[0047] The calculation method of the hyperparameters described in step 3-3 is as follows:

[0048] The marginal likelihood function of the hyperparameters is:

[0049]

[0050] The variance of the noise term is obtained by maximizing the marginal likelihood function and the variance of the hyperparameters , and its calculation formula is:

[0051]

[0052]

[0053] in, For degraded data The p-th column vector of p =1, 2, 3, 4).

[0054] As a further solution of the present invention: the calculation formula of the failure threshold of the thyristor comprehensive health index described in step S3 is as follows:

[0055]

[0056] The failure threshold of the constructed thyristor comprehensive health index is is the failure threshold of the on-state voltage drop, is the failure threshold of the thyristor leakage current, is the failure threshold of the thyristor reverse recovery charge, is the failure threshold of the thyristor gate current.

[0057] As a further solution of the present invention: the process of establishing the thyristor degradation model in step S4 is as follows:

[0058] For any time t , using the scale parameter , the shape parameter is The Gamma process describes the comprehensive health index of thyristors Y ( t ) degradation process, Y ( t ) meets the following conditions:

[0059] (1) Established with probability 1;

[0060] (2) Y ( t ) have stable independent increments;

[0061] (3) For any and , ;

[0062] Furthermore, State Parameter , the scale parameter is The probability density function of the degradation amount is:

[0063]

[0064] in is the Gamma function, and

[0065] .

[0066] As a further solution of the present invention: the method for estimating the unknown parameters of the thyristor degradation model described in step S4 is:

[0067] According to the maximum likelihood method, the unknown parameters of the model are estimated by finding the maximum value of the likelihood function of the model. The likelihood function of the model is:

[0068]

[0069] Taking the derivative of the likelihood function, we get:

[0070]

[0071] in is the digamma function, , , ;

[0072] Solve the above equations, the unknown parameters of the model The expression is:

[0073]

[0074] Solving the following equation gives the scale parameter Estimated value of

[0075]

[0076] Will Substitute the estimated value of In the calculation formula, the estimated value of the location parameter is obtained .

[0077] As a further solution of the present invention: the probability density function of the remaining life of the thyristor in step S4 is , and its calculation formula is as follows:

[0078]

[0079] in, , ;

[0080] The expectation of the probability density function of the remaining life is calculated to obtain the predicted value of the remaining life of the thyristor. The calculation formula for the remaining life of the thyristor is:

[0081]

[0082] in, for The remaining life of the thyristor at time , the degradation of the comprehensive health index at this time is .

[0083] The present application also proposes an electronic device, including a memory and a processor, wherein the memory is used to store a program that supports the processor to execute a method for predicting the remaining life of a high-power thyristor, and the processor is configured to execute the program stored in the memory.

[0084] The present application also proposes a storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of the method for predicting the remaining life of a high-power thyristor are executed.

[0085] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention proposes a thyristor remaining life prediction method integrating multi-dimensional characteristic parameters, uses comprehensive health indicators to describe the degradation process of different characteristic parameters, and more comprehensively characterizes the health status of the thyristor. It solves the problem that there are many thyristor degradation characteristic parameters and it is difficult to predict the remaining life of the thyristor using a small number of characteristic parameters. The remaining life of the thyristor is predicted based on the comprehensive health indicators, which improves the accuracy of the remaining life prediction of the thyristor. It has a wide range of applications and provides more effective guidance for the operation and maintenance of the converter valve. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without creative work. Among them:

[0087] Figure 1 The present invention is a flowchart of a method for predicting the remaining life of a high-power thyristor. DETAILED DESCRIPTION

[0088] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0089] Example 1

[0090] See also Figure 1 As shown, the present invention provides a method for predicting the remaining life of a high-power thyristor, the prediction method comprising the following steps:

[0091] 1. Carry out high temperature blocking test on thyristor and collect degradation data of thyristor on-state voltage drop, leakage current, reverse recovery charge and gate current. The specific method is as follows:

[0092] Set the thyristor in blocking state and apply reverse voltage to the sample. The voltage value is 80% of the maximum reverse bias voltage. Set the ambient temperature to 120℃. Detect the on-state voltage drop, leakage current, reverse recovery charge and gate current of the thyristor every 25 hours. Stop the test after 1000 hours.

[0093] 2. Calculate the acceleration factor, convert the operating time of the thyristor in the test to the operating time under actual working conditions, and calculate the degradation rate of each characteristic parameter:

[0094] The calculation formula of the acceleration factor is as follows:

[0095]

[0096] in, is the acceleration factor, is the voltage applied to the thyristor during the test, is the average voltage of the thyristor under actual working conditions, is the activation energy, with a value of 0.6, k is the Boltzmann constant, which is , is the temperature of the thyristor in the test, is the temperature of the thyristor during normal operation under actual working conditions;

[0097] The calculation formula for converting the operating time of the thyristor in the test to the operating time under actual working conditions is as follows:

[0098]

[0099] in, is the time under actual operating conditions, The time for the high temperature blocking test to run;

[0100] The calculation formula of the degradation rate of each characteristic parameter is as follows:

[0101]

[0102] in , , , They are the thyristor on-state voltage drop, leakage current, reverse recovery charge and gate current. ( k =1,2,…, K ) time degradation rate, , , , They are the thyristor on-state voltage drop, leakage current, reverse recovery charge and gate current. The degradation data at the moment, , , , They are the rated values ​​of the thyristor's on-state voltage drop, leakage current, reverse recovery charge and gate current respectively.

[0103] 3. Construct the comprehensive health index of thyristors based on the high-temperature blocking test data of thyristors, and determine the weight coefficient and failure threshold of the comprehensive health index:

[0104] The calculation formula of the thyristor comprehensive health index is as follows:

[0105]

[0106] in, is the comprehensive health index vector of the thyristor, is the data matrix of the characteristic parameters of the thyristor, in which each characteristic parameter of each sample is tested. Observations, is the weight coefficient vector of the thyristor, , , , are the weight coefficients of thyristor on-state voltage drop, leakage current, reverse recovery charge and gate current, respectively. is the noise vector of the thyristor, where ;

[0107] The calculation of the weight coefficient comprises the following steps:

[0108] 3-1 Initialize the distribution of model parameters;

[0109] 3-2 Based on the Bayesian formula, the posterior distribution of the weight coefficient is constructed, and the mean and standard deviation of the weight coefficient are estimated. The calculation method is as follows:

[0110] According to the sparse Bayesian algorithm, the weight coefficient is set ( p =1, 2, 3, 4) through hyperparameters Sampling is obtained, ,in , , , They are weight coefficients , , , The corresponding hyperparameters, assuming , we can get the posterior distribution function of the weight coefficient as follows:

[0111]

[0112] in

[0113]

[0114]

[0115] in

[0116] in

[0117] I is the 4th-order identity matrix

[0118] According to the above formula, the posterior distribution of the weight coefficient is:

[0119]

[0120] in ,

[0121] It can be seen that the posterior distribution of the weight coefficient follows a Gaussian distribution, so the mean and standard deviation of the weight coefficient are:

[0122] ;

[0123] 3-3 The hyperparameters required for the sparse Bayesian algorithm are calculated by maximizing the marginal likelihood function. The calculation method is as follows:

[0124] The marginal likelihood function of the hyperparameters is:

[0125]

[0126] The variance of the noise term is obtained by maximizing the marginal likelihood function and the variance of the hyperparameters , and its calculation formula is:

[0127]

[0128]

[0129] in, For degraded data The p-th column vector of p =1,2,3,4);

[0130] 3-4 Repeat steps 3-2 and 3-3 until the estimated value of the weight coefficient converges, output the model parameters, and obtain the weight coefficient;

[0131] The failure threshold calculation formula of the thyristor comprehensive health index is as follows:

[0132]

[0133] The failure threshold of the constructed thyristor comprehensive health index is is the failure threshold of the on-state voltage drop, is the failure threshold of the thyristor leakage current, is the failure threshold of the thyristor reverse recovery charge, is the failure threshold of the thyristor gate current.

[0134] 4. Establish a thyristor degradation model based on comprehensive indicators, estimate the unknown parameters of the degradation model, and predict the remaining life of the thyristor:

[0135] The establishment process of the thyristor degradation model is as follows:

[0136] For any time t , using the scale parameter , the shape parameter is The Gamma process describes the comprehensive health index of thyristors Y ( t ) degradation process, Y ( t ) meets the following conditions:

[0137] (1) Established with probability 1;

[0138] (2) Y ( t ) have stable independent increments;

[0139] (3) For any and , ;

[0140] Furthermore, State Parameter , the scale parameter is The probability density function of the degradation amount is:

[0141]

[0142] in is the Gamma function, and

[0143] ;

[0144] The estimation method of the unknown parameters of the thyristor degradation model is:

[0145] According to the maximum likelihood method, the unknown parameters of the model are estimated by finding the maximum value of the likelihood function of the model. The likelihood function of the model is:

[0146]

[0147] Taking the derivative of the likelihood function, we get:

[0148]

[0149] in is the digamma function, , , ;

[0150] Solve the above equations, the unknown parameters of the model The expression is:

[0151]

[0152] Solving the following equation gives the scale parameter Estimated value of

[0153]

[0154] Will Substitute the estimated value of In the calculation formula, the estimated value of the location parameter is obtained ;

[0155] The probability density function of the remaining life of the thyristor is , and its calculation formula is as follows:

[0156]

[0157] in, , ;

[0158] The expectation of the probability density function of the remaining life is calculated to obtain the predicted value of the remaining life of the thyristor. The calculation formula for the remaining life of the thyristor is:

[0159]

[0160] in, for The remaining life of the thyristor at time , the degradation of the comprehensive health index at this time is According to the above-mentioned thyristor remaining life prediction method and the results of the high temperature blocking test, the predicted values ​​of the remaining life of thyristors with different degradation amounts are shown in Table 1 below:

[0161] .

[0162] Example 2

[0163] An electronic device comprises a memory and a processor, wherein the memory is used to store a program supporting the processor to execute a method for predicting the remaining life of a high-power thyristor, and the processor is configured to execute the program stored in the memory.

[0164] Example 3

[0165] A storage medium stores a computer program, which executes the steps of a method for predicting the remaining life of a high-power thyristor when the computer program is run by a processor.

[0166] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for predicting the remaining life of a high-power thyristor, characterized in that: The prediction method includes the following steps: S1. Perform a high temperature blocking test on the thyristor and collect degradation data on the thyristor on-state voltage drop, leakage current, reverse recovery charge and gate current; S2. Calculate the acceleration factor, convert the operating time of the thyristor in the test to the operating time under actual working conditions; calculate the degradation rate of each characteristic parameter; S3. Construct a comprehensive health index of the thyristor based on the thyristor high temperature blocking test data, and determine the weight coefficient and failure threshold of the comprehensive health index; S4. Establish a thyristor degradation model based on comprehensive indicators, estimate unknown parameters of the degradation model, and predict the remaining life of the thyristor; The calculation formula of the thyristor comprehensive health index described in step S3 is as follows: in, is the comprehensive health index vector of the thyristor, is the data matrix of the characteristic parameters of the thyristor, in which each characteristic parameter of each sample is tested. Observation, is the weight coefficient vector of the thyristor, , , , are the weight coefficients of thyristor on-state voltage drop, leakage current, reverse recovery charge and gate current, respectively. is the noise vector of the thyristor, where ; The calculation of the weight coefficient includes the following steps; 3-1 Initialize the distribution of model parameters; 3-2 Construct the posterior distribution of the weight coefficient based on the Bayesian formula and estimate the mean and standard deviation of the weight coefficient; 3-3 Calculate the hyperparameters required for the sparse Bayesian algorithm by maximizing the marginal likelihood function; 3-4 Repeat steps 3-2 and 3-3 until the estimated value of the weight coefficient converges, output the model parameters, and obtain the weight coefficient; The weight coefficient calculation method described in step 3-2 is as follows: According to the sparse Bayesian algorithm, the weight coefficient is set ( p =1, 2, 3, 4) through hyperparameters Sampling is obtained, ,in , , , The weight coefficients are , , , The corresponding hyperparameters, assuming , we can get the posterior distribution function of the weight coefficient as follows: in in in I is the 4th-order identity matrix According to the above formula, the posterior distribution of the weight coefficient is: in , It can be seen that the posterior distribution of the weight coefficient follows a Gaussian distribution, so the mean and standard deviation of the weight coefficient are: The calculation method of the hyperparameters described in step 3-3 is as follows: The marginal likelihood function of the hyperparameters is: The variance of the noise term is obtained by maximizing the marginal likelihood function and the variance of the hyperparameters , and its calculation formula is: in, For degraded data The p-th column vector of p =1, 2, 3, 4).

2. The method for predicting the remaining life of a high-power thyristor according to claim 1, characterized in that: The specific method of the thyristor high temperature blocking test described in step S1 is: Set the thyristor in blocking state and apply reverse voltage to the sample. The voltage value is 80% of the maximum reverse bias voltage. Set the ambient temperature to 120℃. Detect the on-state voltage drop, leakage current, reverse recovery charge and gate current of the thyristor every 25 hours. Stop the test after 1000 hours.

3. The method for predicting the remaining life of a high-power thyristor according to claim 1, characterized in that: The specific steps of step S2 are as follows: The calculation formula of the acceleration factor is as follows: in, is the acceleration factor, is the voltage applied to the thyristor during the test, is the average voltage of the thyristor under actual working conditions, is the activation energy, with a value of 0.6, k is the Boltzmann constant, which is , is the temperature of the thyristor in the test, is the temperature of the thyristor during normal operation under actual working conditions; The calculation formula for converting the operating time of the thyristor in the test to the operating time under actual working conditions is as follows: in, is the time under actual operating conditions, The time for the high temperature blocking test to run; The calculation formula of the degradation rate of each characteristic parameter is as follows: in , , , They are the thyristor on-state voltage drop, leakage current, reverse recovery charge and gate current. ( k =1,2,…, K ) time degradation rate, , , , They are the thyristor on-state voltage drop, leakage current, reverse recovery charge and gate current. The degradation data at the moment, , , , They are the rated values ​​of the thyristor's on-state voltage drop, leakage current, reverse recovery charge and gate current respectively.

4. The method for predicting the remaining life of a high-power thyristor according to claim 1, characterized in that: The calculation formula for the failure threshold of the thyristor comprehensive health index described in step S3 is as follows: The failure threshold of the constructed thyristor comprehensive health index is is the failure threshold of the on-state voltage drop, is the failure threshold of the thyristor leakage current, is the failure threshold of the thyristor reverse recovery charge, is the failure threshold of the thyristor gate current.

5. The method for predicting the remaining life of a high-power thyristor according to claim 1, characterized in that: The process of establishing the thyristor degradation model described in step S4 is as follows: For any time t , using the scale parameter , the shape parameter is The Gamma process describes the comprehensive health index of thyristors Y ( t ) degradation process, Y ( t ) meets the following conditions: (1) Established with probability 1; (2) Y ( t ) have stable independent increments; (3) For any and , ; Furthermore, State Parameters , the scale parameter is The probability density function of the degradation amount is: in is the Gamma function, and 。 6. The method for predicting the remaining life of a high-power thyristor according to claim 5, characterized in that: The method for estimating the unknown parameters of the thyristor degradation model described in step S4 is: According to the maximum likelihood method, the unknown parameters of the model are estimated by finding the maximum value of the likelihood function of the model. The likelihood function of the model is: Taking the derivative of the likelihood function, we get: in is the digamma function, , , ; Solve the above equations, the unknown parameters of the model The expression is: Solving the following equation gives the scale parameter Estimated value of Will Substitute the estimated value of In the calculation formula, the estimated value of the location parameter is obtained .

7. The method for predicting the remaining life of a high-power thyristor according to claim 6, characterized in that: The probability density function of the remaining life of the thyristor described in step S4 , and its calculation formula is as follows: in, , ; The expectation of the probability density function of the remaining life is calculated to obtain the predicted value of the remaining life of the thyristor. The calculation formula for the remaining life of the thyristor is: in, for The remaining life of the thyristor at time , the degradation of the comprehensive health index at this time is .

8. An electronic device, comprising a memory and a processor, characterized in that: The memory is used to store a program that supports a processor to execute the method for predicting the remaining life of a high-power thyristor according to any one of claims 1 to 7, and the processor is configured to execute the program stored in the memory.

9. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for predicting the remaining life of a high-power thyristor according to any one of claims 1 to 7 are executed.

Images