CRSP補(bǔ)償原理

1、 CRSP Supplement Doc. _CRSP Supplement IIUpdated Diagram and ExplanationsTable of ContentsSection 1: Concept of CRSP2Section 2: Target voltage calculation:2Section 3: Percent deviation output calculation:3Section 4: Dynamic effects on the sensor output voltage:4Section 5: Is CRSP working? Is Measure

2、ment Accurate?6Section 6: Where is the error being generated?6Section 7: Correcting for dynamic errors7Section 8: Monitoring The CRSP output verses sensor output for errors:8Section 9: CRSP %Dev. Monitoring:9Section 10: Updating Target Volts:9Section 11: Corrected %Thickness Dev Output:11Section 12:

3、 Coil End Events:11Section 13: CRSP Output Testing:12Section 1: Concept of CRSPThe first use of CRSP was used several years ago, with a DEC based system. At that time CRSP consisted of a special PCB and a data link to the DEC processor.With the advent of MXProline, CRSP functionality is the same but

4、 it is accomplished within the PC and MSS software. There is no special / modified PCBs required for CRSP operation. Within the MSS are two standard N.I. DAQ PCBs, each supplying an analog output voltage representing percent deviation. DAQ1 analog output is used as a CRSP monitoring signal and DAQ2

5、analog output is designated for the mill AGC controller.What is CRSP? CRSP is short for Continues Real time Signal Processor. CRSP is designed to provide a high-speed linear percent deviation output voltage. Where a control system requires an updated error output faster than the PC processing rate,

6、a CRSP signal provides the required update time. (At the time of this document the output update rate is dependent upon processor Q, resulting in a 5-6msec update rate.)The second function of CRSP is to convert the entered target thickness into a non-linear sensor output target voltage. (The sensor

7、response is non-linear verse thickness) This is accomplished within the PC software using the sensor calibration curve data and Sterlings method math function (f(x). Based upon the entered thickness target a conversion is ran, possibly several times, until the converted error is below an entered per

8、centage. Once obtained the target voltage is saved and transferred to the MSS. Within the MSS software the sensor voltage and target voltage are constantly compared, at a 5-6msec rate. A voltage difference represents a thickness error. Using a conversion factor the voltage difference is converted in

9、to a percent deviation. Scaling the percent deviation for a voltage output, the CRSP signal voltage is updated. A mill controller then adjusts for the thickness change.This %-deviation voltage, CRSP output, is a bipolar output voltage, +-10VDC, representing a +-10% deviation and is a linear voltage

10、signal.Section 2: Target voltage calculation:On the coil data entry display the desired target, setpoint, and alloy are entered into the next coil data column; either manually or automatically via an OPC communications link. When the load coil data function is performed the next column data is loade

11、d into the current column data.A sensor standardize will be activated and if the sensor is operating properly a good standardize will occur. Then the current target is used to process a math function for calculating the desired target voltage. Only after a good standardize will CRSP calculate the re

12、quired data. (This process is basically taking the calibration data and processing in reverse to determine the sensor non-linear voltage output at the given target.) This value is then stored into the Target Volts as the initial value, and also downloaded to the MSS as Target Volts. Within the CRSP

13、setup tables is an initial target value. CRSP uses this value for the first time a conversion is ran; a starting point. If this value and the entered target have a large difference, the conversion will fail. CRSP will run approx. 10 times at max. And each successive run the calculated target is chan

14、ged by a fixed percent; in the setup table. A good conversion will occur when the calculated error is less than a limit; in the setup table. If after 10 runs the calculated error is not less than the limit, the conversion is bad and no further calculations are performed. A bad conversion flag is set

15、. But, at the time of this document, there is no video alarm. The evidence of a problem is the mill AGC system is not controlling to target and the CRSP output voltage is not updating; remains at zero.Section 3: Percent deviation output calculation:A block diagram of the CRSP operation is shown in d

16、iagram #1. Using this diagram the deviation output voltage calculation is explained. A more in-detail explanation is shown later in this document.When the sensor is enabled on sheet, the sensor output voltage is read by DAQ #1 analog input, sent to the DAQ #1 PCB and then to the processor PCB. (Ligh

17、t blue line)The target voltage has already been calculated, explained above, and transferred to the MSS.The MSS processor then will calculate the difference between the sensor output voltage and the calculated target voltage. The result is defined as Voltage Error. Multiplying the voltage error time

18、s a Gain conversion factor, %Dev per volts, converts the voltage error into a %Thickness Dev. (At the time of this writing it is not known if the %Thickness Dev is in units of fractional percent or actual percent. The difference would be multiplying by 100.) The deviation value is the sent to DAQ #2

19、 PCB and to the analog output on DAQ #2 interface PCB.At this point the thickness deviation is now available for the AGC control system to read and control the mill. (Assuming all other permissive are correct.) The output is a bipolar signal of +-10VDC that represents a deviation of +-10%, respectfu

20、lly.The Gain value, %Dev per volts, is calculated at the same time initial target volts is calculated. Once calculated the gain value is sent to the MSS via the LAN. The gain value will not change during the running of a coil. It is a fixed value.Deviation output voltage will continue to function so

21、 long as the sensor is on sheet and no gauge read errors are present. If an error exists the deviation voltage output is reset to zero, indicating no thickness error. This will keep the AGC system from causing upsets in the sheet and the AGC system probably receives a signal from MXProline indicatin

22、g a measurement errorSection 4: Dynamic effects on the sensor output voltage:Basically the CRSP deviation output voltage is similar to an open loop function, no feedback to determine if an error exists. The only parameters that calculate the voltage error are the sensor output voltage, target voltag

23、e and gain. At this point the only value that is changing is the sensor raw output voltage; raw indicates UN-compensated or directly from the sensor.Unfortunately there are dynamic factors at work which will effect the sensor thickness readings; and output voltage. Mainly this is air gap temperature

24、 but there is also some effect from aluminum oxide, dirt, within the gap, a slight effect from oil mist and passangle compensation. The CRSP output is calculated within the MSS comparing the sensor output voltage to the target voltage. If there are any conditions that effect the sensor output voltag

25、e (interpreted as thickness changes), a thickness error is generated. (Due to the changing sensor voltage verses target voltage, the CRSP output voltage is changed. The controller will compensate for the different voltage and a thickness error is generated when there is actually no difference in she

26、et thickness) The main dynamic factor that will cause this false error is air gap temperature changes and passangle changes. Compensation is accomplished via adjusting the target voltage. (The x-ray sensor calculated thickness is compensated for temperature, using the air gap temp. sensors, and pass

27、angle. But the actual sensor output voltage, raw voltage, is not compensated and this is the voltage used by the MSS for the high-speed CRSP output voltage.)For an example assume that the sheet thickness remains at a constant value. (Not practical but required for this explanation.) As the air gap t

28、emperature increases, mill running, the mass in the gap is less; lighter air. This will cause an increase in sensor voltage output that is then calculated as a lighter sheet; when actually the sheet has not changed thickness. Also a changing passangle produces a heavier sheet, to correct thickness t

29、o a heavier sheet.Using diagram #2 shows temperature in red and raw sensor voltage in dark blue. As the temperature increases the sensor raw output voltage increases. This is an indication of a lighter sheet, shown in the light blue plot. If the AGC were on the light blue thickness plot would be cen

30、tered on the target line, as shown by the dark green plot.Based upon the thickness deviation output calculation, see diagram #1, as the sensor voltage changes the voltage error changes and the %Thickness Dev changes. The AGC system detects the change and adjusts the actuator to correct for the error

31、. But there is not any real sheet thickness error, just a change in temperature. This occurrence will create a coil that will be off gauge; possibly 3 8 percent on a foil mill.Passangle variations are shown below. Section 5: Is CRSP working? Is Measurement Accurate? With AGC enabled there appears to

32、 be no problem. A chart recorder or trend plot indicates a sheet thickness that is on target and the mill is running very well. But as the coil runs longer the air gap temp. is getting hotter and hotter. This will cause more error and the %Thickness Dev output will keep changing. This in turn causes

33、 the AGC system to make more corrections. Even with the AGC system disabled, the thickness will follow the light blue plot and go light. But at this point the operator will make a manual correction and the thickness is again back on target, so we think.Now maybe the mill is starting to run different

34、. The sheet edges are changing, mill is slowing and payoff tension is lower. All this maybe due to less reduction, sheet is getting heavier. The wise operator has discovered the payoff tension has dropped to a very low value, pressure is less or the mill has slowed down. AGC appears to be operating

35、correctly, so what has happened?Once the coil is finished some samples are taken. It is now discovered that the sheet, coil, is heavy by 5-6 percent! But the thickness displayed on the trend plots indicates exactly on target. More samples indicate the sheet is heavy, but at reduced amounts of error.

36、 At the beginning of the coil, for example the first 5 minutes, there is no error.On the next coil the same thickness change happens. All sensor correction factors are enabled and providing a good, reasonable, value. So now were starting to think something is wrong with the x-ray sensor, AGC system

37、or the deviation signal to the AGC system. The AGC system is performing correctly, so the problem appears to be the generation of the CRSP voltage or the x-ray sensor. All sensor data and corrections factors are valid and applied.Section 6: Where is the error being generated?After many hours of chec

38、king samples, trend plots and misc. data it is determined there is no problem with the x-ray sensor. Also the CRSP deviation output voltage is correct. Now what to do?Let us think for a few minutes. When the air gap heats up the changing temperature is detected. The system software applies a correct

39、ion value to the calculated thickness and the displayed thickness looks good. (A test of temperature correction verifies the measurement is correct) The coil seems to be running good. Displayed as the dark green plot on diagram #2.When sheet samples are taken, again there is a 5-6% error! And we kno

40、w the correction factors are enabled and applying a 5-6% correction. So what is happening?Well the temperature correction appears to be working correctly, since the amount of temp. Correction and the actual sample errors are about the same. Lets see what has happened.1. Sensor is functioning correct

41、ly, no problems.2. The air gap temp. is enabled.3. Coil samples are 5-6% heavy.4. All plotted data indicates a good coil.5. Since the coil samples are heavy and the AGC was enabled, this indicates that the sensor is measuring to the light side; higher output voltage.6. A plot of the %Thickness Dev o

42、utput voltage shows a normal voltage but the voltage is constantly changing.7. Also the control actuator position indicates a constant change in position.Item #6 and #7 are of most interest. There is a constant changing occurring with the control actuator and the deviation voltage; Not a good, norma

43、l, indication. Section 7: Correcting for dynamic errorsIt appears that temperature is our problem, since the temp. correction value is equal to the actual sheet thickness error. The indication being both error and correction are approx. equal. This may indicate no correction is applied.It is known t

44、hat the MXProline displayed thickness is a good value. The raw calculated thickness and all the modifiers, correction factors, are displayed and the final corrected thickness value is good; calculated thickness correlates with the samples. So, the problem must be somewhere else.Let us again look at

45、the MSS calculation. A simple calculation but what is happening with the sensor voltage and target voltage? We now the target voltage is not changing, the gain is not changing so that leaves the sensor voltage. Remember the sensor voltage is the RAW value, un-compensated. As the air gap temperature

46、changed so did the sensor voltage. This produces a changing %Thickness Dev voltage and the AGC system response to the change. But actually there is no sheet thickness change. Shown in diagram #3 are voltage values representing a changing air gap temperature.The nominal target voltage for this produc

47、t is 5.5V. As the temperature increased the sensor output voltage increased, to 7V vice the nominal of 5.5V. The calculated thickness, PC, has the correction factors applied and the calculated percent error is only -1.6%. Now the CRSP output voltage is indicating a measurement percent error of 7.5%.

48、 There is something incorrect because both percent errors should be the same.The error is within the MSS generation of the CRSP voltage. This is displayed above in diagram #3. The nominal target voltage is 5.5V but the sensor output voltage has changed to 7.0V, due to temperature changes. The result

49、ing difference is calculated into a 7.5% deviation signal. AGC Enabled;Now lets turn on the AGC system and observe the control action. A 7.5V error signal is detected and the AGC system adjusts the desired controller. This action will continue until the error voltage, CRSP signal, is at zero volts;

50、indicating no error. The CRSP output voltage is now at zero, no error exist, the controller is happy but the operator is concerned. Why, because now the trend plots are showing a sheet thickness of 7.5% heavy. But the error output voltage shows no error. This is a reversal of what previously happene

51、d. Diagram 4 demonstrates this action.At this point it appears that with AGC on or off there will be a 7.5% thickness error. And this is a true statement. The question is how to compensate for this error? Can a value be modified to compensate for this air gap temperature change? If so, which value?

52、Section 8: Monitoring The CRSP output verses sensor output for errors:From previous test we know that the sensor is operating correctly. Measurement is correct and all correctors are applied correctly. This is demonstrated in diagram #3; as the temperature changed, the sensor output changed, tempera

53、ture change detected, temperature correction factor calculated and applied and the result is a measurement that is 1.6% from target. Looking at diagram #3, the error offset is generated within the MSS. With a target voltage of 5.5V and a changing sensor voltage to 7.0V, the difference of 2.5V is cau

54、sing a 7.5% error. Based upon the equation for % thickness Dev., (sensor volts target volts) * gain = %Dev. (shown in diagram 3 or 4), the only variable is the sensor voltage. The gain and target voltages are constant values. To eliminate the error offset the gain or target voltage requires changing

55、. Gain is a conversion factor and if changed could cause other control issues. This leaves the target voltage. By changing the target voltage, to compensate for temperature changes, the resulting error offset is eliminated. This is simulating what is happening with the sensor output voltage. As the

56、temperature changed the sensor output changed and a target voltage change of the same magnitude should be made.Back on page six it is stated that the sensor calculated %Dev. and the CRSP output voltage, which represents %Dev., should be the same value. If these values are not the same then an offset

57、 error will exist. Based upon this information it should be possible to monitor the CRSP output voltage and compare this %Dev, to the sensor calculated %Dev. The difference between these two values can be used to update the target voltage and as the target changes the offset error is eventually reduced to zero. This difference value is termed “CRSP ERROR” and is displayed on the control display; and other displays.Section 9: CRSP %Dev. Monitoring:A method of updating the target voltage is required. As previously stated if sensor %Dev. and CRSP output %Dev. are compared, this d