CN1723623A - Methods for Manipulating Numeric Data Values - Google Patents

Abstract

The invention relates to a method for processing digital data values, according to which a differential value is formed from a respective current digital data value and a respective predicted value with the aid of a predictor method, compressed values being generated from the differential values in a Rice process that succeeds the predictor method. The aim of the invention is to carry out said method relatively rapidly. To achieve this, the respective predicted value is obtained using a signal model, which describes the expected temporal progression of the digital data values.

Description

Methods for Manipulating Numeric Data Values

technical field

The invention relates to a method for processing digital data values, wherein difference values are established from respective current digital data values and from respective predicted values by means of a predictor method, and in a RICE method following the predictor method Compression values are generated from these differences.

Background technique

For professionals, such a method is disclosed, for example, at the Internet address which can be called at http://www.monkeysaudio.com/theory.html (called on November 13, 2002). There it is explained how digital audio data can be compressed for compression purposes by means of the predictor method and subsequently the RICE method. To calculate the prediction values required in the predictor method, the audio data value immediately preceding the current audio data value is used in each case.

Furthermore, in the regulation and control of industrial plants, measured values of operating processes, such as manufacturing and conversion processes, are usually acquired in digital form and made available as output information for subsequent processing. Subsequent processing can be, for example, a readjustment of process parameters. In this case, the acquired measured values often have to be transferred between different technical devices in the form of digital data values or digital measured values or stored for subsequent processing. In order to reduce the amount of data required for the transmission of the digital measurement data, the digital measurement data are compressed by means of digital data compression before transmission in conjunction with the filling level regulation disclosed in document WO 01/91081. In this way the bandwidth required for data transfer can be reduced.

Contents of the invention

The technical problem to be solved by the present invention is to provide a relatively rapid method for compressing digital data values.

According to the invention, the above-mentioned technical problem is solved by a method of the type mentioned at the outset in that the corresponding predicted values are obtained by means of a signal model which describes the expected temporal variation of the digital data values.

The main advantage achieved with the present invention over the prior art is that the storage space required for the compression values determined from the digital data values is significantly reduced with the aid of the proposed method at a relatively high speed. In this way, digital data values can thus be compressed without loss in real time, ie immediately after their acquisition. Determining the prediction value for the predictor method as quickly and reliably as possible is achieved by determining the prediction value during execution of the predictor method on the basis of a signal model suitable for describing the temporal variation of the digital data values. By using such a signal model, the predicted value can be used to establish the difference without corresponding calculations based on previous data values and thus without a time delay.

Space requirements in memory may be reduced relative to uncompressed data values. The reduction in memory space requirements is achieved in the proposed method with minimal computing power in the processing of digital data values, since only simple computing operations such as addition, subtraction and bit manipulation are required. This makes it possible for the presented method to be used also for processing digital data values in technical devices which provide computing power only on a limited scale by means of suitable processors. The need for compression computing power not available to other functions is minimized.

The compression according to the method according to the invention has the effect here that only data which do not contain information are eliminated. The digital measurement data can thus be completely generated again during decompression, ie without loss.

According to one embodiment of the method according to the invention, the signal is determined on the basis of a sine or cosine function with a constant period and its corresponding harmonics, a decaying e function or a sine function with a decaying e function as an envelope Model. In this way, the corresponding predicted value can advantageously be determined particularly quickly, since no further calculations have to be carried out.

Starting from a method of the type mentioned at the outset, another solution to the technical problem described above is that, according to the invention, digital data values are obtained from periodic signals and used as corresponding predictive values to which the respective current digital data values belong at least in The digital data value collected in the previous cycle of the current digital data value. The required computing power can advantageously be kept low by using data values acquired in each case one period before the current data value for the predicted value. No additional calculation operations are necessary, so that the method can be carried out quickly.

In order to improve the accuracy of the corresponding predicted value, in addition to the digital data value collected exactly one cycle before the corresponding current digital data value, other digital data values collected in integer multiple cycles before the corresponding current digital data value used to form the corresponding predicted values. These data values can be combined with one another by simple mathematical functions, such as averaging and/or weighting, so that the computing power required for this can be kept to a minimum.

In a preferred development of the method according to the invention, in the RICE method, an overflow determined by subtracting the actual data width of the difference from the data width of the current difference predicted by means of the RICE prediction value , is compared with a predeterminable boundary value, and if the overflow exceeds the boundary value, the difference is output as a compressed value according to the predetermined maximum data width. A significant advantage of this extension is that, for continuous difference values which fluctuate strongly with respect to their data width, the RICE method is only used to the extent that the RICE method can be used effectively, ie with little overflow. In the case of high overflow, the corresponding compressed value is output according to the maximum predetermined data width.

Furthermore, in a preferred development of the method according to the invention, the compressed value is transmitted via the data transmission section, and the difference value is subsequently obtained again from the compressed value using the RICE decoding method, and from the difference value and the corresponding prediction The value of the numeric data is determined by adding according to the reverse predictor method. In this case, the transmission of the compressed measured values via the data transmission section includes both wired and wireless transmission methods, for example radio transmission. According to this extension, compressed values can preferably be decompressed remotely from the compression location. For example, compressed values can be transmitted from field devices of an industrial facility via a data bus to a central computer which analyzes the digital data values after decompressing the compressed values.

In order to simplify the subsequent processing of the compressed value or to transmit additional information, a preferred embodiment of the invention can provide header data for the compressed value.

According to a further preferred embodiment of the method according to the invention, the digital data value is formed from an input measurement variable of the field device. A field device can be understood here, for example, to be a protection and/or control technology device which is usually used in industrial (eg energy technology, chemical or petroleum) installations. In contrast to audio technology, in which almost arbitrarily long periods of time are available for compressing digital data values, in field devices the compression has to be carried out particularly quickly due to the limited computing and storage capacity. The method according to the invention can therefore be used particularly advantageously here.

Furthermore, protection and/or control technology devices in energy technology installations can also preferably be used as field devices.

Description of drawings

The present invention will be further described below according to the embodiments. In the picture:

Figure 1 represents a schematic diagram for explaining the method of compressing digital data values,

Fig. 2 shows a schematic diagram for explaining the method of decompressing compressed digital data values.

Detailed ways

FIG. 1 shows a schematic diagram for explaining a method for compressing digital data values acquired with a field device not shown.

Since in the embodiment shown the digital data values are to be formed from (analog) input measurement variables of the field device, they will be referred to below as digital measurement data.

Measurement values converted into digital measurement data in the usual manner are acquired by means of the measuring device 1 . To process the digital measurement data, a predictor method is first implemented. The predictor method is part of a process for compressing digital measurement data. In this case, prediction values for the digital measurement data are determined by means of the predictor device 2 .

By means of the predictor method, the data width (number of bits) required to represent the digital measurement data is firstly reduced by subtracting the difference from the digital measurement data and the appropriate predicted value. In this case, the predicted value used should be as close as possible to the actual digital measurement data, so that the smallest possible difference results when establishing the difference. Further processing is then carried out with significantly smaller difference values than the digital measurement data in terms of the required data width.

In order to form the predicted values for the predictor method, a signal model can be used, for example with a sine or cosine function and its corresponding harmonics, a decaying e function or a sine function with a decaying e function as an envelope Base. These are in each case functions which can be represented and processed computationally with little effort and which are partly periodic. Previous digital measurement data acquired before the current digital measurement data to be compressed are considered when establishing the signal model for determining the predicted value. To form the signal model, it is expedient to use specific properties of the preceding digital measurement data, such as amplitude, period length and decay properties. These properties can be calculated with simple calculation operations, but the predicted values can nevertheless be determined relatively reliably.

In the simplest case, the preceding digital measurement data of exactly one preceding acquisition period can also be used unchanged as the predicted value. The measures described last, in which reference is made to a preceding acquisition period, can be used in particular if the acquired measured values and thus also the digital measurement data have a periodic nature, as is often the case in energy technology installations. When the numerical measurement data are processed for the first time using the predictor method, a starting value must be provided for the first predicted value.

According to FIG. 1 , the digital measurement data of the measuring device 1 and the predicted values determined accordingly by the predictor device 2 are transferred ( 10 ) and ( 20 ) to the subtraction device 3 , where the predicted values are converted from the respective associated digital measurement data subtracted from. As a result of the subtraction a difference is formed which is then sent to the means 4 for carrying out the RICE method.

Through this well-known RICE method, the practitioner can obtain its details from the literature (see for example http://www.monkeysaudio.com/theory.html, invoked on November 13, 2002), reducing the value as a predictor The data width of the difference formed by the result of the method, so as to finally generate and output (40) the compressed value.

In order to transmit and store this difference, the same data width can be used respectively. This data width must at least correspond to the maximum possible data width of the difference. However, since the required data width is usually not constant in the preceding and following differences, but may fluctuate from one difference to the next, a non-negligible memory is wasted in transmitting and storing the difference at the maximum data width space. This is further explained in the following example: Three difference values are transmitted in numerical representation, namely 11010110, 1101 and 10110. When transferring with a constant data width, the maximum data width of the occurring difference must be used, here 8 (data width of the first difference). Therefore, the three differences are transmitted in the form 110101100000110100010110. The addition of zeros in the maximum data width to convey the smaller difference (1101, 10110) unnecessarily wastes storage space because no additional information is conveyed.

An algorithmic method is provided by means of the RICE method, which encodes the difference values produced by the predictor method into compressed values in a suitable manner, ie in a way that optimizes storage space. The basic idea of the RICE method is to compress the difference according to the data width matching the difference respectively. In order to separate the compressed values obtained successively from the difference values and to encode information which might not be representable in too small a data width, so-called RICE codes are introduced. This will be explained below.

For the RICE method, the expected data width for each subsequent difference requires a RICE prediction. The compression values generated by means of the RICE method are then generally stored according to the data width predicted using the RICE prediction value. If the data width of the resulting difference according to the previous predictor method is greater than the data width predicted using the RICE predictor, overflow (highest value bit that can no longer be represented in the predicted data width) is encoded in the RICE code . A RICE code consists of a number of binary 0s and a final binary 1 directly resulting from this overflow. To obtain a compressed value, the RICE code is directly combined with one another, taking into account the predicted data width, from the corresponding difference values.

The RICE prediction value for the data width is generated from the data width values of a defined number of preceding digital measurement data, wherein the preceding digital measurement data are weighted differently depending on the time distance to the currently estimated RICE prediction value.

The RICE method is ineffective if the actual data width of the corresponding difference deviates too much from the data width predicted using the RICE prediction value. For this reason, the RICE method is only used until a certain difference is reached between the predicted and actual data width. If this limit value is exceeded in this method, the largest data width is used instead of the RICE predicted value. The exceeding of the limit value is marked by a special value of the RICE code which does not occur under normal conditions (without exceeding the limit value).

To further explain the compression method, reference is made below to three sequentially following difference values, which may be represented, for example, as a first binary value (11001110110), a second binary value (10110) and a third binary value (1101111) numerical order. The RICE predictor for the data width is given by the data width of each previous difference of the corresponding binary value. The boundary value or the maximum permissible difference between the actual data width of the corresponding binary value and the predicted RICE predicted value is 4, of which the maximum transmittable data width is 16. Under these assumed conditions, the relationship shown in Table 1 results if the first binary value is transmitted in its actual data width.

Table 1

Determine the compressed value for the second binary value (10110) value illustrate RICE Prediction of Data Width 11 The actual data width of the first binary value 11001110110 actual data width 5 The actual data width of the second binary value 10110 overflow RICE code binary 1 no overflow, terminated with 1 result value binary 00000010110 The second binary value of the data width predicted from the RICE predicted value compressed value Binary 100000010110 RICE code + value according to predicted data width

For the code for the third binary value (1101111) value illustrate RICE Prediction of Data Width 5 The actual data width of the second binary value 10110 actual data width 7 The actual data width of the third binary value 1101111 overflow Binary 11, decimal 3 2 bits, binary 11, decimal 3 RICE code binary 0001 number of zeros given by overflow in decimal notation, terminated by 1 result value binary 01111 third binary value without the two most significant bits compressed value binary 00101111 RICE code + result value

That is, the bit sequence as the second and third binary values yields 10000001011000101111. For comparison it may be mentioned that without using the RICE method with the largest data width occurring (11, the data width of the first binary value) 0000001011000001101111 is generated as a bit sequence for the second and third binary value, i.e. 22 bits instead of the 20 bits produced by using the RICE method.

Fig. 2 shows a schematic diagram for explaining decompression. Here, the method explained in conjunction with FIG. 2 works in reverse. The compressed value is sent to the RICE unit 20 (100). As a result of the RICE decoding method implemented by means of the RICE means 20 a difference value is obtained again which is combined (200) in the addition means 21 with the predicted value generated by means of the predictor means 22 for the predictor method and ( 300), so that finally the digital measurement data is output again (400). When decompressing (cf. Fig. 2 ) the start value of the predictor must either be fixedly identical to the predictor used for compression (cf. Fig. 1 ), or be transmitted along with the compressed value.

In order to optimize the processing of compressed values for practical use in industrial installations, in particular energy technology installations, header data can be provided for the compressed values. This header data may include, for example, the data width for the difference, the number of the difference, the type and parameters of the corresponding predictor for the predictor method and the type and parameters of the corresponding RICE predictor for the data width (RICE encoding method) Information. Furthermore, information about the starting values for the predicted values of the predictor method and/or the RICE predicted values can also be contained.

Compression by means of the method described in connection with FIG. 1 can be performed in real time directly after the acquisition of the digital measured values. The decompression of the compressed values can then preferably take place immediately prior to this application for other applications, wherein the (decompressed) digital measurement data can be used, for example, in applications for display or simulation.

Claims (8)

1. method that is used to handle digital data value, wherein, by means of the fallout predictor method from each current digital data value and from each predicted value, set up difference, and from these differences, produce compressed value in the RICE method after this fallout predictor method, it is characterized in that the signal model that changes by means of a kind of time of describing the expection of this digital data value and obtain corresponding predicted value.

2. method according to claim 1 is characterized in that, at the SIN function that has constant cycle and corresponding harmonic wave thereof or cosine function, decay e function or have on the decay basis of e function as the SIN function of envelope, determines described signal model.

3. method that is used to handle digital data value, wherein, by means of the fallout predictor method from each current digital data value and from each predicted value, set up difference, and from these differences, produce compressed value in the RICE method after this fallout predictor method, it is characterized in that, from periodic signal, obtain digital data value, and use the digital data value of gathering in the previous cycle of current digital data value at least as the corresponding predicted value under the corresponding current digital data value.

4. each described method in requiring according to aforesaid right, it is characterized in that, in described RICE method, with definite overflowing by the real data width that will deduct this difference by means of the data width of the current difference of RICE predicted value prediction, compare with the boundary value that can be scheduled to, and, if this overflows and has surpassed this boundary value then this difference is exported as compressed value according to predetermined maximum data width.

5. each described method in requiring according to aforesaid right, it is characterized in that, described compressed value is transmitted section by data to be transmitted, and utilize the RICE coding/decoding method from this compressed value, to obtain this difference once more subsequently, and from this difference and corresponding predicted value, pass through to determine digital data value according to reverse fallout predictor method addition.

6. each described method in requiring according to aforesaid right is characterized in that described compressed digital measurement data have a data.

7. each described method in requiring according to aforesaid right is characterized in that described digital data value is formed by the input measurement parameter of field apparatus.

8. method according to claim 7 is characterized in that, as protection and/or the control technology equipment in the described field apparatus use energy technologies facility.

Images