A method and apparatus for detecting a fault in a DC-DC converter

Abstract

The application relates to a DC-DC converter fault detection method and a detection device, which first collects current ripple signals and PWM signals of a direct current converter; secondly, ripple characteristic signals and the PWM signals are subjected to incremental correlation coefficient operation and real-time discrimination of fault diagnosis. The method for detecting faults by taking the current ripple as a monitoring quantity is first proposed, the magnetic coupling device is used to actively collect the ripple, the ripple signals are subjected to characteristic processing operation to generate the ripple characteristic signals, the method is not disturbed by input voltage flicker and fluctuation, the fault detection is accurate, and the effect is obvious in distributed power supply fault detection. Through recursive calculation of a correlation coefficient formula, the incremental correlation coefficient formula is first proposed. The efficiency of the DSP signal processing circuit for the correlation coefficient operation of the signals is improved, and the real-time fault detection is realized. If the incremental correlation coefficient is detected to be suddenly changed, the power supply of the direct current converter is immediately cut off through an execution circuit, and circuit protection is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to a voltage converter detection method and detection device, in particular to a DC-DC converter fault detection method and detection device based on incremental iterative correlation coefficient. BACKGROUND

[0002] In recent years, distributed energy is widely used in various industrial fields. DC-DC converter is the core component of power conversion, and is also the most vulnerable component. The open circuit and short circuit of the switching tube are the main reasons for its damage. Once the DC-DC converter fails, it will cause the device to fail, and in severe cases, it may cause the system to be paralyzed, and even endanger personal safety. Therefore, fault diagnosis of DC-DC converter can maintain the stability of the system.

[0003] The current fault diagnosis technology of DC-DC converter is based on the condition that the input voltage is stable, and the fault diagnosis is realized by monitoring the inductance current or voltage characteristic quantity. However, the new energy power supply system is easily affected by the region, environment and other external factors, which may cause the input voltage to fluctuate, flicker and other problems, so using the current fault diagnosis technology to detect the new energy power supply system may cause misjudgment and damage the stability of the new energy power supply system.

[0004] T. Park and T. Kim, "Novel fault tolerant power conversion system for hybrid electric vehicles," in Proc. IEEE VPPC 2011, 2011 pp. 1-6. proposed a short-circuit detection method based on the comparison of duty ratio and inductor current size. However, when the input voltage fluctuates, it will interfere with the logic circuit, causing the logic circuit to misjudge.

[0005] M. Shahbazi, E. Jamshidpour, P. Pour, S. Saadate, and M. Zolghadri, "Open- and short-circuit switch fault diagnosis for nonisolated dc-dc converters using fifield programmable gate array," IEEE Trans. on Ind. Electron., vol. 60, no. 9, pp. 4136-4146, 2013. A fault detection method based on inductor current slope was proposed, which contains two algorithms. The method can detect short-circuit and open-circuit faults within two switching cycles. But the logic circuit will be disturbed after the input voltage step change, resulting in misjudgment.

[0006] H.-K. Cho S.-S. Kwak & S.-H. Lee, "Fault diagnosis algorithm based on switching function for boost converters" Deagu Gyeongbuk Institute of Science and Technology, Dal-sunggun, Daegu 711-873, Republic of Korea Accepted author version posted online: 03 Oct 2014. Published online: 16 Oct 2014. A fault detection method based on inductor voltage was proposed. The method directly collects the voltage across the inductor, which is easy to produce sharp pulses and damage components. And when the input voltage step changes, the logic gate of the detection circuit may be damaged, resulting in misjudgment.

[0007] Ehsan Jamshidpour, Member, IEEE, Philippe Poure, “Photovoltaic Systems Reliability Improvement by Real-Time FPGA-Based Switch Failure Diagnosis and Fault-Tolerant DC-DC Converter” DOI 10.1109 / TIE.2015.2421880, IEEE Transactions on Industrial Electronics. This paper proposes a fault diagnosis method based on inductor current waveform. This method directly collects the current across the inductor, which is prone to generating sharp pulses that can damage components. Furthermore, fluctuations in the input voltage can interfere with the logic circuit, causing it to misjudge the circuit. Summary of the Invention:

[0008] The purpose of this invention is to address the shortcomings of existing technologies by providing a DC-DC converter fault detection method based on incremental iterative correlation coefficients.

[0009] Another objective of this invention is to provide a detection device for a DC-DC converter fault detection method.

[0010] Inventive Concept: This invention comprises two stages: The first stage involves acquiring the current ripple signal and PWM signal of the DC-DC converter, performing feature processing on the current ripple signal to generate a ripple feature signal. The second stage involves performing incremental iterative correlation coefficient calculation on the ripple feature signal and the PWM signal, as well as real-time fault diagnosis. In the first stage, the ripple acquisition circuit uses a magnetic coupling element to acquire the current ripple signal, which undergoes feature processing to generate the ripple feature signal. The PWM acquisition circuit uses an impedance element to acquire the PWM signal, measures the phase angle difference of the signals using a phase comparison method, and performs phase shifting on the ripple feature signal based on this phase angle difference to make the signal phases the same. In the second stage, the in-phase ripple feature signal and the PWM signal are subjected to incremental iterative correlation coefficient calculation. A sudden change in the incremental iterative correlation coefficient is used to determine whether a circuit fault has occurred. If a sudden change in the correlation coefficient is detected, a fault has occurred, and the DC-DC converter is protected by the execution circuit.

[0011] The tasks of "incremental" and "iterative" are achieved by the incremental iterative correlation coefficient formula: ΔP and ΔQ are the incremental components of the incremental iterative correlation coefficient; P ω and Q ω It is the iterative step of the incremental iterative correlation coefficient.

[0012] The object of the present application is achieved by the following technical solutions:

[0013] A DC-DC converter fault detection method, comprising the following steps:

[0014] A. The ripple acquisition circuit acquires the current ripple signal, the PWM acquisition circuit acquires the PWM signal, the current ripple signal is processed by the first group of signal conditioning circuits and sent to the DSP signal processing circuit, and the PWM signal is processed by the second group of signal conditioning circuits and sent to the DSP signal processing circuit;

[0015] B. The current ripple signal is processed by the DSP signal processing circuit to generate a ripple characteristic signal;

[0016] C. The ripple characteristic signal and the PWM signal are phase adjusted: the phase adjustment is achieved by measuring the phase angle difference of the signals by the phase comparison method, and the ripple characteristic signal is phase shifted according to the phase angle difference, so that the signals have the same phase;

[0017] D. Incremental iterative correlation coefficient operation is performed on the in-phase ripple characteristic signal and the in-phase PWM signal to generate an incremental correlation coefficient library;

[0018] E. Whether the circuit fails is determined by monitoring whether the incremental iterative correlation coefficient mutates, and if the correlation coefficient mutates, the circuit fails;

[0019] F. The power supply of the DC converter is immediately cut off by executing the circuit, so that the protection circuit is protected.

[0020] In step B, the current ripple is processed as follows:

[0021] β=(i a+1 -i a )f m (1)

[0022]

[0023] Where, β=(i a+1 -i a )f m is the current differential processing formula, i a is the sampling value of the a-th current, i a+1 is the sampling value of the a+1-th current, f m is the sampling frequency, triat(β) is the current characteristic acquisition function, and lim is the critical value of the fluctuation differential amount of the input voltage. The reference value of lim is set to 400, and then the ripple characteristic signal is generated according to the triat(β) function.

[0024] In step D, the incremental iterative correlation coefficient operation formula is as follows:

[0025]

[0026] wherein:

[0027]

[0028] p ω is the memory correlation coefficient, i.e. the correlation coefficient calculated last time; p ω+1 is the real-time correlation coefficient, i.e. the correlation coefficient of the real-time data entering the correlation calculation window according to the principle of first-in first-out; A ω is the number of high levels in the correlation coefficient window; B ω is the number of low levels in the correlation coefficient window; P ω is the covariance calculated last time, and ΔP is the covariance change amount; Q ω is the product of the standard deviations calculated last time, and ΔQ is the product of the standard deviation change amount. x i y i is the object for correlation coefficient calculation; is the average of x i y i in the correlation coefficient window, respectively; is the difference between the average of x y after the real-time data entering the correlation calculation window according to the principle of first-in first-out and the average calculated last time. m is the total number of high and low levels in the correlation coefficient window.

[0029] A detection device for realizing the fault detection method of the DC-DC converter is connected by a ripple collection circuit, a first group of signal conditioning circuits, a DSP signal processing circuit, a PWM signal, a second group of signal conditioning circuits, a DSP signal processing circuit, a DSP signal processing circuit, an execution circuit module, a DC-DC converter, a DC-DC converter connected through magnetic coupling, and a power supply circuit connected with the first group of signal conditioning circuits, the second group of signal conditioning circuits, the DSP signal processing circuit, the PWM collection circuit and the ripple collection circuit.

[0030] After the ripple signal is collected by the magnetic coupling element, it is amplified by the differential amplifier and then input into the signal conditioning circuit for signal conditioning. After the PWM signal is collected by the impedance element, it is amplified by the differential amplifier and then input into the signal conditioning circuit for signal conditioning.

[0031] Beneficial effects: compared with the prior art, the present application firstly proposes a method for fault detection taking current ripple as monitoring quantity, actively collects ripple by using magnetic coupling device, the collection device is simple and safe, and will not damage components and devices by sharp pulse; ripple signal is subjected to feature processing operation to generate ripple feature signal, which is not interfered by input voltage flicker, fluctuation and the like, the accuracy of fault detection is realized, which is of great significance especially in fault detection of distributed power supply; the present application firstly proposes a formula of incremental iterative correlation coefficient by recursively processing correlation coefficient formula, improves the efficiency of correlation coefficient operation of DSP signal processing circuit on signal, and realizes the real-time of fault detection. If the incremental iterative correlation coefficient is detected to be suddenly changed, the power supply of direct current converter is immediately cut off through execution circuit, so that the effect of protecting circuit is achieved. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 . A DC-DC converter fault detection method flow chart;

[0033] Figure 2 . for the attached Figure 1 Incremental iterative correlation coefficient algorithm flow chart;

[0034] Figure 3 . A DC-DC converter fault detection method working process schematic diagram;

[0035] Figure 4 . A DC-DC converter fault detection device structure block diagram;

[0036] Figure 5 . DPS signal processing circuit;

[0037] Figure 6 . Power supply circuit;

[0038] Figure 7 . Execution circuit;

[0039] Figure 8 . PWM acquisition circuit, signal conditioning circuit;

[0040] Figure 9 . Ripple acquisition circuit, signal conditioning circuit;

[0041] Figure 10 . A DC-DC converter fault detection method ripple feature processing diagram;

[0042] Figure 11 . A DC-DC converter fault detection method fault diagnosis simulation diagram. DETAILED DESCRIPTION

[0043] The present application will be further described in detail below in combination with the drawings and examples

[0044] A DC-DC converter fault detection method, comprising the following steps:

[0045] A. The ripple acquisition circuit acquires the current ripple signal, the PWM acquisition circuit acquires the PWM signal, the current ripple signal is processed by the first group of signal conditioning circuits and sent to the DSP signal processing circuit, and the PWM signal is processed by the second group of signal conditioning circuits and sent to the DSP signal processing circuit;

[0046] B. The current ripple signal is processed by the DSP signal processing circuit to generate a ripple characteristic signal;

[0047] The current ripple characteristic processing is as follows:

[0048] β=(i a+1 -i a )f m (1)

[0049]

[0050] where β=(i a+1 -i a )f m is the current differential processing formula, i a is the sampling value of the a-th current, i a+1 is the sampling value of the a+1-th current, f m is the sampling frequency, triat(β) is the current characteristic acquisition function, and lim is the critical value of the differential amount of the input voltage fluctuation. The reference value of lim is set to 400, and then the ripple characteristic signal is generated according to the triat(β) function;

[0051] C. The phase of the ripple characteristic signal and the PWM signal is compared, the difference between the phase angles of the signals is measured by the phase comparison method, the phase of the ripple characteristic signal is shifted according to the difference between the phase angles, and the phases of the signals are the same;

[0052] D. The increment iterative correlation coefficient operation is performed on the ripple characteristic signal and the PWM signal with the same phase to generate an increment iterative correlation coefficient library;

[0053] The increment iterative correlation coefficient fault diagnosis algorithm is as shown in Figure 2 ;

[0054] The ripple characteristic signal and the PWM signal are sampled, S periods are taken as the correlation operation window, the real-time collected data is stored in the window according to the first-in first-out principle, the increment iterative correlation coefficient operation is performed, the reference value range of the correlation coefficient in the normal working state is set to (0.95-1);

[0055] The incremental iteration type correlation coefficient operation formula is as follows:

[0056]

[0057] Wherein:

[0058]

[0059] ρ ω is the memory correlation coefficient, i.e. the correlation coefficient calculated last time; ρ ω+1 is the real-time correlation coefficient, i.e. the correlation coefficient of the real-time data entering the correlation operation window according to the first-in first-out principle; A ω is the number of high levels in the correlation coefficient window; B ω is the number of low levels in the correlation coefficient window; P ω is the covariance of the last operation, and ΔP is the covariance change amount; Q ω is the product of the standard deviations of the last operation, and ΔQ is the product change amount of the standard deviations. x i ,y i are the objects for correlation coefficient calculation; are the average numbers of x i ,y i in the correlation coefficient window respectively; is the difference between the average numbers of x and y after the real-time data entering according to the first-in first-out principle and the average numbers of the last operation. m is the total number of high and low levels in the correlation coefficient window,

[0060] The specific formula derivation is as follows:

[0061] The calculation formula of ρ ω is as follows:

[0062] Wherein cov(x,y) is the covariance of x and y in the processing window; is the product of the standard deviations of x and y in the window, and since ρ ω is a constant, cov(x,y) and cov(x,y) are also constants.

[0063] Expanding ρ ω , the formula

[0064] Wherein is the average number of x; is the average number of y; x i is the value of the ith x; and y i is the value of the ith y.

[0065] The derivation of Δρ is as follows: Δρ = ρω+1 -ρ ω (6)

[0066] From equation (5), it is too complicated to solve Δρ directly. We can solve the change of numerator and denominator respectively,

[0067] Therefore, ρ ω and ρ ω+1 are expressed as follows:

[0068]

[0069] From equation (7) and equation (8), we can get equation (3):

[0070] Therefore, we can calculate the value of ρ ω by ρ ω , P ω , Q ω+1 , ΔP, ΔQ, i.e., ρ ω+1 = ρ ω f(P ω , Q ω , ΔP, ΔQ).

[0071] Next, we will derive ΔP and ΔQ

[0072] Since the correlation coefficient window follows the principle of same progress and same output, in the case of the duty cycle changing constantly, and the number of high and low levels also changes.

[0073] The derivation process of ΔP and ΔQ is as follows:

[0074] The average of the nth number to the nth+m number of x:

[0075]

[0076] The average of the nth+M number to the nth+m+M number of x:

[0077]

[0078] We get

[0079] Similarly, we get

[0080] The derivation process of A ω and B ω is as follows: A ω +B ω = m = sk

[0081] A1 = (D1 + D2 +... + Ds )k; B1 = (S - D1 - D2 -... - D s )k

[0082] A2 = A1 + high level change amount; B2 = B1 + high level change amount

[0083] ...

[0084] A ω+1 = A ω + high level change amount; B ω+1 = B ω + low level change amount

[0085] Where m is the total number of high and low levels in the correlation coefficient window, s is the number of periods, k is the total number of high and low levels in a period, D s is the duty cycle of the s period.

[0086] △P,△Q are derived as follows:

[0087] Let

[0088] The change amount of the covariance:

[0089] The n-th number to the n+m-th number:

[0090]

[0091] The n+M-th number to the n+m+M-th number:

[0092]

[0093] (14) - (13) gives:

[0094]

[0095] The change amount of the product of the standard deviations:

[0096] The n-th number to the n+m-th number:

[0097]

[0098] The n+M-th number to the n+m+M-th number:

[0099]

[0100] (17) - (16) gives:

[0101]

[0102] E. The circuit failure is detected in real time by monitoring whether the incremental iterative correlation coefficient is mutated. If the correlation coefficient is mutated, the failure is detected.

[0103] F. After the failure is detected, the power supply of the DC-DC converter is immediately cut off by the execution circuit to protect the circuit.

[0104] A detection device for realizing the DC-DC converter failure detection method is connected by a ripple acquisition circuit, a first group of signal conditioning circuits, a DSP signal processing circuit, a PWM signal, a second group of signal conditioning circuits, a DSP signal processing circuit, a DSP signal processing circuit, a execution circuit module, a DC-DC converter, a DC-DC converter connected by magnetic coupling, and a power supply circuit connected to the first group of signal conditioning circuits, the second group of signal conditioning circuits, the DSP signal processing circuit, the PWM acquisition circuit and the ripple acquisition circuit.

[0105] After the ripple signal is collected by the magnetic coupling element, it is amplified by the differential amplifier and input into the signal conditioning circuit for signal conditioning. After the PWM signal is collected by the impedance element, it is amplified by the differential amplifier and input into the signal conditioning circuit for signal conditioning.

[0106] The magnetic coupling element can be any one of a Rogowski coil, a current transformer and a general coil, which can collect the current ripple in the DC converter through magnetic coupling. The collected current ripple signal is input into the DSP signal processing circuit after passing through the signal conditioning circuit for ripple feature processing.

[0107] As shown in Figure 4 The failure detection device for the above method includes a ripple acquisition circuit, a PWM acquisition circuit, a signal conditioning circuit, a DSP signal processing circuit, an execution circuit and a power supply circuit. The current ripple signal and the PWM signal are collected by the ripple acquisition circuit and the PWM acquisition circuit respectively, and input into the DSP signal processing circuit after signal conditioning. L1 and L2 are two ports of the magnetic coupling element; P1 and P2 are two ports of the impedance element. The DSP signal processing circuit generates a ripple feature signal after feature processing of the ripple current signal. Then the ripple feature signal and the PWM signal are phase adjusted, and the zero point positions of the signals are made the same by the phase comparison method. Then, with S cycles as the correlation operation window, the real-time collected data is stored in the window in the principle of first-in first-out for incremental iterative correlation coefficient operation. Finally, the failure is judged by detecting whether the incremental iterative correlation coefficient is mutated. If the incremental iterative correlation coefficient is mutated, the failure occurs. The execution circuit cuts off the power supply of the DC-DC converter through the failure signal, thereby achieving the effect of protecting the circuit. The power supply circuit supplies power to the signal conditioning circuit, the DSP signal processing circuit and the execution circuit respectively.

Claims

1. A method of detecting a fault in a DC-DC converter, characterized by, It comprises the following steps: A. The current ripple acquisition circuit acquires the current ripple signal, and the PWM acquisition circuit acquires the PWM signal. The current ripple signal is processed by the first group of signal conditioning circuits and sent to the DSP signal processing circuit. The PWM signal is processed by the second group of signal conditioning circuits and sent to the DSP signal processing circuit; B. The current ripple signal is processed by the DSP signal processing circuit, and the ripple characteristic signal is generated; C. The ripple characteristic signal and the PWM signal are phase adjusted; The phase adjustment is to measure the difference between the phase angles of the signals by the phase comparison method, and to perform phase translation on the ripple characteristic signal according to the difference between the phase angles, so that the phases of the two signals are the same; D. Incremental correlation coefficient operation is performed on the in-phase ripple characteristic signal and the in-phase PWM signal to generate an incremental correlation coefficient library; The incremental correlation coefficient operation formula is as follows: wherein: ρ ω is the memory correlation coefficient, i.e. the correlation coefficient calculated last time; ρ ω+1 is the real-time correlation coefficient, i.e. the correlation coefficient of the real-time data entering the correlation calculation window according to the principle of first-in first-out; A ω is the number of high levels in the correlation coefficient window; B ω is the number of low levels in the correlation coefficient window; P ω is the covariance calculated last time, and ΔP is the covariance change amount; Q ω is the product of the standard deviations calculated last time, and ΔQ is the product of the standard deviation change amount; x i , y i are the objects for which the correlation coefficient is calculated; are the average values of x i , y i in the correlation coefficient window, respectively. for real-time data after the principle of first-in first-out the difference between the average of the last operation; m is the total number of high and low levels in the correlation coefficient window; E. Whether the circuit fails is determined by monitoring whether the incremental correlation coefficient mutates. If the correlation coefficient mutates, the circuit has a fault; F. The power supply of the DC-DC converter is immediately cut off by executing the circuit, so that the protection circuit is protected.

2. A method for detecting a fault in a DC-DC converter according to claim 1, characterized in that, In step B, the current ripple characteristic processing is as follows: β = (i a+1 -i a )f m (1) Wherein, β = (i a+1 -i a )f m is the current differential processing formula, i a is the sampling value of the a-th current, i a+1 is the sampling value from the a+1-th current, f m is the sampling frequency, triat(β) is the current characteristic acquisition function, and lim is the critical value of the fluctuation differential amount of the input voltage. The reference value of lim is set to 400, and then the ripple characteristic signal is generated according to the triat(β) function.

3. A detecting device for implementing the method for detecting failure of the DC-DC converter according to claim 1, characterized in that, The ripple acquisition circuit is connected to the DSP signal processing circuit through the first group of signal conditioning circuits, and the PWM signal is connected to the DSP signal processing circuit through the second group of signal conditioning circuits. The DSP signal processing circuit is connected to the DC-DC converter through the execution circuit module. The DC-DC converter is connected to the ripple acquisition circuit through magnetic coupling. The power supply circuit is connected to the first group of signal conditioning circuits, the second group of signal conditioning circuits, the DSP signal processing circuit, the PWM acquisition circuit and the ripple acquisition circuit.

4. The detection device of the DC-DC converter fault detection method according to claim 3, characterized in that, After the ripple signal is collected by the magnetic coupling element, it is amplified by the differential amplifier and sent to the signal conditioning circuit for signal conditioning. After the PWM signal is collected by the impedance element, it is amplified by the differential amplifier and input to the signal conditioning circuit for signal conditioning.

Images