BACKGROUND

[0001] The present disclosure relates to process variable transmitters used in process control and monitoring systems. More specifically, the present disclosure relates to using a component embedded in a transmitter to detect sensor degradation or failure.

[0002] Process variable transmitters are used to measure process parameters in a process control monitoring system. Microprocessor-based transmitters often include a sensor, an analog- to-digital converter for converting an output from the sensor into a digital form, a microprocessor for compensating the digitized output, and an output circuit for transmitting the compensated output. Currently, this transmission is normally done over a process control loop, such as a 4-20 _t milliamp control loop, or wirelessly.

[0003] Such process variable transmitters are often connected to a host system by the process control loop. Some current host systems use a statistic, referred to as the Standard Error of the Estimate (Se) statistic, in order to analyze measurement data to predict whether sensors or other process components are failing, before such failures become critical. For instance, some such host systems use the Se statistic to determine whether one of the sensors (such as a thermocouple) is degraded but not yet failing. Other host systems use the Se statistic in order to identify certain process failures (such as reactor runaway) before severe damage occurs. In order to provide this type of information with sufficient accuracy, the conventional Se equation must be updated at least once every second, or more frequently.

[0004] More specifically, the Se statistic is used to determine the standard deviation of a sensor signal over a linearization region. This helps to identify how well the data is linearized or how well it is tracking the linearization. One example of the equations used for this statistic are shown below:

1Ζ»∑ί(Π -Τ?

[0005] A Eq. 1

[0006] B Eq. 2

2

[0007] AiAt -\- B Eq. 3 [0008] S _e = Eq. 4

(n-2)

[0009] Where:

[0010] n=number of samples (usually 12 to 20)

[0011] i=sample number(l to n)

[0012] Ti=n average inputs

[0013] t (^calculated input

[0014] 7 -sampled_inputs

[0015] t-update_time

[0016] S _e =Standard Error of Estimate

[0017] These equations are resource intensive for embedded microcontroller systems used on many process transmitters. They require a relatively large amount of random access memory for each sensor that is being analyzed. The resolution of the equations is also quite time intensive, and this can reduce the amount of time that is available to process the normal measurement signal input by the sensor.

SUMMARY

[0018] A process variable transmitter includes a memory that stores a filtered sensor value that is calculated based on a prior received sensor value and a filtered rate of change value that is calculated based on a prior rate of change value. The process variable transmitter also includes a controller that receives a sensor value and compares it to the filtered sensor value to obtain a rate of change value. The controller also compares the rate of change value to the filtered rate of change value to obtain a deviation value, and generates an indication, which may be a sensor failure warning output, based on the deviation value. This is done within the process variable transmitter.

BRIEF DESCRIPTION OF THE DRAWINGS

[0019] FIG. 1 is a simplified block diagram of a process variable transmitter coupled to a host system and sensors in a process.

[0020] FIG. 2 is a flow diagram illustrating one embodiment of the operation of the system show in FIG. 1 . [0021] FIG. 3 is a flow diagram illustrating one embodiment of the operation of the system shown in FIG. 1 , for multiple sensors.

[0022] FIG. 4 is a block diagram illustrating one embodiment of multiple processor variable transmitters and multiple sensors.

[0023] FIG. 5 is a flow diagram illustrating one embodiment of the operation of the system shown in FIG. 4.

DETAILED DESCRIPTION

[0024] FIG. 1 is a simplified block diagram of a transmitter 10 in accordance with one embodiment. Transmitter 10, in the embodiment shown in FIG. 1 , includes analog-to-digital (A/D) converter 12, processor 14, memory 16 and input/output (I/O) component 18. Process variable transmitter 10 also illustratively includes conditioning component 19. Transmitter 10 is shown coupled to a plurality of different process variable (PV) sensors 22 and 24. Transmitter 10 is also shown coupled to host system 20 over control loop 26. Of course, transmitter 10 could be connected to a wireless communication link instead of, or in addition to, process control loop 26. In one embodiment, process control loop 26 provides power to transmitter 10.

[0025] Sensors 22 and 24 are illustratively process variable sensors that receive inputs from process 28 that is being sensed. For example, sensor 22 may illustratively be a thermocouple that senses temperature and sensor 24 may also be a thermocouple, or a different sensor, such as a flow sensor. Other PV sensors can include a variety of sensors, such as pressure sensors, pH sensors, etc. Sensors 22 and 24 illustratively provide an output that is indicative of a sensed process variable to conditioning component 19.

[0026] Conditioning component 19 illustratively conditions the signals received from sensors 22 and 24, such as by amplifying them, linearizing them, etc., and provides them to A/D converter 12. The analog signals provided to A D converter 12 are thus indicative of the sensed parameters. A/D converter 12 converts the analog signals into digital signals and provides them to processor 14.

[0027] In one embodiment, processor 14 is a computer microprocessor or microcontroller that has associated memory 16 and clock circuitry (not shown) and provides information regarding the sensed parameters to I/O component 18. I/O component 18, in turn, provides that information to host system 20 over control loop 26. I/O component 18 can provide the information over control loop 26 either in digital format, or in analog format by controlling current flow (I) through loop 26. In any case, the information related to the sensed parameters is provided over process control loop 26 by transmitter 10.

[0028] In the embodiment shown in FIG. 1 , processor 14 also includes sensor deviation detector 30. Detector 30 detects statistics related to the sensor signals provided by sensors 22 and 24 and identifies whether sensors 22 and/or 24 are degrading, are failing, or are about to fail. FIG. 2 is a flow diagram illustrating the operation of process variable transmitter 10, and specifically sensor deviation detector 30 in identifying sensors that are failing, or that are about to fail.

[0029] As discussed above, some conventional host systems calculated the Standard Error of the Estimate (Se) statistic in order to analyze measurement data in an attempt to identify failing sensors. However, it is believed that this has not been done in a component embedded in transmitter 10. In contrast, sensor deviation detector 30 can process the sensor signals received from A/D converter 12 in order to identify sensors that are degrading, failing or are about to fail. In doing so, sensor 14 first receives the sensor input values from A/D converter 12. This is indicated by block 40 in FIG. 2. Then, instead of calculating the Se value, sensor deviation detector 30 uses a delay operator (or filter) to access memory 16 and compare the incoming data from the sensor to a filtered sensor signal to obtain an estimate of a rate-of-change of the sensor signal. This is done as follows:

[0030] x = y— y _f ; where Eq. 5

[0031] x is an estimate of the rate of change of the sensor signal;

[0032] y is the input signal from the sensor (as digitized by A/D converter 12); and

[0033] y _f is the average of some number of previous sensor input samples for this given sensor and stored in memory 16.

[0034] Comparing the sensor input value (y) to the filtered sensor value to obtain an estimate (x) of the rate of change is indicated by block 42 in FIG. 2.

[0035] Sensor deviation detector 30 then accesses memory 16 and performs a similar filtering operation on the estimate (x) of the rate of change obtained in Eq. 5 above, in order to obtain a deviation value as follows:

[0036] d v = \x - Xi Eq. 6 [0037] where Xi is the average of some number i of previous rate of change calculations performed for this given sensor and stored in memory 16. Comparing the estimate of the rate of change for the current signal to a filtered rate of change value to obtain the deviation value shown in Eq. 6 is indicated by block 44 in FIG. 2.

[0038] Sensor deviation detector 30 then compares the new deviation value (dev) to a threshold value to determine whether the deviation corresponding to the current sensor input is beyond some threshold that would indicate that the sensor is degraded, failing, or is about to fail completely. The specific value for the threshold can be determined empirically, or it can be based on the particular type of sensor being processed. For instance, if the sensor is a thermocouple sensing temperature in a first type of process, then the threshold value may be set to one level, whereas if the sensor is a thermocouple in a different type of process, or if it is a different type of sensor (such as a pH sensor or flow sensor) then the threshold value may be set differently, as desired. Comparing the deviation value to the threshold value is indicated by block 46 in FIG. 2.

[0039] If the deviation value exceeds or surpasses the threshold value at block 46, then sensor deviation detector 30 generates an indication of this to the user. In one embodiment, this indication takes the form of a sensor failure warning 36 that is provided to I/O component 18, for transmission to host 20. This is indicated by block 48 in FIG. 2.

[0040] However, if, at block 46, it is determined that the deviation value does not surpass the threshold value, then sensor deviation detector 30 updates the filtered sensor value and the filtered rate of change estimate in transmitter 10 and stores those updated values in memory 16. This is indicated by block 50 in FIG. 2.

[0041] Processor 14 then determines whether there are more sensors to check. This is indicated by block 52 in FIG. 2. If not, then processing is complete and detector 30 waits until it is time to perform its detection again. However, if, at block 52, it is determined that there are more sensors to check, then a next sensor is selected, as indicated at block 54 in FIG. 2, and processing reverts back to block 40.

[0042] It can thus be seen from the above description that sensor deviation detector 30 can detect whether the sensors are failing, by implementing Equations 5 and 6, and this is all done by components embedded within process variable transmitter 10. This is because, instead of using the Se equations which require multiple processes, with each new data point, and which also include averaging, calculating a trend line, evaluating theoretical values from the trend across a number of previously corrected samples, and using complex mathematics such as multiple squares and a square root, sensor deviation detector 30 simply operates by comparing a new sensor input value to a rolling average of prior values, and does the same for the deviation value. This is much less costly in terms of processor overhead and time, yet still provides similar characteristics to those used by host systems to detect sensor failure. Thus, sensor degradation and failure can be detected within process variable transmitter 10, itself, without the need for calculations at host system 20.

[0043] While FIG. 2 illustrates the operation of the system shown in FIG. 1 for processing sensors on an individual basis, the system can also be used to process inputs from a plurality of different sensors 22 and 24 that either sense variables in process 28 and provide sensor signals to the same process variable transmitter 10, or that provide signals to multiple different process variable transmitters. When receiving signals from multiple different sensors, sensor deviation detector 30 can detect or sense abnormalities by comparing the values from the sensors to an expected relationship among those values. For instance, sensors 22 and 24 are both measuring temperature at different points in process 28, there may be an expected relationship between those two sensor values (such as one being a certain percentage cooler or hotter than the other). Similarly, the values from those sensors can be expected to change in relation to one another according to a given, predetermined, relationship. By way of example, if the sensor signals are expected to be related linearly, and if one of the sensors shows a rapid increase in temperature, but the other does not, then the expected relationship between those two sensors has not been observed, and it may be that one of the sensors is failing, or is about to fail. FIG. 3 is a flow diagram illustrating the operation of the system shown in FIG. 1 in order to perform these types of detections.

[0044] Transmitter 10 first receives and processes multiple sensor inputs as shown in FIG. 2. Of course, the sensor inputs can optionally be from multiple different transmitters, or they can be from multiple different processes, or both. Receiving and processing the sensor inputs as discussed above with respect to FIG. 2 is illustrated by block 60 in FIG. 3.

[0045] Sensor deviation detector 30 then compares the values obtained by processing as described above with respect to FIG. 2 to determine whether there are any relative sensor abnormalities by comparing the values across the multiple different sensor inputs processed at block 60. This is indicated by block 62 in FIG. 3. If there are sensor abnormalities as determined at block 62, then sensor deviation detector 30 outputs an indication of this (such as sensor failure warning 36) as indicated by block 64. If not, processing is complete.

[0046] It should be noted that, in the embodiment shown in FIG. 3, the expected correlation between sensor inputs can be stored in memory 16 and accessed by sensor deviation detector 30 in making the determination as to whether there are any relative sensor abnormalities. The relationship can be stored in memory 16 as an equation, or as a table of values that can be used as a look up table, or in other ways. By way of example only, it may be that transmitter 10 is used on one or more processes where a linear correlation is expected from sensor-to-sensor or from process-to-process. In performing the detection shown in FIG. 3, sensor deviation detector 30 can determine whether there is a relative process or sensor abnormality simply by determining whether the expected linear relationship is observed among the input values. Of course, other relationships can be used as well.

[0047] FIG. 4 is a simplified block diagram showing process variable transmitter 10 along with sensors 22 and 24 and host system 20. FIG. 4 shows that sensors 22 and 24 are sensing process variables in process (or pipe) 28. FIG. 4 shows that an additional process variable transmitter 70, along with another sensor 72, is also disposed to sense a process variable from process (or pipe) 28. Transmitters 10 and 70 are coupled for communications, with one another.

[0048] It will be appreciated that using sensor deviation detector 30 to calculate deviations for multiple transmitters or sensor inputs allows detector 30 to calculate process variable velocity and acceleration, so long as the distance between sensors 22, 24, and 72 is known. The distance can be illustratively input into sensor deviation detector 30 by a user configurable entry, or otherwise.

[0049] More specifically, when the sensors are connected to the same process 28, at different points, as illustrated in FIG. 4, and the distance between the sensors is known, then sensor deviation detector 30 can detect characteristics on each sensor channel that represent a process variable increase or decrease. When this process variable increase or decrease is seen along two or three (or more) channels, sensor deviation detector 30 can calculate both velocity and acceleration for the process variable. For instance, if sensors 22, 24 and 72 are all temperature sensors, then by processing the signals input from those sensors as described above with respect to FIGS. 2 and/or 3, sensor deviation detector 30 can identify characteristics on each sensor channel that represent a temperature increase or decrease and can thus calculate the velocity and acceleration of the temperature changes in process 28. Similarly, sensor deviation detector .30 can identify peaks or valleys in the input from one sensor channel to the next to identify temperature dissipation or heat loss across process 28 by observing the channel-to-channel differences in the peaks or valleys. Sensor deviation detector 30 can also identify the status of a given sensor channel as deviating from the normal relationship, and thus indicate that the accuracy of the signal on that channel is uncertain.

[0050] FIG. 5 is a flow diagram illustrating the operation of the system shown in FIG. 4 in order to perform these types of calculations. Process variable transmitter 10 first receives sensor inputs from sensors 22, 24, and 72 (via process variable transmitter 70) spaced along the same process 28. This is indicated by block 80 in FIG. 5. Sensor deviation detector 30 in process variable transmitter 10 then processes the sensor inputs as described above with respect to FIG. 2. This is indicated by block 82 in FIG. 5.

[0051] Sensor deviation detector 30 then looks for abnormal characteristics corresponding to the sensor inputs. This can be, for example, a deviation from an expected relationship, characteristics corresponding to abnormal velocity and acceleration, characteristics corresponding to abnormal process variable differences (such as temperature dissipation or heat loss) across the process, etc. This is indicated by block 84 in FIG. 5.

[0052] Sensor deviation detector 30 then records the results identified in memory 16 and generates any applicable warnings 36. This is indicated by block 86 in FIG. 5.

[0053] While the description above has given a number of examples for process variables that can be sensed, it will of course be appreciated that a variety of other process variables can be sensed and processed in substantially the same way. Examples of such other process variables include pressure, level, flow or flow rate, etc.

[0054] Although the present disclosure has been described with reference to illustrative embodiments, workers skilled in the art will recognize that changes may be made in form and detail without departing from the spirit and scope of the disclosure.