RU2540449C1 - Method to generate status of intelligent sensor measurement results - Google Patents

Abstract

FIELD: measurement equipment.

SUBSTANCE: upper and lower threshold values of a reference signal are generated, besides, the lower threshold value is less than one, and the upper one is higher. Values of the reference signal are compared with threshold values, besides, if the reference signal value is in the range between the upper and lower threshold values or is equal to them, the measurement is given the status of confirmed, and if it is out of the borders defined by the upper and lower threshold values - as invalid. To generate a reference signal, the sensor outlet signal is divided into two additive components, the first of which does not depend on a noise component at the sensor inlet, and the second one is directly proportionate to it. Energy of the second sensor component is registered, and the reference signal is the ratio of the current energy of the component to the value of its energy at the moment, when the sensor operability was guaranteed.

EFFECT: possibility to assess metrological condition of a sensor of physical value in the mode of continuous technological process.

3 dwg

Description

The invention relates to instrumentation and can be used in instrumentation in the development, manufacture and diagnosis of intelligent sensors and measuring systems of various types.

Currently, due to the increasing requirements for the reliability and efficiency of various types of control systems, requirements for their metrological support are increasing, i.e. to ensure the necessary reliability of the measurement information. The main problems in ensuring the reliability of measurement information are associated with sensors: their components age, the parameters change over time. Sudden defects also occur. All this can lead to errors in control [R. Taymanov, K. Sapozhnikova, Metrological Self-Check and Evolution of Metrology, Measurement, 43, 2010, pp. 869-877]. The main ways to solve the problem of ensuring the reliability of information are to reduce the inter-calibration (inter-verification) interval and various methods of backup and integration of sensors. None of these methods can be considered optimal. The first method dramatically increases the cost of operating the system and it may simply not be applicable due to the peculiarities of its operation. In addition, the method does not guarantee the preservation of system parameters during the inter-verification interval. The use of the second method may be limited by the design features of the system, its tactical and technical characteristics and the requirement of independence of the influence of external conditions on the parameters of the sensor.

According to GOST R 8.673-2009. GSI. “Intelligent sensors and intelligent measuring systems. Terms and definitions ”, page 4, paragraphs 3.11. It is proposed to solve this problem for smart sensors by implementing the self-monitoring function in them. In the same place, on pages 2-3, 7, it is proposed to implement the self-monitoring function by analyzing the reference value obtained during the operation of the sensor (measuring system). In this case, the analysis result can be expressed in the status of the measurement result.

The main problem with this approach is to obtain a reference value. Usually, the method for generating the reference value is determined after a thorough analysis of the sources of the error and determination of the critical component of the error.

A known method of monitoring metrological health [patent RU No. 2321829 C2, IPC G01D 3/00, publ. 10.20.2007, “A method for monitoring the metrological operability of a non-electric quantity measuring transducer and a device for its implementation”], in which parts having different sensitivity to the factor affecting the operability of the transducer are allocated (formed) in the sensitive element. During operation, values from these parts are periodically measured with a sensitivity sufficient to detect an increase in the error of the measuring transducer, and they are used to judge its metrological serviceability, i.e. the reference value is formed by comparing signals from parts of the primary transducer having different sensitivity to the disturbing factor. The disadvantage of this method is the lack of the ability to directly correct the characteristics of the converter during operation, which is due to the fact that in this device the ratio between the errors caused by systematic and random components is unknown.

A known method of monitoring metrological health [patent RU No. 2444707, IPC G01D 3/00, publ. 10/07/2010, “Temperature measuring transducer with metrological operability control”], according to which, for the formation of the reference signal, as in the previous method, the values of the readings of the parts of the sensitive element, which have different dependence on the disturbing factor, in which, however, one part of the sensitive the element is made of several, mainly two, parallel-connected conductors, and the total cross-sectional areas of the sensitive elements are comparable with each other. This increases the reliability of the correction of the result of temperature measurements directly during operation and makes it possible to increase the calibration interval

The known method [Ivanova EP, Kurskaya T.N., Shramko S.V. About some aspects of the use of self-calibrating temperature sensors // TEMPERATURE-2011: 4th All-Russian and COOMET countries conference on thermometry problems. Abstracts (St. Petersburg, April 19-21). P.63-65], based on the use of reference points of phase transitions (melting) of reference metals for self-diagnosis of temperature sensors. The method consists in using an additional heater and ampoules with a reference metal. To check the metrological operability of the thermal converter, an additional heater is switched on, heating the ampoule and the thermal converter itself to a temperature exceeding the melting point of the reference metal. The temperature "plateau" arising at the time of the reference metal melting was used as a reference signal to verify the operation of the primary transducer. The disadvantages of the method are the limited number of control points and the need for a complex selection of reference metals.

A common disadvantage for all of these methods is the orientation to a specific type and design of sensors.

Closest to the claimed solution is a method of test effects [Strelkova O.V., Shestakov A.L. Algorithm for assessing the state of a resistance thermal converter using test influences. // MEASUREMENTS-2008: International scientific and technical conference. Conference proceedings (Penza, October 22-24), pp. 13-16], which consists in the fact that during the operation of the sensor periodically change the value of the parameter controlled by the sensor. The sensor response to these changes is used as a reference value and the basis for the formation of the measurement status. The disadvantages of the method is that it requires the development of a method for generating an impact, a corresponding change in the design of the sensor, and does not fully control the conversion function. In this case, it is necessary to control the stability of the impact, which causes another round of problems.

The problem to which the invention is directed, is to develop a universal method for assessing the metrological state of a physical quantity sensor in a continuous process.

This task is achieved by the fact that, in order to formulate the status of the measurement results of the smart sensor, the upper and lower threshold values of the reference signal are formed, the lower threshold value being less than unity and the upper one being greater, comparing the values of the reference signal with threshold values, and if the value of the reference signal is in the range between the upper and lower threshold values or equal to them, the measurement is assigned the status - confirmed, and in case of exceeding the boundaries defined by the upper and lower threshold values, - unreliable, according to the invention, for generating a reference signal, the sensor output signal is divided into two additive components, the first of which is independent of the noise component at the sensor input, and the second is directly proportional to it, the energy of the second sensor component is recorded, and the reference signal is taken the ratio of the current energy of the component to the value of its energy at the time when the sensor was guaranteed to be in good working order.

The invention is illustrated by the following graphic materials:

Figure 1 is a General block diagram of a sensor of a physical parameter;

Figure 2 is a block diagram of a physical quantity sensor with a deviation of the sensor conversion function from the reference;

Figure 3 is an example implementation of a method for an additive mixture of a narrow-band low-frequency useful signal and white noise.

Consider the task of evaluating the value of a parameter acting on a sensor under conditions of a possible “drift” of sensor characteristics. In the General case, the block diagram of the formation of the output signal of the sensor can be represented in the form shown in figure 1.

Based on the presented block diagram, the signal Y (t) at the output of the sensor will have the form

Where Y (t) is the detected signal at the sensor output, S (i) is the measured signal at the sensor input, U (t) is the signal at the output of the primary converter, ξ (t), ς (t) are the additive noise at the input and output sensor accordingly. F ₀ (x), F ₁ (u) are the conversion functions of the primary converter (FPPP) and the linearizing electronic unit, respectively. Hereinafter, it is understood that the sensor can be considered a non-inertia device in the frequency range of the measured parameter.

, тоIf the transformation F _{0 is} unique and $$F_{\mathrm{one}}=F_0^{-\mathrm{one}}$$

then

Formula 2 is the main formula when using the signal from the sensor in the calculations of control systems and information processing.

However, this is true only if the above conditions are met:

1. The transformation F _{0 is} unique.

.2. $$F_{\mathrm{one}}=F_0^{-\mathrm{one}}$$

.

Consider the case when condition (2) is not satisfied, i.e. FPPP is distorted for any reason relative to the original F ₀ . Then we can assume that F = F ₀ + ΔF, where F is the true transformation function, F ₀ is the transformation function adopted in the calculations, ΔF is the variation of the transformation function satisfying the proximity condition for the functions F and F ₀ , at least order.

Then the variation ΔF can be recounted to the input of the block diagram of figure 1 and it will take the form of figure 2. The difference between FIG. 2 and FIG. 1 is an additional interference ΔS of the sensor. Note that the conversion function is considered unchanged, i.e. relation (2) remains valid with the corresponding correction ξ (t).

The additional interference ΔS is found from the condition

Condition (3) means that the distortion of the input signal with a constant conversion function is similar, from the point of view of further processing, to the distortion of the conversion function with a constant input signal.

Taking ΔS so small that it provides a small ΔF (this will need to be considered separately), we can write

F ₀ (x) + F ' ₀ (x) Δs = F ₀ (x) + ΔF (x).

Hence we obtain the expression for ΔS:

In (4), x is the total signal acting at the input of the primary transducer, i.e.

Considering (2), (4) and (5) and assuming that the input noise significantly exceeds the sensor intrinsic noise (i.e., ς (t) can be neglected), we have

Or

Assuming that ξ (t) is also quite small with respect to the useful signal, we obtain an expression for the sensor output signal taking into account the distortion of the conversion function.

Thus, it can be considered that the deviations of the actual FINN with respect to the reference one can be replaced by the introduction of an additional noise component at the input of the reference sensor.

In expression (8), we can distinguish parts that depend on and are independent of ξ

If nothing is known about the characteristics of the input signal and the interference, then it is not possible to separate the signal (8) into components (9) - (10) and evaluate the distortion of the conversion function. However, at least there are two cases where obtaining this estimate is easy to implement:

- test impact method (when we ourselves form ξ (t)),

- The assumption of a constant additive white noise at the input, which is a common practice in the analysis of information and control systems.

In the first case, to separate Y (t) into the components Y _s (t) and Y _ξ (t), it is possible to use methods for extracting a signal of a known shape ξ (t) against the background of an unknown noise, which in this case is the signal S (t), in second, you can use various filtering methods that allow you to separate signals with different correlation functions (for white noise, ξ (t) is δ is a function, for a signal S (t) is usually a correlation function of a narrow-band process). In both cases, the task of estimating ΔF reduces to solving the system of differential equations (9-10) with respect to the unknown S (t) and ΔF (x).

If the only task is to detect distortions F ₀ , then you can simply control the deviations Y _ξ (t) from a constant when the signal changes. In the first case, this boils down to controlling the constancy of the amplitude of the response to the testing effect, and in the second, to controlling the constancy of the dispersion of the noise signal during operation. Depending on the degree of fluctuation of the signal Y _ξ (t), an estimate of the reliability of the measurement of the signal S (t) can be obtained.

Thus, to solve the technical problem posed, it is proposed to use the deviation of the amplitude (dispersion) of the signal component Y _ξ (t) from a constant value during operation to form the reference quantity.

If the input of the sensor with a nonlinear conversion function is affected by a narrow-band low-frequency signal mixed with stationary white noise (Fig. 3), then the claimed method can be implemented as follows.

The useful signal S (t) in a mixture with noise ξ (t) is fed to the input of the sensor, the primary transducer 1 of which has a nonlinear conversion function. Since the useful signal is assumed to be narrow-band low-pass, then, in the simplest case, it can be distinguished using the simplest low-pass filter 3. Then the difference in the signal at the input and output of the low-pass filter 3 allows us to estimate the noise component of the signal.

, близкую к априорному значению σ₀, а решающее устройство - величину, близкую к единице.If the sensor electronic unit 2 compensates for the nonlinearity of the converter 1 (the sensor is metrologically sound) and the sensor's own noise can be neglected, we can assume that Y _s (t) = S (t) and Y _ξ (t) = ξ (t) Then the signal dispersion calculator Y _ξ (t) will give the value $$\hat{\sigma }$$

close to the a priori value of σ ₀ , and the decisive device is a value close to unity.

As an a priori dispersion value, you can use the value obtained with a guaranteed good working condition of the sensor (the beginning of its operation).

The mismatch of the characteristics of the primary Converter 1 and the electronic unit 2 (the occurrence of a metrological malfunction) relative to the signal Y _ξ (t) will cause two effects:

относительно σ₀,- displacement of the constant component $$\hat{\sigma }$$

relative to σ ₀ ,

синхронно с изменением Y_s(t).- occurrence of modulation $$\hat{\sigma }(t)$$

synchronously with the change in Y _s (t).

от единицы. В этом случае выход отношения дисперсии за установленные значения минимального или максимального порога может служить индикатором появления метрологической неисправности.Both effects will cause a rejection of the relationship. $$\raisebox{1ex}{$\hat{\sigma }$}\!\left/ \!\raisebox{-1ex}{${\sigma }_0$}\right.$$

from one. In this case, the dispersion ratio exceeding the set minimum or maximum threshold values can serve as an indicator of the appearance of a metrological malfunction.

Thus, the above information confirms the possibility of implementing the claimed invention, achieving the specified technical result and solving the problem.

Claims (1)

A method of forming the status of a measurement result of an intelligent sensor by generating a reference signal, determining its upper and lower threshold values, the lower threshold value being less than one and the upper one greater than comparing the current value of the reference signal with threshold values, and if the value of the reference signal is in the range between the upper and lower threshold values or equal to them, the measurement is assigned the status - confirmed, and in case of exceeding the boundaries defined by the upper and lower thresholds values, - unreliable, characterized in that for the formation of the reference signal, the sensor output signal is divided into two additive components, the first of which is independent of the noise component at the sensor input, and the second is directly proportional to it, the energy of the second sensor component is recorded, and as reference signal take the ratio of the current energy of the component to the value of its energy at the time when the sensor was guaranteed to be in good working order.

Images