RU2015557C1 - Multimeasure functional converter - Google Patents

Abstract

FIELD: functional converters. SUBSTANCE: range of changes in input signal SV^N in any layer of multifunctional converter for converter of increments is by an order less at least, than range of change in signals V^N. Circuit of the device has several converters of increments. Any converter has scaling units 7, 13, 19N, delay unit 10, 16N, comparison unit 5, 11, 17N. Input of any unit is connected to the second input of corresponding comparison unit. Precision of simultaneous processing of several signals may be improved. EFFECT: improved precision of signal converter. 1 dwg

Description

The invention relates to functional converters and can be used in accounting systems, planning and operational management in solving problems of economical calculation and storage of moving average values for measurement information signals. The set of measurement information signals V ^I , V ^II , ..., V ^{N is} pre-ordered, for example, in ascending order of their average level, and their ranges of change are close to each other.

 A one-dimensional converter is known that is implemented as a series connection of a delay unit, a first integrator and a comparison unit, a second integrator, the input of which is connected to the input of the delay unit and to the input of the converter, the output of the second integrator connected to the second positive input of the comparison unit, the output of which is connected to the output transducer. Simultaneous processing of several (N) signals is possible if N converters are used [1].

 The disadvantage of this converter is the low accuracy of the conversion.

 The closest in technical essence is a one-dimensional converter [2], implemented in the form of a serial connection of the delay unit, the comparison unit and the integrator, the input of the delay unit being connected to the input of the converter and to the second positive input of the comparison unit, and the output of the integrator connected to the output of the converter. Simultaneous processing of N signals can be carried out using N converters.

 The disadvantage of this Converter is the low accuracy of the conversion. Improving the accuracy of the converter is associated with a complication of its structure, especially with regard to the delay unit. The greatest distribution in the implementation of the delay block was obtained by the schemes corresponding to the Padé or Taylor approximation of the first, second, and higher orders. Moreover, an increase in the degree of the corresponding polynomial leads to a significant complication of the scheme. So, for example, a delay link reproduction scheme corresponding to the fourth-order Padé approximation is realized using six or seven operational amplifiers (see Tetelbaum I.M., Schneider Yu.R. Practice of analog modeling of dynamic systems. Reference manual. M.: Energoatomizdat . 1987, p.113, fig. 1.8.5, 1.8.6).

 The purpose of the invention is to increase the accuracy of the measuring signal signal converter.

 The goal is achieved in that in a multidimensional functional converter containing N delay units, N comparison units, N integrators, the first and second inputs of the i-th comparison unit being connected respectively to the input and output of the i-th delay unit, the output of the i-th comparison unit connected to the input of the i-th integrator, N adders, N additional comparison blocks, the first and second sources of reference voltages, N scaling units are introduced, and the output of the i-th integrator through the i-th scaling unit is connected to the first input of the i-th adder, outputconnected to the i-th output of the converter, the first input of the i-th additional comparison unit is connected to the i-th input of the converter, and the output to the input of the i-th delay unit, the second input of the first additional comparison unit is connected to the output of the first reference voltage source, the second input of the i-th additional comparison block, in addition to the first, is connected to the first input of the (i-1) -th additional comparison block, the second input of the first adder is connected to the output of the second voltage reference source, and the second input of the i-th adder, except the first connected to the output of the (i-1) -th adder.

The introduction of a set of new blocks and relationships allows using the same transducers as in the prototype to significantly increase the accuracy of the simultaneous processing of several measurement information signals due to the fact that on each layer of a multidimensional functional transducer the range of changes of the input signal δ V ^N for least an order of magnitude smaller than the range of variation of the signal V ^N.

- приращения входных сиг-
налов на первом, втором и
N-м слоях многомерного
функционального преобра-
зователя;
δ Y^I, δ Y^II, ..., δ Y^N - выходные сигналы масштабирующих блоков соответственно первого, второго и N-го слоев преобразователя;
Y^I, Y^II, ..., Y^N - выходные сигналы многомерного функционального преобразователя.The drawing shows a diagram of a multidimensional functional Converter, where the following notation:
V ^o is the reference input signal of the multidimensional functional converter, supplied to the first negative input of the first additional comparison unit;
Y is the reference output signal of the multidimensional functional converter, supplied to the first input of the first adder;
V ^I , V ^II , ..., V ^N - input signals of a multidimensional functional converter;

- increments of input signals
the catch on the first, second and
Nth layers multidimensional
functional transformation
the caller;
δ Y ^I , δ Y ^II , ..., δ Y ^N are the output signals of the scaling blocks of the first, second, and Nth layers of the transducer, respectively;
Y ^I , Y ^II , ..., Y ^N are the output signals of a multidimensional functional converter.

 The multidimensional functional converter comprises a second constant signal source 1, a first constant signal source 2, a first additional comparison unit 3, a first delay unit 4, a first comparison unit 5, a first integrator 6, a first scaling unit 7, a first adder 8, a second additional comparison unit 9 , the second delay unit 10, the second comparison unit 11, the second integrator 12, the second scaling unit 13, the second adder 14, the Nth additional comparison unit 15N, the Nth delay unit 16N, the Nth comparison unit 17N, the Nth integrator 18N, Nth scale ablation unit 19N, Nth adder 20N.

The input signals of the measuring information V ^I , V ^II , ..., V ^N are respectively supplied to the first, second and Nth layers of the multidimensional functional transducer. These signals are pre-ordered, for example, in the order of increasing their average levels.
V _cf. ^I <V _cf. ^II <V _cf. ^N.

) , который поступает на вход первого интегратора 6. Выходной сигнал первого интегратора 6, равный

1-l

V^I , где с - скорость интегрирования, поступает на вход первого масштабирующего блока 7, где он умножаетcя на постоянную величину, равную

, формируя тем самым на своем выходе сигнал

1-l

V^I ,о скользящем среднем в диапазоне изменения приращений δ V^I. Этот сигнал поступает на второй вход первого сумматора 8, где суммируется с опорным сигналом Y^o, формируемого с помощью первого источника 2 постоянного сигнала, и вырабатывает на выходе сумматора 8 и соответственно на выходе первого слоя многомерного функционального преобразователя сигнал
Y^I= Y^o +

1-l

V^I, что соответствует скользящему усреднению сигнала V^I при условии, что первый источник 2 постоянного сигнала формирует на своем выходе сигнал Y^o, связанный с сигналом V^o соотношением
Y^o = CV^o.The signal V ^{I is} supplied to the second positive input of the first additional block 3 of comparison, where, comparing with the signal from the first negative input with the reference signal δ V ^o generated by the second constant signal source 1, it generates a difference signal δ V ^I = V ^I - V _o ^o. The signal δ V ^I is supplied to the second positive input of the first comparison unit 5 and through the first delay unit 4, after a time equal to the cutoff interval τ _{from, it} is supplied to the first negative input of the first comparison unit 5. As a result, the signal δV ^-1 (1-e

), which is fed to the input of the first integrator 6. The output signal of the first integrator 6, equal

1-l

V ^I , where c is the integration speed, enters the input of the first scaling unit 7, where it is multiplied by a constant value equal to

thereby forming a signal at its output

1-l

V ^I , about the moving average in the range of variation of increments δ V ^I. This signal is supplied to the second input of the first adder 8, where it is summed with the reference signal Y ^o generated by the first constant signal source 2, and generates a signal at the output of the adder 8 and, accordingly, at the output of the first layer of the multidimensional functional converter
Y ^I = Y ^o +

1-l

V ^I , which corresponds to the rolling averaging of the signal V ^I , provided that the first constant signal source 2 generates at its output a signal Y ^o associated with the signal V ^{o by the} ratio
Y ^o = CV ^o .

1-l

V^II.Simultaneously with the signal V ^{I, the} signal V ^{II is} supplied to the second input of the multidimensional functional converter and, accordingly, to the second positive input of the second additional comparison unit 9. The signal V ^I is subtracted from this signal, which is fed from the first input of the multidimensional functional converter to the first negative input of the comparison unit 9, forming the difference signal δ V ^II = V ^II - V ^I at its output. This signal is input to a chain consisting of a second delay unit 10, a second comparison unit 11, a second integrator 12, a second scaling unit 13, where it is converted in the same way as on the first layer of a multidimensional functional converter, and thereby forms an output the second scaling unit 13 signal
δY ^II =

1-l

V ^II .

1-l

V^II, что соответствует скользящему усреднению сигнала V^II.This signal is fed to the second input of the second adder 14, where it is added to the signal Y ^I , which, in turn, is fed to the first input of the second adder 14 from the first output of the multidimensional functional converter. Thus, at the output of the second adder 14 and, accordingly, at the output of the second layer of the multidimensional functional converter, a signal equal to
Y ^II = Y ^I +

1-l

V ^II , which corresponds to the moving averaging of the signal V ^II .

1-l

V^N, что соответствует скользящему усреднению на интервале τ_от сигнала V^N.Similarly, the V signal is processed on the following, including the last Nth, layers of the multidimensional functional converter. Here, at the output of the N-th additional additional block 15 N comparison, a signal of the difference δ V ^N = V ^N - V ^{N-1 is formed} by subtracting from the signal V ^N coming through the N-th input of the multidimensional functional converter to the second positive input of the N-th additional block 15 N comparison, the signal V ^N-1 supplied to the first negative input of the N-th additional block 15 N comparison with the (N-1) -th input of the multidimensional functional Converter. Ultimately, a signal is generated at the output of the Nth layer of the multidimensional functional converter
Y ^N = Y ^N-1 +

1-l

V ^N , which corresponds to a moving averaging over the interval τ _{from the} signal V ^N.

 The described multidimensional functional converter is built on the basis of analog elements, however, all of the above is true for the case of a multidimensional functional converter constructed using digital elements of limited capacity. In addition, such a converter is operable not only for the case when the signal is converted on each layer using a moving average, but also for any other transformations.

The presence in the device of two sources of a constant signal, several comparison units and several adders allows you to organize the processing of input signals V ^N on each layer in increments relative to the input signals V ^{N-1 of the} previous layer, which ultimately significantly increases the accuracy of their conversion.

 To prove this, the operating conditions of the prototype converter and the proposed multidimensional functional converter are specified.

The input signals V ^I , V ^II , ..., V ^{N of} both converters are ordered in increasing order of their average levels. The signal V ^{o is} characterized by a basic signal level V ^I , the signal Y ^o corresponds to it.

) существенно (по крайней мере на порядок) меньше диапазона изменения сигналов V^N и Y^N((N = 1,

)) (1).Signal range δ V ^N and δ Y ^N (N =

) significantly (at least an order of magnitude) less than the range of variation of the signals V ^N and Y ^N ((N = 1,

)) (1).

The operator of the current average, implemented both by the prototype converter and the proposed multidimensional functional converter, is denoted by (B (S) and it is believed that it is implemented with an error in such a way that its amplitude-phase characteristic module can vary in the range
B (ω) ± Δ B (ω).

The error ε _Y when converting the signals δ V using the moving average operator is completely determined by the value Δ B (ω).

The transformation B (S) is much more complicated than the operations of summation and subtraction. Therefore accept the condition
ε _C ≈ ε _V << ε _Y , (2) where ε _C , ε _V are the signal conversion errors in the sum and subtract blocks, respectively.

For these conditions, the accuracy of the implementation of the signals Y ^I , Y ^II , ..., Y ^{N is} evaluated using the prototype converter and the proposed multidimensional functional converter.

Vⁱ+{B(S)+ΔB(S)}·

+

ε $\genfrac{}{}{0ex}{}{\text{i}}{\text{c}}$ . (5)
Выделяют из выражения (4) составляющую
ε_y= B(S)·

Vⁱ, которая согласно условию (2) обозначена через ε _Y.For the prototype converter, the output signal for any autonomous “layer” is
Y ^N = V ^N {B (S) + Δ B (S)} = V ^N ˙ B (S) +
+ V ^N ˙Δ B (S) = V ^N ˙ B (S) + ε ^N , (3) where ε ^N = V ^N + Δ B (S). (4)
For the proposed multidimensional functional converter, the output signal is
for the first layer
Y ^I = Y ^o + δ V ^I ˙ B (S) + ε ^l , where ε ^I = δ V ^I + Δ B (S) + ε _V ^I (B (S) + Δ B (S)) + ε _C ^I
for the second layer
Y ^II = Y ^o + {V ^II - V ^o } ˙ B (S) + ε ^II , where ε ^II = (δ V ^I + δ V ^II ) Δ B (S) + (ε _V ^I + ε _V ^II ) xx {(B (S) + Δ B (S)} + (ε _C ^I + ε _C ^II ),
For the N-th layer
Y ^N = Y ^o + {V ^N - V ^o } ˙ B (S) + ε ^N , where ε ^N = ΔB (S)

V ⁱ + {B (S) + ΔB (S)} ·

+

ε $\genfrac{}{}{0ex}{}{\text{i}}{\text{c}}$ . (5)
The component is isolated from expression (4)
ε _y = B (S)

V ⁱ , which according to condition (2) is denoted by ε _Y.

Bearing in mind condition (3), it turns out that for the proposed multidimensional functional converter, the error ε ^N for the last Nth layer will be greatest with respect to other layers and the component ε _Y be determined, i.e.

Vⁱ. (6)
Сравнивая выражения (4) и (6) отмечают, что, если ошибка ε ^N для преобразователя-прототипа зависит от абсолютного значения входного сигнала V^N, то для предлагаемого многомерного функционального преобразователя - от суммы приращений всех входных сигналов, т.е.

Vⁱ . Учитывая условия (1), отмечают, что для первых нескольких слоев предлагаемого преобразователя ошибка реализации заданного преобразования существенно меньше, чем для преобразователя-прототипа. Условие практической применимости предлагаемого преобразователя с точки зрения точности его реализации по сравнению с прототипом выглядит следующим образом

Vⁱ< V^N, (7)
В предположении, что δ Vⁱ одинаковы для каждого слоя, выражение (7) записывается
N δ V < V^N.ε ^N = ΔB (S)

V ⁱ . (6)
Comparing expressions (4) and (6), it is noted that if the error ε ^N for the prototype converter depends on the absolute value of the input signal V ^N , then for the proposed multidimensional functional converter it depends on the sum of the increments of all input signals, i.e.

V ⁱ . Given the conditions (1), it is noted that for the first few layers of the proposed Converter, the error in the implementation of the specified conversion is significantly less than for the prototype converter. The condition of practical applicability of the proposed Converter in terms of accuracy of its implementation in comparison with the prototype is as follows

V ⁱ <V ^N , (7)
Assuming that δ V ^{i are the} same for each layer, expression (7) is written
N δ V <V ^N.

.Then the number of layers N, the output signals of which are more preferable in accuracy with respect to the corresponding output signals of the prototype, are determined in accordance with the expression
N <

.

So, for example, if we assume that N = 3 and write in accordance with condition (1) that δ V ^N = 0.1V ^N , then the maximum error ε ^III occurring at the third output of the multidimensional functional converter is at least 3 times less than the error on any "layer" of the prototype converter.

Claims (1)

 A MULTIDIMENSIONAL FUNCTIONAL CONVERTER containing N delay units, N comparison units, N integrators, wherein the first and second inputs of the i-th comparison unit are connected respectively to the input and output of the i-th delay unit, the output of the i-th comparison unit is connected to the input of the i-th integrator, characterized in that, in order to increase the conversion accuracy, N adders, N additional comparison blocks, first and second sources of reference voltages, N scaling units are introduced into it, and the output of the i-th integrator through the i-th scaling unit connected to the first input of the i-th adder, the output of which is connected to the i-th output of the converter, the first input of the i-th additional comparison unit is connected to the i-th input of the converter, and the output is connected to the input of the i-th delay unit, the second input of the first additional the comparison unit is connected to the output of the first reference voltage source, the second input of the i-th additional comparison unit, in addition to the first, is connected to the first input of the (i-1) -th additional comparison unit, the second input of the first adder is connected to the output of the second voltage reference Ia and the second input of i-th adder, except the first, - yield (i-1) -th adder.