A robust open-circuit fault diagnosis method for bidirectional DC / DC converters based on extended Kalman filter multiple correction

Abstract

This invention discloses a robust open-circuit fault diagnosis method for bidirectional DC / DC converters based on extended Kalman filter multiple correction, belonging to the field of fault diagnosis and analysis technology. The method first establishes a global mathematical model, then constructs a global extended Kalman filter as an observer. A residual signal is generated by subtracting the observed value from the actual measured value. The residual generated by the extended Kalman filter due to model error is used as the model error threshold. When the number of consecutive corrections reaches the set correction threshold, an open-circuit fault is determined in the switching transistor. Finally, the degree of correction of the extended Kalman filter by the measured value is used to determine the open-circuit object. This invention can effectively solve the influence of parameter changes on fault detection during the operation of bidirectional DC / DC converters, improving the robustness of open-circuit fault diagnosis for bidirectional DC / DC converter switching transistors.

Description

Technical Field

[0001] This invention relates to the field of fault diagnosis and analysis technology, and in particular to a robust open-circuit fault diagnosis method for bidirectional DC / DC converters based on extended Kalman filter multiple correction. Background Technology

[0002] In electric vehicle energy storage systems, it's difficult for a single energy source to simultaneously meet the requirements of high energy density and high power density. Therefore, a reasonable combination of high-energy-density lithium batteries and high-power-density supercapacitors is necessary. With the support of energy management strategies, the power of the battery and supercapacitor is rationally allocated to fully utilize their energy characteristics and power characteristics. This saves energy while effectively meeting the range and power performance requirements of electric vehicles. Because the charging and discharging characteristics of the two energy sources differ, a bidirectional DC / DC converter needs to be added to the hybrid energy storage system to reduce the stress on the battery during high-power charging and discharging. This converter controls and regulates the charging and discharging current and input / output voltage between the battery and supercapacitor to optimize their operation. Using two bidirectional DC / DC converters allows for more effective and controllable power allocation between the battery and supercapacitor, effectively stabilizing the DC bus voltage.

[0003] Using a bidirectional DC / DC converter not only enables bidirectional energy flow but also allows for current adjustment based on different operating conditions, ensuring the reliability and safety of batteries and supercapacitors in real-world applications. However, during actual operation, power transistors are susceptible to damage from harmonics, high voltage stress, and high current stress. For electric vehicles, related repairs can result in significant economic losses. Currently, there is limited research on fault diagnosis for open-circuit faults in bidirectional DC / DC converter power transistors; therefore, fault diagnosis of bidirectional DC / DC power transistors is of significant research importance. Summary of the Invention

[0004] To address the problems existing in the prior art, this invention provides a robust open-circuit fault diagnosis method for bidirectional DC / DC converters based on extended Kalman filter multiple correction. This diagnostic method has good robustness and is not affected by changes in the parameters of the bidirectional DC / DC converter.

[0005] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:

[0006] This application provides a robust open-circuit fault diagnosis method for bidirectional DC / DC converters based on extended Kalman filter multiple correction, the method comprising the following steps:

[0007] S1. Construct a global mathematical model based on the state equations of the bidirectional DC / DC converters on both sides of the supercapacitor and the battery.

[0008] S2. Based on the state equations of the bidirectional DC / DC converters on both sides of the supercapacitor and the battery, construct the algorithm equations of the global extended Kalman filter. Use the algorithm equations of the extended Kalman filter to observe the inductor current value on the side of the battery, the inductor current value on the side of the supercapacitor, and the voltage value of the DC bus to generate observations.

[0009] S3. The residual signal is obtained by subtracting the observed value from the extended Kalman filter and the measured value of the actual system. The maximum error generated by the global mathematical model during the smooth operation of the permanent magnet synchronous motor is used as the model error threshold. When the residual signal exceeds the model error threshold, the actual measured value is used to replace the observed value in the next calculation of the extended Kalman filter for continuous correction. When the number of continuous corrections exceeds the set correction number threshold, it is determined that the permanent magnet synchronous motor has a fault. Secondly, the side with the open circuit fault is determined based on the number of continuous corrections of the inductor current on the battery side and the supercapacitor side exceeding the correction number threshold. Finally, the open circuit object is determined by the degree of correction of the extended Kalman filter by the actual value.

[0010] Optionally, in some embodiments of this application, the global mathematical model in S1 is as follows:

[0011]

[0012] In the formula, R1, L1, i l1 These represent the resistance, inductance, and current values ​​on the battery side; R2, L2, i l2 These are the resistance, inductance, and current values ​​on the supercapacitor side, respectively; V bat V sc These are the voltage values ​​of the battery and the supercapacitor, respectively; V o For DC bus output voltage value, i o For DC bus output current value, C o For DC bus filter capacitor; μ 01 μ 23 This is the control signal for the bidirectional DC / DC converter of the battery and supercapacitor.

[0013] Optionally, in some embodiments of this application, in S2, the global mathematical model of the global extended Kalman filter, i.e., the algorithm equation, is:

[0014]

[0015] In the formula The observer's estimated value; The value observed at the previous moment; u k-1 For control quantity; K k For Kalman gain; z kHere, is the measured value; Q is the process noise covariance; R is the measurement noise covariance; H is the measured value. k P is the gain matrix; k The mean square error of the state variable estimation; P k - Estimate the mean square error of the uncorrected state variables;

[0016] The observations obtained are The control quantity is u k ={V bat V sc i o} T .

[0017] Optionally, in some embodiments of this application, step S2 specifically includes the following steps:

[0018] S21. Set the initial values ​​for the observer and the initial state estimate. P k The initial state estimate is P0, and the initial values ​​of Q and R are set.

[0019] S22. Calculate the prior knowledge, and Expanding to Taylor first order, that is:

[0020]

[0021] In the formula T s Let B be the sampling time, B be the control matrix, and A be the sampling time. k This is the state transition matrix;

[0022] S23. Calculate the post-calculation, and output voltage U o Substituting the measured value into the extended Kalman filter, the following calculation is obtained:

[0023]

[0024] P k =(IH k K k )P k -

[0025] Optionally, in some embodiments of this application, the specific steps of S3 are as follows:

[0026] S31. Define the residual inductor current on the battery side and the residual inductor current on the supercapacitor side as the error between the observed value and the actual measured value of the extended Kalman filter, that is:

[0027]

[0028] Where e1(t) represents the residual current of the battery pack and e2(t) is the residual current of the inductor on the supercapacitor side;

[0029] When the system is fault-free, the maximum value of the residual evaluation function is selected as the correction threshold e during the period of stable speed. th1 e th2 ;

[0030] S32. Compare e1(t) and e th1 and e2(t) and e th2 Simultaneously set counting functions C1(t) and C2(t); if e1(t) > e th1 Then the measured value is used to perform the next step in the extended Kalman filter. After correction, C1(t) = 1; similarly, if e2(t) > e th2 Then use the measured value to... After correction, C2(t) = 1; if e1(t) <e th1 And e2(t) <e th2 If C1(t) = 0 and C2(t) = 0, then C1(t) = 0.

[0031] S33. Select the maximum number of consecutive corrections during speed change and load change as thresholds C. th1 C th2 Determine the evaluation function for the corrected signal:

[0032]

[0033] Among them, C r1 (t) is the evaluation function for the correction signal of sudden speed change, C r2 (t) is the evaluation function for the correction signal of the load mutation;

[0034] Compare C r1 (t) and C th1 And C r2 (t) and C th2 The size of C; if C r1 (t) <C th1 And C r2 (t) <C th2 No open-circuit faults were found on either the battery side or the supercapacitor side; if C r1 (t)=C th1 The switching transistor on the battery side has an open circuit fault; if C r2 (t)=C th2 If this occurs, an open-circuit fault will occur in the switching transistor on the supercapacitor side.

[0035] S34. Determine the open circuit object based on the degree of correction to the Kalman filter.

[0036] Optionally, in some embodiments of this application, the specific steps of S34 are as follows:

[0037] S341. Based on the different residual trends between observed and actual values ​​when two different power transistors fail in a bidirectional DC / DC converter, the open-circuit object is determined by the correction degree of the Kalman filter, and a correction degree threshold is set.

[0038] S342. After determining that a fault has occurred on the battery side or the supercapacitor side, calculate the pre-fault C value respectively. th1 C th2 The average values ​​of the residuals at each sampling point are E1(t) and E2(t);

[0039] S343. Based on the identified faulty side, compare E1(t) with... and E2(t) and like This indicates a fault in the battery-side discharge power transistor; if This indicates a fault in the charging power transistor on the battery side; similarly, if This indicates a fault in the discharge power transistor on the supercapacitor side; if This indicates a fault in the charging power transistor on the supercapacitor side.

[0040] Compared with the prior art, the beneficial effects of this invention are as follows:

[0041] This invention provides a method for diagnosing open-circuit faults in power transistors of composite energy sources based on extended Kalman filters. This method is not affected by changes in the parameters of bidirectional DC / DC converters and has high robustness.

[0042] Meanwhile, the construction of the global mathematical model diagnoses robust open-circuit faults in bidirectional DC / DC converters. This not only eliminates the need to classify and consider the working modes of BDDC in subsequent fault analysis, greatly simplifying the calculation process, but also allows for the direct construction of a global observer using the global mathematical model, saving computational resources. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a topology diagram of a composite energy source system.

[0045] Figure 2Block diagram of extended Kalman filter structure for bidirectional DC / DC converter in composite energy source system;

[0046] Figure 3 This is a schematic diagram illustrating the fault diagnosis principle based on the extended Kalman filter of this invention. Detailed Implementation

[0047] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this application. It is understood that the accompanying drawings are provided for reference and illustration only, and are not intended to limit this application. The connection relationships shown in the accompanying drawings are only for clear description and do not limit the connection method.

[0048] like Figure 1 As shown, the composite energy system consists of a battery and a supercapacitor, with two independent bidirectional DC / DC converters serving as interface circuits and a five-phase permanent magnet synchronous motor as the load.

[0049] This invention proposes a robust switching fault diagnosis method for bidirectional DC / DC converters based on extended Kalman filter multiple correction, comprising the following steps:

[0050] S1, according to Figure 1 The specific steps for establishing a global mathematical model of the state equations involved in the topology diagram of the complex energy source system are as follows:

[0051] when i l1 >0 and i l2 When the current is greater than 0, that is, when the current in the battery pack and supercapacitor pack flows in the same direction toward the bidirectional DC / DC converter, the mathematical model is as follows:

[0052]

[0053] In the formula, μ0 and μ2 are the control signals of the discharge tube of the bidirectional DC / DC converter on the battery side and the discharge tube of the bidirectional DC / DC converter on the supercapacitor side, respectively.

[0054] when i l1 >0 and i l2 When the current is less than 0, i.e., the current in the supercapacitor bank is directed towards the bidirectional DC / DC converter, and the current in the battery bank is directed away from the bidirectional DC / DC converter, the mathematical model is as follows:

[0055]

[0056] In the formula, μ3 is the control signal of the charging tube of the bidirectional DC / DC converter on the supercapacitor side.

[0057] when i l1 <0 and i l2 When the current is greater than 0, that is, when the current in the supercapacitor bank is away from the bidirectional DC / DC converter and the current in the battery bank is towards the bidirectional DC / DC converter, the mathematical model is as follows:

[0058]

[0059] In the formula, μ2 is the control signal of the charging tube of the bidirectional DC / DC converter on the battery side.

[0060] when i l1 <0 and i l2 When the current is less than 0, that is, when the current in the battery pack and supercapacitor pack flows in the reverse direction toward the bidirectional DC / DC converter, the mathematical model is as follows:

[0061]

[0062] Based on the above formula, switching functions m and n are constructed for the charge and discharge states of the battery and supercapacitor, respectively:

[0063]

[0064] Where μ is defined 01 μ 23 Given the unique control signals for the bidirectional DC / DC converters of the battery and supercapacitor respectively, we have:

[0065] μ 01 =m(1-μ0)+(1-m)μ1…………(7)

[0066] μ 23 =n(1-μ2)+(1-n)μ3…………(8)

[0067] Based on equations (1), (2), (3), (4), (7), and (8), a global mathematical model of the composite energy source can be obtained:

[0068]

[0069] In the formula, R1, L1, i l1 These represent the resistance, inductance, and current values ​​on the battery side; R2, L2, i l2 These are the resistance, inductance, and current values ​​on the supercapacitor side, respectively; V bat V sc These are the voltage values ​​of the battery and the supercapacitor, respectively; V o For DC bus output voltage value, io For DC bus output current value, C o For DC bus filter capacitor; μ 01 μ 23 This is the control signal for the bidirectional DC / DC converter of the battery and supercapacitor.

[0070] S2, such as Figure 2 As shown, a global extended Kalman filter is constructed based on the state equations of the bidirectional DC / DC converters on both sides of the supercapacitor and the battery. The extended Kalman filter is then used to observe the inductor current on the battery side, the inductor current on the supercapacitor side, and the DC bus voltage. Specifically, the following steps are included:

[0071] First, the mathematical model for the extended Kalman filter needs to be generated. The generated global mathematical model is as follows:

[0072]

[0073] In the formula The observer's estimated value; The value observed at the previous moment; u k-1 For control quantity; K k For Kalman gain; z k Here, is the measured value; Q is the process noise covariance; R is the measurement noise covariance; H is the measured value. k P is the gain matrix; k The mean square error of the state variable estimation; P k - Estimate the mean square error of the uncorrected state variables;

[0074] The observations obtained are The control variable is uk = {Vbat; Vsc; io}T.

[0075] S21. Set the initial values ​​of the observer based on its observed and control quantities, wherein the initial state estimate is set to... P k The initial state estimate is P0, and the initial values ​​of Q and R are set.

[0076] S22. Calculate the prior based on the initial state value:

[0077] Specifically, it will Expanding to Taylor first order, that is

[0078]

[0079] In the formula T s Let B be the sampling time, B be the control matrix, and A be the sampling time. k Let be the state transition matrix.

[0080] This prior calculation allows us to estimate the state value of the next state.

[0081] S23. Perform the calculation of the prior axioms after completing the calculation of the posterior axioms:

[0082] Specifically, the output voltage U o Substituting the measured value into the extended Kalman filter, the following calculation is obtained:

[0083]

[0084] P k =(IH k K k )P k -

[0085] By calculating the posterior, the prior is corrected using the measurement model and actual measurements, thus obtaining a state estimate that is closer to the true state.

[0086] S3. After completing the prior and posterior calculations, the residual signal is obtained by subtracting the observed values ​​from the extended Kalman filter and the measured values ​​of the actual system. The maximum error generated by the global mathematical model during the smooth operation of the permanent magnet synchronous motor is used as the model error threshold. When the residual signal exceeds the model error threshold, the actual measured value is used instead of the observed value for the next calculation of the extended Kalman filter for correction. When the number of consecutive corrections exceeds the set correction number threshold, the permanent magnet synchronous motor is judged to have a fault. The side with the open circuit fault is determined based on whether the number of consecutive corrections of the inductor current on the battery side and the supercapacitor side exceeds the correction number threshold. Finally, the degree of correction of the extended Kalman filter by the actual value is used to determine the object with the open circuit. Specifically, the following steps are included:

[0087] S31. Define the residual inductor current on the battery side and the residual inductor current on the supercapacitor side as the error between the observed value and the actual measured value of the extended Kalman filter, that is:

[0088]

[0089] Where e1(t) represents the residual current of the battery pack and e2(t) is the residual current of the inductor on the supercapacitor side;

[0090] When the system is fault-free, the maximum value of the residual evaluation function is selected as the correction threshold e during the period of stable speed. th1 e th2 .

[0091] S32. Compare e1(t) and e th1 and e2(t) and e th2 Simultaneously set counting functions C1(t) and C2(t); if e1(t) > eth1 Then the measured value is used to perform the next step in the extended Kalman filter. After correction, C1(t) = 1; similarly, if e2(t) > e th2 Then use the measured value to... After correction, C2(t) = 1. If e1(t) <e th1 And e2(t) <e th2 If C1(t) = 0 and C2(t) = 0, then C1(t) = 0.

[0092] S33. Select the maximum number of consecutive corrections during speed change and load change as the threshold C. th1 C th2 Determine the evaluation function for the corrected signal:

[0093]

[0094] Wherein, Cr1(t) is the evaluation function of the correction signal for sudden speed change, and Cr2(t) is the evaluation function of the correction signal for sudden load change;

[0095] Compare C r1 (t) and C th1 And C r2 (t) and C th2 If C r1 (t) <C th1 And C r2 (t) <C th2 No open circuit fault occurred; if C r1 (t)=C th1 If C... r2 (t)=C th2 If this occurs, an open-circuit fault will occur in the switching transistor on the supercapacitor side.

[0096] S34. Determine the open-circuit object based on the degree of correction to the Kalman filter. The specific steps are as follows:

[0097] S341. Due to the failure of two different power transistors on the bidirectional DC / DC converter...

[0098] Since the residual trends of the observed and actual values ​​differ, the open circuit object can be identified based on the degree of correction to the Kalman filter. In this case, a correction threshold is set.

[0099] S342. After determining that a fault has occurred on the battery side or the supercapacitor side, calculate the pre-fault C value respectively. th1 C th2 The average values ​​of the residuals at each sampling point are E1(t) and E2(t).

[0100] S343, Based on S342, the faulty side has been identified. Compare E1(t) with... and E2(t) and When the battery-side switching transistor has an open circuit fault, if This indicates a fault in the battery-side discharge power transistor; if This indicates a fault in the charging power transistor on the battery side; similarly, if the switching transistor on the supercapacitor side has an open circuit fault, then... This indicates a fault in the discharge power transistor on the supercapacitor side; if This indicates a fault in the charging power transistor on the supercapacitor side.

[0101] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A robust open-circuit fault diagnosis method for bidirectional DC / DC converters based on extended Kalman filter multiple correction, characterized in that, The method comprises the following steps: S1, constructing a global mathematical model according to a state equation of the bidirectional DC / DC converter on both sides of the super capacitor and the storage battery, the global mathematical model being: ； wherein R1, L1, i l1 are the battery-side resistance value, inductance value, and current value, respectively; R2, L2, i l2 are the super capacitor-side resistance value, inductance value, and current value, respectively; V bat , V sc are the voltage values of the battery and the super capacitor, respectively; V o is the DC bus output voltage value, i o is the DC bus output current value, C o is the DC bus filter capacitor; μ 01 , μ 23 are the control signals of the bidirectional DC / DC converter of the battery and the super capacitor. S2, constructing an algorithm equation of a global extended Kalman filter according to the state equation of the bidirectional DC / DC converter on both sides of the super capacitor and the storage battery, observing the inductor current value on the side of the storage battery, the inductor current value on the side of the super capacitor and the voltage value of the DC bus by using the algorithm equation of the extended Kalman filter to generate an observation, the algorithm equation being: wherein is an observer estimate; is an observation value at the previous time; is an observation-derived observation quantity; is a control quantity; is a Kalman gain; is a measurement value; Q is a process noise covariance; and R is a measurement noise covariance; is a gain matrix; is an error-corrected state variable estimate error mean square value; A K is the state transition matrix; The algorithmic equation of the extended Kalman filter is used to observe the battery-side inductance current value i 11 , the super capacitor-side inductance current value i 12 , and the voltage value V0 of the DC bus to generate an observation; S3, obtaining a residual error signal by subtracting the observation of the extended Kalman filter from the measured value of the actual system; taking the maximum error generated by the global mathematical model when the permanent magnet synchronous motor is running stably as a model error threshold value, when the residual error signal exceeds the model error threshold value, using the actual measured value to replace the observation value to enter the calculation of the next extended Kalman filter for continuous correction, when the number of continuous corrections exceeds a set correction number threshold value, judging that the permanent magnet synchronous motor has a fault; secondly, judging the side which has an open circuit fault according to the number of continuous corrections of the inductor current on the side of the storage battery and the side of the super capacitor; finally, judging the open circuit object by using the correction degree of the actual value to the extended Kalman filter.

2. The robust open-circuit fault diagnosis method for bidirectional DC / DC converter based on EKF-MRC according to claim 1, characterized in that, In S2, the observed quantity is observed = i l1 ; i l2 ; U o} T , and the control quantity is u k = V bat ; V sc ; i o} T , where U0is the output voltage.

3. The robust open-circuit fault diagnosis method for bidirectional DC / DC converters based on extended Kalman filter multiple corrections according to claim 2, characterized in that, The specific steps of S2 are as follows: S21, set the initial value of the observer, initial state estimation value , the initial state estimation value , set the initial value of Q, R; S22, compute the prior, f(x), from Expand to Taylor first order, i.e.: where T s is the sampling time, B is the control matrix, A k is the state transition matrix; S23, calculate the posterior, the output voltage U o As a measure of the extended Kalman filter, the calculation is made: 。 4. The robust open-circuit fault diagnosis method for bidirectional DC / DC converter based on EKF-MRC according to claim 1, characterized in that, The specific steps of S3 are as follows: S31, defining the inductor current residual error on the side of the storage battery and the inductor current residual error on the side of the super capacitor as the error between the observation value of the extended Kalman filter and the actual measured value, i.e. Wherein e1(t) represents the storage battery current residual error, and e2(t) is the inductor current residual error on the side of the super capacitor; When the system is fault-free, the maximum value of the residual error evaluation function during the stable period of the rotating speed is selected as the correction threshold e th1 , e th2 ; S32. Compare e1(t) and e th1 and e2(t) and e th2 Simultaneously set counting functions C1(t) and C2(t); if e1(t) > e th1 Then the measured value is used to perform the next step in the extended Kalman filter. After correction, C1(t) = 1; similarly, if e2(t) > e th2 Then use the measured value to... After correction, C2(t) = 1; if e1(t) <e th1 And e2(t) <e th2 If C1(t) = 0 and C2(t) = 0, then C1(t) = 0. S33, select the continuous maximum correction times during the speed mutation and load mutation as threshold C th1 , C th2 , determine the evaluation function of the correction signal: where C r1 (t) is an evaluation function of the correction signal for a speed jump, C r2 (t) is an evaluation function of the correction signal for a load jump; Comparison C r1 (t) and C th1 and C r2 (t) and C th2 ; if C r1 (t) < C th1 and C r2 (t) < C th2 , neither the battery side nor the super capacitor side has open circuit failure; if C r1 (t) = C th1 , the switch tube of the battery side has open circuit failure; if C r2 (t) = C th2 , the switch tube of the super capacitor side has open circuit failure; S34, judging the open circuit object according to the correction degree of the Kalman filter.

5. The robust open-circuit fault diagnosis method for bidirectional DC / DC converters based on extended Kalman filter multiple corrections according to claim 4, characterized in that, The specific steps of S34 are as follows: S341、based on the two different power tubes on the two-way DC / DC converter, the residual trend of the observed value and the actual value is different, the degree of correction of the Kalman filter is used to judge the open circuit object, and the correction degree threshold is set 1、 2; S342、When it is judged that the battery side has a failure or the super capacitor side has a failure, the average values E1(t), E2(t) of the residual errors of the C th1 , C th2 sampling points before the failure are calculated, respectively. S343、Combining the judged fault side, compare E1(t) with 1 and E2(t) with 2, if E1(t)< 1, then the battery side discharge power tube is faulty; if E1(t)> 1, then the battery side charging power tube is faulty; similarly, if E2(t)< 2, then the super capacitor side discharge power tube is faulty; if E2(t)> 2, then the super capacitor side charging power tube is faulty.

Images