Data acquisition and processing
The data acquisition and the control system are carry on real-time gathering, the test, processing to in the production process or scientific experiment each kind of physical quantity, and may constitute the corresponding quantity output the feedback control system. Data acquisition and control system many and varied, but its basic work process is similar, collects is observed and controlled the object each kind of simulation quantity, transforms through the AD switch into the digital signal, again through components and so on computer, digital signal processing chip after the signal which gathers carries on processing, again transforms the corresponding simulation quantity through the DA switch, realizes control which needs.
The microcomputer controls the data acquisition is advantageous for the massive sensors to carry on the scanning. The scanning speed decided by the signal dynamic characteristic that, namely certain channels by all means must by the really rapid scanning, avoid the aliasing error; But when turns over low velocity scanning other channels to be possible to compare the low velocity scanning, also cannot have the. In certain data record application, the high speed channel sampling speed request reaches as high as each second 100 times, but the slow channel sampling may every 5 minutes one time, the conventional unprogrammable hard segment data recording instrument according to smooth gathering all channels, moreover all channel sampling frequency must be same. information loss nearly. This method causes the mass data the stack, in which some data is nonessential, simultaneously also reduced the whole effective sampling frequency. Based on microcomputer data recording instrument available compared to other low speed channel frequency scanning some high speed channels.
Analog data before recording must transform into the digital form, this needs to use suitable modulus switch (ADC). ADC transforms the result obtains a digital bit class. The digital signal processing is carries on processing to these symbol set. The ADC switching process by the anti-aliasing filter, the sampling, the quantification and digitized these four is composed.
In carries on before the data processing or the comparison, must carry on the survey, the conversion and the linearization to each data recording instrument channel input voltage.
In the computer control system, the data acquisition is the most basic one kind of pattern, is generally through the sensor, the transmitting instrument transforms the production process each physical parameter the electrical signal, then after the A/D channel or the DI channel, sends in the digital quantity in the computer. The computer in carries on before the demonstration and the control to these digital quantity, but also must according to need to carry on the corresponding data processing.
The data processing cannot leave the value computation, but the most basic value computation is mathematical operations. Because in the control system meets the scene environment is different, the gathering data type and the value scope are different, the precision request is dissimilar, each data input method and the expression method are also various. Therefore, in order to satisfy the different system the need, designed many effective data processing technology method, like the pretreatment, the digital filter, the scale transformation, the table look-up and the overstep reported to the police and so on.
Ⅰ.Survey data pretreatment technology
To surveys the data the pretreatment is the computer control system data processing foundation, this includes the digital zero alignment technology, the system calibration technology as well as the input, the output data polarity and the word length pretreatment technology.
1.1 System errors calibrate
Automatically in the control system survey input channel, has component the generally and so on amplifier zero displacements and the drifting, can create enlarges the electric circuit the gain error and the component parameter unstable and so on the phenomena, they can affect the survey data the accuracy, these errors all belong to the system error. Its characteristic is under certain survey condition, its change rule is may grasp, has the erroneous reason also is generally knows. Therefore, the system error is may through the suitable technical method determine and adjusts, uses the software procedure to carry on processing generally, then carries on the automatic calibration to these system errors.
1.2 Data polar pretreatment
Control system processes the signal very are many is bipolarity, like temperature
pressure, position, angle signal and so on. This request when implementation control, not only must consider the signal the scope, but also must consider the signal the polarity. Therefore, in transforms after to A/D the data and D/A transforms before the data to carry on before processing, must carry on the pretreatment first according to the data polarity, can guarantee obtains the correct result. In the system has the input signal is unipolarity, but the output signal request is bipolarity, like current capacity, control loops and so on pressure; Some is requests the input and the output signal all is bipolarity, like position, control loops and so on angle. Below performs on these two kind of situations to discuss separately.
1.3 Data word length pretreatment
Can have the data word length inconsistent situation frequently in the computer control system. If some systems use 12 A/D switch sampling data, but outputs uses 8 D/A switches; Some systems use 8 A/D switches to carry on the sampling, but in order to enhance the computation the precision, uses the pair of byte running program computation. In order to satisfy the different precision request, the data in carries on the digital filter, after the scale transformation and the control operation must process to the digital quantity figure.
Ⅱ. Digital filter method
Because the industrial production scene environment very severe, each kind of noise source are very many, the computer system through inputs the data signal which the channel gathers, although after hardware electric circuit filter processing, but still could mix has the random disturbance noise. Therefore, in order to enhance the system performance, achieved the accurate survey and the control, in the ordinary circumstances also needs to carry on the digital filter.
The digital filter, are the computer system to the input signal sampling many times, then carries on the digital processing with some computational method, weakens the random error which or the filtration disturbance noise creates, thus obtains a real signal the process. This filter method only is the basis predetermined filter algorithm establishment corresponding procedure, in the essence is one kind of procedure filter. Thus the reliability is high, the stability is good, revises the filter parameter to be also easy, moreover one kind of filter subroutine may use in common by many channels, thus the cost is very low. Moreover, the digital filter may to each kind of unwanted signal, even the extremely low frequency rate signal carries on the filter. Its deficiency is needs to take CPU the machine-hour. In brief, the digital filter and the hardware filter compare the merit very many, therefore obtained the universal application. The commonly used digital filter method includes: The mean value filter, the value filter, restrict the filter and the inertia filter and so on. Here main mean value filter.
The mean value filter are carry on the average algorithm to many sampling value, this is the elimination random error most commonly used method. Concrete may divide into following several kinds.
1. Arithmetic mean filter
The arithmetic mean filter are in sampling period T, to surveys signal y to carry on m time the sampling, after m sampling value adding together arithmetic mean value took this time effective sampling value, the sampling number of times m value has decided the signal smoothness and the sensitivity. Enhances m the value, may enhance smoothness, but the system sensitivity reduces along with it, sampling number of times m value along with accusation object difference and difference. In the ordinary circumstances, the current capacity signal may take about 10, the pressure signal may take about 4, the temperature, the ingredient and so on slow changes the signal to be possible to take 2 even not to carry on the arithmetic mean.
When establishment algorithm procedure, m takes 2,4,8 generally and so on 2 integer powers, is advantageous for with shifting replaces the division to obtain the mean value. This algorithm is suitable for to the periodic disturbance signal filter
2. goes to the extreme value average filter
The arithmetic mean filter cannot the obvious accidental pulse interference elimination, only be its average to the sampling result in, thus reduced the measuring accuracy. Goes to the extreme value average filter is carries on the comparison to the continual sampling m data, removes maximum value and the minimum value, then computation in addition m-2 data arithmetic mean value.
When establishment algorithm procedure, is advantageous for with shifting replaces the division to obtain the mean value, m-2 should take 2,4,8 and so on, therefore m takes 4,6,10 and so on, this algorithm is suitable the peaked pulse disturbance signal filter which meets frequently in the industry situation.
3. Weighted average filter
The arithmetic mean filter and go to the extreme value average filter all to have smooth and the sensitivity contradiction. The sampling number of times too are few then the smooth effect is bad, the number of times too are many then the sensitivity drops, to surveys the parameter the change tendency not to be insensitive. For coordinates both relations, may use the weighted average filter, this algorithm can coordinate the system smoothness and the sensitivity contradiction, enhances the sensitivity, is suitable for the pure time delay big object.
4. Running mean filter
The first three kind of average filter algorithm has a common ground: Namely obtains an effective sampling value to have every time to carry on certain sampling continuously. When the system sampling speed is slow or the sampling signal change is quick, system timeliness is unable to obtain the guarantee. The running mean filter are in each sampling period sampling one time, strove for together this sampling value and the past certain sampling value the average, obtained result namely for effective sampling value. The running mean filter algorithm biggest superiority was timeliness is good, enhanced the system speed of response.
Above discussed several digital filter method, has its characteristic respectively. In the practical application, actually uses, as well as which one kind of digital filter uses, all should regard the special details to decide. Possibly has the system does not need to carry on the digital filter or to apply not appropriately, not only could not achieve the filter effect also can reduce the control quality, but some systems have used the compound filter method--Namely unifies several filter methods the use, possibly can obtain the better filter effect.
Ⅲ. Scale transformation algorithm
In production each kind of parameter all has the different dimension and the value, but in computer control system gathering, in the A/D switching process has become the zero dimension the data, when the system in carries on the demonstration, the record, prints when and reports to the police and so on the operations, must the data which obtains these return to original state into the corresponding dimension physical quantity, this needs to carry on the scale transformation.
3.1 Linear type transformations
The linear scale transformation is the most commonly used scale transformation way, its prerequisite is the sensor output signal with was measured between the parameter assumes the linear relations
3.2 Non-linear type transformations
If the sensor output signal with were measured when between the parameter assumes the non-linear relations, above linearity transforms is not suitable, needs to establish the new scale transformation formula. Because the non-linear parameter change rule is various, therefore should act according to the different situation establishment different non-linearity to transform, but the premise is their functional relation available analysis formula expressed.
3.3 Table look-up law
The so-called table look-up law is the data which calculates beforehand or obtains according to certain order establishes the form, the table look-up procedure duty is according to was measured the parameter the value or the intermediate result, find out the result which finally needs.
|
楼主
标题置顶
标题高亮
