Search

US-20260126347-A1 - RESPONSE TIME EVALUATION METHOD

US20260126347A1US 20260126347 A1US20260126347 A1US 20260126347A1US-20260126347-A1

Abstract

This response time evaluation method includes: a step for acquiring time-series data x(i) of an evaluation signal output from an object being controlled when a step input signal is input to the object being controlled; a step for calculating an unbiased variance σ n 2 of the time-series data x(i) over a noise sampling interval of a length N defined in a steady-state interval of the step input signal; a step for calculating sample variance and moving average time-series data σ np 2 (i), x ma (i) of the time series data x(i) over a smoothing interval of a length NP centered at discrete time points i; a step for calculating time-series data y(i) of a smoothed signal on the basis of the time-series data σ np 2 (i), x ma (i); and a step for calculating the response time of the evaluation signal on the basis of the time-series data y(i).

Inventors

  • Toshimichi Takahashi

Assignees

  • MEIDENSHA CORPORATION

Dates

Publication Date
20260507
Application Date
20240312
Priority Date
20230515

Claims (10)

  1. 1 . A response time evaluation method for an evaluation signal output from a control target, the method comprising the steps of: (A) acquiring time-series data x(i) of the evaluation signal output from the control target when a step input signal having a stepwise change is input to the control target, where i is a parameter indicating discrete time; (B) calculating a variance σ n 2 of the time-series data x(i) over a noise sample section defined within a steady section of the step input signal; (C) calculating, over a smoothing section having a length equal to or less than the noise sample section and centered at the discrete time i, time-series data σ np 2 (i) of a variance of the time-series data x(i), time-series data x ma (i) of a moving average, and time-series data y(i) of a smoothed signal defined by Formula (1); and (D) calculating a response time of the evaluation signal based on the time-series data y(i), wherein, in the step (C), a smoothing section length, which is a length of the smoothing section, is optimized such that a peak value σ peak 2 of the time-series data σ np 2 (i) in the steady section falls within a variance setting range determined to include the variance σ n 2 . [ Math . 1 ]  y ⁡ ( i ) = σ np 2 ( i ) - σ n 2 σ np 2 ( i ) ⁢ { x ⁡ ( i ) - x ma ( i ) } + x ma ( i ) ( 1 )
  2. 2 . The response time evaluation method according to claim 1 , wherein, in the step (D), a time from a rise time of the step input signal to a time when the time-series data y(i) exceeds a threshold determined based on the step input signal is calculated as the response time.
  3. 3 . The response time evaluation method according to claim 1 , wherein the steady section is a section before or after the step input signal changes stepwise.
  4. 4 . The response time evaluation method according to claim 1 , wherein the control target includes a rotating body that rotates in response to the input signal.
  5. 5 . (canceled)
  6. 6 . The response time evaluation method according to claim 1 , wherein, in the step (C), a smoothing section length, which is a length of the smoothing section, is optimized such that the peak value σ peak 2 of the time-series data σ np 2 (i) in the steady section falls within the variance setting range determined to include the variance σ n 2 , and that the time-series data y(i) in the steady section falls within a signal setting range determined to include a command value corresponding to the step input signal.
  7. 7 . A response time evaluation method for an evaluation signal in a dynamometer system including a dynamometer, an inverter configured to supply electric power in response an input signal to the dynamometer, and an evaluation signal output unit configured to output the evaluation signal in accordance with a speed or torque of the dynamometer, the method comprising the steps of: (A) acquiring time-series data x(i) of the evaluation signal output from the sensor evaluation signal output unit when a step input signal having a stepwise change is input to the inverter, where i is a parameter indicating discrete time; (B) calculating a variance σ n 2 of the time-series data x(i) over a noise sample section defined within a steady section of the step input signal; (C) calculating, over a smoothing section having a length equal to or less than the noise sample section and centered at the discrete time i, time-series data σ np 2 (i) of a variance of the time-series data x(i), time-series data x ma (i) of a moving average, and time-series data y(i) of a smoothed signal defined by Formula (2); and (D) calculating the response time of the evaluation signal based on the time-series data y(i), wherein, in the step (C), a smoothing section length, which is a length of the smoothing section, is optimized such that a peak value σ peak 2 of the time-series data σ np 2 (i) in the steady section falls within a variance setting range determined to include the variance σ n 2 . [ Math . 2 ]  y ⁡ ( i ) = σ np 2 ( i ) - σ n 2 σ np 2 ( i ) ⁢ { x ⁡ ( i ) - x ma ( i ) } + x ma ( i ) ( 2 )
  8. 8 . The response time evaluation method according to claim 2 , wherein, in the step (C), a smoothing section length, which is a length of the smoothing section, is optimized such that the peak value σ peak 2 falls within the variance setting range, and that the time-series data y(i) in the steady section falls within a signal setting range determined to include a command value corresponding to the step input signal.
  9. 9 . The response time evaluation method according to claim 3 , wherein, in the step (C), a smoothing section length, which is a length of the smoothing section, is optimized such that the peak value σ peak 2 falls within the variance setting range, and that the time-series data y(i) in the steady section falls within a signal setting range determined to include a command value corresponding to the step input signal.
  10. 10 . The response time evaluation method according to claim 4 , wherein, in the step (C), a smoothing section length, which is a length of the smoothing section, is optimized such that the peak value σ peak 2 falls within the variance setting range, and that the time-series data y(i) in the steady section falls within a signal setting range determined to include a command value corresponding to the step input signal.

Description

TECHNICAL FIELD The present invention relates to a response time evaluation method. More specifically, the present invention relates to a response time evaluation method for evaluating a response time of an evaluation signal output from a control target or a dynamometer system. BACKGROUND ART A chassis dynamometer system is used in tests for measuring and evaluating, for example, a power consumption rate, a fuel consumption rate, and exhaust emission purification performance of a vehicle. The chassis dynamometer system includes rollers configured to receive wheels of the vehicle to be tested, and a dynamometer connected to the rollers. Conditions close to actual running conditions are reproduced by applying, to the vehicle running on the rollers, running resistances such as rolling resistance and inertial resistance, which occur during actual running, using the dynamometer and the rollers. In such a test using the chassis dynamometer system, in order to guarantee measurement results and evaluation results obtained in the test, it is also necessary to evaluate whether a response time of the dynamometer is appropriate. The response time is often calculated by inputting, to the control target, an input signal having a stepwise change, and measuring a change in a predetermined evaluation signal output from the control target. CITATION LIST Patent Document Patent Document 1: Japanese Unexamined Patent Application, Publication No. 2022-148844 DISCLOSURE OF THE INVENTION Problems to be Solved by the Invention In many cases, the evaluation signal output from a sensor includes noise. Therefore, in the technique disclosed in Patent Document 1, median filter processing is applied to time-series data of the evaluation signal, thereby removing noise contained in the original evaluation signal while avoiding causing a delay in a rise of the evaluation signal. More specifically, in the technique disclosed in Patent Document 1, a time width of the median filter is optimized such that a signal after the median filter processing falls within a range between a predetermined upper limit and a predetermined lower limit. Accordingly, according to the technique disclosed in Patent Document 1, if the S/N ratio of the evaluation signal is large, the response time of the evaluation signal can be calculated with high accuracy. However, if the S/N ratio becomes small, the median filter processing may cause excessive smoothing, and the response time may not be calculated with high accuracy. An object of the present invention is to provide a response time evaluation method capable of evaluating the response time of the evaluation signal with high accuracy by executing processing based on a magnitude of noise contained in the original evaluation signal. Means for Solving the Problems (1) A response time evaluation method according to the present invention is a response time evaluation method for an evaluation signal output from a control target, in which the method includes the steps of: (A) acquiring time-series data x(i) of the evaluation signal output from the control target when a step input signal having a stepwise change is input to the control target, where i is a parameter indicating discrete time; (B) calculating a variance σn2 of the time-series data x(i) over a noise sample section defined within a steady section of the step input signal; (C) calculating, over a smoothing section having a length equal to or less than the noise sample section and centered at the discrete time i, time-series data σnp2(i) of a variance of the time-series data x(i), time-series data xma(i) of a moving average, and time-series data y(i) of a smoothed signal defined by Formula (1); and (D) calculating the response time of the evaluation signal based on the time-series data y(i). [Math. 1]y⁡(i)=σnp2(i)-σn2σnp2(i)⁢{x⁡(i)-xma(i)}+xma(i)(1) (2) In this case, in the step (D), it is preferable to calculate, as the response time, a time from a rise time of the step input signal to a time when the time-series data y(i) exceeds a threshold determined based on the step input signal. (3) In this case, it is preferable that the steady section be a section before or after the step input signal changes stepwise. (4) In this case, it is preferable that the control target include a rotating body that rotates in response to the input signal. (5) In this case, in the step (C), it is preferable to optimize a smoothing section length, which is a length of the smoothing section, such that a peak value σpeak2 of the time-series data σnp2(i) in the steady section falls within a variance setting range determined to include the variance σn2. (6) In this case, in the step (C), it is preferable to optimize a smoothing section length, which is a length of the smoothing section, such that a peak value σpeak2 of the time-series data σnp2(i) in the steady section falls within a variance setting range determined to include the variance σn2, and that the time-series data y(i) in