JP2013142687A - Specimen parameter estimation device and method for detecting mechanical resonance frequency - Google Patents

Abstract

PROBLEM TO BE SOLVED: To solve such a problem that when resonance suppression control for suppressing mechanical resonance is applied, resetting of a mechanical parameter is required in each change of a specimen and quick estimation of a mechanical model is difficult.SOLUTION: Model board diagram calculation means prepares a plurality of estimation candidate values (J[k] (k=1,2,3 to N)) of specimen inertia moment Jout of model parameters, sets other model parameters of a dynamometer system and calculates model board diagram data H. Evaluation function operation means calculates an evaluation function F by integrating an absolute value of (H-H)/Hin frequencies ωto ωpreviously set by a logarithmic scale by using the model board diagram data Hand a frequency response H. Evaluation function minimizing means determines estimation candidate values of the specimen inertia moment Jfor minimizing the calculated evaluation function F. A transfer function of the model is calculated by using minimizing inertia moment Jand the other model parameters of the dynamometer system.

Description

  The present invention relates to a parameter estimation device for a specimen by a dynamometer and a method for detecting a mechanical resonance frequency.

  In the dynamometer system applied to the engine bench system, the drive bench system, etc., the specimen is changed such that the engine is an engine bench system, the transmission is a drive bench system, and the torque converter is a torque tester. When this is done, the mechanical resonance frequency typically changes. Many recent dynamometer systems apply resonance suppression control that suppresses mechanical resonance. However, each time the specimen is changed, it is necessary to reset the mechanical system parameters, corresponding to the mechanical resonance frequency of the specimen. It is necessary to perform the resonance suppression control. For this purpose, the mechanical resonance frequency must be detected when the specimen is changed. As a method for detecting the mechanical system parameter, Patent Document 1 is known.

  FIG. 6 shows an example of an apparatus for estimating mechanical system parameters in the engine bench system described in Patent Document 1. Reference numeral 1 denotes an engine as a specimen, 2 denotes a dynamometer, and is connected to the engine 1 via a coupling shaft 3. ing. Reference numeral 4 denotes an opening control unit that calculates an opening signal based on a command from the opening command unit 13. A throttle actuator 5 controls the throttle opening based on the opening signal. 6 is a torque meter, and 7 is a speedometer. The torque and speed signals detected by these measuring instruments are subjected to predetermined processing by the detection storage unit 8, and the values are stored.

  Reference numeral 9 denotes a calculation unit composed of a personal computer or the like, which inputs and records various signals of the engine bench system, downloads various parameters to the system, and sets various command values, and constructs a model and executes parameter identification. A parameter identification unit is provided. An inverter 10 executes torque / speed control for the dynamometer 2 based on the output from the torque current control unit 11. A torque current command unit 12 generates a current command value including a DC value and a random value, and outputs the current command value to the torque current control unit 11 and the detection storage unit 8.

Japanese Patent No. 4788627

  FIG. 7 is a model for estimating the mechanical parameters of the two inertia system, the inverter characteristics, and various detection delay characteristics. The torque T2ref output from the torque current command unit 12 is input to the primary delay element 21 of the inverter and is subtracted. The difference between the spring torque generated by the spring element 23 in the portion 27 is calculated. The deviation value is output to the second (dynamometer) inertia moment element 22 in the two-inertia system, and becomes an output ω2det through a dynamometer speed detection dead time element 25 with a predetermined delay. The calculated value of the second moment of inertia element 22 is also output to the subtractor 28, which performs a difference calculation with the output value of the first (specimen) moment of inertia element 24 and outputs the difference signal as a spring element. To 23. The spring element 23 performs a PI calculation based on the difference signal to generate a spring torque, and outputs the value as a torque T12det through the detection dead time element 26 of the shaft torque, and the subtractor 27 and the first inertia moment element 24.

FIG. 8 shows a flowchart of parameter estimation. In step S1, Bode diagram data is calculated by the Bode diagram data calculation means. For this purpose, the calculation unit 9 obtains (ω, H _R ) from the measured Bode diagram data. omega the actual value of the frequency, H _R is a complex number representing the frequency response at omega. In step S2, a parameter initial value _PMO appropriate for the model is set by the model parameter initial value setting means. Parameter variables for setting the initial value _PMO include inverter DC gain K ₁ , inverter response frequency ω ₁ , mechanical inertia moments J ₁ and J ₂ , mechanical rotation losses C ₁ and C ₂ , mechanical spring constant K ₁₂ , mechanical system spring loss C ₁₂ , shaft torque detection dead time T _T , and dynamometer speed detection dead time T _E.

In step S3, it calculates the frequency ω and the frequency response H _M models Bode the model function calculating means. In step S4, the evaluation function F (H _R , H _M ) is obtained by executing the calculation of equation (1) using ω and H _R measured by the evaluation function calculation means and the model H _M.

Here, the meaning of the function F is to integrate the absolute value of (H _M −H _R ) / H _{R from} the preset frequency ω ₀ to ω ₁ on the log scale. In step S5, convergence determination based on nonlinear programming is executed. If the convergence has not been achieved, a new model parameter P _M is calculated by nonlinear programming in step S6 by model parameter calculating means, and model function calculating means. and creating in order to calculate a new H _M.

According to Patent Document 1 as described above, mechanical system parameters, inverter characteristics, and various detection delay characteristics can be accurately estimated. Therefore, the mechanical system model of the dynamometer can be constructed by using the technique of Patent Document 1, and the mechanical resonance frequency of the dynamometer can be calculated by calculating the transfer function of the model.
However, as shown in FIG. 8, Patent Document 1 uses iterative calculation using nonlinear programming or the like with convergence determination inside the arithmetic unit 9. For this reason, when performing resonance suppression control corresponding to the mechanical resonance frequency of the specimen when changing the specimen, it is difficult to estimate the mechanical system model in a short time, and depending on the method of repeated calculation The calculation does not necessarily converge with a significant system.

  It is an object of the present invention to provide a parameter estimation device for a specimen and a method for detecting a mechanical resonance frequency that can estimate the moment of inertia and estimate the mechanical resonance frequency even when the specimen is changed.

According to claim 1 of the present invention, the dynamometer connected to the specimen via the coupling shaft is controlled by an inverter, and the speed and shaft torque of the detected dynamometer are taken out to the calculation unit and measured at the frequency ω. the measured Bode plot from the frequency response H _R, obtains a model Bode diagram data H _M from the model parameters set in advance, evaluation calculation unit by using the frequency response H _R and model Bode diagram data H _M determined function In estimating the parameters of the dynamometer system by calculating the evaluation function with the calculation means,
Among the model parameters, specimen inertia estimated candidate values of moment J ₁ preparing a plurality Model Bode calculating means for calculating the model Bode data H _M set the other model parameters dynamometer system When,
Using the model board diagram data H _M and the frequency response H _R in the evaluation function calculation means, the absolute value of (H _M −H _R ) / H _R is preset at a frequency ω ₀ to ω ₁ in a logarithmic scale. Evaluation function calculation means for calculating the evaluation function F by integrating
Evaluation function minimizing means for obtaining an estimated candidate value of the specimen inertia moment J ₁ for minimizing the calculated evaluation function F;
The model transfer function is calculated using the minimized moment of inertia J ₁ and other model parameters of the dynamometer system.

Claim 2 of the present invention, the evaluation function according to the evaluation function calculation means, the absolute value of the logarithmic values of the ratio of the absolute value of the frequency response H _R each said model Bode diagram data H _M, logarithmic frequency It is characterized by being configured to be integrated by a scale.

According to the third aspect of the present invention, the dynamometer connected to the specimen through the coupling shaft is controlled by the inverter, and the speed and shaft torque of the detected dynamometer are taken out to the calculation unit and measured at the frequency ω. the measured Bode plot from the frequency response H _R, obtains a model Bode diagram data H _M from the model parameters set in advance, evaluation calculation unit by using the frequency response H _R and model Bode diagram data H _M determined function In estimating the parameters of the dynamometer system by calculating the evaluation function with the calculation means,
Among the model parameters, the estimated candidate values of the specimen inertia J ₁ preparing a plurality of sets other model parameters dynamometer system obtains the model Bode data H _M, the evaluation function calculating unit Using the model board diagram data H _M and the frequency response H _R , the absolute value of (H _M −H _R ) / H _R is integrated from the preset frequency ω ₀ to ω ₁ on a logarithmic scale to evaluate the function. F is calculated, an estimated candidate value of the specimen inertia moment J ₁ that minimizes the calculated evaluation function is obtained, and the model is transmitted using the minimized inertia moment J ₁ and other model parameters of the dynamometer system. The function is calculated, and the resonance frequency is calculated from the calculated transfer function of the model.

Claim 4 of the present invention, the evaluation function according to the evaluation function calculation means, the absolute value of the logarithmic values of the ratio of the absolute value of the frequency response H _R each said model Bode diagram data H _M, logarithmic frequency It is obtained by integrating with a scale.

As described above, according to the present invention, it is only necessary to obtain J ₁ [k] that minimizes the N evaluation function values, which is the number of candidates for the estimated value of the moment of inertia J ₁ of the specimen. Therefore, the resonance frequency can be calculated in a short time.

The machine model figure which shows embodiment of this invention. 5 is a flowchart according to the present invention. 4 is another flowchart according to the present invention. The simulation result figure by this invention. The resonance frequency simulation result figure by this invention. The block diagram of the parameter estimation apparatus of an engine bench system. Mechanical model diagram. The flowchart of the conventional parameter estimation.

The present invention, when performing the resonance suppression control corresponding to the mechanical resonance frequency of the specimen, by setting change only mechanical system inertia J ₁ of the specimen, and the estimation of the parameters of the mechanical system model in a short time, the model The transfer function is calculated using the parameters, and the resonance frequency is calculated from the calculated transfer function of the model. This will be described in detail below based on examples.

As shown in FIG. 1, the mechanical system model used in the embodiment of the present invention is the same mechanical system model as that of Patent Document 1 shown in FIG. 7, except that the mechanical system model shown in FIG. Parameters other than the mechanical system moment of inertia J _{1 of the} specimen such as the engine are parameters specific to the dynamometer system, and once measured, calculation is not required when the specimen is changed. Therefore, the portion that changes during specimen change is to calculate the mechanical resonant frequency by focusing that only the mechanical system inertia J ₁ of the specimen. As a dynamometer system, the dynamometer system shown in FIG.

FIG. 2 is a flowchart for calculating the mechanical resonance frequency. In step S11 in the figure, the Bode diagram data is calculated by the Bode diagram data calculating means. Therefore, the calculation unit 9 uses the torque current command based on the actual measurement value stored in the detection storage unit 8 after the measurement is finished, the Bode diagram created using the shaft torque detector, the torque current command and the speed detection value of the dynamometer. (Ω, H _R ) is calculated from the Bode diagram created in the above. omega the actual value of the frequency, H _R is a complex number representing the frequency response at omega.

In step S12, the model parameter initial value setting means causes parameters unique to the dynamometer system, that is, dynamometer moment of inertia J ₂ , spring stiffness K _{12 of} the engine / dynamometer coupling shaft, and spring loss of the engine / dynamometer coupling shaft. Appropriate parameter initial values are set for the models of C ₁₂ , shaft torque detection dead time T _T , dynamometer rotation speed detection dead time T _E , and transfer function K ₁ / (S / ω ₁ ). Step S
In 13, a plurality (J ₁ [k] (k = 1, 2, 3... N)) of estimated value candidates of the moment of inertia J ₁ of the engine serving as a specimen are prepared in advance.

Step S14 In a model function calculation means by calculating the frequency ω and the frequency response H _M models Bode. Next, in step S15, the evaluation function F (H _R , H 1) is calculated by executing the calculation of equation (1) using ω and H _R measured for each J ₁ [k] by the evaluation function calculation means and the model H _M. H _M ), and J ₁ [kmin] that minimizes the evaluation function is obtained in step S16. In step S16, the transfer function of the model in FIG. 7 is obtained from the determined J ₁ [kmin] and the parameters set in step S12, and the resonance frequency of the transfer function is calculated in step S17.

According to this embodiment, since the operation in steps S13~S15 only seek k that minimizes repeat N times is the estimated value candidates number of J _1, enables sequential calculations, the resonance frequency It can be calculated in a short time.

FIG. 3 is a flowchart for calculating the resonance frequency according to the second embodiment.
In this embodiment, the difference from FIG. 2 is that the arithmetic expression in step S15 ′ is executed based on the expression (2), and the others are the same as FIG.

The value of the calculation of the evaluation function performed in step 15 of the first embodiment may change depending on the phase of the Bode diagram. In the second embodiment, in consideration of this point, the evaluation function is a value determined only by the gain characteristic of the Bode diagram. In other words, the evaluation function F (H _R , H _M ) is a log function (logarithmic function) of the ratio of the absolute value of the frequency response H _{M of the} model expressed by complex numbers and the actually measured frequency response H _R expressed by complex numbers. ) The absolute value of the value is obtained by integrating the frequency on the log scale.

According to this embodiment, in addition to the effects of the first embodiment, even when the shaft torque detection dead time T _T and the dynamometer rotation speed detection dead time T _E change due to some condition, an appropriate engine inertia is achieved. The moment J ₁ can be calculated, and as a result, an appropriate resonance frequency can be calculated.

4 and 5 show simulation results according to the present invention.
In FIG. 4, the horizontal axis represents the moment of inertia J ₁ [k] (k = 1, 2, 3... N) of the engine, and the vertical axis represents the value of the evaluation function.
FIG. 5 shows a model function (line A) for J ₁ that minimizes the vertical axis of FIG. 4 and an actually measured transfer function (line B). In the actual measurement indicated by the line B, there is a disturbance of about 100 Hz, so that a peak-like gain fluctuation occurs at a position of about 100 Hz in the transfer function. When the resonance frequency is detected simply by capturing the peak gain, the example shown in FIG. 5 is erroneously detected as the resonance frequency 100 Hz.

However, according to the present invention, it is possible to obtain a model transfer function in which the measured transfer function indicated by line B and the model transfer function indicated by line A overlap each other in the vicinity of 400 Hz, and the resonance frequency of this model transfer function is calculated. This makes it possible to easily estimate the actual transfer function. In addition, it is possible to estimate the moment of inertia J ₁ of the specimen from the obtained model transfer function.

As described above, according to the present invention, it is only necessary to obtain the minimum k by repeating N times that is the number of estimated values of the inertia moment J ₁ of the specimen, so that sequential calculation is possible and the resonance frequency can be shortened in a short time. It can be calculated.
Further, even when the shaft torque detection dead time T _T and the dynamometer rotation speed detection dead time T _E change due to some condition, an appropriate engine moment of inertia J ₁ can be calculated. The resonance frequency can be calculated.

DESCRIPTION OF SYMBOLS 1 ... Specimen 2 ... Dynamometer 3 ... Coupling shaft 4 ... Engine opening control part 5 ... Throttle actuator 6 ... Torque meter 7 ... Speedometer 8 ... Detection and preservation | save part 9 ... Calculation part 10 ... Inverter 11 ... Torque current control part 12 ... Torque current command section

Claims (4)

The dynamometer is connected through a specimen with a binding shaft controlled by an inverter, measured Bode the speed and the shaft torque of the detected stored dynamometer from the frequency response H _R at the frequency ω which is measured by taking out the calculation unit Figure and obtains the model Bode diagram data H _M from the model parameters set in advance, using a frequency response H _R and model Bode diagram data H _M obtained calculating an evaluation function in the evaluation function calculating means calculating section In estimating the parameters of the dynamometer system,
Among the model parameters, specimen estimated candidate values of the moment of inertia J ₁ [k] preparing a plurality models Bode to set the other model parameters dynamometer system calculates the model Bode data H _M Figure calculation means;
Using the model board diagram data H _M and the frequency response H _R in the evaluation function calculation means, the absolute value of (H _M −H _R ) / H _R is preset at a frequency ω ₀ to ω ₁ in a logarithmic scale. Evaluation function calculation means for calculating the evaluation function F by integrating
Evaluation function minimizing means for obtaining an estimated candidate value of the specimen inertia moment J ₁ for minimizing the calculated evaluation function F;
Specimen parameter estimation apparatus characterized by being configured so as to calculate the transfer function of the model using the other model parameters minimizing the moment of inertia J ₁ and the dynamometer system. Configured such that the evaluation function by the evaluation function calculating unit, the absolute value of the logarithmic values of the ratio of the absolute value of the frequency response H _R each said model Bode diagram data H _M, obtained by integrating the frequency on a logarithmic scale The specimen parameter estimation apparatus according to claim 1, wherein The dynamometer is connected through a specimen with a binding shaft controlled by an inverter, measured Bode the speed and the shaft torque of the detected stored dynamometer from the frequency response H _R at the frequency ω which is measured by taking out the calculation unit Figure and obtains the model Bode diagram data H _M from the model parameters set in advance, using a frequency response H _R and model Bode diagram data H _M obtained calculating an evaluation function in the evaluation function calculating means calculating section In estimating the parameters of the dynamometer system,
Among the model parameters, the estimated candidate values of the specimen inertia J ₁ preparing a plurality of sets other model parameters dynamometer system obtains the model Bode data H _M, the evaluation function calculating unit Using the model board diagram data H _M and the frequency response H _R , the absolute value of (H _M −H _R ) / H _R is integrated from the preset frequency ω ₀ to ω ₁ on a logarithmic scale to evaluate the function. F is calculated, an estimated candidate value of the specimen inertia moment J ₁ that minimizes the calculated evaluation function is obtained, and the model is transmitted using the minimized inertia moment J ₁ and other model parameters of the dynamometer system. A method for detecting a resonance frequency, comprising calculating a function and calculating a resonance frequency from a transfer function of the calculated model. Said evaluation function by the evaluation function calculating unit, the absolute value of the logarithmic values of the ratio of the absolute value of the frequency response H _R each said model Bode diagram data H _M, obtained by integrating the frequency on a logarithmic scale The method for detecting a resonance frequency according to claim 3, wherein:

Images