Method for predicting a measured value and conductivity sensor for carrying out the method

Abstract

Method for predicting a measured value of a measured variable of a sensor (10) of process automation technology, comprising the steps - Recording a first measured value x(i - 1) at a first time, - Recording a second measured value x(i) at a second, later time, - Forming a difference value between the second and first measured values, - Filtering the difference value using a filter with infinite impulse response and obtaining a filtered difference value δ f (i) as the output value of the filter, and - Calculate a future measured value y(i) based on the measured value x(i) at the second time, the filtered difference value δ f (i) and a constant τ characterizing the sensor (10) according to the equation y (i) = x (i) + δ f (i) ⋅ τ.

Description

[0001] The invention relates to a method for predicting a measured value of a measured variable of a sensor in process automation technology and a conductivity sensor for carrying out the method.

[0002] In the field of measurement technology, it is desirable to obtain a measured value as quickly as possible. This will be explained below using a temperature measurement as an example. For hygienic design or to protect the sensor from external influences, in particular to protect the interior of the sensor from any harmful, toxic or otherwise detrimental properties for the sensor, the actual measuring element of the sensor is protected in a housing. This housing, however, acts as a thermal insulator. The response time of the sensor is therefore extended because a temperature change in the medium to be measured is transmitted via the housing to the sensor or sensor element. For many measuring principles, such as conductivity measurement, temperature measurement is an auxiliary variable and a necessary prerequisite for determining the main parameter, such as conductivity.For example, DE 197 47 273 A1 discloses an inductive conductivity sensor which includes a temperature measuring device as an auxiliary variable for detecting the temperature.

[0003] US 8 301 408 B2 and US 2013 / 0 158 937 A1 describe a sensor and a method for determining an estimated measured value in real time. This document discloses the use of an infinite impulse response (IIR) filter.

[0004] From JP 2014-178242 A the use of a Kalman filter algorithm to reduce the noise of a measurement signal is known, whereby a parameter of the filter algorithm is adapted to the current measurement signal curve.

[0005] The invention is based on the object of specifying a method which is, on the one hand, robust and simple, but, on the other hand, also takes into account specific properties of a sensor.

[0006] The object is achieved by a method comprising the steps of: detecting a first measured value x(i - 1) at a first time; detecting a second measured value x(i) at a second, later time; forming a difference value of the second and first measured values; filtering the difference value using a filter with infinite impulse response and obtaining a filtered difference value δ f (i) as the output value of the filter; and calculating a future measured value y(i) based on the measured value x(i) at the second time, the filtered difference value δ f (i) and a constant τ characterizing the sensor according to the equation y(i) = x(i) + δ f (i) - τ,

[0007] In an advantageous further development, the measured quantity is the temperature.

[0008] By applying the above-mentioned method, the physically based, slow response time is shortened. This increases the accuracy of dynamic processes, especially in measurement methods or characteristic curves with a high temperature dependence. A good example of this is phase separation processes, such as those that occur before and after cleaning processes in the food industry. The increased accuracy can, for example, reduce the ejection of product that fills the lines after a cleaning process.

[0009] In a further advantageous embodiment, the sensor comprises a computer unit. In the method mentioned, the properties of the sensor are captured by the constants characterizing the sensor, and the measured value is calculated in advance based on sensor-specific properties.

[0010] In an advantageous embodiment, the constant characterizing the sensor is determined under laboratory conditions before use and is permanently stored in the sensor. This allows the sensor manufacturer to determine the optimal constant for the sensor, without the user having to worry about it. In a first variant, the constant is the same for each sensor type, e.g., for each pH or conductivity sensor. In a second variant, the constant is determined individually for each sensor and stored accordingly. Laboratory or standard conditions within the meaning of this invention are constant temperature, constant air pressure, a well-defined amount of medium, and regular stirring of the medium. Typical values ​​for this are room temperature (22 °C), normal air pressure (1020 hPa), and a volume of approximately 20 l.

[0011] In a further advantageous development, a minimum difference value is used when the difference between the second and first measured values ​​falls below a lower threshold, and a maximum difference value is used when the difference between the second and first measured values ​​exceeds an upper threshold. This allows the prediction of the measured value to be made even more accurate, since a minimum or maximum value that is too small or too large would distort the calculation.

[0012] In a preferred embodiment, the filter with infinite impulse response includes a result of the filtering of a previous measured value and the difference value of the measured values.

[0013] The filter is particularly preferred by δf(i)=d−1d⋅δf(i−1)+1d⋅δc(i) calculated with δ f (i) the filtered signal at time i, d the filter depth and δ c the difference between the measured values.

[0014] In a preferred embodiment, the method further comprises the step of filtering the future measured value by means of a second filter for smoothing the signal curve.

[0015] In another advantageous development, this filter is not an IIR filter. An FIR filter can be used instead.

[0016] The object is further achieved by a conductivity sensor comprising a temperature sensor and a computer unit for carrying out an above-mentioned method.

[0017] The invention is explained in more detail with reference to the following figure. It shows Fig. 1 the conductivity sensor according to the invention, and Fig. 2 a schematic illustration of the method according to the invention.

[0018] The inductive conductivity sensor according to the invention in its entirety has the reference number 1 and is shown in Fig. 1. The conductivity sensor 1 is designed for use in process automation.

[0019] The conductivity sensor 1 is mounted on a container 3, for example via a flange 4 (generally via a process connection), in which the medium 2 to be measured is located. The container 3 is a pipe, for example, made of plastic or metal.

[0020] The conductivity sensor 1 comprises a transmitting coil 6 and a receiving coil 7, which are housed in a housing 9. The housing 9 comprises a housing wall 16. The housing 9 is made of a plastic, in particular a thermoplastic. This plastic is approved for use in the food and biotechnology sectors. For example, this is a polyaryletherketone, such as polyetheretherketone (PEEK). This will be discussed later.

[0021] The transmitting coil 6 and the receiving coil 7 are arranged, for example, opposite one another on opposite sides of a printed circuit board (not shown). The transmitting and receiving coils 6 and 7, respectively, designed as rotationally symmetrical toroidal coils, are thus arranged coaxially one behind the other. The printed circuit board comprises conductor tracks that contact the coils and connect the transmitting coil 6 to a driver circuit and the receiving coil 7 to a receiver circuit. The driver circuit and the receiver circuit can be components of a sensor circuit arranged on the printed circuit board. The coils 6, 7 are connected to a data processing unit 5, in Fig. 1 is connected to a measuring transducer. The data processing unit 5 is generally a computer unit. Some of the computer unit's functions can also be performed directly in the sensor, which in turn includes a corresponding data processing unit.

[0022] The housing 9 forms a channel 12 passing through the transmitting coil 6 and the receiving coil 7 along their rotational axes. If the housing 9 is immersed in an electrically conductive medium 2, the latter surrounds the housing 9, or a housing section 8 intended for immersion in the medium 2, and penetrates into the channel 12, so that a closed current path 13 passing through both coils 6, 7 can form in the medium when the transmitting coil 6 is excited or flowed through with an input signal, i.e. an alternating voltage.

[0023] The conductivity sensor functions like a double transformer, whereby the transmitting and receiving coils 6, 7, as mentioned, are inserted into the medium 6 at least far enough to form a closed current path 13 running through the medium 6 and penetrating the transmitting and receiving coils 6, 7. When the transmitting coil 6 is excited with an alternating voltage signal as an input signal, it generates a magnetic field which induces a current path 13 passing through the coils 6 and 7, the strength of which depends on the electrical conductivity of the medium 2. This results in a current path with ionic conduction in the medium 2. Since this alternating electrical current in the medium in turn creates a changing magnetic field surrounding it, an alternating current is induced in the receiving coil 7 as an output signal. This alternating current or current supplied by the receiving coil 7 as an output signala corresponding alternating voltage is a measure of the electrical conductivity of the medium 2.

[0024] The conductivity sensor 1 comprises a temperature sensor 10 for measuring the temperature of the medium 2. The data processing unit 5 determines the conductivity of the medium 2 based on the input signal, the output signal, and the temperature of the medium 2. The temperature sensor 10 is an electrical or electronic component that delivers an electrical signal as a measure of the temperature. It is, for example, a thermistor or PTC thermistor as components whose resistance changes with temperature. Examples include platinum measuring resistors or ceramic PTC thermistors. Alternatively, a component can be used that directly delivers a processable electrical signal, such as a semiconductor temperature sensor, which delivers a current or voltage proportional to the temperature. Further alternatives include a thermocouple or another common temperature measuring element.

[0025] The temperature sensor 10 comprises a temperature element that provides an electrical signal as a measure of the temperature. This is a thermistor, such as a Pt100 or Pt1000. This signal, such as resistance values ​​or a voltage, is transmitted to the measuring transducer 5 via lines 17.

[0026] The method according to the invention, see Fig. 2, for calculating a future measured value, comprises two steps: a first filter is used for prediction, from which a prediction value y(i) is determined from a measured value x(i), and then smoothing is performed using a second filter. The output is the temperature value f(i). The second filter is not mandatory and serves to smooth the signal. The process is carried out in the transmitter or, in whole or in part, by a corresponding computer unit in the sensor.

[0027] First, a difference δ(i) between the current measured value (input value x(i)) and the last measured value x(i - 1) is determined, with i as the respective time: δ(i)=x(i)−x(i−1)

[0028] Outliers in this difference are “cut off” as follows to avoid excessive jumps: δc(i)={δminif δ(i)<δminδ(i)if δ(i)≤δ(i)≤δmaxδmaxif δ(i)<δmax

[0029] The difference value δ thus determined c (i) enters a simple IIR filter (Infinite Impulse Response Filter) with depth d: δf(i)=d−1d⋅δf(i−1)+1d⋅δc(i)

[0030] The prediction value y(i) is calculated from the current measured value x(i), the output value of the filter δ f (i) and the constant τ. y(i)=x(i)+δf(i)⋅τ

[0031] The above-mentioned constant "τ" is specific to each sensor type; for example, conductivity sensors have a value of η1 and pH sensors have a value of η2. This constant can therefore be related to processor performance, memory, clock speed, and / or design.

[0032] This constant is determined in advance through laboratory tests. It is varied for each sensor type until the best value is determined and an accurate and sufficient prediction of the measured value, such as temperature, can be made. This value is then permanently stored in the sensor. Users have no access to it and cannot change it. List of reference symbols 1 conductivity sensor 2 Medium 3 containers 4 Flange 5 Data processing unit 6 transmitting coil 7 Receiving coil 8 for immersion in 2 specific housing section of 1 9 housings 10 Temperature sensor 11 circuit board 12 channel 13 Current path 14 lids 15 rings 16 Housing wall 17 lines x(i) measured value y(i) prediction value f(i) temperature value i Time δ difference δ c Difference, vs. Truncated δ f filtered difference d filter depth τ constant

Claims

[1] Method for predicting a measured value of a measured variable of a sensor (10) of process automation technology, comprising the steps - Recording a first measured value x(i - 1) at a first time, - Recording a second measured value x(i) at a second, later time, - Forming a difference value between the second and first measured values, - Filtering the difference value using a filter with infinite impulse response and obtaining a filtered difference value δ f (i) as the output value of the filter, and - Calculate a future measured value y(i) based on the measured value x(i) at the second time, the filtered difference value δ f (i) and a constant τ characterizing the sensor (10) according to the equation y(i)=x(i)+δf(i)⋅τ. [2] Method according to claim 1, wherein the measured variable is temperature. [3] Method according to claim 1 or 2, wherein the constant (τ) characterizing the sensor (10) is determined under laboratory conditions before use of the sensor (10) and is permanently stored in the sensor (10). [4] Method according to at least one of claims 1 to 3, wherein a minimum difference value is used when the difference between the second and first measured values ​​falls below a lower threshold value, and wherein a maximum difference value is used when the difference between the second and first measured values ​​exceeds an upper threshold value. [5] Method according to at least one of claims 1 to 4, wherein the filter with infinite impulse response includes a result of the filtering of a previous measured value and the difference value of the measured values. [6] Method according to claim 5, wherein the filter is δf(i)=d−1d⋅δf(i−1)+1d⋅δc(i) is calculated with δ f (i) the filtered signal at time i, d the filter depth, and δ c the difference between the measured values. [7] Method according to at least one of claims 1 to 6, wherein the method further comprises the step - Filtering the future measured value using a second filter to smooth the signal curve. [8] The method of claim 7, wherein the second filter is not an IIR filter. [9] Conductivity sensor (1) comprising a temperature sensor (10) and a computer unit for carrying out a method according to at least one of claims 1 to 8.

Images