JP6451533B2 - Current sensor abnormality diagnosis device - Google Patents

Description

  The present invention relates to a current sensor abnormality diagnosis device that diagnoses abnormality of a current sensor.

  2. Description of the Related Art Conventionally, there is known an apparatus that compares outputs of a plurality of current sensors that detect currents flowing in the same current path and diagnoses an abnormality of the current sensor based on an output deviation. For example, in the configuration disclosed in Patent Document 1, in an inverter device that drives and controls a motor generator (AC motor) of a hybrid vehicle, two current sensors that detect a phase current of two phases out of three phases are provided for each phase. It has been. When the output deviation between the two current sensors exceeds a predetermined abnormality determination value, it is determined that the current sensor is abnormal.

JP 2005-160136 A

The current sensor abnormality detection routine of Patent Document 1 is always executed while the motor generator is being driven. By the way, in general, the output characteristics of the current sensor, that is, the offset and gain characteristics of the sensor output with respect to the true current value are affected by disturbances such as temperature changes. Hereinafter, examples of disturbance will be described as temperature changes.
The outputs of the two current sensors that detect the current flowing in the same current path differ depending on the variation in the temperature characteristics of each sensor when the environmental temperature changes, in addition to the variation in the product for each individual. For this reason, if an abnormality determination value is set without taking into account temperature changes during energization, an output deviation caused by temperature characteristics may be erroneously determined to be abnormal although the current sensor is normal. However, if the abnormality determination value is set excessively in order to prevent erroneous determination, there is a problem that the abnormality detection performance at the time of actual abnormality deteriorates.

  The present invention was created in view of the above points, and an object of the present invention is to provide a current sensor abnormality diagnosis device that improves abnormality detection performance while preventing erroneous determination due to the influence of a temperature change or the like. It is in.

  The present invention relates to a current sensor abnormality diagnosis device that compares outputs of a plurality of current sensors that detect currents flowing in the same current path and diagnoses abnormality of the plurality of current sensors based on an output deviation. The current sensor abnormality diagnosis device includes a learning command unit, a current correction unit, an abnormality determination unit, and a diagnosis execution permission unit.

The learning command unit instructs the execution of “reference output learning” for offset correction of the outputs of the plurality of current sensors so that the output deviations of the plurality of current sensors at the time of learning reference are zero.
After the reference output learning, the current correction unit outputs a correction output in which the outputs of the plurality of current sensors are offset-corrected according to the correction in the reference output learning.
The abnormality determination unit performs abnormality diagnosis of the plurality of current sensors based on a correction output deviation that is a deviation between the correction outputs of the plurality of current sensors.
The diagnosis execution permission unit permits the abnormality determination unit to perform abnormality diagnosis during a diagnosis permission period from when the reference output learning is completed until a predetermined time elapses.

  In the present invention, first, reference output learning is performed according to a command from the learning command unit, and offset correction is performed so that the output deviations of a plurality of current sensors at the time of learning reference are zero. Thereby, offset tolerance can be eliminated among tolerances of offset characteristics and gain characteristics of a plurality of current sensors. Note that “0” in “set the output deviation to 0” is not limited to strict 0, but means a value in a range substantially regarded as 0 in consideration of the resolution of the apparatus.

In addition, the diagnosis execution permission unit permits the execution of the abnormality diagnosis only during the diagnosis permission period in which the corrected output deviation after the reference output learning is within a relatively small range. It can be suppressed to a minute range.
Thereby, the cause of the tolerance of the current sensor output characteristic at the time of abnormality diagnosis can be only the gain characteristic tolerance due to the product variation. Therefore, the abnormality determination value for preventing erroneous determination can be set as small as possible. Therefore, it is possible to reliably detect an abnormality of the actual current sensor at an early stage, and it is possible to improve the abnormality detection performance.

1 is a schematic configuration diagram of a motor drive system of a hybrid vehicle to which a current sensor abnormality diagnosis device according to an embodiment of the present invention is applied. The figure explaining the change of the output deviation of two current sensors in the structure which does not implement (a) reference | standard output learning, and (b) reference | standard output learning. The control block diagram of the current sensor abnormality diagnostic apparatus by one Embodiment of this invention. The time chart of the current sensor abnormality diagnosis by one Embodiment of this invention. The time chart of the current sensor abnormality diagnosis of a comparative example. The flowchart of the current sensor abnormality diagnosis by one Embodiment of this invention. The figure explaining the tolerance of the current sensor output when not performing reference output learning. The figure explaining the tolerance of the current sensor output at the time of performing reference output learning and setting a diagnosis permission period.

  Hereinafter, an embodiment of a current sensor abnormality diagnosis device of the present invention will be described with reference to the drawings. The current sensor abnormality diagnosis device of this embodiment detects an abnormality of a current sensor that detects a phase current flowing in a current path from an inverter to a motor generator in a system that drives a motor generator that is a power source of a hybrid vehicle or an electric vehicle. It is a device to do. In the following specification, the motor generator is referred to as “AC motor” or simply “motor”.

(One embodiment)
First, the configuration of the entire motor drive system will be described with reference to FIG. The motor drive system 90 mounted on the hybrid vehicle 100 converts the DC power of the battery 20 into three-phase AC power by the inverter 30 and supplies it to an AC motor 80 (“MG” in the figure) as a “load”. This is a system for driving the motor 80.
The battery 20 is a chargeable / dischargeable secondary battery such as a nickel metal hydride battery or a lithium ion battery. Instead of the battery, an electric double layer capacitor or the like may be used as a DC power source.
The smoothing capacitor 25 smoothes the inverter input voltage.

  In the inverter 30, six switching elements 31 to 36 of upper and lower arms are bridge-connected. Specifically, the switching elements 31, 32, and 33 are upper-arm switching elements of the U phase, the V phase, and the W phase, respectively, and the switching elements 34, 35, and 36 are below the U phase, the V phase, and the W phase, respectively. This is an arm switching element. The switching elements 31 to 36 are made of, for example, an IGBT (insulated gate bipolar transistor), and are connected in parallel with freewheeling diodes that allow a current from the low potential side to the high potential side.

The inverter 30 converts DC power into three-phase AC power by switching operations of the switching elements 31 to 36 according to drive signals UU, VU, WU, UL, VL, WL from the drive circuit 40 by PWM control or the like.
Note that the motor drive system according to another embodiment may include a boost converter that boosts the DC voltage of the battery 20, and the boosted voltage may be input to the inverter 30.

The motor 80 is, for example, a permanent magnet type synchronous three-phase AC motor. In the present embodiment, the motor 80 is mounted on the hybrid vehicle 100 including the engine 91. The motor 80 has a function as an electric motor that generates torque for driving the driving wheels 95 of the hybrid vehicle 100 and a function as a generator that recovers energy by generating electric power transmitted from the engine 91 and the driving wheels 95. It functions as a “motor generator”.
The motor 80 is connected to the axle 94 via a gear 93 such as a transmission. The torque generated by the motor 80 drives the driving wheel 95 by rotating the axle 94 via the gear 93. Although FIG. 1 illustrates a system configuration including one motor 80, the present embodiment can be similarly applied to a system including two or more motor generators.

A U-phase winding 81, a V-phase winding 82, and a W-phase winding 83 are wound around the stator of the motor 80. A control current sensor is provided in the two-phase winding of the three phases. In the example of FIG. 1, phase currents Iv and Iw are detected in the V-phase winding 82 and the W-phase winding 83, respectively, and output to the feedback (“F / B” in the figure) control unit 55 of the MG-ECU 50. Control current sensors 71 and 73 are provided.
The rotation angle sensor 85 is a resolver, for example, and detects the electrical angle θ of the rotor.

  The HV-ECU 10 receives signals such as an accelerator signal, a brake signal, a shift signal, and a vehicle speed signal and information from other ECUs, comprehensively determines the driving state of the vehicle based on the acquired information, and drives the vehicle. To control. Other ECUs include a battery ECU that controls the battery 20, an engine ECU that controls the engine 91, and the like, in addition to the MG-ECU 50 that controls the motor 80. In FIG. 1, illustration of a battery ECU, an engine ECU, and the like is omitted.

  Each ECU is configured by a microcomputer or the like, and includes a CPU, a ROM, an I / O (not shown), a bus line that connects these configurations, and the like. Each ECU executes control by software processing by executing a program stored in advance by the CPU or hardware processing by a dedicated electronic circuit.

The MG-ECU 50 of the present embodiment includes a feedback control unit 55 and a current sensor abnormality diagnosis device 60.
Feedback control unit 55 receives information such as torque command trq ^* from HV-ECU 10, phase currents Iv and Iw from control current sensors 71 and 73, and electrical angle θ from rotation angle sensor 85. For the phase current, the feedback control unit 55 calculates the current of the other one phase (in this example, the U phase) according to Kirchhoff's law, and converts the three-phase current to the dq axis current using the electrical angle θ, for example, by vector control. Convert. The dq axis current is fed back to the current command, or the estimated torque calculated from the dq axis current is torque fed back to the torque command.

Thus, the feedback control unit 55 calculates a voltage command for energizing the motor 80 and outputs it to the drive circuit 40. The drive circuit 40 generates drive signals UU, VU, WU, UL, VL, WL based on the voltage command, and drives the inverter 30. When electric power is supplied from the inverter 30, the motor 80 outputs a torque corresponding to the torque command trq ^* .

Thus, in order for the MG-ECU 50 to appropriately control the driving of the motor 80, it is important to accurately acquire control information such as phase current. Therefore, when the control current sensors 71 and 73 fail, it is necessary to reliably detect the abnormality early.
Therefore, in this motor drive system 90, monitoring current sensors 72 and 74 for monitoring abnormality of the control current sensors 71 and 73 are provided in the V-phase winding 82 and the W-phase winding 83.

  The V-phase control current sensor 71 and the monitoring current sensor 72 detect and output the V-phase current flowing through the V-phase winding 82 that is the same current path. The W-phase control current sensor 73 and the monitoring current sensor 74 detect and output the W-phase current flowing through the W-phase winding 83 which is the same current path. In FIG. 1, the sensor outputs of the control current sensors 71 and 73 are denoted as Iva and Iwa, and the sensor outputs of the monitoring current sensors 72 and 74 are denoted as Ivb and Iwb.

The two current sensors 71 and 72 for detecting the V-phase current and the two current sensors 73 and 74 for detecting and outputting the W-phase current correspond to “a plurality of current sensors” recited in the claims. . In the appended claims, the two corresponding current sensors 71 and 72 are described as “71/72” and the current sensors 73 and 74 are described as “73/74” as reference numerals of “a plurality of current sensors”.
Here, the term “control” or “monitoring” current sensor is merely an example name. Depending on the embodiment, the current sensor for control and the current sensor for monitoring may be switched at any time, or all the current sensors may be used for both control and monitoring.

The current sensor abnormality diagnosis device 60 acquires the sensor outputs Iva and Iwa of the control current sensors 71 and 73 and the sensor outputs Ivb and Iwb of the monitoring current sensors 72 and 74 for each of the V phase and the W phase. Compare. And based on those output deviations, the abnormality of the current sensors 71-74 is diagnosed. Specifically, when the output deviation exceeds a predetermined deviation threshold, it is temporarily determined that the output deviation is abnormal, and when the accumulated value of the time when the output deviation exceeds the predetermined deviation threshold reaches a specified time, one of the current sensors is Judged to be abnormal.
In the present embodiment, the current sensor abnormality diagnosis device 60 acquires the torque command trq ^{* in} order to determine the energization state of the motor 80.

Furthermore, the current sensor abnormality diagnosis device 60 may acquire a reference temperature Temp_cf that can be referred to as the temperature of the current sensors 71 to 74. For example, as shown in FIG. 1, the ambient temperature detected by a dedicated temperature sensor 77 such as a thermocouple or thermistor mounted near the current sensors 71 to 74 may be used as the reference temperature Temp_cf. Alternatively, the temperature of the cooling medium around the system and the ambient temperature of the system detected by another ECU or the like may be used as the reference temperature Temp_cf of the current sensors 71 to 74.
The current sensor abnormality diagnosis device 60 of the present embodiment only acquires the reference temperature Temp_cf and grasps the temperature change state of the current sensors 71 to 74, and does not use it as information for diagnosis.

  Next, the problem of the apparatus for diagnosing abnormality of the current sensor based on the output deviation of the two current sensors that detect the current flowing in the same current path will be described with reference to FIG. In the following description, the control current sensor and the monitoring current sensor are not distinguished from each other, and two current sensors that detect the current flowing in the same current path are commonly used for the V-phase current and the W-phase current. The outputs are represented as a first current sensor output Io1 and a second current sensor output Io2. In FIG. 2, assuming that the current to be detected is a sinusoidal alternating current having an amplitude A, the first current sensor output Io1 is indicated by a solid line, and the second current sensor output Io2 is indicated by a broken line.

First, FIG. 2A shows a change in output deviation in a configuration in which the reference output learning is not performed.
In FIG. 2A, the two current sensor outputs Io1 and Io2 coincide with each other at the initial t0. Thereafter, when the environmental temperature changes with the passage of time during energization, the first current sensor output Io1 is easily affected by the temperature change, and the output Io2 of the second current sensor is not easily affected by the temperature change. It is assumed that there is a variation in temperature characteristics.

Hereinafter, as an example of the temperature change, a case where the temperature gradually increases will be described. Further, it is assumed that the sensor output shifts to the plus side as the temperature rises. The technical idea is the same even when the temperature gradually decreases or when the correlation between the temperature and the sensor signal output is reversed.
Furthermore, in order to simplify the explanation, it is assumed that the output Io2 of the second current sensor is not affected by the temperature change at all. Therefore, as shown in FIG. 2A, the second current sensor output Io2 does not change after the initial t0. On the other hand, in the first current sensor output Io1, the amplitude center B1 gradually shifts to the plus side after the initial t0.

In this case, even if both of the two current sensors are normal, the sensor output deviation, that is, the output deviation ΔIo gradually increases as the reference temperature Temp_cf increases. Therefore, if an abnormality determination value is set without considering the temperature change during energization, it may be erroneously determined that one of the current sensors is abnormal based on the output deviation ΔIo caused by the temperature characteristics.
Therefore, in order to prevent erroneous determination, it is necessary to set a large abnormality determination value in consideration of the maximum output deviation ΔIo that may occur due to temperature characteristics. However, if the abnormality determination value is set excessively, the detection of the actual current sensor abnormality may be delayed or the abnormality may not be detected. That is, the abnormality detection performance is degraded.

On the other hand, the change of the output deviation in the configuration for performing the reference output learning is shown in FIG.
In FIG. 2B, as in FIG. 2A, the amplitude center B1 of the first current sensor output Io1 shifts to the plus side with the lapse of time from the initial t0, and the output deviation ΔIo gradually increases. Thereafter, at time L0 to L2, “reference output learning” is performed to offset-correct the first current sensor output Io1 so that the output deviation ΔIo becomes zero. The detailed process of reference output learning will be described later. By performing this reference output learning, after the output deviation ΔIo increased from the initial t0 is once reset to 0, the corrected output deviation ΔIc starts to newly increase.

In the present embodiment, attention is paid to the point that “the corrected output deviation ΔIc is within a relatively small range immediately after the execution of the reference output learning” in FIG. Then, attention is paid to the fact that the abnormality determination value can be set as small as possible while preventing the erroneous determination caused by the temperature characteristic by performing the abnormality diagnosis only in the range where the corrected output deviation ΔIc is relatively small.
Based on this viewpoint, the current sensor abnormality diagnosis device 60 of the present embodiment is characterized by having a configuration for setting the abnormality determination value small.

Next, the configuration of the current sensor abnormality diagnosis device 60 is shown in FIG. The current sensor abnormality diagnosis device 60 includes a learning command unit 61, a current correction unit 62, a diagnosis execution permission unit 63, and an abnormality determination unit 66.
First, with reference to FIG. 3, the general function of each part will be briefly described.
The learning command unit 61 instructs execution of “reference output learning” to be described later using a signal such as a motor energization start input from the outside as a trigger.

The current correction unit 62 receives the outputs Io1 and Io2 of the two current sensors. Based on the learning command from the learning command unit 61, the current sensor outputs Io1 and Io2 are offset-corrected so that the sensor output deviation ΔIo is set to 0 at the “reflection” stage of the reference output learning.
Further, after the reference output learning, the corrected outputs Ic1 and Ic2 obtained by offset-correcting the outputs Io1 and Io2 of the current sensor are output in accordance with the correction of the reference output learning.

Diagnosis execution permission unit 63 acquires torque command trq ^* from HV-ECU 10. Further, the learning completion timing is acquired from the learning command unit 61, and the diagnosis permission period Pd for permitting the abnormality diagnosis is set based on the information. Then, a diagnosis permission signal is transmitted to the abnormality determination unit 66.

  The abnormality determination unit 66 acquires the correction outputs Ic1 and Ic2 from the current correction unit 62, and calculates a correction output deviation ΔIc that is a deviation between the correction outputs Ic1 and Ic2. Then, during a period in which the diagnosis permission signal is received from the diagnosis execution permission unit 63, it is determined that one of the two current sensors is abnormal based on the corrected output deviation ΔIc, and a diagnosis signal is output.

The current sensor abnormality diagnosis according to the present embodiment will be described with reference to the time chart of FIG. 4 and the time chart of the comparative example of FIG.
4 and 5, like FIG. 2B, the output deviation ΔIo between the two current sensor outputs that are actually normal gradually increases as the reference temperature Temp_cf changes (rises). ing. Specifically, the output Io1 of the first current sensor shifts to the plus side as the temperature rises, and the output Io2 of the second current sensor does not change with temperature.

The learning command unit 61 of the current sensor abnormality diagnosis device 60 commands execution of reference output learning while the motor 80 is being driven. Here, “during motor driving” includes a state immediately after the start of energization of the motor 80, such as when the hybrid vehicle is ready. For example, the reference output learning may be performed as part of the initial check (initial process) immediately after the start of motor energization. Alternatively, the reference output learning may be performed immediately after the shift range is changed from the N (neutral) range to the D (drive) range.
In such a case, it is correct that the current amplitude immediately before the reference output learning in FIGS. 4 and 5 is a waveform of 0 or close to 0. However, in FIGS. 4 and 5, the current immediately before the reference output learning is simply described with the details omitted.

  The reference output learning includes a “recognition” stage from the learning reference time L0 to the intermediate time L1, and a “reflection” stage from the intermediate time L1 to the completion time L2. In the recognition stage, the output deviation ΔIo of the two current sensors at the learning reference time is recognized. In the reflection stage, the current correction unit 62 offset corrects the output of the current sensor so that the recognized output deviation ΔIo is zero. Note that “0” in “set the output deviation ΔIo to 0” is not limited to strictly 0, but means a value in a range that is substantially regarded as 0 in consideration of the resolution of the apparatus.

  In the examples of FIGS. 4 and 5, the offset correction is performed so that the first current sensor output Io1 matches the second current sensor output Io2. However, in another embodiment, offset correction may be performed so that both the first current sensor output Io1 and the second current sensor output Io2 coincide with a common reference value.

After the reference output learning, the current correction unit 62 outputs correction outputs Ic1 and Ic2 in which the output of the current sensor is offset-corrected according to the correction in the reference output learning. In the example of FIGS. 4 and 5, the second current sensor outputs a correction output Ic2 that is the same as the output Io2 before correction. In another embodiment, corrected outputs Ic1 and Ic2 obtained by offsetting the uncorrected outputs Io1 and Io2 may be output for the two current sensors, respectively.
At the time of learning completion L2, the correction output deviation ΔIc, which is the deviation between the correction outputs Ic1 and Ic2, is zero. Thereafter, the correction output deviation ΔIc gradually increases with a change (increase) in the reference temperature Temp_cf.

  First, the operation of the comparative example in which abnormality diagnosis is always performed will be described with reference to FIG. In the abnormality diagnosis, the abnormality determination unit tentatively determines that there is an abnormality at time D1 when the corrected output deviation ΔIc exceeds the deviation threshold value ΔIth, and starts counting the deviation excessive time Tex. Then, at time D2 when the deviation excess time Tex reaches the specified time Tlim, the abnormality determination unit 66 determines that one of the two current sensors is abnormal, and displays a diagnostic signal indicating that the current sensor is abnormal. Output.

  Further, when the corrected output deviation ΔIc fluctuates above and below the deviation threshold value ΔIth and the deviation excessive time Tex appears intermittently, when the accumulated value ΣTex of the deviation excessive time Tex reaches the specified time Tlim, A diagnostic signal is output. From the viewpoint of eliminating instantaneous noise, only the time when the corrected output deviation ΔIc exceeds the deviation threshold value ΔIth continuously for a predetermined time or more may be counted.

  As described above, in the comparative example, although the two current sensors are actually normal, they are erroneously determined to be abnormal due to the influence of the temperature characteristics. Therefore, if the deviation threshold value ΔIth, which is an abnormality determination value, is set excessively to prevent erroneous determination, detection may be delayed or missed when the current sensor actually becomes abnormal. That is, the abnormality detection performance is reduced.

Next, the operation of abnormality diagnosis according to the present embodiment will be described with reference to FIG.
The diagnosis execution permission unit 63 according to the present embodiment permits the abnormality determination unit 66 to perform abnormality diagnosis during the diagnosis permission period Pd until a predetermined time elapses after the completion of the reference output learning L2. The diagnosis permission period Pd is set to a time at which the maximum temperature change that can be assumed during the period does not affect the temperature characteristics of the current sensor.

  Here, note about the time axis of FIG. 4. The length of one cycle of the sine wave current is exaggerated to be several hundred to several thousand times longer than the actual length with respect to the length of the diagnosis permission period Pd. Have been described. That is, in practice, the length of one cycle of the sine wave current that is on the order of several to several tens of ms is described to the same extent as the diagnosis permission period Pd that is on the order of several to several tens of seconds for convenience of illustration. .

  The abnormality determination unit 66 performs abnormality diagnosis only during the diagnosis permission period Pd in which the diagnosis permission signal is received from the diagnosis execution permission unit 63. Therefore, even if the corrected output deviation ΔIc increases due to the temperature rise after the reference output learning, there is no possibility of reaching the deviation threshold ΔIth within the diagnosis permission period Pd. Therefore, it is prevented that the two current sensors are actually normal but erroneously determined to be abnormal, and therefore it is necessary to set the deviation threshold ΔIth excessively in order to prevent erroneous determination. Disappear. Therefore, the abnormality detection performance can be improved by setting the deviation threshold ΔIth as small as possible.

Next, the current sensor abnormality diagnosis process according to the present embodiment will be described with reference to the flowchart of FIG. In the description of the flowchart, the symbol “S” means a step.
In this description, a plurality of current sensors that detect currents flowing through the same current path are described as “V-phase current sensors 71 and 72”.
In S1, the learning command unit 61 commands execution of reference output learning. After the reference output learning, the current correction unit 62 outputs correction outputs Ic1 and Ic2 obtained by offset correcting the outputs of the current sensors 71 and 72 in accordance with the correction in the reference output learning.

After the reference output learning, the diagnosis execution permission unit 63 determines whether the motor 80 is being driven and whether the torque command trq ^* is equal to or greater than a predetermined value (S2, S3). While the motor 80 is stopped, the current becomes 0, and the abnormality diagnosis is impossible. Even when the motor 80 is being driven, the smaller the torque command trq ^* , the smaller the S / N ratio between the current output and the noise, and the more susceptible to noise. Therefore, when the motor 80 is stopped (S2: NO) or the torque command trq ^* is less than a predetermined value (S3: NO), the process returns to before S2 and the steps are repeated.
Here, the torque command trq ^* reflects the actual torque, and the actual torque corresponds to “output generated by the load” described in the claims. Note that in a system in which the motor 80 is provided with a torque sensor or a system that calculates estimated torque from a current value or the like, determination may be made based on detected torque or estimated torque instead of the torque command trq ^* in S3. .

The diagnosis execution permission unit 63 starts counting the diagnosis permission period Pd from the time L2 when the reference output learning is completed. When the motor 80 is being driven (S2: YES) and the torque command trq ^* is equal to or greater than a predetermined value (S3: YES), in S4, the diagnosis execution permission unit 63 is currently within a predetermined time after the reference output learning. It is determined whether it is present, that is, whether it is during the diagnosis permission period Pd.
If the diagnosis permission period Pd is currently in progress (S4: YES), the abnormality determination unit 66 performs abnormality diagnosis of the current sensors 71 and 72 (S5). If it is not currently in the diagnosis permission period Pd (S4: NO), the abnormality determination unit 66 does not perform abnormality diagnosis (S6).

In the abnormality diagnosis, the abnormality determination unit 66 compares the corrected output deviation ΔIc of the two current sensors 71 and 72 with the deviation threshold ΔIth, and when the corrected output deviation ΔIc exceeds the deviation threshold ΔIth (S7: YES) The time when the corrected output deviation ΔIc exceeds the deviation threshold value ΔIth is counted as the deviation excessive time Tex (S8).
When the accumulated value ΣTex of the deviation excessive time Tex reaches the specified time Tlim, that is, when “ΣTex ≧ Tlim” is satisfied (S9: YES), the abnormality determination unit 66 determines which of the two current sensors 71 and 72 Is determined to be abnormal (S10).

  In this case, according to the idea of giving priority to fail safe, the driving of the motor 80 may be completely stopped (shut down). Further, according to the idea of prioritizing evacuation travel, a faulty current sensor may be identified and feedback control may be continued using a detection value from a normal current sensor. Or you may switch to the control which uses the average value of the detected value of the two current sensors 71 and 72 about V phase. In addition, when the W-phase current sensors 73 and 74 are normal, the control may be switched to one-phase control using only the W-phase current detection value (JP 2013-172591 A).

  On the other hand, when the corrected output deviation ΔIc does not exceed the deviation threshold ΔIth during the diagnosis permission period Pd (S7: NO) or during the diagnosis permission period Pd, the accumulated value ΣTex of the excessive deviation time does not reach the specified time Tlim. When (S9: NO), the abnormality determination unit 66 determines that both of the two current sensors 71 and 72 are normal (S11). In the feedback control unit 55, feedback control using the detection value of the control current sensor 71 is performed.

(effect)
(1) The first effect of the current sensor abnormality diagnosis device 60 of the present embodiment will be described with reference to FIGS.
First, FIG. 7 shows the tolerance of the current sensor output when the reference output learning is not performed. Tolerance factors are mainly divided into those due to product variations and those due to temperature characteristics, and include gain characteristic tolerance and offset tolerance, respectively. The gain characteristic tolerance and offset tolerance due to product variation are represented as “Gp, Bp”, and the temperature characteristic gain characteristic tolerance and offset tolerance are represented as “Gt, Bt”. Here, regarding the offset tolerance, “B” is substituted for the meaning of base shift in order to avoid misreading the initial letter “O” of the offset.

  As shown in FIG. 7B, the gain characteristic tolerances Gp and Gt are proportional to the current value, and the offset tolerances Bt and Bp are constant regardless of the current value. The sum of these four types of tolerances is the overall tolerance of the current sensor. Therefore, in order not to erroneously determine that a normal current sensor is abnormal, it is necessary to set the abnormality determination value larger than the sum of tolerances in the maximum current value.

Further, as shown in FIG. 7A, for the current sensor output characteristics with respect to the current value, the output range in consideration of the tolerance is the sum of the offset tolerance (Bp + Bt) and the sum of the gain characteristics tolerance (Gp + Gt) with respect to the ideal characteristics. ) Is added and subtracted.
If the output characteristics of the monitoring current sensors 72 and 74 match the ideal characteristics, the range of the output deviation ΔIo between the control current sensors 71 and 73 and the monitoring current sensors 72 and 74 is the tolerance range.

Next, FIG. 8 shows the tolerance of the current sensor output when the reference output learning is performed according to this embodiment and the diagnosis permission period Pd for abnormality diagnosis is set. Again, assuming that the output characteristics of the monitoring current sensors 72 and 74 match the ideal characteristics, it is considered that the tolerance range of the control current sensors 71 and 73 corresponds to the range of the output deviation ΔIo.
In the present embodiment, the offset tolerances Bt and Bp become 0 by performing the reference output learning and performing the offset correction so that the output deviation ΔIo becomes 0. Further, since the abnormality determination unit 66 permits the abnormality diagnosis to be performed only during the diagnosis permission period Pd in which the corrected output deviation ΔIc after the reference output learning is within a relatively small range, the gain characteristic of the temperature characteristic at the time of abnormality diagnosis The tolerance Gt can be suppressed to a minute range.

As a result, as shown in FIG. 8B, only the gain characteristic tolerance Gp due to product variation remains as the tolerance of the current sensor. Therefore, the abnormality determination value (deviation threshold value ΔIth) for preventing erroneous determination can be set smaller than in the case of non-learning shown in FIG.
Further, as shown in FIG. 8A, with respect to the current sensor output characteristic with respect to the current value, the output range in consideration of the tolerance is a range obtained by adding / subtracting only the gain characteristic tolerance Gp due to product variation to the ideal characteristic.

  As described above, the current sensor abnormality diagnosis device 60 of this embodiment eliminates offset tolerance and gain characteristic tolerance due to temperature characteristics as much as possible in current sensor abnormality diagnosis, and sets the deviation threshold ΔIth to be small. Can do. Therefore, it is possible to reliably detect an abnormality of the actual current sensor at an early stage, and it is possible to improve the abnormality detection performance.

In addition, other effects (2) to (4) of the present embodiment will be described.
(2) The control current sensors 71 and 73 that are the targets of the abnormality diagnosis in the present embodiment detect currents used for feedback control in the motor drive system 90. In such a form, since the detection value of the current sensor directly affects the controllability of the system, the demand for the reliability of the current sensor increases. Therefore, the effect (1) is effectively exhibited.
In particular, in a system that drives a motor generator that is a power source of a hybrid vehicle or an electric vehicle, in order to achieve good drivability while the rotational speed and required torque of the motor 80 vary greatly depending on the driving state of the vehicle, High precision control is required. Therefore, it is particularly effective to accurately detect the abnormality of the current sensors 71 to 74 using this embodiment.

(3) Since the abnormality diagnosis is performed in an area where the torque command trq ^* is equal to or greater than a predetermined value and a minimum S / N ratio can be secured, the abnormality diagnosis can be appropriately performed.
(4) The diagnosis permission period Pd is set within a predetermined time from the reference output learning completion time L2. Since time is used as a reference, the logic for setting the period becomes simple.

(Other embodiments)
(A) The plurality of current sensors to be subjected to abnormality diagnosis are not limited to those that detect an alternating current supplied from an inverter, but also include those that detect a direct current.
Moreover, in the form in which the current sensor detects the supply current from the inverter, the load of the inverter is not limited to a motor generator for a hybrid vehicle or other AC motor.
Furthermore, the current detected by the current sensor may not be used for feedback control.

  (A) The arrangement of the plurality of current sensors in the “same current path” may not be adjacent as shown in FIG. In a part of an electric circuit where currents of the same magnitude are flowing synchronously, even if they are separated from each other, they are physically used like the primary and secondary windings of the transformer. Even if they are separated, they are interpreted as “the same current path”.

(C) Depending on the nature of the system to which the present invention is applied, the timing of reference output learning, the specific length of the diagnosis permission period Pd, and the like may be set as appropriate.
(D) The plurality of current sensors for detecting the current flowing in the same current path is not limited to two, but may be three or more. When the abnormality determination unit determines that any of the three or more current sensors is abnormal, the abnormal current sensor is identified by majority determination, and the motor drive control is performed using the detection value of the normal current sensor. May be continued.

(E) After the reference output learning, disturbances such as atmospheric pressure and humidity can be assumed as factors affecting the output characteristics of the current sensor, not limited to temperature changes.
As mentioned above, this invention is not limited to the said embodiment at all, In the range which does not deviate from the meaning of invention, it can implement with a various form.

30: Inverter,
60 ... Current sensor abnormality diagnosis device,
61... Learning instruction section,
62 ... current correction unit,
63 ... diagnosis execution permission part,
66 ... an abnormality determination unit,
71, 73 ... (for control) current sensor,
72, 74 ... (for monitoring) current sensor,
80: Motor, AC motor (load).

Claims (8)

A current sensor abnormality diagnosis device that compares outputs of a plurality of current sensors (71/72, 73/74) that detect currents flowing in the same current path and diagnoses abnormality of the plurality of current sensors based on an output deviation. Because
A learning command section (61) for instructing execution of reference output learning for offset correction of outputs of the plurality of current sensors so that output deviations of the plurality of current sensors at the time of learning reference are set to zero;
After the reference output learning, a current correction unit (62) that outputs a correction output obtained by offset-correcting the outputs of the plurality of current sensors according to the correction in the reference output learning;
An abnormality determination unit (66) that performs an abnormality diagnosis of the plurality of current sensors based on a correction output deviation that is a deviation between the correction outputs of the plurality of current sensors;
A diagnosis execution permission unit (63) that permits execution of an abnormality diagnosis by the abnormality determination unit in a diagnosis permission period from the completion of the reference output learning until a predetermined time elapses;
A current sensor abnormality diagnosis device comprising:   The abnormality determination unit determines that any one of the plurality of current sensors is abnormal when the accumulated value of the excessive deviation time, which is a time when the corrected output deviation exceeds a deviation threshold, reaches a specified time. The current sensor abnormality diagnosis device according to claim 1, wherein:   The plurality of current sensors detect a phase current flowing in a current path between an inverter (30) for converting DC power into multiphase AC power and a load (80). Or the current sensor abnormality diagnosis device according to 2;   The current sensor abnormality diagnosis device according to claim 3, wherein the diagnosis execution permission unit permits the abnormality diagnosis to be performed when an output generated by the load is equal to or greater than a predetermined value.   The current sensor abnormality diagnosis device according to claim 3 or 4, wherein the load of the inverter is an AC motor (80).   The current sensor abnormality diagnosis apparatus according to claim 5, wherein the AC motor is used as a power source of a hybrid vehicle or an electric vehicle.   The current sensor abnormality diagnosis device according to claim 3, wherein the reference output learning is executed as an initial process when energization of the load is started.   8. The current sensor abnormality diagnosis device according to claim 1, wherein at least one of the plurality of current sensors detects a current used for feedback control. 9.

Images