A Fault Diagnosis Method for Power Converters under Multi-Noise Mixing and Multi-Operating Modes

Abstract

The present invention discloses a fault diagnosis method for a power converter under multi-noise mixing and multi-operation modes, belonging to the technical field of fault diagnosis. Aiming at the actuator fault of the power converter, the present invention establishes a state space model of the power converter, designs a state estimator of the power converter by minimizing the mean square error and the polytope space. When the operation mode of the power converter changes, first, it is judged whether the operation mode of the power converter changes. If it is detected that the measured output is not within the output estimation interval, further, the system fault and the system operation mode change are distinguished by judging whether the measured output is within the output intervals of all operation modes of the power converter. If the system fails, a fault estimator of the power converter is constructed to estimate the system fault. Compared with the existing fault diagnosis method combining double filtering of random and unknown but bounded mixed noise, the method of the present application solves the problems of operation state recognition and fault diagnosis of the power converter under different operation modes.

Description

Technical Field

[0001] The present invention relates to a fault diagnosis method for a power converter under multi-noise mixing and multi-operation modes, belonging to the technical field of fault diagnosis. Background Art

[0002] A power converter is a device that converts a DC power supply at the input end into a DC power supply with different characteristics at the output end. With the development of technology, power converters are widely used in more and more fields such as new energy vehicles, industrial instruments, and medical equipment. In the actual application of power converters, the safety and reliability of power converters have always faced huge challenges. Therefore, in order to ensure the safe and reliable operation of the power converter system, it is necessary to perform real-time and effective fault diagnosis on the power converter.

[0003] The working environment of the power converter is complex and it is easily affected by various noise interferences during operation. Traditional fault diagnosis methods based on the power converter model usually assume that the noise existing in the operation process of the power converter system follows a single Gaussian distribution. However, in actual applications, the noise and interference in the power converter system not only include random noise with unknown probability characteristics, but may also include bounded noise with unknown boundaries. Therefore, the results obtained by such fault diagnosis methods for a single noise scenario have low accuracy. In the prior art, for a system containing unknown but bounded noise, a method based on set membership filtering is usually adopted for fault diagnosis. For example, CN114155117A discloses a filtering fault diagnosis method for a power converter under multi-operation modes, that is, applying the polytope filtering fault diagnosis method capable of performing fault diagnosis based on unknown but bounded noise to the power converter regression model. However, this method is based on the power converter regression model and can only diagnose parameter faults (such as capacitor failure, inductor failure, etc.) in the power converter by estimating the parameters of the power converter, but cannot diagnose actuator faults (including circuit faults, switch faults, etc.) in the power converter. At the same time, the multi-operation modes involved in this method are caused by changes in devices such as resistors and capacitors in the circuit, so a regression model based on known power converter parameters can be adopted. And for the actuator fault diagnosis of the power converter, since the parameters are unknown, a regression model cannot be used for diagnosis.

[0004] In summary, although CN114155117A that performs fault diagnosis based on the regression model can perform fault diagnosis on a power converter under multi-operation modes, the power converter system it targets is only in a single noise mode, and the types of faults it targets are limited to faults such as capacitor failure and inductor failure, and it cannot effectively diagnose other types of faults in the power converter system under the multi-noise mixing mode. Summary of the Invention

[0005] To solve the above problems, the present invention provides a method for fault diagnosis of a power converter under multi-noise mixing and multi-operation modes. The method includes:

[0006] Step 1: For a power converter system with mixed unknown but bounded noise and Gaussian noise, discrete power converter models are established for its different working modes respectively. The power converter system has two working modes, denoted as working mode 1 and working mode 2 respectively;

[0007] Step 2: Design a power converter state estimator to obtain the state estimation value of the power converter at time k, and determine the state estimation interval of the power converter at time k according to the state estimation value;

[0008] Step 3: According to the state estimation value of the power converter at time k, obtain the output estimation interval at time k;

[0009] Step 4: Measure the output of the power converter at time k through a voltage sensor and a current sensor, denoted as the measured output y k , and judge whether the working mode of the power converter changes according to whether the measured output y k is within the output estimation interval at time k;

[0010] If the measured output y k is within the output estimation interval, it is determined that the working mode of the power converter at time k has not changed, and jump to Step 8; otherwise, go to Step 5;

[0011] Step 5: Judge whether the measured output y k of the power converter at time k is within the output intervals of all working modes of the power converter at time k:

[0012] If y k is not within the output intervals of all working modes of the power converter at time k, it is determined that the power converter has a fault at time k, and jump to Step 7;

[0013] If y k is within the output interval of a certain working mode of the power converter at time k, it is determined that the working mode of the power converter at time k has changed, and jump to Step 6;

[0014] Step 6: Initialize the state estimator at time k - 1, obtain the state estimator and the state estimation interval at time k, and jump to Step 8;

[0015] Step 7: Construct a power converter fault estimator to obtain the fault estimation interval of the power converter at time k;

[0016] Step 8: Set k = k + 1, and jump to Step 2; until the operation of the power converter ends;

[0017] Execute the above steps two to eight in a loop to complete the fault diagnosis process of the power converter without interruption.

[0018] Optionally, step one includes:

[0019] Establish a model for the Buck circuit in operating mode 1:

[0020]

[0021] Establish a model for the Buck circuit in operating mode 2:

[0022]

[0023] where \(i\) is the inductor current, \(u\) c is the capacitor voltage, \(L\), \(C\), and \(r\) are the inductor, electrolytic capacitor, and load resistor respectively, and \(E\) is the input voltage;

[0024] Establish a discrete power converter model with a mixture of unknown but bounded noise and Gaussian noise based on the models of the above Buck circuit operating mode 1 and operating mode 2:

[0025]

[0026] where represents the state vector of the power converter at time \(k\), and the state vector of the power converter at time \(k\) is a vector composed of the true inductor current and capacitor voltage; \(u(k)\) represents the input vector of the power converter at time \(k\), and the input vector is the input voltage of the power converter, represents the output vector of the power converter at time \(k\), and the output vector of the power converter at time \(k\) is a vector composed of the inductor current and capacitor voltage actually measured by the current sensor and voltage sensor respectively;

[0027] \(\sigma(k)\in\{1,2\}\) represents the switching mode in which the power converter system is currently in, and the power converter switches between two modes; \(A\) σ(k) represents the state space matrix of the system in the \(\sigma(k)\) mode, \(B\) σ(k) represents the input matrix of the system in the \(\sigma(k)\) mode, \(C\) σ(k) represents the output matrix of the system in the \(\sigma(k)\) mode, represents the actuator fault, \(w\) k \(\in\langle0,W\rangle\) represents the unknown but bounded disturbance noise at time \(k\), represents the disturbance noise that follows a Gaussian distribution at time \(k\), \(v\) k \(\in\langle0,V\rangle\) represents the unknown but bounded measurement noise at time \(k\), represents the measurement noise that follows a Gaussian distribution at time \(k\).

[0028] Optionally, step two includes:

[0029] 2.1 Construct a power converter state estimator to obtain the state estimate of the power converter at time k:

[0030]

[0031] Wherein, represents the state estimate at time k, L(k) represents the estimator gain at time k, represents the state prediction at time k;

[0032] Determine the state prediction at time k according to Equation (1)

[0033]

[0034] Determine the prediction error at time k according to Equation (1) and Equation (3)

[0035]

[0036] Wherein, Δ(k|k-1) GD represents the prediction error of the Gaussian distribution part at time k The covariance matrix P(k|k-1) is:

[0037]

[0038] Wherein, P(k-1) is the covariance matrix of the estimation error of the Gaussian distribution at time k-1, and Q is the covariance matrix of the perturbation noise of the Gaussian distribution;

[0039] Δ(k|k-1) UBB represents the prediction error of the unknown but bounded part at time k, which is represented by the fully symmetric polytope <0, H(k|k-1)>, and the generation matrix H(k|k-1) is:

[0040] H(k|k-1) = [A σ(k) H(k-1) W] (8)

[0041] Wherein, H(k-1) is the estimation error of the unknown but bounded part at time k-1, and W is the generation matrix of the unknown but bounded perturbation noise;

[0042] Determine the estimation error at time k according to Equation (1) and Equation (2)

[0043]

[0044] Wherein, Δ(k) GDDenote the Gaussian distribution at time k The estimation error of the part, and the covariance matrix P(k) is:

[0045] P(k) = (I - L(k)C σ(k) )P(k|k - 1)(I - L(k)C σ(k) ) T + L(k)RL T (k) (10)

[0046] where I is the identity matrix, and R is the covariance matrix of the measurement noise that follows the Gaussian distribution;

[0047] Δ(k) UBB Denote the estimation error of the unknown but bounded part at time k, which is represented by the fully symmetric polytope <0, H(k)>, and the generation matrix H(k) is:

[0048] H(k) = [(I - L(k)C σ(k) )H(k|k - 1) L(k)V] (11)

[0049] where V is the generation matrix of the unknown but bounded measurement noise;

[0050] According to (10) and (11), the optimal criterion is:

[0051] J = (1 - η(k))tr(P(k)) + η(k)tr(H(k)H T (k)) (12)

[0052] where tr(P(k)) represents the trace of P(k), and η(k) represents the optimal weight coefficient at time k:

[0053]

[0054] Solve equation (13) to obtain the state estimator gain L(k):

[0055]

[0056] where M(k) = [H(k|k - 1) 0], a T (k) = [-C σ(k) H(k|k - 1) V];

[0057] 2.2 Determine the state estimation interval of the power converter at time k:

[0058]

[0059] where, and respectively represent the upper and lower bounds of the estimation error interval of the unknown but bounded part, and respectively represent the upper and lower bounds of the confidence interval of the estimation error that follows a Gaussian distribution.

[0060] Optionally, step three includes:

[0061] Determine the estimated output at time k based on the state estimate value of the power converter at time k:

[0062]

[0063] Obtain the output estimation interval at time k based on the estimated output at time k:

[0064]

[0065] where and respectively represent the upper and lower bounds of the estimated output, and v represents the boundary of the measurement noise v(k).

[0066] Optionally, step four includes:

[0067] Judge whether the measured output y of the power converter at time k is within the output estimation interval at time k according to whether Equation (18) holds; k

[0068]

[0069] If Equation (18) holds, it is determined that the operating mode of the power converter at time k has not changed, and jump to step eight; otherwise, go to step five.

[0070] Optionally, step five includes:

[0071] Obtain the output intervals of all operating modes of the power converter at time k:

[0072]

[0073] where m ∈ {1, 2}, and represent the upper and lower bounds of the output in mode m of the power converter, and m = 1 or 2;

[0074] Judge whether the measured output y of the power converter at time k is within the output intervals of all operating modes of the power converter at time k according to Equation (20); k

[0075]

[0076] If Equation (20) holds, it is determined that the power converter has changed at time k and jump to step six; ​​

[0077] Otherwise, it is determined that the power converter fails at time k, and jumps to Step 7.

[0078] Optionally, Step 7 includes:

[0079] 7.1 Expand the actuator fault in the power converter to the state vector of the power converter, and obtain the augmented power converter model according to Equation (1):

[0080]

[0081] where represents the unknown but bounded disturbance noise vector of the augmented power converter system at time k, represents the disturbance noise vector of the augmented power converter system at time k that follows a Gaussian distribution;

[0082] Construct a power converter fault estimator according to the following formula:

[0083]

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091] where is the generating matrix of

[0092] 7.2 Obtain the fault estimation interval of the power converter at time k:

[0093]

[0094] where f(k) - and f(k) + are the upper and lower bounds of the fault estimation, determined according to Equation (30) and

[0095]

[0096] Optionally, the fault is an actuator fault, including a circuit fault and a switch fault.

[0097] The present invention also provides a fault diagnosis system for a power converter under multi-noise mixing and multi-operation modes. The fault diagnosis system is provided with a voltage sensor and a current sensor to measure the circuit current and the inductor voltage; the fault diagnosis system uses the above-mentioned fault diagnosis method for fault diagnosis.

[0098] The beneficial effects of the present invention are:

[0099] In view of the actuator fault of the power converter, the present application establishes a state space model of the power converter, designs a state estimator of the power converter by minimizing the mean square error and the polytope space. When the operation mode of the power converter changes, first, it is determined whether the operation mode of the power converter changes by detecting whether the measured output of the system is within the output estimation interval. If it is detected that the measured output is not within the output estimation interval, further distinguish between system faults and system operation mode changes by determining whether the measured output is within the output intervals of all operation modes of the power converter. If the system fails, construct a fault estimator of the power converter to estimate the system fault. Compared with the existing fault diagnosis method that combines random and unknown but bounded hybrid noise double filtering, the method of the present application solves the problem of identifying the operation state and fault diagnosis of the power converter under different operation modes. Description of the Drawings

[0100] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0101] Figure 1 It is a flowchart of a fault diagnosis method for a power converter under multi-noise mixing and multi-operation modes disclosed in an embodiment of the present invention.

[0102] Figure 2 It is a topological structure diagram of a power converter provided in an embodiment of the present invention.

[0103] Figure 3 It is a simulation diagram of the upper and lower bounds of the inductor current of the Buck converter estimated by the fault diagnosis method provided in an embodiment of the present invention and the true state value when the converter is fault-free.

[0104] Figure 4 It is a simulation diagram of the upper and lower bounds of the capacitor voltage of the Buck converter estimated by the fault diagnosis method provided in an embodiment of the present invention and the true state value when the converter is fault-free.

[0105] Figure 5 It is a simulation diagram of the upper and lower bounds corresponding to a fault diagnosis method provided by an embodiment of the present invention and the true parameter values when a fault f1 occurs in the converter.

[0106] Figure 6 It is a simulation diagram of the upper and lower bounds corresponding to a fault diagnosis method provided by an embodiment of the present invention and the true parameter values when a fault f2 occurs in the converter. Detailed implementation manners

[0107] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0108] Embodiment 1:

[0109] This embodiment provides a fault diagnosis method for a power converter under multi-noise mixing and multi-operation modes, which is mainly used for diagnosing actuator faults of the power converter, including circuit faults and switch faults. Refer to Figure 1 The method includes:

[0110] Step 1: For a power converter system with mixed unknown but bounded noise and Gaussian noise, discrete power converter models are established for its different working modes. The power converter system has two working modes, which are respectively denoted as working mode 1 and working mode 2;

[0111] The two working modes enable the power converter system to achieve the working effects of boosting or bucking.

[0112] Step 2: Design a power converter state estimator to obtain the state estimation value of the power converter at time k, and determine the state estimation interval of the power converter at time k according to the state estimation value;

[0113] In this application, the power converter state estimator is designed by minimizing the mean square error and the polytope space;

[0114] Step 3: According to the state estimation value of the power converter at time k, obtain the output estimation interval at time k;

[0115] Step 4: Measure the output of the power converter at time k through a voltage sensor and a current sensor, denoted as the measured output y k , and determine whether the working mode of the power converter has changed according to whether the measured output y k is within the output estimation interval at time k;

[0116] If the measured output y k is within the output estimation interval, it is determined that the working mode of the power converter at time k has not changed, and jump to step 8; otherwise, go to step 5;

[0117] Step 5: Determine the measured output y of the power converter at time k k whether it is within the output range of all operating modes of the power converter at time k:

[0118] If y k is not within the output range of all operating modes of the power converter at time k, it is determined that the power converter has a fault at time k, and jump to Step 7;

[0119] If y k is within the output range of a certain operating mode of the power converter at time k, it is determined that the operating mode of the power converter has changed at time k, and jump to Step 6;

[0120] In this step, the output ranges of all operating modes of the power converter can be measured in advance by a voltage sensor and a current sensor;

[0121] Step 6: Initialize the state estimator at time k - 1, obtain the state estimator and the state estimation range at time k, and jump to Step 8;

[0122] Step 7: Construct a fault estimator for the power converter and obtain the fault estimation range of the power converter at time k;

[0123] Step 8: Set k = k + 1 and jump to Step 2; until the operation of the power converter ends;

[0124] Loop through Steps 2 to 8 above to complete the fault diagnosis process of the power converter without interruption.

[0125] Embodiment 2:

[0126] This embodiment provides a method for fault diagnosis of a power converter under multi-noise mixing and multi-operation modes. This embodiment is introduced by taking a Buck converter as an example. The method includes:

[0127] Step 1: For a power converter system with mixed unknown but bounded noise and Gaussian noise, establish discrete power converter models for its different operating modes. The power converter system has two operating modes, denoted as operating mode 1 and operating mode 2 respectively;

[0128] Establish the model of the Buck circuit under operating mode 1:

[0129]

[0130] Establish the model of the Buck circuit under operating mode 2:

[0131]

[0132] where i is the inductor current, uc is the capacitor voltage, L, C, and r are the inductor, electrolytic capacitor, and load resistor respectively, and E is the input voltage;

[0133] Based on the models of the above-mentioned Buck circuit operating modes 1 and 2, a discrete power converter model with a mixture of unknown but bounded noise and Gaussian noise is established:

[0134]

[0135] where, represents the state vector of the power converter at time k, and the state vector of the power converter at time k is a vector composed of the actual inductor current and capacitor voltage; u(k) represents the input vector of the power converter at time k, and the input vector is the input voltage of the power converter, represents the output vector of the power converter at time k, and the output vector of the power converter at time k is a vector composed of the inductor current and capacitor voltage actually measured by the current sensor and voltage sensor respectively;

[0136] σ(k) ∈ {1, 2} represents the switching mode in which the power converter system is currently in, and the power converter switches between two modes; A σ(k) represents the state space matrix of the system in the σ(k) mode, B σ(k) represents the input matrix of the system in the σ(k) mode, C σ(k) represents the output matrix of the system in the σ(k) mode, represents the actuator fault, w k ∈ <0, W> represents the unknown but bounded disturbance noise at time k, represents the disturbance noise that follows a Gaussian distribution at time k, v k ∈ <0, V> represents the unknown but bounded measurement noise at time k, represents the measurement noise that follows a Gaussian distribution at time k.

[0137] Step 2: Design a state estimator for the power converter to obtain the state estimate value of the power converter at time k, and determine the state estimate interval of the power converter at time k according to the state estimate value:

[0138] 2.1 Construct a state estimator for the power converter to obtain the state estimate value of the power converter at time k:

[0139]

[0140] where, represents the state estimate value at time k, L(k) represents the estimator gain at time k, represents the state prediction value at time k;

[0141] According to formula (1), the state prediction value at time k is determined

[0142]

[0143] According to equations (1) and (3), the prediction error at time k is determined

[0144]

[0145] Among them, Δ(k|k-1) GD Indicates that k moment follows Gaussian distribution The partial prediction error, covariance matrix P(k|k-1) is:

[0146]

[0147] Among them, P(k-1) is the estimation error covariance matrix that obeys Gaussian distribution at time k-1, and Q is the covariance matrix of disturbance noise that obeys Gaussian distribution;

[0148] Δ(k|k-1) UBB It represents the prediction error of the unknown but bounded part at time k, which is represented by the fully symmetric polytope <0,H(k|k-1)>, and the generating matrix H(k|k-1) is:

[0149] H(k|k-1)=[A σ(k) H(k-1)W] (8)

[0150] Among them, H(k-1) is the estimated error of the unknown but bounded part at time k-1, and W is the generator matrix of the unknown but bounded disturbance noise;

[0151] According to equations (1) and (2), the estimated error at time k is determined

[0152]

[0153] Where Δ(k) GD Indicates that k moment follows Gaussian distribution The estimation error of the part, the covariance matrix P(k) is:

[0154] P(k)=(IL(k)C σ(k) )P(k|k-1)(IL(k)C σ(k) ) T +L(k)RL T (k) (10)

[0155] Where I is the identity matrix, R is the measurement noise covariance matrix that obeys the Gaussian distribution;

[0156] Δ(k) UBB represents the estimation error of the unknown but bounded part at time k, which is represented by the fully symmetric polytope <0, H(k)>. The generation matrix H(k) is as follows:

[0157] H(k) = [(I - L(k)C σ(k) )H(k|k - 1)L(k)V] (11)

[0158] where V is the generation matrix of the unknown but bounded measurement noise;

[0159] According to (10) and (11), the optimal criterion is obtained as:

[0160] J = (1 - η(k))tr(P(k)) + η(k)tr(H(k)H T (k)) (12)

[0161] where tr(P(k)) represents the trace of P(k), and η(k) represents the optimal weight coefficient at time k:

[0162]

[0163] Solving equation (12) gives the state estimator gain L(k):

[0164]

[0165] where M(k) = [H(k|k - 1) 0], a T (k) = [-C σ(k) H(k|k - 1)V];

[0166] 2.2 Determine the state estimation interval of the power converter at time k:

[0167]

[0168] where, and represent the upper and lower bounds of the estimation error interval of the unknown but bounded part respectively, and represent the upper and lower bounds of the confidence interval of the estimation error that follows a Gaussian distribution respectively.

[0169] Step 3: Obtain the output estimation interval at time k according to the state estimation value of the power converter at time k;

[0170] Determine the estimated output at time k according to the state estimation value of the power converter at time k:

[0171]

[0172] Obtain the output estimation interval at time k according to the estimated output at time k:

[0173]

[0174] Wherein, and respectively represent the upper and lower bounds of the estimated output, and v represents the boundary of the measurement noise v(k).

[0175] Step 4: Judge whether the measured output y of the power converter at time k k is within the output estimation interval at time k;

[0176]

[0177] If equation (18) holds, it is determined that the operating mode of the power converter at time k has not changed, and jump to Step 8; otherwise, go to Step 5;

[0178] Step 5: Judge whether the measured output y of the power converter at time k k is within the output intervals of all operating modes of the power converter at time k:

[0179] If y k is not within the output intervals of all operating modes of the power converter at time k, it is determined that the power converter has a fault at time k and jump to Step 7;

[0180] If y k is within the output interval of a certain operating mode of the power converter at time k, it is determined that the operating mode of the power converter has changed at time k and jump to Step 6;

[0181] Obtain the output intervals of all operating modes of the power converter at time k:

[0182]

[0183] Wherein, m ∈ {1, 2}, and represent the upper and lower bounds of the output in mode m of the power converter; mode m represents operating mode 1 or operating mode 2.

[0184] Judge whether the measured output y of the power converter at time k according to equation (20) k is within the output intervals of all operating modes of the power converter at time k;

[0185]

[0186] If equation (20) holds, it is determined that the power converter has changed at time k and jump to Step 6;

[0187] Otherwise, it is determined that a fault occurs in the power converter at time k, and the process jumps to Step Seven.

[0188] The output intervals of all working modes are measured in advance by voltage sensors and current sensors.

[0189] Step Six: Initialize the state estimator at time k-1, obtain the state estimator and the state estimation interval at time k, and jump to Step Eight;

[0190] Step Seven: Construct a power converter fault estimator to obtain the fault estimation interval of the power converter at time k:

[0191] 7.1 Expand the actuator fault in the power converter to the state vector of the power converter, and obtain the augmented power converter model according to Equation (1):

[0192]

[0193] where represents the unknown but bounded disturbance noise vector of the augmented power converter system at time k, represents the disturbance noise vector of the augmented power converter system at time k that follows a Gaussian distribution;

[0194] Construct the power converter fault estimator according to the following equation:

[0195]

[0196]

[0197]

[0198]

[0199]

[0200]

[0201]

[0202]

[0203] where is the generating matrix of

[0204] 7.2 Obtain the fault estimation interval of the power converter at time k:

[0205]

[0206] where f(k)- and f(k) + are the upper and lower bounds of the fault estimation, which are determined according to Equation (30) and

[0207]

[0208] Step Eight: Set k = k + 1, and jump to Step Two; until the operation of the power converter ends;

[0209] Loop through the above Steps Two to Eight to complete the fault diagnosis process of the power converter without interruption.

[0210] To verify the accuracy and rapidity of the power converter fault diagnosis method under multi-noise mixing and multi-operation modes proposed in this application, the following simulation experiments are carried out:

[0211] Set that when k ∈ {1, 10} and k ∈ {21, 30}, the Buck converter operates in Mode 1; set that when k ∈ {11, 20} and k ∈ {31, 40}, the Buck converter operates in Mode 2; set that when k ∈ {32, 50}, the Buck converter has a fault.

[0212] Figure 3 and Figure 4 respectively show the change situations of the upper and lower bounds of the estimation of the inductor current and capacitor voltage and the change situations of the true values when the Buck converter has no fault. From Figure 3 and Figure 4 it can be seen that the power converter state recognition and fault diagnosis technology based on multi-cell hybrid filtering in this application can well estimate the operation state of the power converter.

[0213] Figure 5 and Figure 6 respectively show the change situations of the upper and lower bounds of the fault estimation and the change situations of the true values when the Buck converter has a fault. From Figure 3 and Figure 4 it can be seen that the power converter fault diagnosis method under multi-noise mixing and multi-operation modes in this application can effectively identify faults.

[0214] Some steps in the embodiments of the present invention can be implemented by software, and the corresponding software program can be stored in a readable storage medium, such as an optical disc or a hard disk, etc.

[0215] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A fault diagnosis method for a power converter under multi-noise mixing and multi-operation modes, characterized in that, The method includes: Step 1: For a power converter system with a mixture of unknown but bounded noise and Gaussian noise, discrete power converter models are established for its different operating modes. The power converter system has two operating modes, denoted as operating mode 1 and operating mode 2 respectively; Step 2: Design a power converter state estimator to obtain the state estimate value of the power converter at time k, and determine the state estimate interval of the power converter at time k according to the state estimate value; Step 3: According to the state estimate value of the power converter at time k, obtain the output estimate interval at time k; Step 4: Measure the output of the power converter at time k through the voltage sensor and current sensor, denoted as the measured output y k , and based on the measured output y k judge whether the operating mode of the power converter has changed according to whether it is within the output estimation interval at time k; If the measured output y k is within the output estimation interval, it is determined that the operating mode of the power converter at time k has not changed, and the process jumps to Step 8; otherwise, it goes to Step 5. Step Five: Determine the measured output y of the power converter at time k k whether it is within the output range of all operating modes of the power converter at time k: If y k is not within the output range of all operating modes of the power converter at time k, it is determined that a fault has occurred in the power converter at time k, and the process jumps to Step Seven; If y k is within the output range of a certain operating mode of the power converter at time k, it is determined that the operating mode of the power converter changes at time k, and the process jumps to Step Six; Step 6: Initialize the state estimator at time k - 1, obtain the state estimator and state estimate interval at time k and jump to Step 8; Step 7: Construct a power converter fault estimator to obtain the fault estimate interval of the power converter at time k; Step 8: Set k = k + 1, and jump to Step 2; until the power converter operation ends; Loop through Steps 2 to 8 above to complete the fault diagnosis process of the power converter without interruption; Establish a discrete power converter model with a mixture of unknown but bounded noise and Gaussian noise according to the models of Buck circuit operating mode 1 and operating mode 2: Among them, represents the state vector of the power converter at time k. The state vector of the power converter at time k is a vector composed of the true inductor current and capacitor voltage; u(k) represents the input vector of the power converter at time k, and the input vector is the input voltage of the power converter, represents the output vector of the power converter at time k. The output vector of the power converter at time k is a vector composed of the inductor current and capacitor voltage actually measured by the current sensor and voltage sensor respectively; σ(k) ∈ {1, 2} represents the switching mode in which the power converter system is currently operating. The power converter switches between two modes; A σ(k) represents the state - space matrix of the system in the σ(k) mode, B σ(k) represents the input matrix of the system in the σ(k) mode, C σ(k) represents the output matrix of the system in the σ(k) mode, represents the actuator fault, w k ∈ <0, W> represents the unknown but bounded disturbance noise at time k, represents the disturbance noise that follows a Gaussian distribution at time k, v k ∈ <0, V> represents the unknown but bounded measurement noise at time k, represents the measurement noise that follows a Gaussian distribution at time k.

2. The method according to claim 1, wherein The said Step 1 includes: Establish a model for Buck circuit operating mode 1: Establish a model for Buck circuit operating mode 2: where \(i\) is the inductor current, \(u\) c is the capacitor voltage, \(L\), \(C\), and \(r\) are the inductor, electrolytic capacitor, and load resistance respectively, and \(E\) is the input voltage.

3. The method according to claim 2, wherein The said Step 2 includes: 2.1 Construct a power converter state estimator to obtain the state estimate value of the power converter at time k: Among them, represents the state estimation value at time k, and L(k) represents the estimator gain at time k, represents the state prediction value at time k; Determine the state prediction value at time k according to Equation (1). Determine the prediction error at time k according to Equation (1) and Equation (3). where, Δ(k|k - 1) GD represents the prediction error of the Gaussian distribution part at time k, and the covariance matrix P(k|k - 1) is as follows: ​ where P(k - 1) is the estimated error covariance matrix obeying Gaussian distribution at time k - 1, and Q is the covariance matrix of the perturbation noise obeying Gaussian distribution; Δ(k|k - 1) UBB represents the prediction error of the unknown but bounded part at time k, which is represented by the fully symmetric polytope <0, H(k|k - 1)>. The generation matrix H(k|k - 1) is as follows: H(k|k - 1) = [A σ(k) H(k - 1)W] (8) where H(k - 1) is the estimated error of the unknown but bounded part at time k - 1, and W is the generation matrix of the unknown but bounded perturbation noise; Determine the estimation error at time k according to Equation (1) and Equation (2) where, Δ(k) GD represents the estimation error of the Gaussian distribution part at time k and the covariance matrix P(k) is:[[]]END]] P(k) = (I - L(k)C σ(k) )P(k|k - 1)(I - L(k)C σ(k) ) T + L(k)RL T (k)(10) where I is the identity matrix, and R is the measurement noise covariance matrix obeying Gaussian distribution; Δ(k) UBB represents the estimation error of the unknown but bounded part at time k, which is represented by the fully symmetric polytope <0, H(k)>. The generation matrix H(k) is as follows: H(k) = [(I - L(k)C σ(k) )H(k|k - 1)L(k)V ] (11) where V is the generation matrix of the unknown but bounded measurement noise; The optimal criterion is obtained according to (10) and (11): J = (1 - η(k))tr(P(k)) + η(k)tr(H(k)H T (k)) (12) where tr(P(k)) represents the trace of P(k), and η(k) represents the optimal weight coefficient at time k: Solve equation (13) to obtain the state estimator gain L(k): where M(k) = [H(k|k - 1) 0], a T (k) = [-C σ(k) H(k|k - 1) V ]; 2.2 Determine the state estimate interval of the power converter at time k: wherein, and respectively represent the upper and lower bounds of the estimated error interval of the unknown but bounded part, and respectively represent the upper and lower bounds of the confidence interval of the estimated error that follows a Gaussian distribution.

4. The method according to claim 3, characterized in that The said Step 3 includes: According to the state estimate value of the power converter at time k, determine the estimated output at time k: According to the estimated output at time k, obtain the output estimate interval at time k: where, and represent the upper and lower bounds of the estimated output respectively, and v represents the bound of the measurement noise v(k).

5. The method according to claim 4, wherein The said Step 4 includes: Determine whether the measured output y of the power converter at time k based on whether Equation (18) holds k is within the output estimation interval at time k; If equation (18) holds, it is determined that the operating mode of the power converter has not changed at time k, and jump to Step 8; otherwise, go to Step 5.

6. The method according to claim 5, characterized in that, The said Step 5 includes: Obtain the output intervals of all operating modes of the power converter at time k: where m ∈ {1, 2}, and represent the upper and lower bounds of the output in the m-th mode of the power converter, where m = 1 or 2; Determine the measured output y of the power converter at time k according to Equation (20). k Whether it is within the output range of all operating modes of the power converter at time k; If equation (20) holds, it is determined that the power converter has changed at time k and jump to Step 6; Conversely, it is determined that the power converter has failed at time k, and jump to Step 7.

7. The method according to claim 6, characterized in that, The said Step 7 includes: 7.1 Extend the actuator fault in the power converter to the state vector of the power converter, and obtain the augmented power converter model according to equation (1): wherein, represents the unknown but bounded disturbance noise vector of the augmented power converter system at time k, represents the disturbance noise vector of the augmented power converter system at time k that follows a Gaussian distribution; Construct a power converter fault estimator according to the following formula: Among them, is the generating matrix of 7.2 Obtain the fault estimation interval of the power converter at time k: where f(k) - and f(k) + are the upper and lower bounds of the fault estimation, determined according to Equation (30) and 8. The method according to claim 7, characterized in that The fault is an actuator fault, including circuit faults and switch faults.

9. A fault diagnosis system for a power converter under multi-noise mixing and multi-operation modes, characterized in that, The fault diagnosis system is provided with a voltage sensor and a current sensor to measure the circuit current and the inductor voltage; the fault diagnosis system uses the fault diagnosis method according to any one of claims 1-8 for fault diagnosis.

Images