Data processing: measuring – calibrating – or testing – Measurement system
Reexamination Certificate
2000-11-22
2003-04-29
Barlow, John (Department: 2857)
Data processing: measuring, calibrating, or testing
Measurement system
C702S189000
Reexamination Certificate
active
06556939
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to the monitoring of physical processes for early detection of impending equipment failure or process disturbance and on-line, continuous validation of sensor operation. More particularly, the invention relates to systems and methods for the generation of replacement signals for failed sensors or for inferred signals for physical parameters that are not directly instrumented.
BACKGROUND OF THE INVENTION
Monitoring the performance of almost any process (such as in refining, chemicals, steel, energy production) requires the use of sensors to assure that operation is maintained within prescribed constraints and that equipment is performing within specifications to assure acceptable product quality and yield. Performance monitoring and optimization of equipment and machines (automobile systems, jet engines, discrete manufacturing, etc.) similarly relies on sensors to ensure safe operation and peak performance. A plethora of sensors have been developed to measure electrical, thermal, chemical and physical parameters of processes and equipment. Types of sensors include thermocouples, accelerometers, mass flow meters, acoustic sensors, stress and strain indicators, vibration sensors, and so on.
For most important process and equipment monitoring and control applications, sensors are nowadays electrically powered, and provide an electrical indication (either analog or digital) of the parameter that is sought to be measured. Furthermore, in many circumstances, sensors are connected via a bus or network, and may contain sufficient processing power on-board to packetize sensor data and transmit it across a network. In some cases, sensors are connected with or contain wireless transmitters or transceivers for transmission of sensor data to a remote location.
Sensor data can be used in processes or in equipment operation in a variety of ways. Sensors provide validation that control settings have taken effect, and a typical practice is to indicate an alarm when a sensor reading exceeds or drops below a safety or tolerance threshold. Sensor data can also be streamed to a data repository for off-line analysis and trending, which is used to schedule maintenance or refine a process. A further use of sensor data is to provide feedback for continuous control of the operation of a process of piece of equipment. In an automobile engine, for example, a number of subsystems use sensor data to compute downstream settings for optimal engine performance, or to meet certain minimum clean air requirements.
There are a variety of circumstances in which it is difficult or impossible to employ a sensor to measure a desired parameter. The environment in which the sensor is placed may be hostile to the longevity or even proper functioning of a sensor, as for example in measuring the flow of a gas containing a problematically high concentration of corrosive acid. Alternatively, the environment may require a sensor that is prohibitively expensive or hard to come by. In another alternative circumstance, the measurement sought may be impossible to reasonably measure directly, as in attempting to determine the remaining empty volume of an unusually shaped chamber partially filled with a liquid. In yet other circumstances, the deployment of a sensor may adversely weaken or otherwise impact the process or system being monitored. For example, in a closed fluid system such as a hydraulic system, placement of a sensor through the wall of the system to directly measure a property of the fluid presents a point of weakness and potential failure in the closed system. What is needed is a way of indirectly measuring the parameter in question.
Under such circumstances, one may attempt to measure one or more other parameters in order to infer the desired parameter. This may require outfitting the process or equipment with additional sensors, and using computing resources to compute the inferred parameter. However, it is generally difficult to successfully do this. Furthermore, it usually requires a great deal of study and knowledge of the process or equipment, or an understanding of the “first-principles” dynamics of the system, which may not be readily obtained without an unreasonable amount of research time and cost. What is needed is an effective way of inferring a hard-to-measure parameter from other measured parameters of a system that correlate in some way, without requiring a complete knowledge of the dynamics of the system and the parameters involved.
Such a need also exists for the circumstance of manufacturing an instrumented product, such as an engine or other machine, which uses sensors for feedback control, safety, or performance optimization. It is highly desirable to reduce the cost of producing the product by not outfitting the product with a sensor for every parameter, but instead inferring some parameters based on readings from other sensors. Such an inference may be possible using a subset of sensors for the machine or engine, coupled with extensive knowledge of the behavior of all the parameters in tandem. However, the requisite knowledge can be difficult and costly to develop. Furthermore, the cost of additional computing power that may be required on-board the product to calculate the inferred sensor values may outweigh the cost savings of removing sensors in the first place. What is needed is a computationally efficient way of inferring values for sensors “removed” from the production units from values of sensors that are in fact built into the production units of the product.
An additional difficulty is presented with the failure of sensors. As an example, sensors may be used to monitor a process or equipment to detect when it deviates from “normal” or correct operation. Normal can mean an acceptable functioning state, or it can be the most preferred of a set of various acceptable states. However, in practice the deviation can be due to a change in the underlying parameter measured by the sensor, or to a faulty sensor. Hence, it is essential that the health of these sensors is also known, and disturbances initiated by sensor faults should be identified and differentiated from process deviations. Often, even though a sensor has failed, it is desirable to continue process operation and the failed sensor must be replaced with a replacement or “virtual” sensor providing the same information. What is needed is a way of providing an output or estimate for a failed sensor within a system to enable continued operation.
“First principles” techniques are known in the art for generating “virtual” sensor data based on other real sensor data. Maloney et al. describe in “Pneumatic And Thermal State Estimators For Production Engine Control And Diagnostics”, Electronic Engine Controls 1998, estimator algorithms implemented in a production grade speed-density Engine Management Systems (EMS). A critical and basic need in the design of EMS control and diagnostic algorithms is the availability of information describing the state of the engine. The estimator algorithms provide engine mass flow, pressure, and temperature estimates for general use by control, diagnostic, and other estimator algorithms. Maloney et al. describe the development of such first principles models with fully instrumented engines in the laboratory, to compute virtual signals off of real sensor signals. The development is involved and highly specific to the application presented. It would thus be desirable to provide a general method for the generation of missing values or virtual signals without have to resort to developing first principles models.
In a related trend, processes or machines are monitored by software-based systems that monitor correlated sensor values in aggregate. Such a system is described in U.S. Pat. No. 5,764,509 to Gross et al., the teachings of which are hereby incorporated by reference. Such a system for monitoring or providing control over a process or machine is superior to traditional threshold-type sensor-based monitoring and control because it can generally differentiate
Michael Best & Friedrich LLC
SmartSignal Corporation
Washburn Douglas
LandOfFree
Inferential signal generator for instrumented equipment and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Inferential signal generator for instrumented equipment and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Inferential signal generator for instrumented equipment and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3077504