JP4557143B2 - Power steering control device, method, and program - Google Patents

Description

  The present invention relates to a power steering control apparatus, method, and program capable of automatically correcting a detected current offset by learning a detected current offset.

  As an auxiliary steering device for an automobile, an electric power steering device using the torque of an electric motor is used. This power steering device includes a torque sensor that detects a steering operation by a driver, a power steering control device (ECU) that calculates an auxiliary steering force based on a detection signal from the torque sensor, and a rotational torque based on an output signal from the ECU. And an electric motor that generates the torque and a reduction gear that transmits the rotational torque to the steering mechanism.

  For example, the ECU described above is configured as shown in FIG. In this figure, ECU 90 receives signals from torque sensor 91 and vehicle speed sensor 92, and ECU 90 has a function of controlling electric motor 93 based on these signals. The ECU 90 includes a current command value calculation unit 901, an adder 902, a current control unit 903, a drive circuit 904, and a current detection circuit 905. The current command value calculation unit 901 calculates a current command value I based on signals from the torque sensor 91 and the vehicle speed sensor 92. The adder 903 calculates a deviation current ΔI between the current value command value I and the detection current i detected by the current detection circuit 905, and the current control unit 903 has a motor control signal such that the error current ΔI becomes zero. It is for calculating. The motor drive circuit 904 has a function of supplying a three-phase current to the electric motor 93 based on the motor control signal. The current detection circuit 905 has two current sensors, and detects the other two current values based on one of the three-phase currents. The detected current i is input to the adder 902 as a negative signal.

  In the ECU 90 configured as described above, negative feedback control is performed so that the detected current i becomes equal to the current command value I. However, when an error exists in the detection system such as the current detection circuit 905, the error appears as it is in the three-phase current. If there is an error in the offset current or the detection gain in the two current sensors constituting the current detection circuit 905, an error occurs in each current value of the three-phase current output from the drive circuit 904. As a result, the electric motor 93 Torque ripple will occur. Such torque ripple adversely affects handle feeling and the like.

  As a conventional technique capable of reducing torque ripple generated by the detection system, for example, an electric power steering control device described in Japanese Patent Application Laid-Open No. 2002-29432 has been devised. This control device learns the detected current when the command value of the three-phase current is zero as an offset value, and corrects the command value of the three-phase current with this offset value.

  However, since this control device does not fix the duty command value to a predetermined value during offset learning, a minute current flows through the current detection circuit, and an error is likely to occur in offset learning. Further, since offset learning is performed only when the current value becomes zero, the frequency of establishment of the offset learning condition is low, and it is difficult to perform sufficient offset learning.

  Furthermore, the offset learning result is reflected not only in the current control unit but also in all the functions of the ECU using the detected current. This may affect the function of the.

As another conventional technique, an electric power steering control device described in Japanese Patent Laid-Open No. 2003-164192 has been devised. This device performs learning of the current detection offset when the duty ratio of each phase of the current command value is 50%, but does not fix the duty command value to a predetermined value during offset learning. Detection error will occur. In addition, as in the above-described prior art, the conditions for offset learning are extremely limited, and sufficient offset learning cannot be performed.
JP 2002-29432 A JP 2003-164192 A

  The present invention has been made in view of the above problems, and a first problem to be solved by the present invention is to fix the duty command value at the time of learning the offset of the detected current to a predetermined value in the power steering control device. This is to improve the offset learning accuracy. A second problem of the present invention is to realize sufficient offset learning by optimally setting learning conditions in offset learning. Furthermore, the third problem of the present invention is to minimize the influence on other functions even during insufficient offset learning by reflecting the result of offset learning only in the current control unit.

In order to solve the above-described problems, the present invention calculates a torque sensor that detects a steering torque applied to a steering, and a current command value for controlling a multiphase drive current based on the detected steering torque. A current command value calculation unit that calculates a multiphase duty command value that is pulse-width-modulated based on the current command value, and a drive current that is based on the duty command value for generating steering assist torque. A drive circuit to be applied to a multiphase motor, a current detection circuit for detecting the value of the drive current in the multiphase motor and outputting a current detection signal, A / D converting the current detection signal, and outputting a current detection value an a / D converter for, when the driving current is substantially zero, and determines whether to start the offset learning, when determining the start of the offset learning, the Dew A learning condition determining unit that sets the I command value duty ratio of 50%, when the start of the offset learning is determined, detects the current detection value and estimates the offset current value of the current detection circuit An offset estimation unit; and an offset correction unit configured to correct each phase of the current detection value based on the offset current value . The newly estimated first offset current value and the previously estimated second When the difference from the offset current value is equal to or greater than a predetermined value, the offset estimation unit corrects the first offset estimated value so that the difference is reduced .

The learning condition determination unit determines whether or not to start offset learning based on at least one of the current command value, the duty command value, the rotation speed of the motor, or the vehicle speed.
The learning condition determination unit determines the start of offset learning after a predetermined time has elapsed since the previous offset learning.
The offset estimation unit estimates the offset current value by sampling the current detection value a plurality of times.

  The power steering control device further includes means for amplifying the plurality of current detection signals represented by digital values and calculating an average value thereof.

Furthermore, according to the present invention, the learning condition determination unit outputs a duty command value having a duty ratio of 50% when the start of offset learning is determined when the drive current becomes substantially zero. In this state, the offset estimation unit estimates the offset current value of the current detection circuit, and the offset correction unit corrects each phase of the current detection value based on the estimated offset current value. That is, the current flowing to the motor during offset learning can be made almost zero, and as a result, the offset current can be accurately detected. In addition, when the difference between the newly estimated first offset current value and the previously estimated second offset current value is equal to or greater than a predetermined value, the offset estimation unit reduces the difference. Modify the first offset estimate. Thereby, it is possible to prevent the offset correction value from changing suddenly, and natural offset learning is possible.

  Further, the learning condition determination unit determines whether or not to start offset learning based on at least one of the current command value, the duty command value, the motor rotation speed, or the vehicle speed. Therefore, it is possible to accurately estimate the state in which the drive current is zero, and to perform highly accurate offset learning.

  The learning condition determination unit determines the start of offset learning after a predetermined time has elapsed since the previous offset learning. For this reason, it can avoid that offset learning is performed more than necessary, and it can prevent that assist control is prevented by offset learning.

  Since the offset estimation unit estimates the offset current value by sampling the current detection value a plurality of times, it is possible to suppress variation in the current detection value and accurately estimate the offset current.

  The power steering control device further includes means for amplifying the plurality of current detection signals represented by digital values and calculating an average value thereof. For this reason, the resolution per 1 LSB of the A / D converter can be substantially increased, and the conversion error in the A / D conversion of the current detection value can be reduced.

The best mode for carrying out the present invention will be described below with reference to the drawings.
FIG. 1 is a schematic diagram of an electric power steering apparatus according to the present embodiment. In this figure, a steering 61 is connected to a rack and pinion 66 via a steering shaft 62, universal joints 63 and 64, and a shaft 65. Further, the rack and pinion 66 is provided with a wheel tie rod 67, and the rotational movement of the handle 61 is converted into the axial movement of the tie rod 67.

  A torque sensor 4 is provided on the shaft 65, and the torque sensor 4 can detect a steering torque applied to the steering 61 and output a torque signal. Further, the reduction gear 31 and the motor 3 are attached to the shaft 65, and the rotational torque of the motor 3 is transmitted to the shaft 65 via the reduction gear 31.

  The ECU 1 calculates the auxiliary steering torque based on the torque signal from the torque sensor 4 and the vehicle speed signal from the vehicle speed sensor 2 as described above, and sends the drive current based on the calculation result to the motor 3. A power supply 5 is connected to the ECU 1 via an ignition key 5a, and a current is supplied to the ECU 1 by turning on the ignition key 5a.

  FIG. 2 is a block diagram illustrating a hardware configuration of the ECU 1. The ECU 1 includes a bus 100, an A / D converter 110, an interface 111, a clock generation circuit 112, a CPU 113, a ROM 114, a RAM 115, an RD converter 116, a PWM controller 117, a motor drive circuit 118, and motor current detection circuits 120 and 121. Has been.

  The bus 100 is for transmitting and receiving data among the A / D converter 110, the I / F 111, the clock generation circuit 112, the CPU 113, the ROM 114, the RAM 115, and the like. The A / D converter 110 outputs the main torque signal and the sub torque signal output from the torque sensor 4, the detected current from the motor current detection circuits 120 and 121, the motor rotation angle signal from the RD converter 116, and the voltage across the terminals of the motor 3. Input and convert to digital signal.

  The torque sensor 4 described above includes two output signals, a main torque signal and a sub torque signal, and the total voltage of these signals is set to have a cross characteristic that is a constant voltage (for example, 5 V). That is, when no torque is applied to the steering wheel, the main torque signal and the sub torque signal are each at a torque neutral voltage of 2.5 V, and when some torque is applied to the steering wheel, the main torque signal and the sub torque signal are at a neutral voltage of 2. .5V changes in opposite directions with reference to 5V.

  The interface 111 counts vehicle speed pulses from the vehicle speed sensor 2 and converts them into digital signals. The ROM 114 is used as a memory for storing a control program for the motor 3, a PWM calculation program, a fail safe program, and the like, and the RAM 115 is used as a work memory for operating the program.

  The PWM controller 117 is for converting a signal representing the torque of the motor 3 into pulse command modulated duty command values W, V, U, Wb, Vb, Ub. Here, the pulse signals W, V, and U represent positive-phase three-phase signals, and the pulse signals Wb, Vb, and Ub represent negative-phase three-phase signals.

  The motor drive circuit 118 is composed of three inverter circuits for generating a WVU three-phase current, and each inverter circuit has a switching transistor on the power supply voltage side (upper stage) and a switching transistor on the ground potential side (lower stage). ing. The normal phase duty command values W, V, and U are input to the gates of the upper switching transistors, and the negative phase duty command values Wb, Vb, and Ub are input to the lower switching transistors. That is, the upper and lower switching transistors are complementarily connected, and drive currents Iu, Iv, and Iw having desired pulse widths are generated by alternately repeating on and off operations. In order to prevent the upper switching transistor and the lower switching transistor from being turned on at the same time, the time when both of them are turned off (dead time) is the normal phase duty command value W, V, U and the reverse phase duty command value Wb. , Vb, Ub are provided before and after the on-time. In this way, by providing the dead time, it is possible to avoid a short circuit between the upper and lower switching transistors. The RD converter 116 has a function of supplying an exciting current to the resolver 31 and outputting an output signal from the resolver 31 to the A / D converter 110 as a rotation angle signal.

  The motor current detection circuits 120 and 121 are composed of current-voltage conversion elements such as resistors, and are used to detect drive currents Iu and Iw to the motor 3 and output a current detection value having a voltage corresponding to the current. is there. The detected current value is corrected by a power supply voltage correction circuit (not shown) and then converted into a digital signal by the A / D converter 110. The current detection value is amplified by a hardware amplification circuit before A / D conversion, and the resolution of the current detection value in the A / D converter 110 is about 0.332 A / LSB. The resolution of the current detection value is further enhanced by amplification and averaging processing on software described later.

  FIG. 3 shows a functional block diagram of the ECU 1. The main torque signal and the sub torque signal from the torque sensor 4 are A / D converted and then input to the current command value calculation unit 11. The current command value calculation unit 11 has a function of calculating a current command value based on a torque signal and a vehicle speed signal. The current command value calculation unit 12 has handle return compensation and motor maximum current control. For example, the handle return compensation performs control for restoring the handle 61 to the neutral position. In general, in the electric power steering apparatus, the self-aligning torque tends to be weak due to the influence of the reduction gear 31 and the like, and thus the handle 61 is difficult to return to the neutral position. Therefore, by detecting the motor terminal voltage and the motor current when the motor 3 is rotated by the action of the self-aligning torque, the motor angular velocity is detected, or the motor angular velocity is detected from the angle difference of the angle sensor. It is possible to calculate a compensation current value for restoring the handle to the neutral position.

  The current command value limiting unit 12 limits the current command value based on the motor rotation speed, and the vector control unit 13 outputs a three-phase current command value representing the current command value of each phase of UVW based on the motor angle. This current command value is input to the adder 14, and negative feedback control is performed so that the detected current value matches the current command value.

  The motor current detection circuits 120 and 121 detect U-phase and W-phase currents supplied to the motor 3, and output current detection values having voltage values corresponding to the current values. The detected current value is A / D converted, amplified by the amplifying unit 124, and further input to the adder 16 and the offset learning unit. The amplifying unit 124 is for increasing the gain on software by shifting the A / D converted current detection value by several bits. According to the present embodiment, the resolution of the current detection value per 1 LSB can be increased by increasing the gain of the current detection value and performing an averaging process described later. Further, the resolution per 1 LSB of the current detection value may be increased by combining the gain increase processing and the integration processing in the current control unit 15. For example, by amplifying the current detection value, the resolution of 0.332 A / LSB can be increased to 0.049 A / LSB, and highly accurate offset learning is possible.

The offset learning unit learns offset current correction and outputs an offset correction value to the adder 16 when the operating state of the power steering apparatus matches a predetermined condition.
The adder 16 subtracts the offset correction value from the current detection value, and outputs the current detection value after correcting the U phase and the W phase. The V-phase current estimator 17 is for correcting the gain of the detected current value of the V-phase to the U-phase and the W-phase. The UVW phase current detection value corrected in this way is compared with the current command value in the adder 14, and the control deviation Δi is given to the current control unit 15. The current control unit 15 performs so-called PI control that combines proportional control and integral control based on the control deviation Δi. An output signal from the current control unit 15 is output to the drive circuit 117, and a three-phase drive current is output to the motor 3.

  The failure determination unit 18 is for performing initial diagnosis and fail-safe processing of the power steering apparatus by monitoring torque signals, power supply voltages, and the like. For example, when the failure diagnosis unit 15 detects an abnormality in the motor 3, the torque sensor 4, etc., an instruction can be given to the current command value calculation unit 11 so as to perform a gradual reduction process of the auxiliary torque.

  FIG. 4 is a block diagram of the offset learning unit. The offset learning unit includes a learning condition determination unit 21, an offset estimation unit 22, and an offset correction unit 23. The learning condition determination unit 21 is for determining whether or not to perform offset correction learning based on a plurality of learning conditions. When the learning condition determination unit determines that the offset correction learning is performed, the learning condition determination unit 21 outputs a duty 50% command for fixing the current command value to the duty 50%, and also represents the learning execution The flag is output to the offset estimation unit 22.

  When the learning condition determination unit 21 determines offset learning, the offset estimation unit 22 has a function of reading the current detection value at the time of 50% duty output and estimating the offset current. When the offset current estimating unit 22 executes the offset current estimation, the offset current estimating unit 22 outputs an offset estimated value to the offset correcting unit 23 together with a predetermined state flag. The offset correction unit 22 to which these signals are input updates the previous offset correction value to the offset estimated value. The updated offset correction value is output to the adder 16 shown in FIG. 3, and offset correction of the current detection value is performed.

As described above, according to the present embodiment, the current detection value subjected to the offset learning is given only to the current control unit 15, and the current detection value not subjected to the offset learning is given to the other functions. For this reason, even when the offset learning is insufficient, it is possible to avoid adverse effects on functions other than the motor control.
The detailed configuration of each part of the offset learning unit 20 will be described below.

<Learning condition determination unit>
FIG. 5 is a block diagram of the learning condition determination unit 21. The learning condition determination unit 21 includes a determination unit 210 that determines whether or not offset learning is possible based on a plurality of learning conditions, and a learning frequency counter 211 that records a time stamp of the previous offset learning. As the learning conditions, an assist state, a vehicle speed signal, a duty command value, a current command value, a motor rotation speed, a learning elapsed time from the previous time, and the like are input to the determination unit 210. The determination unit 210 refers to these conditions, and determines the start of offset learning when the steering is not rotating and almost no current flows through the motor 3.

  Details of the offset learning condition are shown in FIG. In this figure, “AND” indicates that all of a plurality of conditions are satisfied, and “OR” indicates that any of the plurality of conditions is satisfied. In the learning condition column, for example, “the first time when the vehicle speed is 0 km / h AND assist is permitted” indicates that both conditions are satisfied simultaneously. Furthermore, “vehicle speed 0 km / h AND assist permission state AND 300 seconds or more have passed since the previous learning” indicates that all these three conditions need to be satisfied. When any of these conditions is satisfied, and the absolute value of the duty command value is 80 dec or less, the absolute value of the motor speed is 5 dec or less, and the absolute value of the current command value is 20 dec or less is satisfied. The offset learning start condition is satisfied.

  Further, as shown in the confirmation condition column of the figure, after the start condition described above is satisfied, 60 seconds have passed since the entry of the assist permission state, or when the assist permission state is entered for the first time. Learning conditions are fixed. As described above, by executing offset learning after the start condition has been satisfied for a predetermined time, the detected current value in a stable state can be read, and the accuracy of offset learning can be improved.

  In this way, when the learning condition is confirmed, the process shown in the item of the process at the time of confirmation is executed. That is, the determination unit 210 outputs a command to fix the duty command value to 50% duty to the current control unit 15 and initializes past calculation values in the integration element of the current control unit 15. Further, the determination unit 210 sets a state flag indicating a confirmed state of learning determination to “1”. Upon receiving this status flag, the offset estimation unit 22 starts offset learning.

  The end condition column in the figure represents a condition for ending the offset learning confirmation condition. That is, the offset learning process ends after a maximum of 20 msec has elapsed since the confirmation condition is established. As shown in the processing column at the end, when the offset learning ends, the determination unit 210 sets the status flag to “0” and clears the value of the learning frequency counter 211.

  Further, as shown in the column of the sampling period in the figure, the current detection value is sampled every 250 μsec. If the learning condition is no longer satisfied as the offset learning interruption condition, the offset learning is interrupted and the status flag is set to “0”.

  As described above, the absolute value of the current command value is required to be 20 dec or less as one of the offset learning start conditions. According to the measurement results, the current command value was within the range of ± 10 dec when the hand was released from the steering, and the current command value was within the range of ± 20 dec when the steering was lightly steered. From these results, in this embodiment, the current command value as the offset learning start condition is set to ± 20 dec.

  Further, the motor may rotate somewhat depending on the vehicle body vibration and the steering state. For example, when the hand is released from the steering wheel, the motor rotation speed is 0 rpm. When the hand is attached to the steering wheel, the motor rotation speed is ± 5 rpm (1 dec / 1 rpm). From this result, the motor rotation speed is set to ± 5 rpm as an offset learning start condition.

  Furthermore, the duty command value is set within a range of ± 80 dec as an offset start condition. Since the current starts to flow in the motor 3 after the ON time of the switching transistor of the PWM drive circuit 117 exceeds the dead time, almost no current flows in the motor 3 if the duty command value is equal to or less than the dead time. In general, since the dead time of the switching transistor of the PWM drive circuit 117 is about ± 100 dec, in consideration of the variation of the switching transistor, ± 80 dec is set as the offset start condition of the duty command value in this embodiment.

  The vehicle speed as the offset start condition is set to 0 km / h and 1 dec / 2 km / h in this embodiment. This is because offset learning is allowed only during a stop in consideration of driver safety.

<Offset estimation unit>
FIG. 6 is a block diagram of the offset estimation unit 22. The offset estimation unit 22 includes a determination processing unit 220 that detects an offset current, and a sampling averaging processing unit 221 that averages the detected offset current. When a current detection value is input to the determination processing unit 220 and the current detection value is within a predetermined range, the determination processing unit 220 reads the current detection value as an offset current. The sampling averaging processing unit 221 has a function of performing an averaging process by sampling the offset current read by the determination processing unit 220 a plurality of times. As described above, the offset current variation can be reduced by performing the averaging process of the offset current. That is, when the offset current is A / D converted, the digital value fluctuates within a range of 1 bit, and it is difficult to grasp an accurate offset current only by sampling the offset current once. Therefore, according to the present embodiment, the offset current is sampled 64 times to reduce variations in the offset current.

  10 and 11 show the relationship between the number of samplings and the variation in offset current. In these figures, it can be confirmed that the noise component of the offset current is reduced by sequentially increasing the number of sampling times from 0 to 64 times. According to the present embodiment, a sufficient noise reduction effect can be obtained by performing sampling 64 times.

  FIG. 8 shows details of the offset estimation process. As shown in the start condition column of the figure, when the status flag is “1” and the absolute value of the current command value is smaller than 16 dec, the start condition for the offset estimation process is satisfied. The offset estimation unit 22 samples the detected current value of the UW phase and reads this value as an offset current. Further, the offset estimation unit 22 samples and adds the offset current as it is if the absolute value of the read offset current is 8 dec or less, and ± 8 dec if the absolute value of the offset current is within the range of 9 to 15 dec. Is added as an offset value. In this way, the offset estimation unit 22 performs 64 additions, and estimates the result of dividing the addition result by 64 as an offset value. As described above, the gain of the current detection value is increased on the software, and the resolution per ALSB in the A / D conversion is substantially improved by further averaging the current detection value. be able to.

  As shown in the figure, the condition for interrupting the offset estimation process is when the status flag is “0” and the current command value is 16 dec or more. When the offset estimation processing is interrupted, the determination processing unit 220 returns the offset estimation value to the original state, returns the state flag to “0”, and clears the added value of the sampling averaging processing unit 221.

<Offset correction unit>
FIG. 9 shows details of the offset correction value update processing. The offset correction value update process is for updating the above-described offset estimated value as an offset correction value. The start condition of the offset correction value update process is when the status flag is “2”, the previous offset value is not equal to the estimated offset value, and the estimated offset value is 8 dec or less. When this condition is satisfied, an offset correction value update process is performed. That is, when the absolute value of the offset estimated value−the previous offset value is 4 dec or less, the offset correction unit 23 replaces the offset estimated value with the offset correction value. When the absolute value of the offset estimated value−the previous offset value is larger than 4 dec, the offset correction unit replaces the value of (offset estimated value−offset previous value) / 2 with the offset correction value. That is, when the difference between the offset estimated value and the previous offset value is large, the offset correction value is updated step by step. Thereby, a sudden change in control can be avoided. When the offset correction value update process ends, the offset correction unit 23 returns the status flag to “0”. The offset correction value updated in this way is output to the adder 16 in FIG. 3, and offset correction of the current detection value is performed.

<Operation>
Next, the operation of the power steering control device according to this embodiment will be described. FIG. 12 is a main flowchart showing the operation of the power steering control device. First, when the ignition switch 5a is turned on, the ECU 1 starts its operation and executes initial settings such as initial diagnosis of the power steering device (step S1). When the initial setting ends normally and the ECU 1 permits the assist operation of the power steering (YES in step S2), the assist control is started (step S3). That is, the current command value calculation unit 11 calculates a current command value based on a torque signal or the like, and the vector control unit 13 outputs a three-phase current command value. The current detection circuits 120 and 121 detect a drive current to the motor 3 and output a current detection value. The current control unit 15 outputs a PWM signal to the drive circuit 117 so that the detected current value becomes equal to the current command value. The PWM drive circuit 117 supplies a three-phase drive current to the motor 3, and the auxiliary steering torque is applied to the steering 61. To be applied.

  The offset learning unit 20 of the ECU 1 determines an offset learning condition while monitoring the current detection value and the like, and if the offset condition is satisfied, performs an offset current correction on the current detection value (step S4). Until the ignition switch 5a is turned off (YES in step S5), the ECU 1 repeatedly executes the above-described processing.

  13 and 14 are flowcharts showing details of the above-described offset learning process (step S4). First, the learning condition determination unit 21 determines whether or not an offset learning start condition is satisfied by monitoring an assist state, a vehicle speed signal, a duty command value, a current command value, a motor rotation number, and the like (step S101). ). The offset learning start conditions are as shown in the table of FIG. In other words, the start condition is satisfied immediately after entering the assist permission state or after 300 seconds have passed since the last learning, when almost no current flows through the motor 3 and the vehicle speed is 0 km / h. For example, as shown in FIG. 15, after the ignition switch 5a is turned on, when the vehicle stops in the assist-permitted state (time t0), or when the vehicle stops after 300 seconds have passed since the previous learning (time t1, t2, At t3), the offset learning start condition is satisfied. Further, as shown in FIG. 16, the establishment of the learning condition continues until time t14 when a maximum of 20 msec elapses from time t10.

  If the learning condition is not satisfied (NO in step S101), the status flag is set to “0” (step S107), and the process returns to the main flowchart of FIG. On the other hand, when the learning condition is satisfied (YES in step S101), the learning determination unit 21 further determines whether or not the offset learning determination condition is satisfied (step S102). As described above, a definite condition for offset learning is satisfied after 60 seconds have elapsed since entering the assist-permitted state, or immediately after entering the assist-permitted state. If the confirmation condition for offset learning is not satisfied (NO in step S102), the process returns to the main flowchart in FIG. 11. If the confirmation condition is satisfied (YES in step S102), the learning condition determination unit 21 performs step. The processing after S103 is executed (time t11 in FIG. 16).

  That is, the learning condition determination unit 21 sets the state flag to “1” (step S103), and instructs the current control unit 15 to output a duty command value with a duty ratio of 50% (step S104). Furthermore, the learning condition determination unit 21 clears the past value of the integration element in the current control unit 15 (step S105). As a result, the drive current to the motor 3 becomes almost zero, and the current detection values from the current detection circuits 120 and 121 are only offset components. Furthermore, the learning condition determination unit 21 clears the past value of the integral control element of the current control unit 15 and prevents the current from flowing to the motor 3 due to the past value (step S105).

  Subsequently, the offset estimation unit 22 determines whether or not an offset estimation condition is satisfied (step S106). That is, when the state flag is “1” and the absolute value of the detected current value is smaller than 16 dec, the offset estimation unit 22 determines that the offset estimation condition is satisfied (YES in step S106), and estimates the offset current. The process is started (time t12 in FIG. 16). On the other hand, if the offset estimation condition is not satisfied (NO in step S106), the status flag is returned to "0" (step S107), the offset estimated value is returned to the previous value, and the process returns to the main flowchart.

  When the offset estimation condition is satisfied (YES in step S106), the U-phase and W-phase current detection values are sampled as offset current values (step S108). If the absolute value of the sampled offset current is 8 dec or less, the offset estimation unit 22 determines the offset current as an offset value. On the other hand, if the absolute value of the offset current is larger than 16 dec, the offset estimation unit 22 determines the upper limit or lower limit of ± 8 dec as the offset value (step S109). The offset estimation unit 22 samples the offset value determined in this way 64 times, and sets the average value as the offset estimation value (step S110). In this way, after the offset value estimation process is completed, the offset estimation unit 22 sets the status flag to “2” (step S115).

  Subsequently, the offset correction unit 23 determines whether an offset value update condition is satisfied (step S111). That is, if the status flag is “2”, the previous offset value is not equal to the estimated offset value, and the estimated offset value is 8 dec or less, the offset correction unit 23 determines that the offset value update condition is satisfied ( YES in step S111). On the other hand, if the offset value update condition is not satisfied, the process returns to the main flowchart.

  In step S112, when the absolute value of the offset estimated value−the previous offset value is 4 dec or less, the offset correction unit 23 replaces the offset estimated value with the offset correction value. When the absolute value of the estimated offset value−the previous offset value is larger than 4 dec, the offset correction unit 23 replaces the value of (offset estimated value−offset previous value) / 2 with the offset correction value.

  In this way, when the offset correction value update process is completed (step S113), the offset correction unit 23 returns the status flag to “0” (step S114), and the process returns to the main flowchart (time t13 in FIG. 16). The offset correction value updated in this way is output to the adder 16 in FIG. 3, and offset correction of the current detection value is performed.

  As described above, according to the present invention, it is possible to realize highly accurate offset correction by learning the offset current in a state where the duty command value is fixed. That is, since the drive current to the motor becomes zero, only the offset current in the current detection circuit can be detected.

  Moreover, the state where the drive current to the motor becomes zero can be estimated by taking into consideration the motor rotation speed, the current command value, the vehicle speed and the like. Thereby, it is possible to execute highly accurate offset learning even during the assist operation.

  Furthermore, according to the present invention, the offset correction value by offset learning is used only by current control. Therefore, even if offset learning does not function sufficiently, the influence on other functions can be minimized.

  Although the present embodiment has been described above, the present invention is not limited to the above-described configuration, and can be changed without departing from the spirit of the present invention. For example, the power steering control device according to the present embodiment can be applied to a hydraulic power steering device regardless of a column type or a rack type. Further, the form of the program is not limited to the above-described flowchart, and can be changed as long as the same function can be realized. Moreover, it is needless to say that various numerical values such as the above-described threshold value, duration time, and parameter are merely examples, and other numerical values can be used.

It is a schematic diagram of a power steering device concerning an embodiment of the present invention. It is a block diagram of a power steering control device concerning an embodiment of the present invention. It is a functional block diagram of the power steering control device concerning the embodiment of the present invention. It is a block diagram of the offset learning part which concerns on embodiment of this invention. It is a block diagram of the learning condition determination part which concerns on embodiment of this invention. It is a block diagram of the offset estimation part which concerns on embodiment of this invention. It is a figure showing the detail of the learning determination process which concerns on embodiment of this invention. It is a figure showing the detail of the offset estimation process which concerns on embodiment of this invention. It is a figure showing the detail of the offset update process which concerns on embodiment of this invention. It is a graph showing the offset value for every sampling frequency concerning the embodiment of the present invention. It is a graph showing the offset value for every sampling frequency concerning the embodiment of the present invention. It is a flowchart showing operation | movement of the power steering control apparatus which concerns on embodiment of this invention. It is a flowchart showing the detail of the offset learning process which concerns on embodiment of this invention. It is a flowchart showing the detail of the offset learning process which concerns on embodiment of this invention. It is a timing chart of the offset learning process which concerns on embodiment of this invention. It is a timing chart of the offset learning process which concerns on embodiment of this invention. It is a block diagram of the conventional power steering control device.

1 ECU
3 Motor 4 Torque sensor 11 Current command value calculation unit 13 Vector control unit 15 Current control unit 20 Offset learning unit 21 Learning condition determination unit 22 Offset estimation unit 23 Offset correction units 120 and 121 Current detection circuits 124 and 125 Amplification circuit

Claims (5)

A torque sensor for detecting a steering torque applied to the steering;
A current command value calculation unit that calculates a current command value for controlling the multiphase drive current based on the detected steering torque;
Based on the current command value, a current control unit for calculating a pulse width-modulated multiphase duty command value;
A drive circuit for applying a drive current based on the duty command value to a multiphase motor for generating steering assist torque;
A current detection circuit that detects a value of the drive current in the multiphase motor and outputs a current detection signal;
An A / D converter for A / D converting the current detection signal and outputting a current detection value;
When the drive current becomes substantially zero, and determines whether to start the offset learning, when determining the start of the offset learning, learning condition for setting the duty command value to the duty ratio of 50% A determination unit;
If the start of offset learning is determined, an offset estimation unit that detects the current detection value and estimates an offset current value of the current detection circuit;
An offset correction unit configured to correct each phase of the current detection value based on the offset current value ;
When the difference between the newly estimated first offset current value and the previously estimated second offset current value is greater than or equal to a predetermined value, the offset estimation unit reduces the difference. A power steering control device for correcting the first offset estimated value . The learning condition determination unit, the current command value, the duty command value, the rotational speed of the motor, or based on at least one of vehicle speed, claim 1, wherein the determining whether to start the offset learning The power steering control device described in 1. The power steering control device according to claim 1 , wherein the learning condition determination unit determines the start of offset learning after a predetermined time has elapsed since the last offset learning. The power steering control device according to claim 1 , wherein the offset estimation unit estimates the offset current value by sampling the current detection value a plurality of times. 2. The power steering control device according to claim 1 , further comprising means for amplifying the plurality of current detection signals represented by digital values and calculating an average value thereof.

Images