RU2705929C1 - Measurement channel diagnostic method - Google Patents

Abstract

FIELD: measuring equipment.

SUBSTANCE: invention relates to measurement equipment and can be used for diagnostics of measuring channel (MC), which measures object controlled parameter, in mode of continuous technological process. Diagnostics method of the measurement channel (MC) involves the following operations: receiving and recording into the archive at the current time T the readings R(T) of the diagnosed channel and the Si(T) readings of the adjacent channels, measuring parameters, indirectly dependent on parameter measured by diagnosed MC; simultaneously with the receipt and recording of readings in the archive at each current moment in time T, recording the presence of changes in the current value of the readings R(T) of the diagnosed MC relative to their archive value R(T-ΔTR) at the previous moment in time T-ΔTR, as well as presence of changes of current value of Si(T) readings of each adjacent MC relative to their archive value Si(T-ΔTSi) at the previous moment in time T-ΔTSi; if changes in the readings of the diagnosed MC are fixed in the absence of changes in the readings of adjacent MC, or changes in the readings of the adjacent MCs are recorded in the absence of changes in the diagnosed MC, a decision is made on the failure of the MC. At that, changes of the diagnosed MC and adjacent MC readings are fixed in case they exceed preset drift boundaries R_dr readings of diagnosed MC and Si_dr drift readings of adjacent MC, respectively.

EFFECT: high efficiency of diagnosing MC relative to known methods, due to allowance for changes in monitored parameters, indirectly related to the parameter measured by the diagnosed MC, use for redundant and non-redundant MC in continuous process mode.

1 cl, 1 dwg

Description

The invention relates to measuring equipment and can be used to diagnose a measuring channel (IR) that performs the measurement of a controlled parameter of an object in a continuous process.

Potential areas of application are nuclear, thermal and hydropower, chemical and processing industries, as well as other industries where channels are used that measure indirectly dependent controlled parameters.

Currently, with the introduction and rapid development of software and hardware systems (PTC), there is an increase in the level of automation and an increase in the number of IR at newly designed and commissioned industrial facilities. So, at modern electric power facilities, the number of measuring channels, which included software and hardware systems, increased on average 2.5 times in relation to the 2000-2003 projects. and amounted to over 22 thousand units of infrared energy per unit. At the same time, an increase in the number of IR defects began to be observed due to malfunctions in the operation of software and hardware systems, leading to the absence of an IR reaction to real changes in the controlled parameter. Untimely detection of such defects can affect the increase in production risks and the development of critical situations.

One of the main tools for identifying such defects is the online diagnostic methods for IR. To date, the industry of the Russian Federation is widely used methods for online diagnosis of redundant IR, based on tracking individual deviations of the IR from the arithmetic mean of their readings. These methods detect the above defects only in cases where malfunctions of the PTC lead to deviations between the readings of the reserved IR. For non-reserved IR, these methods are not a priori applicable. For comparison, the amount of redundant infrared energy at modern electric power facilities is no more than 59% of the total number of infrared energy units. Hence, the applied methods can diagnose the above defects for no more than 59% of the IR power unit.

There is a need to develop such diagnostic methods that would allow timely detection of the above defects for both redundant and non-redundant IR in the continuous process mode.

One of the directions in the diagnosis of IR are systems using neural networks.

The well-known "Device and method for monitoring a technical installation containing many systems, in particular the installation of a power plant" (RU No. 2313815, G05B23 / 02, publ. 12/27/2007), the well-known "Method for diagnosing mechanisms and systems with electric drive" (RU No. 2431152 G01R 31/34, published May 27, 2011, Bull. No. 15).

A known group of inventions "Method and device for the technical diagnosis of complex technological equipment based on neural networks" (RU No. 2563161, G06N 3/08, publ. 09/20/2015, Bull. No. 26). The essence of the method using neural networks is that the diagnostics of complex technological equipment is carried out by registering and processing signals from sensors located in the working area of the equipment, after which the neural network is trained and based on it a dynamic model is obtained. After that, the signals are recorded during the operation of the equipment, and additional training of the neural network is performed, revealing additional parameters and new relationships between them. A device that implements the method includes sensors, a computer system, and diagnostic signal display devices. The computing system contains a module implemented with the possibility of intelligent analysis and containing a dynamic model that is implemented on a trained neural network, and a module implemented with the possibility of additional training of the neural network and the selection of active and redundant neurons.

The advantage of systems using neural networks is that they provide continuous diagnostics and are applicable for a wide class of tasks with the ability to adapt to changing parameters.

However, the disadvantage is the complexity of the system, requiring large computing power to implement diagnostics. The main thing is the error / inaccuracy of the dependencies found by the neural network, in some cases, can be large enough that will not allow you to effectively perform IR diagnostics. To control the inaccuracy of the dependencies found by the neural network is far from always a trivial task.

          The well-known "Method of monitoring the metrological health of an intelligent measuring instrument" (RU No. 2491510 G01D 3/00, publ. 08/27/2013, Bull. No. 24).

The essence of the method lies in the fact that during operation, the measured value and the controlled parameter of the measuring instrument are periodically determined, the obtained values of the controlled parameter are compared with the adopted reference value, each received value of the measured value and its corresponding current value of the controlled parameter are stored, the differences between the last received the value of the measured value and its values obtained earlier, for the values of the measured value, the difference It exceeds the triple permissible error of measurements, compares with each other the corresponding current values of the monitored parameter and, based on the results of the comparison, judges the metrological serviceability of the intelligent measuring instrument.

The advantage of the invention is the provision of the possibility of periodic (almost continuous) monitoring of the metrological serviceability of an intelligent measuring instrument during its operation (without interrupting regular measurements).

The disadvantage is that the method is applicable to devices with the function of metrological diagnostic self-monitoring, which include several sensitive elements, "having different sensitivity to the factor affecting the metrological health of the measuring transducer." Therefore, the method has a highly specialized field of application and is not adapted to solve the problem.

          The well-known "Method for diagnosing a measurement sensor" (RU No. 2587635, G01F 25/00, publ. 06/20/2016). The essence of the method lies in the fact that the signal from the output of the diagnosed sensor is compared with control typical signals. In this case, the physical quantity measured by the diagnosed sensor is additionally measured by at least three sensors performing measurements in different ways. Next, for each pair of sensors, the value of the hypothesis test criterion is calculated on the equality of the distribution centers of two independent samples, consisting of the results of multiple measurements of a physical quantity. The obtained value of the criterion is compared with the normalized value, and if there is a significant discrepancy in the readings of a pair of sensors, a conclusion is made about the presence of a metrological failure of the sensor.

The use of duplicate measurements provides an increase in the metrological reliability and reliability of the results of the diagnosis of sensors.

However, the method does not allow an objective assessment of IR malfunction, since it does not provide for the recognition of cases where the discrepancy is caused by the characteristics of the process, and when its cause is a real malfunction of technical equipment from the IR structure. The main disadvantage is that the method is not applicable for the diagnosis of unreserved IR.

          The well-known "Centralized control device" (RU No. 2141722, H04B 3/46, publ. 20.11.1999), including channel sensors (parameters) of the control object, a multi-channel block normalization (unification) of sensor signals, a multi-channel unit for comparison and indication, as well as virtual a standard that can be performed using a computer that performs a given algorithm:

где y - выходное эталонное значение; N - число каналов (параметров); i - порядковый номер канала X_i - значение выходного сигнала i-го канала (параметра) блока нормализации..

where y is the output reference value; N is the number of channels (parameters); i - channel serial number X _i - output signal value of the i-th channel (parameter) of the normalization block.

If the channels (parameters) deviate during operation, we obtain signals X _i ≠ X _o and, accordingly, y ≠ y _o .

The advantage of the device is the possibility of its operation without stopping the process and dismantling the sensors.

However, this device is used "for monitoring multi-channel receiving paths, multi-channel communication systems ...", where a single calculation of the arithmetic mean (checksum) is enough to detect defects.

In relation to the diagnostic task, the implementation of such calculations does not allow to identify defects even of redundant IR, since constant changes in the readings of the sensors will constantly lead to a change in the arithmetic mean of their readings. For unreserved IRs, this method is not applicable.

           The well-known "Method of automatic control of metrological characteristics of measuring instruments (SI) of the mass of oil or liquid petroleum products (NP) when they are dispensed at fuel bases" (RU No. 2593446, G01F 25/00, published on 08.20.2016, bull. No. 22), selected as a prototype.

The essence of the method lies in the fact that before and at the end of each holiday operation, the results of measuring the mass of oil or oil products (NP) are automatically recorded and an automatic comparative analysis of the results of measurements of the mass of released oil or NP is performed according to at least three measuring instruments (SI) . According to the data of the automatic measurement system in tanks, according to the fuel dispensers and according to the data of the automatic measurement system in the receiving capacities and tanks of vehicles with accumulation of statistics, the facts of exceeding the limiting measurement errors by individual SIs judge the possibility of further operation or the need for unscheduled verification of SI. To analyze the results of three non-equal measurements of the mass of the released oil or oil, the method of comparing the measurement results with the determination of the total arithmetic mean is used, and for each SI used in the tempering operation, the actual deviation from the general arithmetic mean with the maximum permissible deviation is compared.

The use of duplicate measurements provides an increase in metrological reliability and reliability of the results of the diagnosis of measuring instruments. The advantage of the method is that the control of SI can be carried out in real conditions of its operation in the continuous process.

To perform the diagnosis of nonreserved IR method is not applicable. In addition, it does not allow an objective assessment of the malfunction of even redundant IRs, since it does not provide for the recognition of cases when the discrepancy is caused by the specifics of the process, and when its cause is a real malfunction of technical equipment from the IR structure. Therefore, the algorithm of actions provided by this invention, as applied to the task, is not enough to perform an effective diagnosis of IR.

          The objective of the invention is to develop an effective method that allows you to perform diagnostics for an extended class of IR.

          The technical result consists in increasing the efficiency of the diagnostic method by taking into account changes in the readings of adjacent channels measuring parameters indirectly related to the parameter measured by the diagnosed IR, while expanding the scope of application of IR diagnostics in relation to known methods because of the possibility of its use as for redundant, and non-reserved IR in the continuous process.

The claimed technical result is achieved using the proposed method.

A method for diagnosing a measuring channel (IR) is to perform the following operations:

- receive and record in the archive at the current time T the readings R (T) of the diagnosed channel and the readings Si (T) of adjacent channels, measuring parameters indirectly dependent on the parameter measured by the diagnosed IR;

- simultaneously with the reception and recording of readings in the archive at each current time T, the presence of changes in the current value of the readings R (T) of the diagnosed IR relative to their archive value R (T-ΔT _R ) at the previous time T-ΔT _R, as well as the presence of changes in the current value of the Si (T) readings of each adjacent IR relative to their archive value Si (T – ΔT _Si ) at the previous time point T-ΔT _Si ;

- if changes in the readings of the diagnosed IR are recorded in the absence of changes in the readings of adjacent IRs, or changes are recorded in the readings of adjacent IRs in the absence of changes in the diagnosed IR, then a decision is made about the IR failure.

In order to avoid recording minor changes in IR parameters that cannot cause a IR malfunction, changes in the readings of the diagnosed IR and adjacent IRs are recorded if they exceed the predetermined drift boundaries R _{dr of} the diagnosed IR and drift Si _dr readings of adjacent IR, respectively.

        Improving the diagnostic efficiency is achieved due to the fact that the diagnostic process takes into account not only the readings of the diagnosed IR, but also the presence of changes in the readings of adjacent IRs, which makes it possible to recognize cases when the change in the IR readings is caused by real changes in the controlled parameter, and when false changes in the IR readings occur, or a reaction There is no IR at all, for example, due to malfunctions of the PTC. At the same time, this expands the scope of the method and allows the diagnosis of both redundant and non-redundant IRs.

          From the prior art, no sources of information have been revealed that reveal the essence of the proposed method for diagnosing a measuring channel.

          The following is an example of a specific implementation of the method for diagnosing a measuring channel in practice.

In FIG. Archival readings of one diagnosed channel and three archived readings of adjacent channels are presented.

A method for diagnosing a measuring channel is implemented as follows.

First, by means of a computer, they continuously receive and record in the archive the readings of the diagnosed IR, as well as the readings of adjacent IR. As adjacent IRs, all IRs that measure physical parameters that are indirectly dependent on the parameter measured by the diagnosed IR can be selected.

Indirectly dependent parameters are those parameters whose changes cause changes in the parameter measured by the diagnosed IR, and vice versa. For example, indirectly dependent parameters are:

- pressure and temperature of saturated steam (the relationship of the parameters is due to thermodynamic relations for saturated steam);

- pressure and flow rate at the pressure of the pump unit (the relationship is characterized by the laws of hydrodynamics);

- pressure at the head and pressure at the inlet of the pump unit (the relationship is due to the pressure-flow characteristic of the pump unit);

- flow rate in lines suitable for the collector and flow rate in the collector (the relationship is due to the law of conservation of mass and energy / balance ratios);

- flow rate on the medium supply line to the tank and the liquid level in the tank (the relationship is due to the law of conservation of mass and energy).

Then, the increment ΔR of the readings of the diagnosed IR for a predetermined time interval ΔT _{R is} determined continuously in time over time:

ΔR = R (T) –R (T – ΔT _R )

R (T) - the value of the diagnosed IR at the current time T

R (T – ΔT_R) Is the value of the diagnosed IR at time T-ΔT_R. The value of R (T –ΔT_R) is taken from archived data.

ΔT _{R is} recommended to be set so that its value is not less than the triple damping time of the measuring signals at the test object, and does not exceed 1/5 of the duration of the observed average transient processes. So, for electric power facilities, the ΔT value for IR pressure, flow rate and temperature is recommended to be taken from 20 s to 30 s.

If the found increment ΔR modulo exceeds a predetermined boundary drift readings R _dr , then the fact of the presence of changes in the readings of the diagnosed IR:

│ΔR│≥ R _dr

The value of R _{dr is} recommended to be selected so that it is not lower than the allowable drift of the sensor readings, and does not exceed 1/4 of its maximum permissible measurement error. For example, for a pressure sensor with an accuracy class of 0.1, with a measurement scale of 0 to 100 kPa, the value of R _{dr is} recommended to choose 25 Pa.

At the same time, the presence of changes in the readings of adjacent IRs is similarly monitored. To do this, check the condition:

│ΔSi│≥ Si _dr ,

Where

ΔSi = Si (T) –Si (T – ΔT _Si )

Si (T) - the value of the readings of the i-th adjacent IR at the current time T

Si (T – ΔT _Si ) is the value of the readings of the i-th adjacent IR at time T - ΔT _Si . This value is taken from archived data.

ΔSi is the increment of the readings of the i-th adjacent IR for a predetermined time interval ΔT _Si,

Si _dr - a predetermined boundary drift readings for the i-th adjacent IR.

Si value_dr, by analogy with R_dr, it is recommended to choose in such a way that it is not less than the allowable drift of the sensor readings of the i-th adjacent IR, and does not exceed 1/4 of its maximum permissible measurement error.

The value ΔT _Si is set based on the duration of the dynamic processes occurring at the control object. It is recommended that the ΔT _Si value be no lower than ΔT _R , and no higher than 1/3 of the duration of the observed average statistical transients. In particular, for electric power facilities for IR pressure, temperature and flow rate, ΔT _{S is} recommended to be selected in the range from 30 seconds to 1 minute.

If changes in the readings of the diagnosed IR are recorded in the absence of changes in the readings of adjacent IR or vice versa (i.e. only one of the above mathematical conditions is observed), then the IR malfunction is recorded.

If no changes in the readings occur (all the above mathematical conditions are not satisfied), then the IR malfunction is not fixed.

If both the readings of the diagnosed IR and the readings of adjacent IRs (all the above mathematical conditions are satisfied) are changed, then the IR failure is also not fixed.

          In the event that the number of adjacent IRs is more than two, an additional identification of the faulty IR is performed. In this case, the diagnosed IR is identified as faulty if:

- changes in the readings of the diagnosed IR were not detected, and changes in the readings of at least two adjacent IRs were detected,

- changes in the readings of the diagnosed IR were detected, changes in the readings of adjacent IRs (the number of which is at least two) were not detected.

One of the adjacent IRs is identified as defective if:

- if there are no changes in the readings of a particular adjacent IR, changes in the readings of the diagnosed and other adjacent IRs are detected,

- if changes in the readings of a specific adjacent IR are detected, changes in the readings of the diagnosed and other adjacent IR are not detected.

The method is implemented on the basis of a computer that receives signals from the measuring channels of the control object.

          The following is an example of the results of testing the proposed method in practice with the presentation of the diagnosed IR with a detected defect in the archive trends presented in FIG.

As the diagnosed, the IR pressure at the pressure of the pump unit was chosen. Archival readings of this channel are shown in the drawing under the number 1 (indicated by a dotted line).

Three redundant IR flows were selected as adjacent IR. Archival readings of these IRs are shown in FIG. under numbers 2, 3, 4.

The following values were assigned to the input static parameters:

R _dr = 5 Pa,

Si _dr = 0.75 kg / m ³ ,

ΔT _R = 20 s,

ΔT _Si = 45 s.

Detailing the results of the proposed diagnostic method:

For the diagnosed IR pressure, at each moment of time, the increment ΔR was calculated and the obtained value was compared with the specified limit of the allowable drift R _dr . Similarly, for each adjacent IR flow rate, the increment ΔSi was continuously calculated and the obtained value was compared with the Si boundary, _etc. By exceeding the permissible drift limit, changes in readings were recorded at all adjacent IR flow rates. There were no changes in the readings of the diagnosed IR. As a result, an IR malfunction was recorded, while the diagnosed IR was identified as a malfunction.

Subsequently, according to the obtained diagnostic results, a malfunction in the digital module was detected, in order to restore the working state of the IR, it required a reboot. Regular self-diagnosis tools did not reveal this defect.

         Thus, the results of using the method in real conditions confirmed its effectiveness.

In relation to the electric power industry, according to preliminary estimates, the proposed method can be applied to 94% of the measuring channels of the power unit. Given that at modern electric power facilities the number of infrared energy reaches over 22 thousand units, the introduction of the proposed method is of particular relevance.

Claims (5)

1. A method for diagnosing a measuring channel (IR), which consists in performing the following operations: - take and write to the archive at the current time T the readings R (T) of the diagnosed channel and the readings Si (T) of adjacent channels measuring parameters indirectly dependent on the parameter measured by the diagnosed IR; - simultaneously with the reception and recording of readings in the archive at each current time point T, the presence of changes in the current value of the readings R (T) of the diagnosed IR relative to their archive value R (T-ΔT_R) at the preceding moment of time T-ΔT_R as well as the presence of changes in the current value of Si (T) readings of each adjacent IR relative to their archive value Si (T –ΔT_Si) at the preceding moment of time T-ΔT_Si; - if changes in the readings of the diagnosed IR are recorded in the absence of changes in the readings of adjacent IRs, or changes are recorded in the readings of adjacent IRs in the absence of changes in the diagnosed IR, then a decision is made about the IR failure. 2. The method according to claim 1, characterized in that the changes in the readings of the diagnosed IR and adjacent IRs are recorded if they exceed the predetermined drift boundaries R _{dr of the} readings of the diagnosed IR and drift Si _dr readings of adjacent IR, respectively.

Images