Patent Application
20060178857
The present invention is related to sensing systems. More particularly, the present invention relates to so-called smart sensors.
Many sensing systems employ redundant sensing to improve the accuracy and robustness of their measurements. Such systems are characterized by a plurality of substantially identical sensors configured to measure a predetermined parameter. Essentially, redundancy is based upon simple repetition of the functionality of the same type of sensor either in the same or in different locations.
As illustrated in
In such redundant sensor arrangements, data fusion and validation are also performed by the microprocessor or PLC, thereby consuming even more of the limited processor resources and being subject to processor cycle delays and throughput constraints. Validating individual sensors among only a pair of sensors is difficult where no information about the sensors is available apart from the data at the inputs of the microprocessor or PLC. Simpler, analog systems may employ arithmetic averaging techniques to fuse the data from a pair of sensors. More robust systems employ three sensors and validate individual sensor data with covariance techniques. Of course, as alluded to, more sensors require more of the already limited input points and processor resources, consume additional space and provide additional system cost.
So-called smart sensors may alleviate some of the burden on the microprocessor or PLC by performing much of the signal processing in-situ (e.g. signal conditioning and filtering, analog-to-digital (A/D) conversion, error and offset compensations, linearization, data storage) and additionally provide communication and data buffering to and from the microprocessor or PLC. While this may eliminate the need for most custom post-processing at the microprocessor or PLC, a redundant sensor arrangement of such smart sensors still suffers from certain shortfalls and requires processor or PLC level validation. For example, conventional validation techniques are subject to generic influences upon the sensors, such as radio frequency interference (RFI) and electromagnetic interference (EMI), and will fail to diagnose such common mode issues. Spatial diversity of redundant sensors (i.e. diversity of sensor locations) has been employed in an attempt to address such generic influences. However, no practical degree of spatial diversity will lessen the influence of homogeneously distributed generic influences 115. If generic influences are concentrated or focused 120, spatial diversity may provide some relief; however, spatial diversity may not be practical or may introduce other sources of measurement error, particularly in spatially critical sensing applications such as localized parameter measurements wherein distribution of sensors is generally undesirable, impractical or irrational.
The present invention overcomes the shortfalls of redundant sensing and spatial diversity. In accordance with the present invention, a sensor assembly for measuring a predetermined parameter includes a plurality of sensing elements. The sensing elements are integrated within a unitary sensor package. Each of the sensor elements is operative in accordance with a unique sensing principle to provide a respective measurement signal corresponding to the predetermined parameter. A signal processor is integrated within the unitary sensor package and is effective to fuse the respective measurement signals. The signal processor is also effective to provide a single sensor output signal based upon the measurement signals provided by the plurality of sensing elements that is indicative of the predetermined parameter. Each of the sensing elements is substantially immune from common mode effects due to influences which may operate upon all sensor elements. The signal processor may also provide conditioning and validation of the sensor element signals.
In accordance with a preferred implementation, the signal processor includes micro-controller circuitry including a storage medium having a computer program encoded therein. The computer program includes code for acquiring sensing element signals, code for conditioning sensing element signals, code for validating sensing element signals, and code for fusing the sensing element signals to provide an integrated sensor signal.
An exemplary embodiment of a temperature sensing application includes, for example, a thermistor, a thermocouple and a pyrometer as sensing elements. Preferably, the sensing element complement includes a non-contacting-type sensing element (e.g. pyrometer, thermal imagers and ratio thermometers) and a contacting-type sensing element (e.g. thermistor, thermocouple, and thermopile).
A method for sensing a predetermined parameter in accordance with the present invention includes providing a plurality of sensing elements within an integrated sensing package. At least two of the plurality of sensing elements are characterized by disparate sensing principles to provide respective sensing element signals corresponding to the predetermined parameter. The method also includes fusing the sensing element signals with processing circuitry within the integrated sensing package, and may further include validating the sensing element signals.
The drawings, which are understood to be exemplary of a preferred embodiment of the present invention and not limiting thereof, are now referred wherein:
The present invention will now be described with reference to
With particular reference to
Of course, more complex variants of simple sensors such as those exemplified above may be implemented as the sensing elements in accordance with the present invention, it being understood that the exhibition of disparate measurement principles should be retained. For example, a thermopile comprising a plurality of thermocouples may be used in place of or in conjunction with a single thermocouple. Also, a variety of pyrometer-based sensors includes two-dimensional thermal imagers and ratio thermometers, each of which may be used in place of or in conjunction with a simple pyrometer.
Signal processor circuitry 305 within the smart sensor 301 integrated package 310 provides for signal conditioning and filtering, analog-to-digital (A/D) conversion (as required), error and offset compensations, linearization, etc. of the plurality of sensors (S) signals. Additionally, data storage and communication and data buffering to and from the microprocessor or PLC may be provided by circuitry 305. Circuitry 305 may be implemented in completely analog fashion in certain applications. However, circuitry 305 is preferably microcontroller-based with conventional control and logic circuitry as required by the particular sensor application and includes a CPU, read-only and read-write memory devices in which are stored a plurality of routines for carrying out operations in accordance with the present invention, including routines for signal conditioning and filtering, error and offset compensations, linearization, etc. of the signals from the plurality of sensors (S). Circuitry 305 may also include, for example, such common input/output (I/O) circuitry including A/D and D/A converters, non-volatile memory devices, digital signal processors, mixed-mode circuitry, etc. Being processor-based, such circuitry can be custom programmed to satisfy specific system requirements and later reprogrammed or re-calibrated as needed.
Independent measurements from the plurality of sensors (S) are validated and fused inside the sensor in order to provide a reliable source of information to the controller 210. Such distributed processing relieves such processing functions from the controller 210 and advantageously eliminates the attendant throughput constraints and delays.
Beginning first with block 410, sensor element data acquisition includes steps necessary to read the individual sensors (Sa-Sc). Such steps may be performed on a regular basis such as through a conventional timer interrupt loop or through other irregular interrupts such as event based interrupts. The frequency of data acquisition will vary in accordance with such factors as the parameter being sensed and the measurement principle of the sensing element. This operation may further include provision of voltage or current to the sensor, for example a control current to a thermistor to enable acquisition of a resultant voltage. Additionally, multiplexing of the various sensor elements to a single input stage would require coordination and management at this point if employed.
Block 420 represents the conditioning of the sensor element data so acquired. For example, signal conditioning comprising conventional “debouncing”, filtering, averaging, error and offset compensations, linearization etc. are performed on the acquired data. Analog to digital conversion is also performed on the data as part of the signal conditioning. However, such A/D conversion may be performed at various points in the conditioning—and even validation—of the sensed data since often times certain operations are more complex in the digital domains and it may be preferable to process the data in the analog domain. Eventually, however, it is preferable to digitize analog sensor element data.
Next, block 430 represents validation of the individual sensor element data whereat the health of a particular sensor element may be checked. Such operation may include rationality checks based on stored data tables, recent historical sensor element data or quasi-covariance relative to the other commonly packaged sensor elements or a true co-variance relative to other similar sensor elements in a system employing redundant such sensor elements either as additional sensor elements either part of or apart from the same integrated package 301.
Validated sensor data can then be fused in any variety of known manners to achieve an integrated sensor output as illustrated at block 440. Various fusion frameworks ranging in complexity from simple correlative, through analytical to empirically learned, or hybrids thereof, can be utilized to fuse the sensor element data using, for example, Dempster-Shafer or Bayesian data fusion to aggregate signals acquired from different sources and even at different times. If desired, additional outputs are synthesized at this point also as required. For example, a power measurement can be obtained indirectly by measuring the current through and the voltage across an electric circuit or element and determining the electrical power as a function of current and voltage.
Block 450 next represents storage of data which may include individual sensor element data, fused and synthesized sensor data and any other data which may be used in the sensor operation, diagnostics and prognostics. Finally, block 460 represents communication management and data transfer between the smart sensor 301 and control 210 or other busses or networks 215.
The invention has been described with respect to certain preferred embodiments that are intended to be taken by way of illustration of the invention and not by way of limitation. For example, while the invention has been described with respect to an automotive engine temperature sensing application, it is equally applicable, with appropriate modifications, to other sensing applications.
