Online monitoring method for aging state of T-type three-level inverter

Abstract

The invention discloses a T-type three-level inverter aging state on-line monitoring method, which belongs to the technical field of power electronics, and comprises the following steps: S1, constructing an inverter mathematical model based on a switch state; s2, synchronous acquisition of multi-physical-quantity data is carried out; s3, calculating on resistance through an active voltage vector method; s4, conducting resistance is measured in real time through an in-situ monitoring method; and S5, carrying out weighted fusion on the on-resistance multi-source data. According to the online monitoring method for the aging state of the T-type three-level inverter, respective advantages of an active voltage vector monitoring method and an in-situ monitoring method are integrated, and the aging feature extraction model with a fault-tolerant verification mechanism is constructed by fusing multi-physical-quantity measurement information and operation state data, so that the aging state of the T-type three-level inverter is monitored. The precision bottleneck of a traditional single data source monitoring method can be broken through, the on-resistance change of the power device is accurately monitored in real time, the aging state of the power device is evaluated, and a basis is provided for system maintenance.

Description

Technical Field

[0001] The present invention relates to the technical field of power electronics, and particularly relates to an online monitoring method for the aging state of a T-type three-level inverter. Background Art

[0002] In the prior art, as a typical representative of multilevel converters, the T-type three-level inverter has been widely used in scenarios such as new energy power generation systems, high-power motor drives, and electric vehicle powertrains due to its high efficiency, low harmonic output, and excellent voltage stress characteristics. Among them, power semiconductor devices (IGBTs, MOSFETs, etc.), which are the core components of the inverter, will cause aging failures due to thermo-mechanical stress, electromigration effects, etc. during long-term operation, resulting in the drift of key parameters such as on-resistance. The on-state loss of the aged power device increases significantly, and the abnormal increase in the junction temperature caused thereby will reduce the system reliability.

[0003] Currently, there are mainly three types of technical routes for the aging monitoring technology of power devices: one is the off-line monitoring method based on a sensor network, which realizes parameter acquisition through external current / voltage / temperature sensors. Although the measurement accuracy is high, it is necessary to transform the original circuit hardware, increasing the system complexity and economic cost; the second is the life prediction method based on a reliability model, which calculates the remaining life by establishing a failure physics model. However, due to the complexity of the device degradation mechanism and the time-variability of the working conditions, there are inherent defects in the model generalization ability; the third is the direct monitoring method based on electrical-thermal characteristic parameters, which realizes state evaluation by online extracting characteristic quantities such as on-resistance and threshold voltage. Research results show that under the action of power cycle and thermal cycle aging effects, the on-resistance of power devices will have a significant drift of 3% - 20%, and its change amplitude is much higher than the on-state voltage drop and has a strong linear correlation with the aging degree, which is the optimal characteristic parameter for characterizing the device degradation state.

[0004] However, the existing monitoring methods still face significant technical bottlenecks: the monitoring scheme based on single-sensor data is vulnerable to interference from working conditions such as load fluctuations and temperature changes, resulting in an increase in the error of characteristic quantity extraction; the method based on parameter identification requires the support of an accurate device model, and it is difficult to avoid the problem of model parameter mismatch in practical applications; the data-driven method can avoid the modeling problem, but it requires a large number of fault samples to support, and the engineering implementation cost is high. These problems seriously restrict the application of online monitoring technology. Especially in application scenarios such as offshore wind power generation and aviation motor drives with complex working conditions and strict reliability requirements, accuracy, reliability, and engineering applicability are several important indicators for evaluating the system aging state monitoring method.

[0005] Therefore, there is an urgent need for a solution that can integrate multi-source measurement data to improve the accuracy of the aging state monitoring of T-type three-level inverters and enhance the fault tolerance of on-line monitoring technology. Based on this need, the present invention discloses an on-line monitoring method for the aging state of T-type three-level inverters based on multi-source data fusion. Summary of the Invention

[0006] The object of the present invention is to provide an on-line monitoring method for the aging state of T-type three-level inverters, which can break through the accuracy bottleneck of traditional single data source monitoring methods, monitor the change of the on-resistance of power devices in real time and accurately, evaluate its aging state, and provide a basis for system maintenance.

[0007] To achieve the above object, the present invention provides an on-line monitoring method for the aging state of T-type three-level inverters, including the following steps: S1. Construct an inverter mathematical model based on the switching state; S2. Synchronously collect multi-physical quantity data; S3. Calculate the on-resistance by the active voltage vector method; S4. Measure the on-resistance in real time by the in-situ monitoring method; S5. Perform weighted fusion of multi-source data of the on-resistance.

[0008] Preferably, in S1, according to the topological structure characteristics of the T-type three-level inverter, an accurate mapping relationship between the switching state and the output voltage is established; the switching functions of the three-phase bridge arms are defined as ， , and its value range is {1, 0, -1}, corresponding to: if , then the upper bridge arm of this phase conducts, and the output is P level, + V dc / 2; if , then the midpoint bridge arm of this phase conducts, and the output is O level, 0; if , then the lower bridge arm of this phase conducts, and the output is N level, - V dc / 2; Thus, the mathematical expression of the output voltage on the three-phase AC side is obtained: ; wherein, V dc is the DC bus voltage, is the output voltage on the three-phase AC side; The synthetic space vector expression of the output voltage of the T-type three-level inverter is: ; In the formula, is the synthesized space vector of the output voltage of the T-type three-level inverter. The specific value of the synthesized space vector is related to the combination of the switching functions of the three-phase bridge arms S a , S b , S c , and there are 27 in total.

[0009] Preferably, the synthesized space vector presents a regular hexagon structure in space, including 19 basic voltage vectors, namely: 6 large vectors, whose amplitude is , and each large vector uniquely corresponds to 1 switching state; 6 medium vectors, whose amplitude is , and each medium vector uniquely corresponds to 1 switching state; 6 small vectors, whose amplitude is , and each small vector corresponds to 2 redundant switching states; 1 zero vector, whose amplitude is 0, and each zero vector corresponds to 3 switching states; the switching state is characterized by the combination of the output levels P, O, and N of the bridge arm.

[0010] Preferably, in S2, the following measurement data are obtained in real time through the body voltage / current sensor network of the T-type three-level inverter: DC bus voltage V dc , three-phase load current and , load line voltage V ab , V bc ; after all signals are processed by the isolation conditioning circuit, they are synchronously recorded into the controller at the same sampling rate to construct a multi-dimensional data matrix with time alignment.

[0011] Preferably, in S3, based on the transformation relationship between the three-phase stationary coordinate system and the two-phase rotating coordinate system, a mathematical calculation model of the two-phase rotating coordinate system of the power devices of the T-type three-level inverter is constructed: The Clarke transformation is used to convert the measured load line voltage into the voltage vector in the α - β coordinate system. The Clarke transformation formula is: ; Among them, V ab , V bc are the line voltages between phases a-b and b-c on the load side respectively; V α , V β are respectively the α - β axis of the load line voltage in the coordinate system, α axis,β Axis component; Under different switch states, V α 、 V β The values of are different. Further introduce to represent the vector Under the action of α - β Axis and α Axis components of the load voltage in the coordinate system, where β Axis components, where ; Vector Has a one-to-one correspondence with the switch state ( , S b , S c ).

[0012] Preferably, define And Are the conduction voltages of the power devices in phases a, b, and c respectively, where And Are the numbers of the three-phase switch devices corresponding to the vector state, ; 23 represents the switch devices T x2 And T x3 ; 、 V ob And V oc Represent the load voltages of phases a, b, and c including the conduction resistance respectively; And Represent the load voltages of phases a, b, and c without the conduction resistance respectively, as follows: ; Use the Clarke transformation to transform the And In the a-b-c coordinate system to α - β Coordinate system, and obtain And : ; Replace the conduction voltages of the power devices And With the product form of the conduction resistance and the load current, we get: ; Among them, And Represent the conduction resistances of the switch devices in phases a, b, and c respectively. When or When the value is 1 or 4, it represents the on-resistance of a single switch 1 or 4. When the value is 23, it represents the on-resistance of the two switches 2 and 3 in series. All represent load current, the subscript Indicates the voltage vector number of the load current measured by the sensor at the current moment. The letter part indicates the current current flowing through the power device. Phase load current, ; Infer the circuit connection mode under each switching state and obtain the corresponding mathematical expression between the voltage vector, on-resistance and load current; Combine the mathematical expressions of different switch states to solve the on-resistance of the power device and get and The values are: ; The subscript part and All are vector state numbers. .

[0013] Preferably, in S4, a measurement circuit is added near the power device for monitoring, and the measurement circuit consists of two parts: a clamping circuit and a parameter acquisition circuit; the clamping circuit includes: Schottky diodes D1 and D2, a voltage regulator diode DZ2, a current limiting resistor R, a capacitor C and a DC voltage source; the parameter acquisition circuit includes: differential operational amplifiers OP1 and OP2; The on-state current is obtained indirectly by connecting a precision sampling resistor in series with the power device. The on-state voltage drop is divided by the on-state current to obtain the on-state resistance of the power device to be measured.

[0014] Preferably, in S5, the on-resistance values obtained by the active voltage vector method and the in-situ monitoring method are respectively recorded as R 1. R 2. The calculation formula is as follows: ; in, σ 1 and σ 2 are random errors of the resistance values obtained. R 1 and R 2 are independent of each other, R 0 is the actual value of the resistor, assuming R Estimated value of 0 With the obtained value R 1 and R 2 is linearly related, and yes R The unbiased estimate of , then: ; Among them, is the weight coefficient of the values obtained by the two methods: ; In the process of calculation using test data, it is assumed that the initial errors of each method are the same, that is, the weight coefficient = = 0.5; In actual operation, the R values calculated by each method are checked against the regular test data, and the weight coefficients of each method are corrected; Applying the data of the active voltage vector method and the in-situ monitoring method, the weighted fusion resistance value obtained is: ; In the formula, and are the mean square errors calculated through prior knowledge or test data of the active voltage vector method and the in-situ monitoring method.

[0015] Preferably, to prevent the final resistance value R F from being untrustworthy due to the failure of a certain sensor, the criterion formula used is: ; Among them, R 0 is the rated on-resistance of the power device under test, is the actual allowable error; Only when the criterion formula holds can the weighted fusion of multi-source data be used to calculate the final resistance value R F ; Otherwise, when , it is determined that the resistance value R 1 obtained by the active voltage vector method is untrustworthy, and at this time is taken, that is ; When , it is determined that the resistance value R 2 obtained by the in-situ monitoring method is untrustworthy, and at this time is taken, that is ; Among them, R Fp is the average on-resistance value calculated from the previous n data windows, and the initial value is taken as R 0; When the resistance value satisfies , it is determined that the power device is faulty and an alarm is issued, otherwise wait for the calculation of the next cycle.

[0016] Therefore, the beneficial effects of the present invention adopting the above-mentioned online monitoring method for the aging state of a T-type three-level inverter are: (1) By integrating the system operating state data (load current, DC bus voltage, load voltage, switch state, etc.) with the in-situ monitoring data of the power device conduction resistance, the present invention constructs a high-precision extraction method for aging characteristics with a redundant verification mechanism, effectively overcoming the defect that the traditional single-sensor data is vulnerable to the interference of working condition fluctuations, and improving the credibility and robustness of the monitoring results of the method.

[0017] (2) The present invention uses the inverter body sensors (load current sensor, DC bus voltage sensor, load voltage sensor) and the in-situ monitoring circuit of the power device as multi-source data inputs, without the need to add dedicated monitoring hardware, reducing the hardware transformation cost compared with the traditional external sensor scheme.

[0018] (3) Through multi-source data fusion, the present invention can update the aging parameters online, realize early warning of the remaining life of the device, and support predictive maintenance.

[0019] The technical solution of the present invention will be further described in detail below with reference to the drawings and embodiments. Description of the Drawings

[0020] Figure 1 [[ID= :15]] is a schematic diagram of the online monitoring and control system for the aging state of the T-type three-level inverter in the present invention; Figure 2 is a synthetic space vector distribution diagram of the output voltage of the T-type three-level inverter in the present invention; Figure 3 is a schematic diagram of the conduction resistance monitoring based on the active voltage vector method in the present invention; Figure 4 is a schematic diagram of the connection state between the inverter and the load under the action of the voltage vector V1(POO) in the present invention; Figure 5 is a wiring schematic diagram of the in-situ monitoring method in the present invention; Figure 6 is a schematic diagram of the in-situ monitoring measurement circuit in the present invention; Figure 7 is a schematic diagram of the weighted fusion algorithm for multi-source data of the conduction resistance in the present invention; Figure 8 is a flowchart of the online monitoring method for the aging state of the T-type three-level inverter in the present invention.

[0021] Reference Signs 1. DC voltage source; 2. T-type three-level inverter circuit; 3. Load; 4. Controller; 5. Online monitoring module for aging state. Specific Embodiments

[0022] The technical solution of the present invention will be further described below with reference to the drawings and embodiments.

[0023] Unless otherwise defined, the technical terms or scientific terms used in the present invention shall have the ordinary meanings as understood by those of ordinary skill in the art to which the present invention pertains. The "first", "second" and similar terms used in the present invention do not denote any order, quantity or importance, but are only used to distinguish different components. The terms such as "comprising" or "including" mean that the elements or objects appearing before this term cover the elements or objects listed after this term and their equivalents, without excluding other elements or objects. The terms such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "upper", "lower", "left" and "right" are only used to represent relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0024] Embodiment 1 As Figure 1 shown, the present invention provides an online monitoring and control system for the aging state of a T-type three-level inverter, which includes a DC voltage source 1, a T-type three-level inverter circuit 2, a load 3, a controller 4 and an online monitoring module 5 for the aging state. The controller 4 is connected to the T-type three-level inverter circuit 2. The controller 4 collects analog signals such as load current, DC bus voltage, and load voltage signals, and after calculation, sends switch state instructions to the T-type three-level inverter circuit 2 in the form of space vector pulse width modulation (SVPWM). The online monitoring module 5 for the aging state is connected to the T-type three-level inverter circuit 2 to realize the reading of in-situ monitoring measurement data. The controller 4 is connected to the online monitoring module 5 for the aging state to realize the reading of system operation state data (load current, DC bus voltage, load voltage, switch state); finally, the online monitoring module 5 for the aging state realizes multi-source data fusion and provides high-precision monitoring results.

[0025] The present invention also provides an online monitoring method for the aging state of a T-type three-level inverter, including the following steps: S1. Construct an inverter mathematical model based on switch states.

[0026] According to the topological structure characteristics of the T-type three-level inverter, an accurate mapping relationship between switch states and output voltages is established. Define the switch functions of the three-phase bridge arms as , whose value range is {1, 0, -1}, corresponding respectively: if , then the upper arm of this phase conducts and outputs a P level (+ V 1] dc / 2); if , then the midpoint arm of this phase conducts and outputs an O level (0); if , then the lower arm of this phase conducts and outputs an N level (- V dc / 2).

[0027] Thus, the mathematical expression of the three-phase AC side output voltage is obtained: (1); Wherein, V dc is the DC bus voltage, is the three-phase AC side output voltage.

[0028] The synthetic space vector expression of the output voltage of the T-type three-level inverter is: (2); In the formula, is the synthetic space vector of the output voltage of the T-type three-level inverter. The specific value of the synthetic space vector is related to the combination of the three-phase bridge arm switching functions S a , S b , S c . The switching state of the T-type three-level inverter is actually the combination of switching functions (Sa, Sb, Sc), and there are 27 possibilities, so there are 27 such synthetic space vectors.

[0029] According to Equation (2), the synthetic space vector distribution diagram of the output voltage of the T-type three-level inverter as shown in Figure 2 can be generated. The synthetic space vectors present a regular hexagon structure in space, including 19 basic voltage vectors, which are: 6 large vectors with an amplitude of , and each vector uniquely corresponds to 1 switching state; 6 medium vectors with an amplitude of , and each vector uniquely corresponds to 1 switching state; 6 small vectors with an amplitude of , and each vector corresponds to 2 redundant switching states; 1 zero vector with an amplitude of 0, corresponding to 3 switching states. Therefore, the switching state is characterized by the combination of the arm output levels P (+ V dc / 2), O (0), N (- V dc / 2).

[0030] During the operation of the T-type three-level inverter, the controller selects the voltage vector to be output at the current moment according to the SVPWM principle to determine the on or off of the power devices, and at the same time converts the selected voltage vector into the corresponding switching state and submits it to the aging state online monitoring module 5 for the calculation of the on-resistance. S2. Synchronously collect multi-physical quantity data.

[0031] S2. Synchronously collect multi-physical quantity data.

[0032] The following measurement data are obtained in real time through the body voltage / current sensor network of the T-type three-level inverter: (1) DC bus voltage V dc ; (2) Three-phase load currents , and ; (3) Load line voltages V ab , V bc .

[0033] After all signals are processed by the isolation conditioning circuit, they are synchronously recorded into the controller at the same sampling rate to construct a time-aligned multi-dimensional data matrix.

[0034] S3. Calculate the on-resistance through the active voltage vector method.

[0035] As Figure 3 shown, the aging state online monitoring module 5 reads the system operation state data (load current, DC bus voltage, load voltage, switch state) transmitted by the controller, combines different switch states with the corresponding current paths, and calculates the real-time monitoring value of the on-resistance of the power device through the constraint relationship of Kirchhoff's voltage equation.

[0036] Based on the transformation relationship between the three-phase stationary coordinate system and the two-phase rotating coordinate system, a two-phase rotating coordinate system mathematical calculation model of the power device of the T-type three-level inverter is constructed.

[0037] In this embodiment, the Clarke transformation is used to convert the measured load line voltage into α - β the voltage vector in the coordinate system. The Clarke transformation formula is: (3); Among them, V ab , V bc are the line voltages between phases a-b and b-c on the load side respectively; V α , V β are the α - β axis and α axis components of the load line voltage in the coordinate system respectively. The load line voltages β here V ab , V bc contain the on-resistance information of the power device and are converted to α - β in the coordinate systemV α , V β also includes the on-resistance information of the power device.

[0038] Under different switching states, V α , V β the values of are different. Therefore, and are further introduced to represent the , components of the load voltage in the α-β coordinate system under the action of the vector , where α -axis, β -axis components, and .

[0039] The vector V i has a one-to-one correspondence with the switching states ( , S b , S c ) as follows: V 1 (1, 0, 0), V 2 (1, 0, -1), V 3 (1, -1, -1), V 4 (1, -1, 0), V 5 (1, -1, 1), V 6 (1, 0, 1), V 7 (1, 1, 1), V 8 (1, 1, 0), V 9 (1, 1, -1), V 10 (0, 1, -1), V 11 (0, 1, 0), V 12 (, 1, 1), V 13 (0, 0, 1), V 14 (0, 0, 0), V 15 (0, 0, -1), V 16 (0, -1, -1), V 17 (0, -1, 0), V 18 (0, -1, 1), V 19 (-1, -1, 1), V20 (-1, -1, 0), V 21 (-1, -1, -1), V 22 (-1, 0, -1), V 23 (-1, 0, 0), V 24 (-1, 0, 1), V 25 (-1, 1, 1), V 26 (-1, 1, 0), V 27 (-1, 1, -1).

[0040] The basic principle of the active voltage vector method: Define and as the conduction voltages of the power devices in phases a, b, and c respectively, where and are the numbers of the three-phase switching devices corresponding to the vector states, ; 23 represents the switching devices T x2 and T x3 ; 、 V ob and V oc represent the load voltages of phases a, b, and c including the conduction resistance respectively; and represent the load voltages of phases a, b, and c without the conduction resistance respectively, which are expressed as follows: (4); Use the Clarke transformation to transform and in the a-b-c coordinate system to α - β coordinate system, and obtain and : (5); Replace the power device conduction voltages and with the product form of the conduction resistance and the load current, and get: (6); Among them, and represent the conduction resistances of the switching devices in phases a, b, and c respectively. When or When the value is 1 or 4, it represents the on-resistance of single tube 1 or 4. When the value is 23, it represents the on-resistance of the two tubes in series of switch tubes 2 and 3. All represent load current, the subscript Indicates the voltage vector number of the load current measured by the sensor at the current moment. The letter part indicates the current current flowing through the power device. Phase load current, .

[0041] Infer the circuit connection mode under each switching state and obtain the corresponding mathematical expression between the voltage vector, on-resistance and load current; Combine the mathematical expressions of different switch states to solve the on-resistance of the power device and get and The values are: (7); The subscript part and All are vector state numbers. .

[0042] like Figure 4 As shown, the switch state V Take 1 (1, 0, 0) as an example, V TA1 、 V TB2 、 V TB3 、 V TC2 and V TC3 They are power devices T A1 、T B2 、T B3 、T C2 and T C3 The on-state voltage; 、 V ob and V oc Respectively represent the load voltages of phase a, phase b, and phase c including on-resistance; V 1a 、 V 1b and V 1c They represent the load voltages of phases a, b, and c, excluding on-resistance, as follows: (8); Use Clarke transformation to transform the abc coordinate system V 1a 、 V1b and V 1c Converted to the α-β coordinate system, we get V 1α and V 1β : (9); Replace the conduction voltages of the power devices V TA1 , V TB2 , V TB3 , V TC2 and V TC3 with the product form of the on-resistance and the load current, we get: (10); wherein R TA1 , R TB23 and R TC23 respectively represent the on-resistances of the single transistor T A1 , the series connection of the transistors T B2 and T B3 , and the series connection of the transistors T C2 and T C3 . All represent the load current. The numerical part of the subscript represents the voltage vector number of the load current measured by the sensor at the current moment, and the letter part represents the phase load current flowing through the power device . For example, represents the a-phase load current measured under the action of the V 1 vector, and this current only flows through the power device T A1 .

[0043] The V 1α , V 1β on the left side of the equation set (10) are obtained by converting the load line voltages V ab , V bc through Equation (3), while the DC bus voltage V dc on the right side, as well as the load currents and are all obtained by synchronously collecting sensor data. Therefore, the equation set (10) is also about the on-resistances R TA1 ,R TB23 and R TC23 A system of three linear equations.

[0044] According to the above method, the circuit connection mode in other switching states can be obtained in the same way, and the mathematical expressions between the corresponding voltage vector, on-resistance and load current can be obtained. By combining the mathematical expressions of multiple switching states, the on-resistance of the power device can be solved. For example, by analyzing the voltage vector V 1 and V 23 The circuit wiring can be obtained R TB23 and R TC23 The values are: (11); in, and V 23β They are V 23 The load current and load line voltage of phase b and phase c under the action of vector β Axis component.

[0045] The on-resistances of the remaining power devices are calculated in the same way as above.

[0046] S4. Measure the on-resistance in real time through the in-situ monitoring method.

[0047] The wiring diagram of the in-situ monitoring method is as follows Figure 5 As shown in the figure, it is necessary to add a measurement circuit near the power device (take IGBT as an example) for monitoring.

[0048] The specific structure of the measurement circuit is as follows: Figure 6 As shown in Figure 1, the circuit consists of a clamping circuit and a parameter acquisition circuit. The clamping circuit includes Schottky diodes D1 and D2, a Zener diode DZ2, a current-limiting resistor R, a capacitor C, and a DC voltage source (+5V). The parameter acquisition circuit includes differential operational amplifiers OP1 and OP2.

[0049] The clamping circuit is responsible for clamping the high voltage when the IGBT is turned off; when the IGBT is turned on, only a low voltage is passed, that is, the conduction voltage drop V on。The Schottky diode D1 is connected to the collector of the IGBT, allowing only current to flow from the measurement circuit to the collector of the IGBT. Taking the emitter of the IGBT as a reference: when the IGBT is turned off, the potential of its collector is higher than that of the emitter. At this time, the Schottky diode D1 is reverse-biased and disconnects the collector from the measurement circuit; when the IGBT is turned on, there is a small potential difference between the collector and the emitter of the IGBT, which forward-biases the Schottky diode D1, and the circuit forms a closed loop through the IGBT. The voltage at node 1 is the conduction voltage drop of the IGBT V on and the forward voltage drop of the Schottky diode D1 V D The sum of. Subtract the forward voltage drop of the Schottky diode from the absolute value of the voltage at node 1 V D to obtain the conduction voltage drop of the power device V on 。

[0050] Generally, the forward voltage drop of the Schottky diode depends on the current flowing through it and the current junction temperature. In theory, if the current junction temperature is known, the forward voltage drop of the Schottky diode can be estimated; however, accurate and reliable junction temperature estimation is difficult in an actual circuit. To overcome this problem, another Schottky diode D3 with exactly the same parameters is connected in series at the rear stage of the Schottky diode D1, and its position is close to D1, so that they operate at approximately the same junction temperature. When the IGBT is turned on, the same current flows through D1 and D2, ensuring that D1 and D2 have the same forward voltage drop

[0051] In the parameter acquisition circuit, the differential operational amplifier OP1 measures the forward voltage drop across the Schottky diode D3, and the output voltage of OP1 is V D , which is also the voltage at node 2. The differential operational amplifier OP2 measures the voltage difference between node 1 and node 2. According to the above analysis, the output voltage of OP2 (i.e., the voltage at node 3) is the conduction voltage drop of the power device to be measured V on 。

[0052] The conduction current can be indirectly obtained by connecting a precision sampling resistor in series at the emitter of the IGBT. Divide the conduction voltage drop by the conduction current to finally obtain the conduction resistance of the power device to be measured

[0053] S5. Perform weighted fusion of multi-source data of the conduction resistance

[0054] The schematic diagram of the weighted data fusion algorithm is as Figure 7 shown, and its core idea is to minimize the total mean square error and, according to the measurement data obtained by each sensor Find its corresponding optimal weighting factor Make the fused estimated value reach the optimum.

[0055] In this embodiment, the active voltage vector method and the in-situ monitoring method are respectively denoted as Method 1 and Method 2. The on-resistance values obtained by Method 1 and Method 2 are respectively denoted as R 1, R 2, and the calculation formulas are as follows: (12); Wherein, σ 1 and σ 2 are both random errors of the obtained resistance values, R 1 and R 2 are independent of each other, R 0 is the true resistance value. Assume that R the estimated value of 0 and the obtained value R 1 and R 2 are linearly related, and is R an unbiased estimate of, then there is: (13); Wherein, is the weight coefficient of the values obtained by the two methods: (14); In the process of calculating using the test data, it is assumed that the initial errors of each method are the same, that is, the weight coefficient . In actual operation, the R values calculated by each method are checked against the regular test data, and the weight coefficients of each method are corrected.

[0056] Apply the weighted data fusion method to the on-line monitoring of the on-resistance. The flow of the on-line monitoring method for the aging state of the designed inverter is as Figure 8 shown. Applying the data of Method 1 and Method 2, the weighted fused on-resistance value is: (15); In the formula, and are the mean square errors calculated through prior knowledge or through the test data of the active voltage vector method and the in-situ monitoring method, and are the weight coefficients of Method 1 and Method 2 respectively.

[0057] To prevent the final on-resistance value R F from being untrustworthy due to the failure of a certain sensor, the criterion formula used is: (16); Wherein, R 0 is the rated on-resistance of the power device under test (which can be obtained by querying the data sheet), is the actual allowable error.

[0058] Only when equation (16) holds can the weighted fusion of multi-source data be used to calculate the final resistance value R F . Otherwise, when ( R Fp is the average on-resistance value calculated from the first n data windows, and the initial value is taken as R 0), the resistance value R 1 obtained by judgment method 1 is not credible. At this time, take , that is ; when , the resistance value R 2 obtained by judgment method 2 is not credible. At this time, take , that is .

[0059] When the resistance value satisfies , it is judged that the power device is faulty and an alarm is issued. Otherwise, wait for the calculation of the next cycle.

[0060] Therefore, the present invention adopts the above-mentioned online monitoring method for the aging state of a T-type three-level inverter, which can break through the accuracy bottleneck of the traditional single data source monitoring method, monitor the change of the on-resistance of the power device in real time and accurately, evaluate its aging state, and provide a basis for system maintenance.

[0061] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that: they can still modify or equivalently replace the technical solutions of the present invention, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. An on-line monitoring method for the aging state of a T-type three-level inverter, characterized in that It includes the following steps: S1. Construct an inverter mathematical model based on the switch states; S2. Synchronously collect multi-physical quantity data; S3. Calculate the on-resistance by the active voltage vector method; S4. Measure the on-resistance in real time by the in-situ monitoring method; S5. Perform weighted fusion of multi-source data of the on-resistance.

2. An online monitoring method for the aging state of a T-type three-level inverter according to claim 1, characterized in that: In S1, according to the topological structure characteristics of the T-type three-level inverter, an accurate mapping relationship between the switching state and the output voltage is established; the switching functions of the three-phase bridge arms are defined as ， , and its value range is {1, 0, -1}, corresponding respectively to: if , then the upper bridge arm of this phase conducts, outputs the P level, + V dc / 2; if , then the midpoint bridge arm of this phase conducts, outputs the O level, 0; If , then the lower arm of this phase conducts, outputting N level, - V dc / 2; Thus, the mathematical expression of the output voltage on the three-phase AC side is obtained: ； Among them, V dc is the DC bus voltage, is the three-phase AC side output voltage; The synthetic space vector expression of the output voltage of the T-type three-level inverter is: ； In the formula, is the synthesized space vector of the output voltage of the T-type three-level inverter. The specific value of the synthesized space vector is related to the combination of the switching functions S a , S b , S c , and there are 27 of them.

3. An on-line monitoring method for the aging state of a T-type three-level inverter according to claim 2, characterized in that: The synthesized space vector presents a regular hexagon structure in space and includes 19 basic voltage vectors, namely: 6 large vectors with an amplitude of , and each large vector uniquely corresponds to one switching state; 6 medium vectors with an amplitude of , and each medium vector uniquely corresponds to one switching state; 6 small vectors with an amplitude of , and each small vector corresponds to two redundant switching states; 1 zero vector, whose amplitude is 0, and each zero vector corresponds to 3 switch states; The switch state is characterized by the combination of the arm output levels P, O, and N.

4. The on-line monitoring method for the aging state of a T-type three-level inverter according to claim 2, characterized in that: In S2, the following measurement data are obtained in real time through the body voltage / current sensor network of the T-type three-level inverter: DC bus voltage V dc , three-phase load current , and , load line voltage V ab , V bc ; after all signals are processed by the isolation conditioning circuit, they are synchronously recorded into the controller at the same sampling rate to construct a multi-dimensional data matrix with time alignment.

5. An on-line monitoring method for the aging state of a T-type three-level inverter according to claim 4, characterized in that: In S3, based on the transformation relationship between the three-phase stationary coordinate system and the two-phase rotating coordinate system, construct a two-phase rotating coordinate system mathematical calculation model for the power devices of the T-type three-level inverter: Use the Clarke transformation to convert the measured load line voltage into a voltage vector in the α-β coordinate system. The Clarke transformation formula is: ； Among them, V ab , V bc are the line voltages between phases a-b and b-c on the load side, respectively; V α , V β are the α-axis and β-axis components of the load line voltage in the α-β coordinate system, respectively. Introduction and are used to represent the under the action of the vector α - β axis and α axis components of the load voltage in the coordinate system, where β ; the vector has a one-to-one correspondence with the switch state ( , , S b , S c ).

6. The on-line monitoring method for the aging state of a T-type three-level inverter according to claim 5, characterized in that: Definition 、 and are the conduction voltages of the power devices of phase a, phase b, and phase c, respectively, where , and are the numbers of the three-phase switch devices corresponding to the vector states; ; 23 represents the switch devices T x2 and T x3 ; 、 and represent the load voltages of phase a, phase b, and phase c including the on-resistance, respectively; 、 and represent the load voltages of phase a, phase b, and phase c without the on-resistance, respectively, as shown below: ； Use the Clarke transformation to transform the , and from the a-b-c coordinate system to the α - β coordinate system, obtaining and : ； Replace the on-state voltage of the power device , and with the product of the on-resistance and the load current, resulting in: ； Among them, , and respectively represent the on-resistances of the switching devices of phase a, phase b, and phase c. When , or takes a value of 1 or 4, it represents the on-resistance of single transistor 1 or 4, and when taking a value of 23, it represents the on-resistance of the series connection of transistors 2 and 3; both represent the load current. The subscript represents the voltage vector number of the load current at the current moment measured by the sensor, and the letter part represents the phase load current flowing through the power device, ; Deduce the circuit connection modes in each switch state, and obtain the mathematical expressions between the corresponding voltage vectors, on-resistances, and load currents; Combine the mathematical expressions of different switch states to solve the on-resistance of the power device and get 、 and The values are: ； where the subscript part and are both vector state numbers, .

7. An online monitoring method for the aging state of a T-type three-level inverter according to claim 1, characterized in that: In S4, add a measurement circuit near the power device for monitoring. The measurement circuit consists of two parts: a clamping circuit and a parameter acquisition circuit; The clamping circuit includes: Schottky diodes D1 and D2, zener diode DZ2, current-limiting resistor R, capacitor C, and DC voltage source; The parameter acquisition circuit includes: differential operational amplifiers OP1 and OP2; The on-current is indirectly obtained by connecting a precision sampling resistor in series with the power device. Divide the on-voltage drop by the on-current to finally obtain the on-resistance of the power device to be measured.

8. An on-line monitoring method for the aging state of a T-type three-level inverter according to claim 1, characterized in that: In S5, the on-resistance values obtained by the active voltage vector method and the in-situ monitoring method are respectively denoted as R 1、 R 2, and the calculation formulas are as follows: ； wherein, σ 1 and σ 2 are both random errors of the obtained resistance values, R 1 and R 2 are independent of each other, R 0 is the true value of the resistance. Assuming that R the estimated value of 0 and the obtained values R 1 and R 2 are linearly related, and is R an unbiased estimate of, then there is: ； Among them, is the weight coefficient of the values obtained by the two methods: ； In the process of calculation using test data, it is assumed that the initial errors of each method are the same, that is, the weight coefficient ; In actual operation, the R values calculated by each method are checked against the regular test data, and the weight coefficients of each method are corrected; Applying the data of the active voltage vector method and the in-situ monitoring method, the weighted fusion resistance value is: ； wherein, and are the mean square errors calculated from prior knowledge or test data by the active voltage vector method and the in-situ monitoring method.

9. An online monitoring method for the aging state of a T-type three-level inverter according to claim 8, characterized in that: To prevent the final resistance value from being untrustworthy due to a failure of a certain sensor R F The criterion formula used is: ； Among them, R 0 is the rated on-resistance of the power device under test, is the actual allowable error; Only when the criterion formula holds can the weighted fusion of multi-source data be used to calculate the final resistance value R F ; otherwise, when the resistance value obtained by the active voltage vector method is judged R 1 is not credible. At this time, take , that is ; when the resistance value obtained by the in-situ monitoring method is judged R 2 is not credible. At this time, take , that is ; where R Fp is the average on-resistance value calculated from the first n data windows, and the initial value is taken as R 0; When the resistance value meets , determine that the power device is faulty and issue an alarm; otherwise, wait for the calculation of the next cycle.

Images