A filter fault diagnosis method for power converters in multiple operating modes

Abstract

This invention discloses a filtering fault diagnosis method for power converters operating under multiple modes, belonging to the field of fault diagnosis technology. This method applies a polytopic set-membership filtering approach to encapsulate feasible parameter sets. When system parameters change, it first determines whether the power converter's operating mode has changed by detecting the compatibility between the feasible parameter set and the data samples. If incompatibility is detected, it further distinguishes between system faults and system operating mode changes by judging the consistency between the data samples and all operating modes of the power converter. If a system fault occurs, model matching is performed to determine the fault type. If no fault type is matched, it indicates that a new fault type has occurred in the power converter, and the new fault type is added to the fault database. Compared to existing fault diagnosis methods based on fully symmetric polytopes, this method solves the problem of identifying and diagnosing the operating status of power converters with multiple operating modes.

Description

Technical Field

[0001] The invention relates to a filter fault diagnosis method for a power converter in multiple operation modes, belonging to the technical field of fault diagnosis. Background Art

[0002] With the continuous development of power electronics technology, power converters are playing an important role in more and more fields, such as energy, agriculture, etc. At the same time, inductor failure and capacitor failure are common faults of power converters, which can easily lead to failure of converter boost or buck. Therefore, in order to ensure the safe and reliable operation of power converters, it is very necessary to perform online estimation of power converter parameters and real-time fault diagnosis.

[0003] The working environment of the power converter is complex and it is easily affected by various noise interferences during operation. Most of the current fault diagnosis methods based on the power converter model assume that the noise and interference conform to a certain probability distribution, such as normal distribution, Poisson distribution, etc., to achieve fault diagnosis of the power converter. However, in practical applications, the statistical characteristics of noise and interference in the power converter system are difficult to obtain, that is, the actual noise and interference may not conform to the assumed probability distribution. Therefore, the results obtained by this type of fault diagnosis method based on assumptions are not accurate. Compared with the above-mentioned fault diagnosis method based on the power converter model, the fault diagnosis method based on set membership filtering only requires that the noise satisfies the unknown but bounded condition, that is, it does not require that the noise and interference meet a certain probability distribution, so its fault diagnosis accuracy is improved. Among the existing fault diagnosis methods based on set membership filtering, the more commonly used methods are: set membership filtering method based on fully symmetric polytopes, set membership filtering method based on intervals, and method based on ellipsoids. Among them, the set membership filtering method based on fully symmetric polytopes can transform the solution of polyhedral space into simple matrix operations. Compared with the methods based on intervals and ellipsoids, it has less computational complexity and less conservative estimation results, making the fault diagnosis results more accurate and faster.

[0004] The existing fault diagnosis method based on fully symmetric polytopes is mainly aimed at power converters with a single operating mode. It determines whether the system is faulty by judging whether the intersection of the fully symmetric polytope that encapsulates the system parameters at time k-1 and the strip space formed by the system input and output data at time k is an empty set. However, in some application scenarios, such as the formation production link of batteries, in order to achieve different formation effects, that is, to achieve different boost effects, power converters with multiple operating modes are required. For power converters with multiple operating modes, whether a fault occurs or the operating mode changes, the intersection of the fully symmetric polytope that encapsulates the system parameters at time k-1 and the strip space formed by the system input and output data at time k is an empty set. Therefore, the existing fault diagnosis method based on fully symmetric polytopes cannot distinguish whether a power converter with multiple operating modes has a fault or a change in operating mode. Summary of the invention

[0005] In order to solve the problem that the existing fault diagnosis method cannot determine whether a fault occurs or the operation mode changes when performing fault diagnosis on a power converter with multiple operation modes, the present invention provides a filter fault diagnosis method for a power converter in multiple operation modes, the method comprising:

[0006] Step 1: Establish a regression model of the power converter and obtain a fault library of the power converter;

[0007] Step 2: Set the fully symmetric polyhedron corresponding to the parameter vector to be estimated of the power converter at the initial moment And obtain the real parameter vector to be estimated for each operating mode The fully symmetric polytope of

[0008] Step 3: Collect data at time k {y k ,Φ k}, and according to the fully symmetric polytope at time k-1 With the data {y k ,Φ k} Whether it is compatible or not, determine whether the operation mode of the power converter has changed:

[0009] If they are compatible, it is determined that the operation mode of the power converter has not changed at time k, and the process jumps to step 4;

[0010] If not compatible, go to step five;

[0011] Step 4: Determine the fully symmetric polytope of the feasible parameter set of the parameters to be estimated at time k based on the power converter regression model established in step 1 Then jump to step nine;

[0012] Step 5: Determine the data {y at time kk ,Φ k}Consistency with all operating modes of the power converter:

[0013] If the data {y k ,Φ k} is inconsistent with all the operating modes of the power converter, it is determined that the power converter fails at time k and jumps to step 7;

[0014] If the data {y k ,Φ k} is consistent with a certain operation mode of the power converter, it is determined that the operation mode of the power converter at time k is changed and the process jumps to step 6;

[0015] Step 6: For the fully symmetric polytope at time k-1 Initialize and obtain the fully symmetric polytope of the feasible parameter set of the parameters to be estimated at time k And jump to step nine;

[0016] Step 7: Initialize the fully symmetric polytope Identify the fault type through model matching and determine whether the fault is a known fault in the fault library;

[0017] Step 8: If it is determined in step 7 that a new fault type occurs in the power converter, the fault type is added to the fault library and the process goes to step 9;

[0018] Step 9: Set k=k+1, update the corresponding data information, and jump to step 3 until k>N, where N is the predetermined data length;

[0019] The above steps 3 to 9 are executed cyclically to complete the state identification and fault diagnosis process of the power converter without interruption.

[0020] Optionally, the regression model of the power converter established in step 1 is:

[0021]

[0022] Among them, θ σ(k) =[a σ(k),1 ,a σ(k),2 ,b σ(k),1 ,b σ(k),2 ] T represents the parameter vector to be estimated for the σ(k)th operating mode of the power converter, e k is the system noise, |e k |≤δ, δ represents the known system noise boundary; y k represents the output data at time k, that is, the output voltage at time k; u krepresents the input data at time k, that is, the input voltage at time k;

[0023] Parameter to be estimated a σ(k),1 ,a σ(k),2 ,b σ(k),1 ,b σ(k),2 They are:

[0024]

[0025] in, T s is the sampling time; L, C, and R are the inductance, electrolytic capacitor, and load resistance in the equivalent circuit of the power converter, respectively.

[0026] Optionally, it is characterized in that the step 2 comprises:

[0027] 2.1 Set the fully symmetric polyhedron corresponding to the parameter vector to be estimated of the power converter at the initial moment

[0028] For the power converter at time k, its operating mode is σ(k), and the fully symmetric polyhedron at time k is defined as:

[0029]

[0030] Among them, p k,σ(k) is the fully symmetric polyhedron corresponding to the parameter vector to be estimated at time k The center of is the corresponding generator matrix, B is the unit interval, that is, [-1,1], B r is the r-dimensional box space composed of unit intervals;

[0031] 2.2 Obtain the actual estimated parameter vector of the power converter in each operating mode The fully symmetric polytope of

[0032] is the actual parameter vector to be estimated when the power converter operates in the σ(k) mode, and the package is defined as The fully symmetric polytope is:

[0033]

[0034] p σ(k) is the fully symmetric polytope of the real parameter vector to be estimated in the σ(k) mode The center of is the corresponding generator matrix; g σ(k),i is the corresponding generator.

[0035] Optionally, the step three includes:

[0036] According to whether formula (6) is valid, we can judge the fully symmetric polyhedron at time k-1. With the data {y k ,Φ k}Is it compatible?

[0037]

[0038] If formula (6) holds true, it is determined that the operation mode of the power converter has not changed at time k, and the process jumps to step 4; otherwise, the process jumps to step 5.

[0039] Optionally, the step 4 includes:

[0040] After determining that the operation mode of the power converter has not changed at time k, update the fully symmetric polytope at time k.

[0041] Fully symmetric polytope at time k Update according to the following formula:

[0042]

[0043] Optionally, the step five includes:

[0044] According to formula (8), the data {y k ,Φ k}Consistency with all operating modes of the power converter;

[0045] If it is detected that formula (8) holds true, it is determined that the operation mode of the power converter has changed at time k, and the process jumps to step 6;

[0046]

[0047] On the contrary, if it is detected that the following inequality (9) holds, it is determined that the power converter fails at time k, and the process goes to step 7;

[0048]

[0049] Where ‖·‖1 represents the 1-norm.

[0050] Optionally, the model matching method in step seven includes:

[0051] Define the k-time parameter estimation sequence Parameter estimation error at time k When the parameter estimation error Δθ k If l is less than a certain threshold ε within a certain time length, the fault is separated, the fault identification time is recorded and the fault type is displayed by the filter;

[0052] Define the number of fault types as M, and the j-th fault type at time k satisfies:

[0053]

[0054] in, F M is the fault type in the fault library, and fault identification filters f1, f2, …, f are designed for each of the M fault types. M , when the system fails, if the jth fault identification filter is not empty, it means that the current fault type matches the type in the fault library, that is, the fault type at this time is determined to be F j ;

[0055] If the current fault type does not match the type in the fault library, the fault is determined to be a new fault and is added to the fault library.

[0056] Optionally, the fault types in the fault library obtained in step 1 include inductor failure and capacitor failure.

[0057] Optionally, the regression model of the power converter established in step 1 includes:

[0058] According to the working principle of the power converter, the circuit model of the power converter is established as follows:

[0059]

[0060] Where i is the inductor current, u o is the output voltage;

[0061] The circuit model of the power converter is approximated as a second-order system:

[0062]

[0063] Where s is a complex variable;

[0064] Take the sampling time T s =0.01, and the formula (2) is discretized by bilinear transformation to obtain:

[0065]

[0066] Let b σ(k),0 =1, and taking into account the system noise e k ,get:

[0067] A σ(k) (z)y k =B σ(k) (z)u k +e k (4)

[0068] Among them, A σ(k) (z) = 1 + a σ(k),1 z -1 +a σ(k),2 z -2 , B σ(k) (z) = 1 + b σ(k),1 z -1 +b σ(k),2 z -2 .

[0069] Optionally, the operating mode of the power converter at time k is σ(k)∈{1,2,…S}, where S is the number of operating modes of the power converter.

[0070] The beneficial effects of the present invention are:

[0071] The present application proposes a filter fault detection method for a power converter with multiple operating modes, including obtaining a fault library of the power converter, establishing a regression model of the power converter, and using a polytope set member filtering method to envelop a feasible parameter set. When the system parameters change, firstly, by detecting whether the parameter feasible set is compatible with the data sample, it is determined whether the power converter operating mode has changed. If it is detected that the parameter feasible set is incompatible with the data sample, the consistency of the data sample with all the operating modes of the power converter is further determined to distinguish between system faults and system operating mode changes. If a system fault occurs, model matching is performed to determine the fault type. If no fault type is matched within a certain period of time, it means that a new fault type has occurred in the power converter, and the new fault type is added to the fault library. Compared with the existing fault diagnosis method based on fully symmetric polytopes, the method of the present application solves the problem of operating state identification and fault diagnosis of power converters with multiple operating modes. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. 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 these drawings without creative work.

[0073] Figure 1 It is a flow chart of a filter fault diagnosis method for a power converter under multiple operating modes provided in an example of the present invention.

[0074] Figure 2 This is a topological structure diagram of a power converter provided by an embodiment of the present invention.

[0075] Figure 3is the Boost converter parameter b estimated by the fault diagnosis method provided by an embodiment of the present invention when the converter has no fault σ(k),1 Simulation plots of the corresponding upper and lower bounds and true parameter values.

[0076] Figure 4 is the Boost converter parameter b estimated by the fault diagnosis method provided by an embodiment of the present invention when the converter has no fault σ(k),2 Simulation plots of the corresponding upper and lower bounds and true parameter values.

[0077] Figure 5 is the Boost converter parameter b estimated by the fault diagnosis method provided by an embodiment of the present invention when a converter fails. σ(k),1 Simulation plots of the corresponding upper and lower bounds and true parameter values; Figure 5 Two different parameter failures are assumed to occur in the time periods k∈{700,1000} and k∈{1800,2000}.

[0078] Figure 6 is the Boost converter parameter b estimated by the fault diagnosis method provided by an embodiment of the present invention when a converter fails. σ(k),2 Simulation plots of the corresponding upper and lower bounds and true parameter values; Figure 6 Two different parameter failures are assumed to occur in the time periods k∈{700,1000} and k∈{1800,2000}.

[0079] Figure 7 is the Boost converter parameter b estimated by the fault diagnosis method provided by an embodiment of the present invention when a converter fails. σ(k),2 The corresponding upper and lower bounds and simulation diagrams of the true parameter values, Figure 7 Two different parameter failures are assumed to occur in the time periods k∈{700,1000} and k∈{1800,2000}. DETAILED DESCRIPTION

[0080] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0081] Embodiment 1:

[0082] This embodiment provides a filter fault diagnosis method for a power converter in multiple operation modes. Figure 1 , the method comprising:

[0083] Step 1: Establish a regression model of the power converter and obtain a fault library of the power converter;

[0084] Step 2: Set the fully symmetric polyhedron corresponding to the parameter vector to be estimated of the power converter at the initial moment And obtain the real parameter vector to be estimated for each operating mode The fully symmetric polytope of

[0085] Step 3: Collect data at time k {y k ,Φ k}, and according to the fully symmetric polytope at time k-1 With the data {y k ,Φ k} Whether it is compatible or not, determine whether the operation mode of the power converter has changed:

[0086] If they are compatible, it is determined that the operation mode of the power converter has not changed at time k, and the process jumps to step 4;

[0087] If not compatible, go to step five;

[0088] Step 4: Determine the fully symmetric polytope of the feasible parameter set of the parameters to be estimated at time k based on the power converter regression model established in step 1 Then jump to step nine;

[0089] Step 5: Determine the data {y at time k k ,Φ k}Consistency with all operating modes of the power converter:

[0090] If the data {y k ,Φ k} is inconsistent with all the operating modes of the power converter, it is determined that the power converter fails at time k and jumps to step 7;

[0091] If the data {y k ,Φ k} is consistent with a certain operation mode of the power converter, it is determined that the operation mode of the power converter at time k is changed and the process jumps to step 6;

[0092] Step 6: For the fully symmetric polytope at time k-1 Initialize and obtain the fully symmetric polytope of the feasible parameter set of the parameters to be estimated at time k And jump to step nine;

[0093] Step 7: Initialize the fully symmetric polytope Identify the fault type through model matching and determine whether the fault is a known fault in the fault library;

[0094] Step 8: If it is determined in step 7 that a new fault type occurs in the power converter, the fault type is added to the fault library and the process goes to step 9;

[0095] Step 9: Set k=k+1, update the corresponding data information, and jump to step 3 until k>N, where N is the predetermined data length;

[0096] The above steps 3 to 9 are executed cyclically to complete the state identification and fault diagnosis process of the power converter without interruption.

[0097] Embodiment 2

[0098] This embodiment provides a filter fault diagnosis method for a power converter in multiple operating modes. This embodiment is described by taking a Boost converter as an example. The method includes:

[0099] Step 1: Establish a regression model of the power converter and obtain a fault library of the power converter;

[0100] 1.1 Establish a regression model for power converter;

[0101] Figure 2 The topology diagram of the Boost converter, where S is a fully controlled switch tube, D is a power diode, i is the inductor current, and u is o is the output voltage, L, C, R are inductance, electrolytic capacitor and load resistance respectively, and E is the input voltage.

[0102] When the switch S is disconnected, according to the working principle of the Boost converter, the diode will be turned on instantly, and the Boost circuit model is established as follows:

[0103]

[0104] Approximate the Boost circuit model as a second-order system:

[0105]

[0106] Among them, y=u o , u=E, s is a complex variable.

[0107] Take the sampling time T s =0.01, and the formula (2) is discretized by bilinear transformation to obtain:

[0108]

[0109] Among them, a σ(k),1 , a σ(k),2 , b σ(k),0 , b σ(k),1 , bσ(k),2 It is the parameter to be estimated obtained by discretizing the system model.

[0110] Use regularization method to deal with b σ(k),0 =1, and taking into account the system noise e k ,get:

[0111] A σ(k) (z)y k =B σ(k) (z)u k +e k (4)

[0112] Among them, A σ(k) (z) = 1 + a σ(k),1 z -1 +a σ(k),2 z -2 , B σ(k) (z) = 1 + b σ(k),1 z -1 +b σ(k),2 z -2 ,

[0113]

[0114] Therefore, the parameter to be estimated for the Boost converter is a σ(k),1 , a σ(k),2 , b σ(k),1 , b σ(k),2 , σ(k) represents the operating mode of the Boost converter at time k. Different inductors L can achieve different degrees of boost effect, making the operating mode of the Boost converter different. σ(k)∈{1,2,…S}, S is the number of operating modes of the Boost converter, u k represents the input data at time k, that is, the input voltage at time k, y k represents the output data at time k, that is, the output voltage at time k, and the system noise e k The probability distribution of |e is unknown but bounded, k |≤δ, where δ represents the known system noise boundary.

[0115] Convert the discretized model into a regression model as follows:

[0116]

[0117] Among them, θ σ(k) =[a σ(k),1 ,a σ(k),2 ,b σ(k),1 ,b σ(k),2 ] T represents the parameter vector to be estimated for the σ(k)th operating mode of the power converter,

[0118] 1.2 Obtain the fault library of the power converter;

[0119] Inductor failure and capacitor failure are common faults of Boost converters. The two fault parameters are shown in Table 1:

[0120] Table 1 Parameter description under fault conditions

[0121]

[0122] Step 2: Set the fully symmetric polyhedron corresponding to the parameter vector to be estimated of the power converter at the initial moment And obtain the real parameter vector to be estimated for each operating mode The fully symmetric polytope of

[0123] 2.1 Set the fully symmetric polyhedron corresponding to the parameter vector to be estimated of the power converter at the initial moment

[0124] For the power converter at time k, its operating mode is σ(k), and the fully symmetric polyhedron at time k is defined as:

[0125]

[0126] p k,σ(k) is the fully symmetric polyhedron corresponding to the parameter vector to be estimated at time k The center of is the generating matrix, B is the unit interval, that is, [-1,1], B r is the r-dimensional box space consisting of unit intervals.

[0127] 2.2 Obtain the real parameter vector of the power converter for each operating mode The fully symmetric polytope of

[0128] is the real parameter vector of the power converter operating in the σ(k) mode, and is defined as The fully symmetric polytope is:

[0129]

[0130] p σ(k) A fully symmetric polyhedron The center of To generate the matrix.

[0131] Step 3: Collect data at time k {y k ,Φ k}, according to the fully symmetric polytope at time k-1 With the data {y k ,Φ k} Whether it is compatible or not, judging whether the operation mode of the power converter has changed;

[0132] If the fully symmetric polytope at time k-1 With the data {y k ,Φ k} is compatible, it is determined that the operation mode of the power converter at time k has not changed, and jump to step 4; if the fully symmetric polytope at time k-1 With the data {y k ,Φ k}Incompatible, jump to step 5;

[0133] Specifically, the fully symmetric polytope at time k-1 is determined based on whether formula (6) holds true. With the data {y k ,Φ k}Is it compatible?

[0134]

[0135] If it is detected that formula (6) is established, it is determined that the operation mode of the power converter has not changed at time k, and the process jumps to step 4; otherwise, the process jumps to step 5.

[0136] Step 4: Determine the fully symmetric polytope of the feasible parameter set of the parameters to be estimated at time k according to the power converter model Then jump to step nine;

[0137] After determining that the operation mode of the power converter has not changed at time k, update the fully symmetric polytope at time k.

[0138] Fully symmetric polytope at time k Update according to the following formula:

[0139]

[0140] Step 5: Determine the data {y at time k k ,Φ k}Consistency with all operating modes of the power converter;

[0141] In step 3, determine the fully symmetric polytope at time k-1. With the data sample {y k ,Φ k} is incompatible, it is necessary to determine the data {y k ,Φ k}Consistency with all operating modes of the power converter;

[0142] If the data {y k ,Φ k} is inconsistent with all the operating modes of the power converter, it is determined that the power converter fails at time k and jumps to step 7;

[0143] If the data {y k ,Φ k} is consistent with a certain operation mode of the power converter, it is determined that the operation mode of the power converter at time k is changed and the process jumps to step 6;

[0144] Specifically, according to formula (8), the data {y k ,Φ k}Consistency with all operating modes of the power converter;

[0145] If it is detected that formula (8) holds true, it is determined that the operation mode of the power converter changes at time k;

[0146]

[0147] On the contrary, if it is detected that the following inequality (9) holds, it is determined that the power converter fails at time k;

[0148]

[0149] Step 6: For the fully symmetric polytope at time k-1 Initialize and obtain the fully symmetric polytope of the feasible parameter set of the parameters to be estimated at time k And jump to step nine;

[0150] If step 5 determines that the operation mode of the power converter changes at time k, it means that the parameter estimation for the next time can no longer be performed according to the original operation mode. Therefore, it is necessary to perform a full symmetric polyhedron at time k-1. Initialize and reset the fully symmetric polyhedron corresponding to the replaced mode.

[0151] Step 7: Initialize the fully symmetric polytope Identify the fault type through model matching; and determine whether the fault occurs belongs to the known faults in the fault library;

[0152] If step five determines that the power converter fails at time k, it is necessary to match the fault parameters with the fault information in the fault library obtained in step one to determine whether it is a known type of fault;

[0153] Specific:

[0154] If the match is successful within a certain period of time, the fault information is recorded and the process goes to step nine;

[0155] If the match fails within a certain period of time, it is determined that a new fault type has occurred in the system and the process jumps to step eight;

[0156] The model matching method is:

[0157] Define the k-time parameter estimation sequence Parameter estimation error at time k When the parameter estimation error Δθ k The fault is separated by satisfying that l is less than a certain threshold ε within a certain length of time, recording the fault identification time and using the filter to display the fault type. The number of fault types is defined as M, and the jth fault type at time k satisfies:

[0158]

[0159] in, F M is the fault type in the fault library, and fault identification filters f1, f2, …, f are designed for each of the M fault types. M When a fault occurs in the system, if the jth fault identification filter is not empty, it means that the current fault type matches the type in the fault library, that is, the fault type at this time is determined to be F j .

[0160] It should be noted that, when performing fault matching, the length of the matching time can be set by the technician according to the actual situation.

[0161] Step 8: Determine whether a new fault type occurs in the system, add the fault type to the fault library, and jump to step 9;

[0162] If the match fails within the simulation time, it means that the current fault does not belong to any fault in the fault library and is a new fault type. Then the fault type is added to the fault library and the process goes to step nine.

[0163] Step nine: Set k=k+1, update relevant data information, and jump to step three until k>N, where N is the data length.

[0164] By executing the method cyclically, the state identification and fault diagnosis process can be completed without interruption.

[0165] In order to verify the effectiveness and feasibility of the power converter operating state identification and fault diagnosis method based on multi-cell set member filtering proposed in this application, the following simulation experiments are carried out:

[0166] For the Boost converter in which the switch S is disconnected, the Boost converter achieves different degrees of boost effect by setting two different inductance L values, thereby working in two operating modes. That is, when the Boost converter has no fault, the converter has two operating states:

[0167] When k∈{1,500} and k∈{1001,1500}, L=50μH, the actual parameter value of Boost converter is [0.1192,0.0064,1.5079,0.0677] T ;

[0168] When k∈{501,1000} and k∈{1501,2000}, L=500μH, the actual parameter value of the Boost converter is [0.128,0.0064,0.9410,0.0625] T .

[0169] It is assumed that two different parameter faults occur in the time periods of k∈{700,1000} and k∈{1800,2000}, namely, fault 1 inductor failure and fault 2 capacitor failure. That is, the system parameters will change differently from the normal working mode in these two time periods.

[0170] Figure 3 and Figure 4 The parameters to be estimated b are shown respectively when the Boost converter is fault-free. σ(k),1 and b σ(k),2 The changes in the upper and lower bounds and the true value are given by Figure 3 and Figure 4 It can be seen that the filter fault diagnosis method for the power converter under multiple operating modes provided in the present application can well estimate the true value of the system parameter, that is, effectively identify the operating state of the Boost converter.

[0171] Figure 5 and Figure 6 The parameters to be estimated b are shown respectively when the Boost converter fails. σ(k),1 and b σ(k),2 The changes in the upper and lower bounds and the true value are given by Figure 5 and Figure 6 It can be seen that the filter fault diagnosis method for the power converter under multiple operating modes provided in the present application can quickly detect the occurrence of a fault, reinitialize the system under the fault, and track and estimate the fault parameters.

[0172] Figure 7 The fault diagnosis results are shown by Figure 7(a) It can be seen that when there is no fault, the fault detection filter is empty. When a fault occurs in the system, the fault detection signal immediately jumps between k = 701 and k = 1301, indicating that the system has a fault. The system is promptly detected. Figure 7 (b) It can be seen that after the fault signal is detected, the fault identification filter takes effect. When k = 710, only filter f1 is not empty, indicating that only fault 1 occurs at this time, that is, the inductance of the Boost circuit fails, and Figure 7 (c) It can be seen that when k=1813, only filter f2 is not empty, which means that only fault 2 occurs at this time, that is, the Boost circuit capacitor fails.

[0173] When the existing fault diagnosis method for a single operating mode performs fault diagnosis on a converter with multiple operating modes, if the operating mode of the power converter changes, the existing fault diagnosis method will determine that the converter has failed, and then isolate and identify the fault of the converter, but in fact the converter has not failed, which will affect the actual operation of the converter.

[0174] Some steps in the embodiments of the present invention may be implemented using software, and the corresponding software program may be stored in a readable storage medium, such as a CD or a hard disk.

[0175] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A filter fault diagnosis method for a power converter in multiple operating modes, characterized in that: The method comprises: Step 1: Establish a regression model of the power converter and obtain a fault library of the power converter; Step 2: Set the fully symmetric polyhedron corresponding to the parameter vector to be estimated of the power converter at the initial moment And obtain the real parameter vector to be estimated for each operating mode The fully symmetric polytope of Step 3: Collect data at time k {y k ,Φ k }, and according to the fully symmetric polytope at time k-1 With the data {y k ,Φ k } Whether it is compatible or not, determine whether the operation mode of the power converter has changed: If they are compatible, it is determined that the operation mode of the power converter has not changed at time k, and the process jumps to step 4; If not compatible, go to step 5; Step 4: Determine the fully symmetric polytope of the feasible parameter set of the parameters to be estimated at time k based on the power converter regression model established in step 1 Then jump to step nine; Step 5: Determine the data {y at time k k ,Φ k }Consistency with all operating modes of the power converter: If the data {y k ,Φ k } is inconsistent with all the operating modes of the power converter, it is determined that the power converter fails at time k and jumps to step 7; If the data {y k ,Φ k } is consistent with a certain operation mode of the power converter, it is determined that the operation mode of the power converter at time k is changed and the process jumps to step 6; Step 6: Fully symmetric polytope at time k-1 Initialize and obtain the fully symmetric polytope of the feasible parameter set of the parameters to be estimated at time k And jump to step nine; Step 7: Initialize the fully symmetric polytope Identify the fault type through model matching and determine whether the fault is a known fault in the fault library; Step 8: If it is determined in step 7 that a new fault type occurs in the power converter, the fault type is added to the fault library and the process goes to step 9; Step 9: Set k=k+1, update the corresponding data information, and jump to step 3 until k>N, where N is the predetermined data length; The above steps 3 to 9 are executed cyclically to complete the state identification and fault diagnosis process of the power converter without interruption.

2. The method according to claim 1, characterized in that: The regression model of the power converter established in step 1 is: Among them, θ σ(k) =[a σ(k),1 ,a σ(k),2 ,b σ(k),1 ,b σ(k),2 ] T represents the parameter vector to be estimated for the σ(k)th operating mode of the power converter, e k is the system noise, |e k |≤δ, δ represents the known system noise boundary; y k represents the output data at time k, that is, the output voltage at time k; u k represents the input data at time k, that is, the input voltage at time k; Parameter to be estimated a σ(k),1 ,a σ(k),2 ,b σ(k),1 ,b σ(k),2 They are: in, T s is the sampling time; L, C, and R are the inductance, electrolytic capacitor, and load resistance in the equivalent circuit of the power converter, respectively.

3. The method according to claim 2, characterized in that The second step comprises: 2.1 Set the fully symmetric polyhedron corresponding to the parameter vector to be estimated of the power converter at the initial moment For the power converter at time k, its operating mode is σ(k), and the fully symmetric polyhedron at time k is defined as: Among them, p k,σ(k) is the fully symmetric polyhedron corresponding to the parameter vector to be estimated at time k The center of is the corresponding generator matrix, B is the unit interval, that is, [-1,1], B r is the r-dimensional box space composed of unit intervals; 2.2 Obtain the actual estimated parameter vector of the power converter in each operating mode The fully symmetric polytope of is the actual parameter vector to be estimated when the power converter operates in the σ(k) mode, and the package is defined as The fully symmetric polytope is: p σ(k) is the fully symmetric polytope of the real parameter vector to be estimated in the σ(k) mode The center of is the corresponding generator matrix; g σ(k),i is the corresponding generator.

4. The method according to claim 3, characterized in that The step three comprises: According to whether formula (6) is valid, we can judge the fully symmetric polyhedron at time k-1. With the data {y k ,Φ k }Is it compatible? If formula (6) holds true, it is determined that the operation mode of the power converter has not changed at time k, and the process jumps to step 4; otherwise, the process jumps to step 5.

5. The method according to claim 4, characterized in that The fourth step comprises: After determining that the operation mode of the power converter has not changed at time k, update the fully symmetric polytope at time k. Fully symmetric polytope at time k Update according to the following formula:

6. The method according to claim 5, characterized in that The step five comprises: According to formula (8), the data {y k ,Φ k }Consistency with all operating modes of the power converter; If it is detected that formula (8) holds true, it is determined that the operation mode of the power converter has changed at time k, and the process jumps to step 6; On the contrary, if it is detected that the following inequality (9) holds, it is determined that the power converter fails at time k, and the process goes to step 7; Where ‖·‖1 represents the 1-norm.

7. The method according to claim 6, characterized in that The model matching method in step seven includes: Define the k-time parameter estimation sequence Parameter estimation error at time k When the parameter estimation error Δθ k If l is less than a certain threshold ε within a certain time length, the fault is separated, the fault identification time is recorded and the fault type is displayed by the filter; Define the number of fault types as M, and the j-th fault type at time k satisfies: in, F M is the fault type in the fault library, and fault identification filters f1, f2, …, f are designed for each of the M fault types. M , when a fault occurs in the system, if the jth fault identification filter is not empty, it means that the current fault type matches the type in the fault library, that is, the fault type at this time is determined to be F j ; If the current fault type does not match the type in the fault library, the fault is determined to be a new fault and is added to the fault library.

8. The method according to claim 7, characterized in that The fault types in the fault library obtained in step 1 include inductor failure and capacitor failure.

9. The method according to claim 1, characterized in that: The regression model of the power converter established in step 1 includes: According to the working principle of the power converter, the circuit model of the power converter is established as follows: Where i is the inductor current, u o is the output voltage; The circuit model of the power converter is approximated as a second-order system: Where s is a complex variable; Take the sampling time T s =0.01, and the formula (2) is discretized by bilinear transformation to obtain: Let b σ(k),0 =1, and taking into account the system noise e k ,get: A σ(k) (z)y k =B σ(k) (z)u k +e k (4) Among them, A σ(k) ψ(z) = 1 + a σ(k),1 z -1 + a σ(k),2 z -2 , B σ(k) ψ(z) = 1 + b σ(k),1 z -1 + b σ(k),2 z -2 .

10. The method according to claim 1, characterized in that The operating mode of the power converter at time k is σ(k)∈{1,2,…S}, where S is the number of operating modes of the power converter.

Images