CN1465956A

CN1465956A - Error correcting method with sectional correction according to check point

Info

Publication number: CN1465956A
Application number: CNA02133398XA
Authority: CN
Inventors: 李伯祥
Original assignee: Individual
Current assignee: Individual
Priority date: 2002-07-03
Filing date: 2002-07-03
Publication date: 2004-01-07
Anticipated expiration: 2022-07-03
Also published as: CN1266455C

Abstract

The invention is an error-correcting method according to check point segment correction, in whole measurement range, according to the actual condition of error distribution, selecting the error curve bend to set check points. According to the actual check point, making segment, two adjacent points a segment, and calculating each point's correction value in each segment. The correction includes three steps: variable error filtering, correcting according to check value of production factory and correcting according to that of user.

Description

The error correcting method of segmentation correction according to check point

This method relates to the measuring error correction of microcomputer detection-control equipment.

Existing microcomputer detection-control equipment as system controlled by computer electronics universal testing machine, to the modification method of nonlinearity erron is: gamut by 0～5%, 5%～10%, be divided into 5 grades so successively, every grade can multiply by a correction factor, and a selected reference point is revised.Its weak point is every grade can only accurately be revised a bit, and reference point preferably.This has correction effect preferably concerning changing comparatively simple error.To changing comparatively complicated error, as: at same shelves A point is positive error, and D point is a negative error, then revised the then D error increase of ordering of A point, has revised the then A error increase of ordering of D point.

The invention provides a kind of error correcting method that also can revise the various errors that change complexity.

This method is in the gamut scope, and Select Error curve corner is as necessary effect adjusting point, and the centre can be inserted the effect adjusting point as required, and the check point of insertion is many more, and the precision of correction is high more.According to the check point segmentation of reality, every adjacent 2 is one section then, every section error by following formula correction each point.Y=YT+ Δ YT formula 1 Δ YTB=Δ Y, (N-1)+[Δ Y, (N)-Δ Y, (N-1)] [YT-Y, (N-1)]/[Y, (N)-Y, (N-1)] formula 2 Δ YTA=Δ Y, (N-1)+[Δ Y, (N)-Δ Y, (N-1)] [YT+ Δ YTB-Y, (N-1)]/[Y, (N)-Y, (N-1)] formula 3 Δ YT=Δ Y, (N-1)+[Δ Y, (N)-Δ Y, (N-1)] [YT+ Δ YTA-Y, (N-1)]/[Y, (N)-Y, (N-1)] formula 4

Y is revised measured value, and YT is the measured value before revising, and Δ YT is a modified value, and N is a serial number, and Y (N) is the effect value (standard value) that N is ordered, and Δ Y (N) is the modified value of this check point, and Δ YTB, Δ YTA are the excessive variable of formula.The meaning of formula 2 is that the difference of the modified value between 2 changes along with the variation of this section range.Because the YT in the formula 2 is a measured value, Y (N-1) is a standard value, there is error between standard value and the measured value, so also there is error in the modified value Δ YTB that calculates, YT+ Δ YTB in the formula 3 is used for preliminary correction to YT, also further accurately, the substitution formula 4 again, and the modified value of calculating can be met the demands for the modified value Δ YTA that draws.

Y in the formula 2 (N), Y (N-1) also can use the measured value substitution.

Below we illustrate the error correction process of this method with example.If there is a puller system as follows first grade effect value and modified value:

Serial number N	????1	????2	????3	????4
Serial number N	????1	????2	????3	????4	Effect value (ox)	????100	????180	????200	????300
Error	????3	????-2	????-3	????0.2	Effect value (ox)	????100	????180	????200	????300
Error	????3	????-2	????-3	????0.2	Modified value	????-3	????2	????3	???-0.2

To equal 3 be that effect is worth near the max value of error 100 Ns to error in the table,-3 is near the max value of error 200 Ns, is necessary effect adjusting point, if the error correction of maximum is not fallen, then correction effect can be subjected to very big influence, and the 180, the 300th, middle insertion point.The measured value that selects N=2, N-1=1 to calculate before revising between them at 2 is respectively 103,130,160,178 o'clock modified value.ΔYTB＝ΔY(N-1)+[ΔY(N)-ΔY(N-1)][YT-Y(N-1)]/[Y(N)-Y(N-1)]ΔYTB＝3+[2-(-3)](103-100)/(180-100)

＝-2.8ΔYTA＝ΔY(N-1)+[ΔY(N)-ΔY(N-1)][YT+ΔYTB-Y(N-1)]/[Y(N)-Y(N-1)]ΔYTA＝(-3)+[2-(-3)][103+(-2.8)-100]/(180-100)

＝-2.99ΔYT＝ΔY(N-1)+[ΔY(N)-ΔY(N-1)][YT+ΔYTA-Y(N-1)]/[Y(N)-Y(N-1)]ΔYT＝(-3)+[2-(-3)][103+(-2.99)-100]/(180-100)

=-3 modified values that in like manner can calculate at 130,160,178 are respectively-1.2,0.79,2.Substitution formula 1 obtains revised measured value and is respectively 100,128.8,160.79,180.Use the same method and to revise other error of 3 sections.

After choosing check point and obtaining effect value and modified value, the requirement of follow procedure is carried out data processing with effect value and modified value input computing machine by computing machine, realizes the correction to error.

Finish revise for the first time after, check error maximum point everywhere once more, finding to have needs the place revised, then inserts new effect adjusting point, number of times so repeatedly is many more, then revised precision is high more, inspection correction that at least will be once.Because check point is provided with incorrectly, cause the unfavorable situation of correction effect to prevent.

The error correction process of this method was divided into for three steps: variable error filtering → and according to factory's effect modified value correction → according to the correction of user's effect modified value.By variable error filtering eliminating undesired signal and random fluctuation error, employing be the method for moving average.Computing formula is: YT=[YX (1)+YX (2)+... + YX (N)]/N formula 5YT: be the revised measurement data of filtering, YX (1): be nearest (last) measurement data.N: the measurement number of times of moving average.Factory mainly is the error maximum point of selecting everywhere during effect, takes into account the actual needs point and carries out the effect correction, and 10～20 of check points can be set, and the various errors that can make measuring system like this are basically by the standard correction.

User's effect makeover process is mainly used in and revises measuring equipment environment for use difference, uses measured value drift, the different caused errors of effect standard for a long time.Mainly effect correction according to actual needs, check point can be set at 5～10.

Prior art is that manufacturer and user revise on the same software page, manufacturer separately uses the different software pages with the user in the new method, can use identical and different effect values and modified value separately, so both can satisfy the needs of producer's high precision correction, can satisfy the requirement of user again by the actual needs correction.

Fig. 1 is a computer program process flow diagram of realizing this error correcting method, and the effect process of factory is abbreviated as effect 1, and user's effect process abbreviates effect 2 as.With the effect value of factory and modified value by formula 1,2,3,4, the process revised calls effect 1 and revises, user's effect value and modified value by formula 1,2,3,4, the process revised calls effect 2 and revises.

Claims

The error correcting method of 1 segmentation correction according to check point is with effect data input computing machine, carries out data processing by computing machine, realizes the correction to error; It is characterized in that in the gamut scope Select Error curve corner is as necessary effect adjusting point, other effect adjusting point can be inserted in the centre; Check point segmentation according to reality, every adjacent 2 is one section, every section modified value of calculating each point with following formula: Δ YTB=Δ Y (N-1)+[Δ Y (N)-Δ Y (N-1)] [YT-Y (N-1)]/[Y (N)-Y (N-1)] formula 2 Δ YTA=Δ Y (N-1)+[Δ Y (N)-Δ Y (N-1)] [YT+ Δ YTB-Y (N-1)]/[Y (N)-Y (N-1)] formula 3 Δ YT=Δ Y (N-1)+[Δ Y (N)-Δ Y (N-1)] [YT+ Δ YTA-Y (N-1)]/[Y (N)-Y (N-1)] formula 4YT is the measured value before revising, N is a serial number, Y (N) is the effect value (standard value) that N is ordered, Δ Y (N) is the modified value of this check point, Δ YTB, Δ YTA is the excessive variable of formula, and Δ YT is a modified value; After before finishing, once revising, can insert new effect adjusting point again, can be so repeatedly repeatedly; The error correction process of this method was divided into for three steps: variable error filtering, according to factory's effect modified value correction, according to the correction of user's effect modified value; Manufacturer separately uses the different software pages during with user's effect, separately can be with identical with different effect value and modified values.
2 by the described error correcting method of claim 1, it is characterized in that Y (N), the Y (N-1) in the formula 2 uses the measured value substitution.