阅读说明：本技术 使用人工智能在过程工业中进行生产核算的方法和系统 (Method and system for production accounting in process industry using artificial intelligence ) 是由 什里坎特·巴特 拉胡尔·库玛-维杰 于 2020-04-20 设计创作，主要内容包括：本发明涉及一种使用人工智能(AI)在过程工业中进行生产核算的方法和系统。更具体地,本发明涉及过程工厂中的多个测量仪器(102)和过程设备中的故障检测。来自测量仪器(102)的多个测量信号通过过程控制系统接收,并且从多个测量信号中提取噪声。所提取的噪声使用基于AI的数据分析技术而与从多个参考信号中提取的噪声进行关联。进一步地,过程控制系统标识一个或多个参数中的偏差。过程控制系统使用所关联的噪声和一个或多个参数的所标识的偏差来检测多个测量仪器(102)或过程设备的故障。(The present invention relates to a method and system for production accounting in the process industry using Artificial Intelligence (AI). More particularly, the present invention relates to fault detection in a plurality of measurement instruments (102) and process devices in a process plant. A plurality of measurement signals from a measurement instrument (102) are received by a process control system and noise is extracted from the plurality of measurement signals. The extracted noise is correlated with noise extracted from a plurality of reference signals using AI-based data analysis techniques. Further, the process control system identifies a deviation in one or more parameters. The process control system uses the associated noise and the identified deviation of the one or more parameters to detect a fault of the plurality of measurement instruments (102) or the process device.)

1. A method for detecting faults in a plurality of measurement instruments (102) and process devices in a process plant, wherein the plurality of measurement instruments (102) are configured to monitor one or more parameters associated with a process, wherein a plurality of measurement signals are generated based on the monitoring, the method being performed by a process control system, the method comprising:

receiving the plurality of measurement signals from the plurality of measurement instruments (102);

extracting noise present in the plurality of measurement signals;

correlating the extracted noise from the plurality of measurement signals with noise extracted from a plurality of reference signals obtained without faults in the plurality of measurement instruments (102);

identifying a deviation in the one or more parameters; and

detecting a fault in at least one of the plurality of measurement instruments (102) and the process device using at least one of the associated noise and the identified deviation of the one or more parameters, wherein the detected fault is corrected for use in controlling the process in the process plant.

2. The method of claim 1, wherein correlating the plurality of extracted noises with a plurality of reference noises comprises using one or more Artificial Intelligence (AI) -based data analysis techniques.

3. The method of claim 1, wherein identifying a deviation comprises associating the one or more parameters with a predefined threshold range to determine a deviation in the one or more parameters, wherein the one or more parameters comprise at least one of: a mass of a material, an energy of the material, and a flow rate of the material.

4. The method of claim 1, wherein the detection of the fault comprises: identifying at least one of: sensor failure, sensor drift, sensor calibration problems, material leaks in the process equipment of the process plant.

5. The method of claim 1, wherein the detected fault is verified by an operator and the verified fault is used in subsequent fault detection.

6. A process control system for detecting faults in a plurality of measurement instruments (102) and process devices in a process plant, comprising:

a processor; and

a memory communicatively coupled to the processor, wherein the memory stores processor instructions that, when executed, cause the processor to:

receiving a plurality of measurement signals from the plurality of measurement instruments (102);

extracting noise present in the plurality of measurement signals;

correlating the extracted noise from the plurality of measurement signals with noise extracted from a plurality of reference signals obtained without faults in the plurality of measurement instruments (102);

identifying a deviation in the one or more parameters; and

detecting a fault in at least one of the plurality of measurement instruments (102) and the process device using at least one of the associated noise and the identified deviation of the one or more parameters, wherein the detected fault is corrected for use in controlling the process in the process plant.

7. The process control system of claim 6, wherein the processor being configured to correlate the plurality of extracted noises with a plurality of reference noises comprises: one or more Artificial Intelligence (AI) -based data analysis techniques are used.

8. The process control system of claim 6, wherein the processor being configured to identify a deviation comprises: associating the one or more parameters with a predefined threshold range to determine a deviation in the one or more parameters, wherein the one or more parameters include at least one of: a mass of a material, an energy of the material, and a flow rate of the material.

9. The process control system of claim 6, wherein the processor being configured to detect a fault comprises: identifying at least one of: sensor failure, sensor drift, sensor calibration problems, material leaks in the process equipment of the process plant.

10. The process control system of claim 6, wherein the operator verifies the detected fault and the verified fault is used in subsequent fault detection.

Technical Field

The present invention relates generally to industrial/process plants and, more particularly, to production accounting using artificial intelligence in process plants.

Background

Typically, material inventory verification/production accounting in a process plant involves verifying the actual inventory using inventory recorded in the system. Measurements from sensors associated with the process device are used to record inventory present in the process device. In practice, it is observed that there is a deviation between the recorded inventory and the actual inventory. The problems of inventory verification are mainly due to calibration problems of the sensors, leaks in the process equipment, sensor failures, drift in sensor measurements, etc. It is important to have a system for identifying and predicting faults in real time. Therefore, the manufacturing productivity is improved.

Existing solutions for detecting faults involve standard data reconciliation and coarse error detection techniques. These techniques allow for spatial redundancy, such as mass and energy balance of materials in the process plant, to be used to detect faults.

The coarse error detection technique is based on historical data. Due to the statistical nature of the algorithm, any slow drift in the average measurement instrument can be ignored and averaged.

Further, if the coarse errors are indeed outliers rather than reflecting leaks or instrument deviations, they may be averaged over a good measurement without being detected by statistical techniques (which may be erroneous due to probabilistic nature). Furthermore, some good measurements may be falsely identified as coarse errors, and therefore, the accuracy of the reconciled data may be affected.

Further, if the average measurement containing the coarse error is not eliminated and used for reconciliation, fault detection may be missed.

A problem with existing solutions is that due to the statistical nature of the algorithms, the possibility of multiple faults in the measurement instrument and process equipment may not be detected.

In view of the above, there is a need to address at least one of the above limitations and to propose a method and system to overcome the above problems.

Disclosure of Invention

In one embodiment, the present invention relates to a method and system for detecting faults in a plurality of measurement instruments and process equipment in a process plant. In one embodiment, a plurality of measurement instruments are configured to monitor one or more parameters associated with a process. In one embodiment, a plurality of measurement signals are generated based on the monitoring. In one embodiment, a process control system is configured to receive a plurality of measurement signals from a plurality of measurement instruments. Still further, the process control system is configured to extract noise present in the plurality of measurement signals. Still further, the process control system is configured to correlate the extracted noise from the plurality of measurement signals with noise extracted from the plurality of reference signals. In the absence of a fault in the plurality of measurement instruments, a plurality of reference signals are obtained. Thereafter, the process control system is configured to identify a deviation in one or more parameters. Finally, the process control system is configured to detect a fault in at least one of the plurality of measurement instruments and the process device using at least one of the associated noise and the identified deviation of the one or more parameters. The detected faults are corrected for use in controlling the process in the process plant.

In one embodiment, the process control system correlating the plurality of extracted noises to a plurality of reference noises comprises: one or more Artificial Intelligence (AI) -based data analysis techniques are used.

In one embodiment, identifying the deviation includes: the one or more parameters are correlated with a predefined threshold range to determine a deviation in the one or more parameters. Further, the one or more parameters include at least one of: mass of the material, energy of the material, and flow rate of the material.

In one embodiment, the detection of the fault includes: identifying at least one of: sensor failure, sensor drift, sensor calibration problems, material leaks in process equipment of a process plant.

In one embodiment, the detected fault is verified by an operator, and the verified fault is used in subsequent fault detection.

Systems of varying scope are described herein. In addition to the aspects and advantages described in this summary, further aspects and advantages will become apparent by reference to the drawings and by reference to the following detailed description.

Drawings

The subject matter of the invention is explained in more detail in the following text with reference to preferred exemplary embodiments which are illustrated in the drawings, in which

FIG. 1 illustrates an example environment of a process plant according to an embodiment of this disclosure;

FIG. 2 illustrates an example process control system according to an embodiment of this disclosure;

FIG. 3 illustrates an example flow diagram for detecting faults in measurement instruments and process devices in accordance with an embodiment of this disclosure;

FIG. 4 illustrates an example fault detection of a leak in a process device of a process plant according to an embodiment of this disclosure; and

FIG. 5 illustrates an example fault detection of drift in measurements of a flow sensor of a process plant according to an embodiment of this disclosure.

Detailed Description

The invention discloses a method and a system for carrying out production accounting in process industry by using artificial intelligence.

FIG. 1 illustrates an exemplary environment of a process plant (100). The process plant (100) includes one or more process devices, for example, tanks (101A, 101B) for storing materials, mixers for mixing the materials of the one or more tanks (101A, 101B), pipes for interconnecting the one or more tanks (101A, 101B) and the one or more mixers, valves for controlling the flow of materials into and out of the tanks (101A, 101B), pumps connected to the tanks (101A, 101B) for pumping materials from one tank (101A, 101B) to another, and instrumentation (102A, 102B) for monitoring one or more parameters associated with the process devices, the instrumentation (102A, 102B) including: a temperature sensor; a pressure sensor; a weight sensor for measuring the amount of material stored in the tank, the composition of one or more materials stored in the tank (e.g., 101A); and a flow meter for measuring the material flow rate. Those skilled in the art will appreciate that a process plant may include "N" tanks, which may be represented as a plurality of tanks (101A, 101N). Hereinafter, for simplicity, the plurality of canisters is designated with reference numeral 101. Reference to a particular tank is indicated using a corresponding reference numeral (e.g., (101A)). Further, those skilled in the art will appreciate that a process device may be associated with multiple measurement instruments (102A, … … 102N). Hereafter, for simplicity, the measuring instrument is denoted with reference numeral 102. Reference to a particular measurement instrument is denoted using a corresponding reference numeral (e.g., (102A)). Further, one or more measurement signals from the measurement instruments (102) are sent to a summation unit (103) for aggregating the measurement signals. The aggregated measurement signals are given to a process control system for analysis and fault detection in a measurement instrument (102) or process device.

In one embodiment, a tank (101A) in a process plant includes an inlet for receiving one or more materials from one or more tanks (101). A tank (101A) in a process plant includes an outlet for pumping material stored in the tank (101A) to one or more tanks (101) in the process plant. Further, a measurement instrument (102) for measuring one or more signals may be associated with the process device, for example, inside the process device, below the process device, or on an external surface of the process device.

In one embodiment, the aggregate signal received from the summing unit (103) is used to extract one or more parameters of the process. Further, an operator may use the extracted one or more parameters to perform data reconciliation and detect faults in the measurement instrument (102) and the process device using the process control system.

FIG. 2 illustrates an example process control system. In one embodiment, a process control system (200) may be used to implement a method for detecting faults in measurement instruments and process equipment in a process plant. The process control system (200) may include a central processing unit ("CPU" or "processor") (202). The processor (202) may include special purpose processing units such as integrated system (bus) controllers, memory management control units, floating point units, graphics processing units, digital signal processing units, and so forth. The processor (202) may be arranged to communicate with one or more input/output (I/O) devices (not shown) via the I/O interface (201). Using the I/O interface (201), the process control system (200) may communicate with one or more I/O devices. In some embodiments, a process control system (200) is connected to a service operator through a communication network (206). The processor (202) may be arranged to communicate with a communication network (206) via a network interface (203). The network interface (203) may communicate with a communication network (206). The memory (205) may store a collection of programs or database components including, but not limited to, a user interface (206), an operating system (207), a web server (208), and the like. In some embodiments, the process control system (200) may store user/application data (206), such as data, variables, records, and the like, described in this disclosure.

In one embodiment, a process control system may receive a plurality of measurement signals from one or more measurement instruments associated with process devices of a process plant. Further, the process control system extracts noise present in the plurality of measurement signals. Further, the process control system correlates the extracted noise with a plurality of noises extracted from the reference signal. In the absence of a fault, a reference signal is recorded and stored in the process control system. Thereafter, deviations of one or more parameters associated with the process are identified. Finally, the identified deviations and associated noise are used to detect faults in measurement instruments and process equipment of the process plant.

FIG. 3 illustrates an exemplary flow chart for detecting faults in a measurement instrument (102) and a process device. At step 301, a measurement instrument (102) associated with a process device of a process plant monitors one or more parameters. The process control system receives a plurality of measurement signals from one or more measurement devices via a summing unit (103). A summing unit (103) aggregates a plurality of signals from one or more measurement devices.

At step 302, the process control system extracts noise present in the plurality of measurement signals. Noise extraction is accomplished by standard signal processing techniques.

At step 303, the extracted noise is correlated with noise from multiple reference signals. Further, the correlation of the plurality of extracted noises with the plurality of noises from the reference signal is accomplished using one or more Artificial Intelligence (AI) -based data analysis techniques (e.g., time series analysis). In the absence of a fault in the process plant, a plurality of reference signals are obtained and stored in the process control system. The plurality of reference signals are stored based on manual verification done by an operator. An example is detailed in fig. 3 in the later description.

In one embodiment, periodic measurements of a plurality of measurement signals from one or more measurement instruments (102) have an inherent autocorrelation. The autocorrelation indicates a similarity between the plurality of measurement signals and the delayed plurality of measurement signals. Any faults associated with one or more measurement instruments (102) or process devices are reflected in the noise associated with the corresponding measurement. Thus, the autocorrelation in the noise of the plurality of measurement signals changes or is affected. Further, such changes that identify the correlation of noise in the plurality of measurement signals are used to verify faults in the process device or the one or more measurement instruments (102).

At step 304, the process control system identifies a deviation in one or more parameters. Process plants typically use closed loop control systems to maintain a desired quality or yield of a product. In a closed loop control system, there is a clear correlation between a fault in some measurement signals and its effect on one or more other parameters (associated with the process of the process plant). An example is detailed in fig. 4 in the later description. This association can affect the desired quality or yield of the product. Accordingly, deviations with respect to one or more parameters associated with the process are identified based on the association.

In one embodiment, identifying a deviation in one or more parameters comprises: one or more parameters are associated with a predefined threshold range. The threshold range of a process device may indicate a maximum and a minimum amount of material stored in the process device or a maximum and a minimum amount of material flow from one process device to another process device. The predefined threshold ranges may vary from one process device to another and from one process plant to another. The one or more parameters may include at least one of: mass of the material, energy of the material, and flow rate of the material.

Further, in one embodiment, Artificial Intelligence (AI) -based data analysis techniques (e.g., time series analysis) may be used to identify deviations in one or more parameters of the process plant.

At step 305, the process control system detects a fault in the measurement instrument (102) or the process device using the one or more associated noises at step 303 and the identified deviation at step 304. The process control system may use standard statistical techniques (e.g., kalman filtering and principal component analysis to detect outliers) to detect faults.

In one embodiment, a fault detected by the process control system is verified by an operator. Based on the faults detected by the process control system, an operator may manually verify or authenticate the faults in the process plant and the authentication is updated to the process control system. Based on the verification updated by the operator, the process control system may increase the probability of fault detection by incorporating appropriate learning of the AI techniques used at steps 303 and 304.

FIG. 4 illustrates an example fault detection of a leak in a process device of a process plant. The tank (e.g., 101A) is connected to one or more tanks (101D and 101F). Further, as shown in fig. 4, the tank 101D is connected to 101G and 101H, and the tank 101F is connected to 101H and 101I. A measurement instrument (e.g., 102A) associated with a tank (e.g., 101A) measures a plurality of signals and sends them to a process control system for fault detection. Let there be a leak in the flow from tank 101C to tank 101E (401). The leak (401) affects the mass balance between the flow from the tank 101C to the tank 101E (further, the flow from the tank 101E to the tank 101H and the flow from the tank 101E to the tank 101I). Further, the leak (401) affects the accumulation of material in the tank 101E. Noise extracted across one or more measurement signals during a leak (401) is correlated with noise extracted from a reference signal obtained without a fault or leak using one or more AI-based data analysis techniques. For example, due to the leak (401), the noise correlation in the flow from the tank 101E to the tank 101H and the tank 101I may be higher. Thus, the obtained noise correlation, along with the traditional data reconciliation, identifies a fault or leak (401).

FIG. 5 illustrates an example fault detection of drift in measurements of a flow sensor of a process plant. The tank (e.g., 101A) is connected to one or more tanks (101D and 101F). Further, as shown in fig. 5, the tank 101D is connected to 101G and 101H, and the tank 101F is connected to 101H and 101I. A measurement instrument (e.g., 102A) associated with a tank (e.g., 101A) measures a plurality of signals and sends them to a process control system for fault detection. There is a drift in the signal measured by the flow sensor (501) associated with the flow from the tank 101A to the tank 101D. This results in less material for formulation in tank 101H. To achieve the desired quality or yield of product, the flow rate from tank 101E to tank 101H may be higher to compensate for less material in tank 101H for formulation. Based on the closed loop system analysis, the process control system identifies deviations in the measured flow rates of the material from the tanks 101A and 101E by comparing the measured flow rate of the flow sensor (501) to a predefined threshold range. Thus, the identified deviations in the measured flow, along with the traditional data reconciliation, identify faults or drift in the flow sensor (501).

This written description uses examples to describe the subject matter herein, including the best mode, and also to enable any person skilled in the art to make and use the subject matter. The patentable scope of the subject matter is defined by the claims, and may include other examples that occur to those skilled in the art. Such other examples are intended to be within the scope of the claims if they have structural elements that do not differ from the literal language of the claims, or if they include equivalent structural elements with insubstantial differences from the literal languages of the claims.

Reference numerals

101-tank;

102-a measuring instrument;

103-a summing unit;

200-a process control system;

201-I/O interface;

202-a processor;

203-network interface;

204-a storage interface;

205-a memory;

206-a user interface;

207-an operating system;

208-a web server;

206-a communication network;

210-an input device;

211-an output device;

212 — a remote device;

401 — leakage;

501-flow sensor.