US20140019077A1 - Deconvolution method for emissions measurement - Google Patents

Deconvolution method for emissions measurement Download PDF

Info

Publication number
US20140019077A1
US20140019077A1 US14/007,111 US201214007111A US2014019077A1 US 20140019077 A1 US20140019077 A1 US 20140019077A1 US 201214007111 A US201214007111 A US 201214007111A US 2014019077 A1 US2014019077 A1 US 2014019077A1
Authority
US
United States
Prior art keywords
function
response
recited
convolution function
convolution
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Abandoned
Application number
US14/007,111
Inventor
Frank Berghof
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
AVL Test Systems Inc
Original Assignee
AVL Test Systems Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by AVL Test Systems Inc filed Critical AVL Test Systems Inc
Priority to US14/007,111 priority Critical patent/US20140019077A1/en
Assigned to AVL TEST SYSTEMS, INC. reassignment AVL TEST SYSTEMS, INC. ASSIGNMENT OF ASSIGNORS INTEREST (SEE DOCUMENT FOR DETAILS). Assignors: BERGHOF, FRANK
Publication of US20140019077A1 publication Critical patent/US20140019077A1/en
Assigned to RBS CITIZENS, N.A., AS AGENT reassignment RBS CITIZENS, N.A., AS AGENT SECURITY INTEREST Assignors: AVL CALIFORNIA TECHNOLOGY CENTER, INC., AVL MICHIGAN HOLDING CORPORATION, AVL PEI EQUIPMENT, LLC, AVL POWERTRAIN ENGINEERING, INC., AVL PROPERTIES, INC., AVL STRATEGIC ANALYTIC SERVICES, INC., AVL TEST SYSTEMS, INC., AVL TSI EQUIPMENT, LLC
Abandoned legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01NINVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
    • G01N33/00Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
    • G01N33/0004Gaseous mixtures, e.g. polluted air
    • G01N33/0006Calibrating gas analysers
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01DMEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
    • G01D18/00Testing or calibrating apparatus or arrangements provided for in groups G01D1/00 - G01D15/00
    • G01D18/008Testing or calibrating apparatus or arrangements provided for in groups G01D1/00 - G01D15/00 with calibration coefficients stored in memory
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01DMEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
    • G01D3/00Indicating or recording apparatus with provision for the special purposes referred to in the subgroups
    • G01D3/02Indicating or recording apparatus with provision for the special purposes referred to in the subgroups with provision for altering or correcting the law of variation
    • G01D3/022Indicating or recording apparatus with provision for the special purposes referred to in the subgroups with provision for altering or correcting the law of variation having an ideal characteristic, map or correction data stored in a digital memory

Definitions

  • Emissions analyzers or measurement instruments, measure certain gaseous constituents within a sample of exhaust, or aerosol, as a function of time, or are configured to measure particulate matter, such as soot, within an exhaust sample, as examples.
  • the response of the instrument may be uncorrected for the convolution of the measurement with some other signal representative of the transfer function, or the transient response, of the instrument.
  • Deconvolution is a process used to reverse, or correct, the effects of convolution.
  • the response of an instrument is recorded online in the time domain.
  • Deconvolution of the recorded signal is performed offline in post-processing by (1) decomposing the recorded data, via a Fourier transform, into the frequency domain, (2) using a model to remove the effects convolution, and then (3) constructing a convolution corrected signal, via an inverse Fourier transform, back into the time domain.
  • the method includes determining an inverse convolution function, the inverse convolution function being in the time domain.
  • the method further includes recording a response of an instrument to an exhaust sample as a function of time, and convolving the recorded response with the inverse convolution function, the result being a convolution corrected instrument response.
  • the method includes determining an idealized convolution function, the idealized convolution function being in the time domain.
  • the idealized convolution function is transformed from the time domain to the frequency domain, and a regularizing filter function is divided by the transformed idealized convolution function. The result of the division is the inverse convolution function in the frequency domain.
  • the inverse convolution function is then transformed from the frequency domain to the time domain.
  • FIG. 1 illustrates an example system including a measurement instrument configured to respond to exhaust.
  • FIG. 2 illustrates an example method of correcting a response of a measurement instrument.
  • FIG. 3 is representative of the details of the first step from FIG. 2 .
  • FIG. 4 is representative of the details of the second step from FIG. 2 .
  • FIG. 5 is representative of the details of the third step from FIG. 2 .
  • FIG. 1 illustrates an example system 10 including an engine 12 and an exhaust pipe 14 downstream thereof.
  • the engine 12 could be an engine of a vehicle, or could be a stand-alone engine in a lab, as examples.
  • the engine 12 could further be any type of engine, including a diesel engine.
  • a portion 24 b of the sample 24 a is directed toward a measurement instrument 26
  • another portion 24 c is directed toward a filter box 28 in parallel with the measurement instrument 26 .
  • the filter box 28 need not be present, however.
  • the measurement instrument 26 is a soot sensor, such as the AVL 483 Micro Soot Sensor (MSS), for example.
  • the response (or, signal) from the measurement instrument 26 is indicative of a concentration of soot, as a function of time, within the portion 24 b of the sample 24 a.
  • a controller 30 which may be any type of known computer, is in communication with the measurement instrument 26 to record the response thereof.
  • the controller 30 could include a processor (or, CPU), screen, hard drive, mouse, keyboard, etc.
  • the controller 30 is further configured to perform each of the calculations in the steps described below, and may be configured to communicate with other various components in the system 10 .
  • a reference gas source 32 selectively in communication with the sampling line 22 b by way of an adjustable valve 34 .
  • the controller 30 in one example, is configured to adjust the valve 34 , however the valve 34 could be manually adjustable.
  • the reference gas is a gas having a known soot concentration.
  • the reference gas source 32 can include an appropriate reference gas, however, as will be appreciated from the below.
  • this disclosure extends to other types of measurement instruments.
  • this disclosure extends to gas analyzers configured to measure a quantity (e.g., a concentration) of one or more gaseous constituents within a sample of exhaust, such as of CO 2 , CO, NO, NO 2 , NO x , CH 4 , HC, O 2 , NH 3 , and N 2 O, as examples.
  • the disclosed method can further be used to deconvolute data from any measurement instrument for which a convolution curve can be determined, such as temperature, pressure, flow rate, speed and torque measurements, as examples.
  • the system 10 is likewise non-limiting, and this disclosure extends to other system set-ups, including those mounted for use on-road or in a lab.
  • FIG. 2 shows a high-level overview of the steps in one example of the disclosed method.
  • the response of the system 10 (specifically, the response of the measurement instrument 26 ) to a step input signal change is measured, at 100 .
  • idealized and inverse convolution functions are then determined at 200 and 300 , respectively. Steps 100 , 200 , 300 can be performed offline, before acquiring data during engine operation.
  • steps 100 - 300 are then used in the fourth step, at 400 , to deconvolute data acquired by the measurement instrument during engine operation.
  • this data is acquired during an emissions test.
  • the deconvoluted data can be further refined in an optional fifth step, at 500 . Steps 100 - 500 are discussed in detail below.
  • n(t) a function in the time domain
  • N(f) the same function in the frequency domain
  • FIG. 3 shows the detail of step 100 .
  • a sample of reference gas which has a known quantity of a measurable exhaust component, is connected to the measurement instrument, via positioning of the valve 34 , and an uncorrected response of the instrument x(t) is recorded.
  • the reference gas would have a known soot concentration
  • a reference gas with a known HC concentration would be selected.
  • times T A , T B , and T C are determined. As generally noted, these times are times at which the amplitude of the recorded signal is at three different percentage values relative to the known signal. This is indicative of the attenuation caused by the measurement instrument and other measurement equipment. In this example, 10%, 50%, and 90% are used, for T A , T B , and T C , respectively.
  • FIG. 4 is representative of the details of step 200 , the result of which is the determination of h(t), the idealized convolution function.
  • This function generally represents an approximation of the real convolution function, using a model consisting of the Gauss function convoluted with the impulse response function:
  • g(t) is the Gaussian function
  • i(t) is the impulse response function, defined as:
  • a look-up table is used to determine the ratio.
  • the inputs to an example look-up table are T A , T B , and T C .
  • the normalized convolution function h n (t) is calculated.
  • the normalized convolution function is:
  • h n ( t ) g n ( t )* i n ( t )
  • a scaling factor k is determined at step 206 , and is defined as:
  • T A,n is the time at which ⁇ h n (t) reaches A % of its maximum value (in this example, 10%)
  • T C,n is the time at which ⁇ h n (t) reaches C % of ats maximum value (in this example 90%).
  • scaling factor k can be used to determine the parameters ⁇ , ⁇ , and ⁇ of the idealized convolution function h(t) based on the following equations:
  • T B,n is the time at which ⁇ h n (t) reaches B % of its maximum value (in this example, 50%). Having solved for these parameters, the idealized convolution function h(t) can then be determined by solving for g(t) and i(t), above.
  • step 200 the idealized convolution function h(t) could be approximated as the first derivative of the uncorrected instrument response x(t).
  • FIG. 5 generally illustrates the steps for determining the inverse convolution function k(t).
  • the idealized convolution function h(t) is transformed into the frequency domain by Fourier transformation, as follows:
  • H ( f ) F ( h ( t )).
  • a regularizing filter function R(f) is calculated from the following equation:
  • H MAG (f) is the magnitude, or absolute value, of H(f)
  • is a positive adjustable filter parameter.
  • is a constant, positive real value.
  • is a function of frequency, however a constant value is typically sufficient.
  • can be tuned to adjust the convolution corrected instrument response y(t).
  • K ( f ) R ( f )/ H ( f ).
  • R(f) and H(f) may include complex numbers, and thus, in one example, the above division follows the rules for division of two complex numbers and can be performed by dividing the magnitude of R(f) (e.g., R MAG (f)) by the magnitude of H(f) (e.g., H MAG (f)) and subtracting the phase angle of H(f) (e.g., H PHA (f)) from the phase angle of R(f) (e.g., R PHA (f))
  • the inverse convolution function K(f) is converted into the time domain by way of an inverse Fourier transformation to determine an initial inverse convolution function k init (t):
  • the regularizing filter function R(f) depends from a positive adjustable filter parameter ⁇ , which may be a constant value, and need not be frequency dependent.
  • the positive adjustable filter parameter ⁇ is generally representative of a signal to noise ratio.
  • the uncorrected instrument response x(t) recorded in step 100 is convolved with k init (t) to construct an convolution corrected instrument response y(t), at step 310 , as follows:
  • the convolution corrected instrument response y(t) is then be evaluated relative to the known reference gas signal from step 100 , at step 312 .
  • this evaluation is performed by graphically comparing the two signals, however this could also be performed using a one-dimensional optimization algorithm to minimize the sum of squares of the deviations between the deconvoluted response and the signal representative of the known data.
  • the positive adjustable filter parameter a can further be adjusted, or “tuned,” to increase the accuracy of the inverse convolution function k init (t), thus increasing the accuracy of the convolution corrected instrument response y(t) relative to the reference gas signal from step 100 .
  • Tuning is dependent on varying the constant positive adjustable filter parameter ⁇ , from which k init (t) depends.
  • the dynamic response (or, slope) of y(t) is assessed at 314 , while overshoots and undershoots (e.g., amplitude) of y(t) are accounted for at 316 .
  • increasing ⁇ would reduce the slope of y(t) (e.g., worse recovery of dynamic response) but also lower the over and undershoots.
  • e.g., a value for ⁇ representing an acceptable compromise between error in slope and error due to over/undershoots is determined
  • the corresponding inverse convolution function is saved as k(t), at 320 , for later use in step 400 .
  • step 400 the k(t) saved at 320 is used for deconvolution of the uncorrected instrument response m(t).
  • the system 10 would be arranged as shown in FIG. 1 , for example, such that valve 34 is adjusted so that the sample 24 a sourced from the engine 12 is directed toward the instrument 26 .
  • the controller 30 executes the convolution of m(t) with k(t) by way of the following Riemann sum:
  • y i is the i-th value of the convolution corrected instrument response vector
  • m i ⁇ (j ⁇ 1) is the i ⁇ (j ⁇ 1)-th value of the uncorrected measured instrument response vector
  • k′ is the flipped inverse convolution function in the time domain (as used herein, “flipped” means that the order of the values in the vector is reversed)
  • n is the number of values in the inverse convolution function vector
  • j is the running index of the inverse convolution function vector
  • i is the running index of the uncorrected instrument response vector.
  • the calculation at step 400 can be done quickly and efficiently relative to other methods, such methods require transformations between the time and frequency domains. Post processing is thus not necessary with this disclosed method, and the convolution corrected instrument response y(t) can be determined online, during engine operation. Again, as noted above, the controller 30 can be used calculate the convolution corrected instrument response y(t).
  • the convolution corrected instrument response y(t) can be further refined to eliminate deviations that may be present at step changes.
  • this further refinement called a derivative corrected instrument response p(t) can be calculated by solving for p(t) using the following equation:
  • is a constant
  • k(t) is the inverse convolution function from 320
  • y(t) is the convolution corrected instrument resulting from 400 .
  • p(t) is solved for iteratively, using y(t) as an initial estimate for p(t). Again, this fifth step is optional, and need not be included.

Abstract

Disclosed is a method of correcting a response of an instrument. The method includes determining an inverse convolution function, the inverse convolution function being in the time domain. A response of an instrument to an exhaust sample is recorded as a function of time. The recorded response is then convolved with the inverse convolution function, the result being a convolution corrected instrument response.

Description

    RELATED APPLICATIONS
  • This application claims priority to U.S. Provisional Application No. 61/468,112, filed 28 Mar. 2011.
  • BACKGROUND
  • Emissions analyzers, or measurement instruments, measure certain gaseous constituents within a sample of exhaust, or aerosol, as a function of time, or are configured to measure particulate matter, such as soot, within an exhaust sample, as examples. The response of the instrument, however, may be uncorrected for the convolution of the measurement with some other signal representative of the transfer function, or the transient response, of the instrument. Deconvolution is a process used to reverse, or correct, the effects of convolution.
  • In one known method, the response of an instrument is recorded online in the time domain. Deconvolution of the recorded signal is performed offline in post-processing by (1) decomposing the recorded data, via a Fourier transform, into the frequency domain, (2) using a model to remove the effects convolution, and then (3) constructing a convolution corrected signal, via an inverse Fourier transform, back into the time domain.
  • SUMMARY
  • Disclosed is a method of correcting a response of an instrument. The method includes determining an inverse convolution function, the inverse convolution function being in the time domain. The method further includes recording a response of an instrument to an exhaust sample as a function of time, and convolving the recorded response with the inverse convolution function, the result being a convolution corrected instrument response.
  • Further disclosed is a method of determining an inverse convolution function. The method includes determining an idealized convolution function, the idealized convolution function being in the time domain. The idealized convolution function is transformed from the time domain to the frequency domain, and a regularizing filter function is divided by the transformed idealized convolution function. The result of the division is the inverse convolution function in the frequency domain. The inverse convolution function is then transformed from the frequency domain to the time domain.
  • These and other features of the present disclosure can be best understood from the following drawings and detailed description.
  • BRIEF DESCRIPTION OF DRAWINGS
  • The drawings can be briefly described as follows:
  • FIG. 1 illustrates an example system including a measurement instrument configured to respond to exhaust.
  • FIG. 2 illustrates an example method of correcting a response of a measurement instrument.
  • FIG. 3 is representative of the details of the first step from FIG. 2.
  • FIG. 4 is representative of the details of the second step from FIG. 2.
  • FIG. 5 is representative of the details of the third step from FIG. 2.
  • DETAILED DESCRIPTION
  • FIG. 1 illustrates an example system 10 including an engine 12 and an exhaust pipe 14 downstream thereof. The engine 12 could be an engine of a vehicle, or could be a stand-alone engine in a lab, as examples. The engine 12 could further be any type of engine, including a diesel engine.
  • Exhaust 20 generated from the engine 12 flows downstream of the engine 12, is tapped at 22 a, and a sample 24 a of the exhaust 20 is directed to a sampling line 22 b. A portion 24 b of the sample 24 a is directed toward a measurement instrument 26, whereas another portion 24 c is directed toward a filter box 28 in parallel with the measurement instrument 26. The filter box 28 need not be present, however.
  • In this example, the measurement instrument 26 is a soot sensor, such as the AVL 483 Micro Soot Sensor (MSS), for example. The response (or, signal) from the measurement instrument 26 is indicative of a concentration of soot, as a function of time, within the portion 24 b of the sample 24 a.
  • A controller 30, which may be any type of known computer, is in communication with the measurement instrument 26 to record the response thereof. As those in the art would appreciate, the controller 30 could include a processor (or, CPU), screen, hard drive, mouse, keyboard, etc. The controller 30 is further configured to perform each of the calculations in the steps described below, and may be configured to communicate with other various components in the system 10.
  • A reference gas source 32 selectively in communication with the sampling line 22 b by way of an adjustable valve 34. The controller 30, in one example, is configured to adjust the valve 34, however the valve 34 could be manually adjustable. In this example, the reference gas is a gas having a known soot concentration. The reference gas source 32 can include an appropriate reference gas, however, as will be appreciated from the below.
  • Notably, while a soot sensor is shown, this disclosure extends to other types of measurement instruments. For example, this disclosure extends to gas analyzers configured to measure a quantity (e.g., a concentration) of one or more gaseous constituents within a sample of exhaust, such as of CO2, CO, NO, NO2, NOx, CH4, HC, O2, NH3, and N2O, as examples. The disclosed method can further be used to deconvolute data from any measurement instrument for which a convolution curve can be determined, such as temperature, pressure, flow rate, speed and torque measurements, as examples. The system 10 is likewise non-limiting, and this disclosure extends to other system set-ups, including those mounted for use on-road or in a lab.
  • FIG. 2 shows a high-level overview of the steps in one example of the disclosed method. As shown, the response of the system 10 (specifically, the response of the measurement instrument 26) to a step input signal change is measured, at 100. Then, idealized and inverse convolution functions are then determined at 200 and 300, respectively. Steps 100, 200, 300 can be performed offline, before acquiring data during engine operation.
  • The results from steps 100-300 are then used in the fourth step, at 400, to deconvolute data acquired by the measurement instrument during engine operation. In one example, this data is acquired during an emissions test. The deconvoluted data can be further refined in an optional fifth step, at 500. Steps 100-500 are discussed in detail below.
  • As those in the art would immediately acknowledge, a function in the time domain is represented as n(t), for example, while the same function in the frequency domain would be represented as N(f). This notation is used throughout the application.
  • FIG. 3 shows the detail of step 100. At 102-106 a sample of reference gas, which has a known quantity of a measurable exhaust component, is connected to the measurement instrument, via positioning of the valve 34, and an uncorrected response of the instrument x(t) is recorded. As noted, in the example where the measurement instrument 26 is a soot sensor, the reference gas would have a known soot concentration Likewise, if the measurement instrument was configured to measure HC, a reference gas with a known HC concentration would be selected.
  • At 108, times TA, TB, and TC are determined. As generally noted, these times are times at which the amplitude of the recorded signal is at three different percentage values relative to the known signal. This is indicative of the attenuation caused by the measurement instrument and other measurement equipment. In this example, 10%, 50%, and 90% are used, for TA, TB, and TC, respectively.
  • FIG. 4 is representative of the details of step 200, the result of which is the determination of h(t), the idealized convolution function. This function generally represents an approximation of the real convolution function, using a model consisting of the Gauss function convoluted with the impulse response function:

  • h(t)=g(t)*i(t)
  • where g(t) is the Gaussian function is defined as:
  • 1 σ 2 π - ( ( t - μ ) 2 2 σ 2 )
  • and where i(t) is the impulse response function, defined as:
  • 1 τ - ( t τ ) , for t 0.
  • At step 202 the ratio
  • τ σ
  • is determined, which is needed to calculate the normalized convolution function hn(t) in step 204. In one example the ratio
  • τ σ
  • is determined from the following equation:
  • T B - T A T C - T A .
  • In another example, a look-up table is used to determine the ratio. The inputs to an example look-up table are TA, TB, and TC.
  • At step 204 the normalized convolution function hn(t) is calculated. The normalized convolution function is:

  • h n(t)=g n(t)*i n(t)
  • where gn(t) is the Gaussian function g(t) from above, with μ=0 and σ=1:
  • 1 2 π - ( t 2 2 )
  • and where in(t) is the impulse response function i(t) from above, with τn equal to the ratio
  • τ σ
  • determined in step 202:
  • 1 τ n - ( t τ n ) , for t 0
  • A scaling factor k is determined at step 206, and is defined as:
  • k = T C - T A T C , n - T A , n
  • where TA,n is the time at which ∫hn(t) reaches A % of its maximum value (in this example, 10%), and where TC,n is the time at which ∫hn(t) reaches C % of ats maximum value (in this example 90%).
  • At 208, scaling factor k can be used to determine the parameters σ, μ, and τ of the idealized convolution function h(t) based on the following equations:

  • σ=k

  • μ=kTB,n

  • τ=kτn
  • where TB,n is the time at which ∫hn(t) reaches B % of its maximum value (in this example, 50%). Having solved for these parameters, the idealized convolution function h(t) can then be determined by solving for g(t) and i(t), above.
  • As an alternative to step 200, the idealized convolution function h(t) could be approximated as the first derivative of the uncorrected instrument response x(t).
  • FIG. 5 generally illustrates the steps for determining the inverse convolution function k(t). At 502 the idealized convolution function h(t) is transformed into the frequency domain by Fourier transformation, as follows:

  • H(f)=F(h(t)).
  • Next, as 304, a regularizing filter function R(f) is calculated from the following equation:
  • R ( f ) = ( H MAG ( f ) ) 2 ( H MAG ( f ) ) 2 + α
  • where HMAG(f) is the magnitude, or absolute value, of H(f), and where α is a positive adjustable filter parameter. In one example, α is a constant, positive real value. In another example, α is a function of frequency, however a constant value is typically sufficient. As noted below, in the example where α is a constant, α can be tuned to adjust the convolution corrected instrument response y(t).
  • At 306 the inverse convolution function K(f) is calculated by:

  • K(f)=R(f)/H(f).
  • Notably, R(f) and H(f) may include complex numbers, and thus, in one example, the above division follows the rules for division of two complex numbers and can be performed by dividing the magnitude of R(f) (e.g., RMAG(f)) by the magnitude of H(f) (e.g., HMAG(f)) and subtracting the phase angle of H(f) (e.g., HPHA(f)) from the phase angle of R(f) (e.g., RPHA(f))
  • At 308, the inverse convolution function K(f) is converted into the time domain by way of an inverse Fourier transformation to determine an initial inverse convolution function kinit(t):

  • k init(t)=F −1(K(f)).
  • The regularizing filter function R(f) depends from a positive adjustable filter parameter α, which may be a constant value, and need not be frequency dependent. The positive adjustable filter parameter α is generally representative of a signal to noise ratio.
  • Once kinit(t) is determined, the uncorrected instrument response x(t) recorded in step 100 is convolved with kinit(t) to construct an convolution corrected instrument response y(t), at step 310, as follows:

  • y(t)=x(t)*k init(t).
  • The convolution corrected instrument response y(t) is then be evaluated relative to the known reference gas signal from step 100, at step 312. In one example this evaluation is performed by graphically comparing the two signals, however this could also be performed using a one-dimensional optimization algorithm to minimize the sum of squares of the deviations between the deconvoluted response and the signal representative of the known data.
  • As represented in steps 314-318, the positive adjustable filter parameter a can further be adjusted, or “tuned,” to increase the accuracy of the inverse convolution function kinit(t), thus increasing the accuracy of the convolution corrected instrument response y(t) relative to the reference gas signal from step 100.
  • Tuning is dependent on varying the constant positive adjustable filter parameter α, from which kinit(t) depends. The dynamic response (or, slope) of y(t) is assessed at 314, while overshoots and undershoots (e.g., amplitude) of y(t) are accounted for at 316. As an example, increasing α would reduce the slope of y(t) (e.g., worse recovery of dynamic response) but also lower the over and undershoots. Once a desirable α is found (e.g., a value for α representing an acceptable compromise between error in slope and error due to over/undershoots is determined), the corresponding inverse convolution function is saved as k(t), at 320, for later use in step 400.
  • In step 400, the k(t) saved at 320 is used for deconvolution of the uncorrected instrument response m(t). In step 400, the system 10 would be arranged as shown in FIG. 1, for example, such that valve 34 is adjusted so that the sample 24 a sourced from the engine 12 is directed toward the instrument 26.
  • To construct the convolution corrected instrument response y(t), the uncorrected instrument response m(t) is convolved with k(t):

  • y(t)=m(t)*k(t)
  • In one example, the controller 30 executes the convolution of m(t) with k(t) by way of the following Riemann sum:
  • y i = j = 1 n k j m i - ( j - 1 )
  • where yi is the i-th value of the convolution corrected instrument response vector, mi−(j−1) is the i−(j−1)-th value of the uncorrected measured instrument response vector, k′ is the flipped inverse convolution function in the time domain (as used herein, “flipped” means that the order of the values in the vector is reversed), n is the number of values in the inverse convolution function vector, j is the running index of the inverse convolution function vector, and i is the running index of the uncorrected instrument response vector.
  • Since the convolution corrected instrument response y(t) is calculated, at 400, entirely in the time domain using multiplication and summation, the calculation at step 400 can be done quickly and efficiently relative to other methods, such methods require transformations between the time and frequency domains. Post processing is thus not necessary with this disclosed method, and the convolution corrected instrument response y(t) can be determined online, during engine operation. Again, as noted above, the controller 30 can be used calculate the convolution corrected instrument response y(t).
  • At optional step 500, the convolution corrected instrument response y(t) can be further refined to eliminate deviations that may be present at step changes. In one example, this further refinement, called a derivative corrected instrument response p(t), can be calculated by solving for p(t) using the following equation:
  • p ( t ) + β · ( p t ) * k ( t ) = y ( t )
  • where β is a constant, k(t) is the inverse convolution function from 320, and y(t) is the convolution corrected instrument resulting from 400. In one example, p(t) is solved for iteratively, using y(t) as an initial estimate for p(t). Again, this fifth step is optional, and need not be included.
  • Although the different examples have the specific components shown in the illustrations, embodiments of this invention are not limited to those particular combinations. It is possible to use some of the components or features from one of the examples in combination with features or components from another one of the examples.
  • One of ordinary skill in this art would understand that the above-described embodiments are exemplary and non-limiting. That is, modifications of this disclosure would come within the scope of the claims. Accordingly, the following claims should be studied to determine their true scope and content.

Claims (20)

What is claimed is:
1. A method of correcting a response of an instrument comprising:
determining an inverse convolution function, the inverse convolution function being in the time domain;
recording a response of an instrument to an exhaust sample as a function of time; and
convolving the recorded response with the inverse convolution function, the result being a convolution corrected instrument response.
2. The method as recited in claim 1, wherein the determining step includes determining an idealized convolution function, the idealized convolution function being in the time domain.
3. The method as recited in claim 2, wherein the idealized convolution function is the first derivative of a response of the instrument to a reference exhaust sample.
4. The method as recited in claim 2, wherein the idealized convolution function is calculated by convolving a Gaussian function with an impulse response function.
5. The method as recited in claim 4, wherein the Gaussian and impulse response functions are based on a scaling factor, the scaling factor determined based on a normalized convolution function and on a response of the instrument to a reference exhaust sample.
6. The method as recited in claim 5, wherein the normalized convolution function is calculated by convolving a normalized Gaussian function with a normalized impulse response function.
7. The method as recited in claim 6, wherein the impulse response function is based on values from the response of the instrument to the reference exhaust sample.
8. The method as recited in claim 2, wherein the determining step includes transforming the idealized convolution function from the time domain to the frequency domain.
9. The method as recited in claim 8, wherein the determining step includes dividing a regularizing filter function by the transformed idealized convolution function, the result being the inverse convolution function in the frequency domain.
10. The method as recited in claim 9, wherein the determining step includes transforming the inverse convolution function from the frequency domain to the time domain.
11. The method as recited in claim 9, wherein the regularizing filter function is based on the transformed idealized convolution function and a positive adjustable filter parameter.
12. The method as recited in claim 11, wherein the positive adjustable filter parameter is a constant value independent of frequency.
13. The method as recited in claim 12, wherein the determining step includes adjusting the positive adjustable filter parameter to adjust overshoots, undershoots, and a dynamic response of the inverse convolution function.
14. The method as recited in claim 1, wherein the instrument is a gas analyzer configured to measure a concentration of a gaseous constituent of the exhaust sample as a function of time.
15. The method as recited in claim 1, further including calculating a derivative corrected instrument response to eliminate noise at step changes in the convolution corrected instrument response.
16. The method as recited in claim 1, wherein a derivative corrected instrument response is calculated by solving for p(t) using the following equation:
p ( t ) + β · ( p t ) * k ( t ) = y ( t )
where p(t) is the derivative corrected instrument response, β is a constant, k(t) is the inverse convolution function, and y(t) is the convolution corrected instrument response.
17. A method of determining an inverse convolution function comprising:
determining an idealized convolution function, the idealized convolution function being in the time domain;
transforming the idealized convolution function from the time domain to the frequency domain;
dividing a regularizing filter function by the transformed idealized convolution function, the result being the inverse convolution function in the frequency domain; and
transforming the inverse convolution function from the frequency domain to the time domain.
18. The method as recited in claim 17, wherein the idealized convolution function is calculated by convolving a Gaussian function with an impulse response function.
19. The method as recited in claim 17, wherein the regularizing filter function is based on the transformed idealized convolution function and a positive adjustable filter parameter, and wherein the positive adjustable filter parameter is a constant value independent of frequency.
20. The method as recited in claim 19, further including adjusting the positive adjustable filter parameter to adjust overshoots, undershoots, and a dynamic response of the inverse convolution function.
US14/007,111 2011-03-28 2012-03-14 Deconvolution method for emissions measurement Abandoned US20140019077A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US14/007,111 US20140019077A1 (en) 2011-03-28 2012-03-14 Deconvolution method for emissions measurement

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
US201161468112P 2011-03-28 2011-03-28
US14/007,111 US20140019077A1 (en) 2011-03-28 2012-03-14 Deconvolution method for emissions measurement
PCT/US2012/029020 WO2012134815A2 (en) 2011-03-28 2012-03-14 Deconvolution method for emissions measurement

Related Parent Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2012/029020 A-371-Of-International WO2012134815A2 (en) 2011-03-28 2012-03-14 Deconvolution method for emissions measurement

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US16/036,458 Continuation US10520480B2 (en) 2011-03-28 2018-07-16 Deconvolution method for emissions measurement

Publications (1)

Publication Number Publication Date
US20140019077A1 true US20140019077A1 (en) 2014-01-16

Family

ID=46932235

Family Applications (2)

Application Number Title Priority Date Filing Date
US14/007,111 Abandoned US20140019077A1 (en) 2011-03-28 2012-03-14 Deconvolution method for emissions measurement
US16/036,458 Active US10520480B2 (en) 2011-03-28 2018-07-16 Deconvolution method for emissions measurement

Family Applications After (1)

Application Number Title Priority Date Filing Date
US16/036,458 Active US10520480B2 (en) 2011-03-28 2018-07-16 Deconvolution method for emissions measurement

Country Status (6)

Country Link
US (2) US20140019077A1 (en)
EP (2) EP3101573B1 (en)
JP (1) JP5932018B2 (en)
CN (1) CN104303017B (en)
CA (1) CA2831593A1 (en)
WO (1) WO2012134815A2 (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2016112269A1 (en) * 2015-01-09 2016-07-14 Avl Test Systems, Inc. System and method for detecting a leak in an exhaust gas sampling apparatus
US20170010201A1 (en) * 2015-07-06 2017-01-12 Ngk Spark Plug Co., Ltd. Particulate detection apparatus and particulate detection system
US10520480B2 (en) 2011-03-28 2019-12-31 Avl Test Systems, Inc. Deconvolution method for emissions measurement

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102016223489A1 (en) * 2016-11-25 2018-05-30 Robert Bosch Gmbh Method and device for filtering a sensor signal of a sensor
EP3396398B1 (en) * 2017-04-27 2020-07-08 Rohde & Schwarz GmbH & Co. KG Signal correction method, system for correcting a measured signal, as well as oscilloscope
CN110657864B (en) * 2019-10-08 2020-12-18 三门核电有限公司 Sensor response time measuring method

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6271522B1 (en) * 1998-05-16 2001-08-07 Deutsches Zentrum Fur Luft-Und Raumfahrt E.V. Process for the quantitative analysis of gas volumes, specifically exhaust and waste gases from combustion systems or incineration plants, as well as systems for performing these processes
US20030080295A1 (en) * 2001-08-16 2003-05-01 Webber Michael E. Gas sensor for ammonia, carbon dioxide and water
US20030112980A1 (en) * 2001-12-17 2003-06-19 Siemens Vdo Automotive, Inc. Digital filter modeling for active noise cancellation
US20050016871A1 (en) * 2001-10-01 2005-01-27 Compton Richard Guy Determining gas concentration
US20050066744A1 (en) * 2003-07-16 2005-03-31 Mario Kupnik Ultrasonic gas flowmeter as well as device to measure exhaust flows of internal combustion engines and method to determine flow of gases
US20070029477A1 (en) * 2005-04-29 2007-02-08 Sionex Corporation Compact gas chromatography and ion mobility based sample analysis systems, methods, and devices
US20080006775A1 (en) * 2006-06-22 2008-01-10 Arno Jose I Infrared gas detection systems and methods
US20100073173A1 (en) * 2005-01-14 2010-03-25 Unitada Europe Limited Particulate detector

Family Cites Families (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4646274A (en) * 1983-12-21 1987-02-24 Atlantic Richfield Company Method and apparatus for correcting distorted seismic data
JP3185413B2 (en) * 1992-11-25 2001-07-09 ソニー株式会社 Orthogonal transform operation and inverse orthogonal transform operation method and apparatus, digital signal encoding and / or decoding apparatus
US5838598A (en) * 1995-12-12 1998-11-17 Analog Devices, Inc. Method and apparatus for gain correction of a sigma-delta converter
US6907383B2 (en) * 1996-03-28 2005-06-14 Rosemount Inc. Flow diagnostic system
JP3889883B2 (en) * 1998-07-21 2007-03-07 浜松ホトニクス株式会社 Method and apparatus for measuring internal information of scattering medium
JP3950243B2 (en) * 1998-11-05 2007-07-25 浜松ホトニクス株式会社 Method and apparatus for measuring internal information of scattering medium
JP4116197B2 (en) * 1999-06-24 2008-07-09 浜松ホトニクス株式会社 Signal waveform measuring method and apparatus
AU2002227768B2 (en) * 2001-02-09 2006-08-17 Commonwealth Scientific And Industrial Research Organisation Lidar system and method
US7217266B2 (en) * 2001-05-30 2007-05-15 Anderson R Rox Apparatus and method for laser treatment with spectroscopic feedback
JP3782724B2 (en) * 2001-12-05 2006-06-07 株式会社堀場製作所 Exhaust gas measurement system
JP2003280169A (en) * 2002-03-26 2003-10-02 Seiko Epson Corp Photomask and method for producing the same
JP2004271498A (en) * 2003-02-20 2004-09-30 Gigaphoton Inc Computing method of spectral index value of laser beam, arithmetic unit of spectral index value of laser beam, and measuring device of spectral waveform
CN1985266B (en) * 2004-07-26 2010-05-05 奥普提克斯晶硅有限公司 Panoramic vision system and method
FR2898411B1 (en) * 2006-03-08 2008-05-16 Inst Francais Du Petrole REAL-TIME ESTIMATION METHOD OF ENGINE COMBUSTION PARAMETERS FROM VIBRATORY SIGNALS
GB2454835B (en) * 2006-07-13 2011-03-02 Exxonmobil Upstream Res Co Removing air wave noise from electromagnetic survey data
DE102006045722B4 (en) * 2006-09-27 2014-11-27 Siemens Aktiengesellschaft Method of correcting scattered radiation in projection radiography and computer tomography and apparatus therefor
CN101257469B (en) * 2008-01-11 2010-09-08 清华大学 Method for using system information to inhibit phase noise in orthogonal frequency division multiplexing system
JP5613974B2 (en) * 2008-10-29 2014-10-29 富士通株式会社 Temperature measurement method
US8675791B2 (en) * 2009-01-09 2014-03-18 Universite D'angers Method and an apparatus for deconvoluting a noisy measured signal obtained from a sensor device
WO2011004378A1 (en) * 2009-07-08 2011-01-13 Technion Research And Development Foundation Ltd. Method and system for super-resolution signal reconstruction
CN104303017B (en) 2011-03-28 2017-05-17 Avl测试系统公司 Deconvolution method for emissions measurement

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6271522B1 (en) * 1998-05-16 2001-08-07 Deutsches Zentrum Fur Luft-Und Raumfahrt E.V. Process for the quantitative analysis of gas volumes, specifically exhaust and waste gases from combustion systems or incineration plants, as well as systems for performing these processes
US20030080295A1 (en) * 2001-08-16 2003-05-01 Webber Michael E. Gas sensor for ammonia, carbon dioxide and water
US20050016871A1 (en) * 2001-10-01 2005-01-27 Compton Richard Guy Determining gas concentration
US20030112980A1 (en) * 2001-12-17 2003-06-19 Siemens Vdo Automotive, Inc. Digital filter modeling for active noise cancellation
US20050066744A1 (en) * 2003-07-16 2005-03-31 Mario Kupnik Ultrasonic gas flowmeter as well as device to measure exhaust flows of internal combustion engines and method to determine flow of gases
US20100073173A1 (en) * 2005-01-14 2010-03-25 Unitada Europe Limited Particulate detector
US20070029477A1 (en) * 2005-04-29 2007-02-08 Sionex Corporation Compact gas chromatography and ion mobility based sample analysis systems, methods, and devices
US20080006775A1 (en) * 2006-06-22 2008-01-10 Arno Jose I Infrared gas detection systems and methods

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10520480B2 (en) 2011-03-28 2019-12-31 Avl Test Systems, Inc. Deconvolution method for emissions measurement
WO2016112269A1 (en) * 2015-01-09 2016-07-14 Avl Test Systems, Inc. System and method for detecting a leak in an exhaust gas sampling apparatus
US10656050B2 (en) 2015-01-09 2020-05-19 Avl Test Systems, Inc. System and method for detecting a leak in an exhaust gas sampling apparatus
US20170010201A1 (en) * 2015-07-06 2017-01-12 Ngk Spark Plug Co., Ltd. Particulate detection apparatus and particulate detection system
US10101257B2 (en) * 2015-07-06 2018-10-16 Ngk Spark Plug Co., Ltd. Particulate detection apparatus and particulate detection system

Also Published As

Publication number Publication date
EP2691901A2 (en) 2014-02-05
US10520480B2 (en) 2019-12-31
WO2012134815A2 (en) 2012-10-04
JP2014516404A (en) 2014-07-10
US20180321205A1 (en) 2018-11-08
WO2012134815A3 (en) 2014-04-10
EP3101573B1 (en) 2018-08-29
CN104303017B (en) 2017-05-17
EP2691901A4 (en) 2015-03-25
EP2691901B1 (en) 2016-08-10
EP3101573A1 (en) 2016-12-07
CA2831593A1 (en) 2012-10-04
CN104303017A (en) 2015-01-21
JP5932018B2 (en) 2016-06-08

Similar Documents

Publication Publication Date Title
US10520480B2 (en) Deconvolution method for emissions measurement
CN107329932B (en) Time-frequency domain modal parameter identification method based on nonlinear frequency modulation component decomposition
Li et al. Real-time infrared gas detection based on an adaptive Savitzky–Golay algorithm
EP2863187A2 (en) Flow rate measurement method and apparatus
CN109708861A (en) A kind of automobile exhaust pipe thermal vibration detection method and detection system, computer program
Henry Spectral analysis techniques using Prism signal processing
US20110218742A1 (en) Calculating device and flow meter equipped with calculating device
CN108254794B (en) Magnetic resonance denoising method and device based on modeling reverse recovery technology
JP2012112716A (en) Air flow rate measuring device
CN101556173A (en) Coriolis mass flowmeter digital resolver based on wavelet de-noising and method thereof
US6463388B1 (en) Method and system for attenuation of acquisition geometry imprint in seismic data
Volosnikov et al. Dynamic measurement error evaluation and minimization based on FIR-filter
US20120089357A1 (en) Method and apparatus for identifying anomalies of a signal
Emad Optimal frequency-response corrections for eddy covariance flux measurements using the Wiener deconvolution method
Chen et al. A phase difference measurement method based on the extended kalman filter for Coriolis mass flowmeters
Mahadevan et al. Recovery of tail pipe species concentrations and its effect on emissions calculations from raw exhaust gas streams during chassis dynamometer tests
Pham et al. Evaluation of partial flow dilution systems for very low PM mass measurements
CN109374969B (en) Down-sampling sinusoidal signal phase difference measurement method based on coherent accumulation DFT
JP3782724B2 (en) Exhaust gas measurement system
Aquino et al. Comparison of Time-Domain and Time-Frequency-Domain System Identification Methods on Tall Building Data with Noise
Lassahn et al. Errors in cross-correlation peak location
US7233967B2 (en) Reflection filter
Meng et al. Method for direction diagnosis of multiple fluctuation sources on the flow standard facility
CN110485222B (en) Dynamic data inversion method and device
JP2770046B2 (en) Method and apparatus for estimating resolution bias error

Legal Events

Date Code Title Description
AS Assignment

Owner name: AVL TEST SYSTEMS, INC., MICHIGAN

Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:BERGHOF, FRANK;REEL/FRAME:031559/0480

Effective date: 20131030

AS Assignment

Owner name: RBS CITIZENS, N.A., AS AGENT, MICHIGAN

Free format text: SECURITY INTEREST;ASSIGNORS:AVL MICHIGAN HOLDING CORPORATION;AVL POWERTRAIN ENGINEERING, INC.;AVL TEST SYSTEMS, INC.;AND OTHERS;REEL/FRAME:033549/0585

Effective date: 20140620

STCB Information on status: application discontinuation

Free format text: ABANDONED -- AFTER EXAMINER'S ANSWER OR BOARD OF APPEALS DECISION