阅读说明：本技术 用于检测与过程工厂中的气体色谱装置关联的故障的方法和控制系统 (Method and control system for detecting faults associated with gas chromatography devices in a process plant ) 是由 J·古加利亚 A·K·古普塔 于 2019-08-20 设计创作，主要内容包括：本申请公开用于检测与过程工厂中的气体色谱装置(100)关联的故障的方法和服务器(111)。气体色谱装置(100)与数据库(110)关联,该数据库(110)配置成存储所测量的色谱图和历史色谱图。最初,服务器从数据库接收所测量的色谱图。在接收所测量的色谱图时,服务器配置成检测用于所测量的色谱图的至少一个实时症状。可通过将具有预定配置数据的历史色谱图与所测量的色谱图进行比较来检测实时症状。在检测实时症状时,服务器配置成确定与气体色谱装置关联的故障。通过映射从数据库所接收的实时症状和故障签名数据来确定故障。使用机器学习模型来生成故障签名数据,该机器学习模型通过提供故障和历史气体色谱图来训练。(The present application discloses a method and a server (111) for detecting a fault associated with a gas chromatography device (100) in a process plant. The gas chromatography device (100) is associated with a database (110), the database (110) being configured to store measured chromatograms and historical chromatograms. Initially, the server receives the measured chromatogram from the database. Upon receiving the measured chromatogram, the server is configured to detect at least one real-time symptom for the measured chromatogram. Real-time symptoms can be detected by comparing historical chromatograms with predetermined configuration data with measured chromatograms. Upon detecting the real-time symptom, the server is configured to determine a fault associated with the gas chromatography apparatus. The fault is determined by mapping the real-time symptom and fault signature data received from the database. Fault signature data is generated using a machine learning model that is trained by providing fault and historical gas chromatograms.)

1. A method for detecting, by at least one server (111), a fault associated with a gas chromatography device (100) in a process plant, the at least one server (111) being operatively coupled with the gas chromatography device (100), wherein the gas chromatography device (100) is further associated with at least one database (110), the at least one database (110) being configured to store at least one of historical gas chromatograms and real-time gas chromatograms derived from the gas chromatography device (100) by performing one or more gas chromatographs of a gas mixture, the method comprising:

receiving a real-time gas chromatogram for a real-time gas chromatogram performed on the gas mixture from the at least one database (110), wherein the real-time gas chromatogram is a graph representing values of composition of each of a plurality of compounds and retention times associated with corresponding compounds in a predefined sequence;

detecting at least one real-time symptom for the real-time gas chromatogram using predetermined configuration data associated with the real-time gas chromatogram, wherein the at least one real-time symptom is indicative of a change in the graph relative to the predetermined configuration data; and

determining one or more faults from a plurality of faults associated with the gas chromatography device (100) using the at least one real-time symptom and fault signature data, wherein the fault signature data is generated using a machine learning model trained by providing a historical gas chromatogram received from the at least one database (110) and the plurality of faults, wherein the historical gas chromatogram is derived for one or more previous gas chromatograms of the gas chromatography device (100), wherein the fault signature data indicates an association of each of the plurality of faults with a plurality of symptoms.

2. The method of claim 1, further comprising determining a confidence score for each of the one or more faults, wherein the confidence score indicates a probability of occurrence of the corresponding fault in the gas chromatography apparatus (100).

3. The method of claim 1, wherein the predetermined configuration data indicates an exact value of a composition of each of the plurality of compounds and the holding time.

4. The method of claim 1, wherein detecting the at least one real-time symptom comprises:

detecting a plurality of peaks in the graph;

comparing the plurality of peaks to the predetermined configuration data to identify at least one of normal peaks from the plurality of peaks and one or more false peaks in the graph; and

detecting the at least one real-time symptom associated with a peak other than the normal peak using one or more image processing techniques.

5. The method of claim 1, wherein determining the one or more faults comprises mapping the at least one real-time symptom from the plurality of symptoms in the fault signature data to the plurality of faults of the fault signature data, wherein the fault signature data is a lookup table comprising rows and columns as the plurality of faults and the plurality of symptoms.

6. A control system for detecting a fault associated with a gas chromatography device in a process plant, wherein the control system comprises at least one database and at least one server (111) operatively coupled with the gas chromatography device (100), wherein the at least one database (110) is configured to store at least one of historical gas chromatograms and real-time gas chromatograms derived from the gas chromatography device (100) by performing one or more gas chromatographs of a gas mixture, wherein the at least one server (111) is configured to:

receiving a real-time gas chromatogram for a real-time gas chromatogram performed on the gas mixture from the at least one database (110), wherein the real-time gas chromatogram is a graph representing values of composition of each of a plurality of compounds and retention times associated with corresponding compounds in a predefined sequence;

detecting at least one real-time symptom for the real-time gas chromatogram using predetermined configuration data associated with the real-time gas chromatogram, wherein the at least one real-time symptom is indicative of a change in the graph relative to the predetermined configuration data; and

determining one or more faults from a plurality of faults associated with the gas chromatography device (100) using the at least one real-time symptom and fault signature data, wherein the fault signature data is generated using a machine learning model trained by providing a historical gas chromatogram received from the at least one database (110) and the plurality of faults, wherein the historical gas chromatogram is derived for one or more previous gas chromatograms of the gas chromatography device (100), wherein the fault signature data indicates an association of each of the plurality of faults with a plurality of symptoms.

7. The control system of claim 6, wherein the at least one server (111) is further configured to:

determining a confidence score for each of the one or more faults, wherein the confidence score indicates a probability of occurrence of the corresponding fault in the gas chromatography device (100).

8. The control system of claim 6, wherein the predetermined configuration data indicates an exact value of the composition of each of the plurality of compounds and the holding time.

9. The control system of claim 6, wherein the at least one server (111) detects the at least one real-time symptom by performing the steps of:

detecting a plurality of peaks in the graph;

comparing the plurality of peaks to the predetermined configuration data to identify at least one of normal peaks from the plurality of peaks and one or more false peaks in the graph; and

detecting the at least one real-time symptom associated with a peak other than the normal peak using one or more image processing techniques.

10. The control system of claim 6, wherein the at least one server (111) determines the one or more faults by mapping the at least one real-time symptom from the plurality of symptoms in the fault signature data with the plurality of faults of the fault signature data, wherein the fault signature data is a lookup table comprising rows and columns as the plurality of faults and the plurality of symptoms.

Technical Field

The present disclosure relates generally to gas chromatography devices in process plants. More particularly, the present disclosure relates to a method for detecting faults associated with Gas Chromatography (GC) devices in a process plant and a server thereof.

Background

A process plant (e.g., a nuclear power plant, a chemical plant, a natural gas plant, a refinery, etc.) includes a GC device configured to analyze a gas/liquid mixture. By analysis, the composition of the compounds in the gas mixture is measured. This measurement is critical for certain processes due to the presence of harmful gases. Moreover, evaluation (evaluation) of gas mixtures may require such measurements. Ensuring the accuracy of the measurement along with the measurement may also be crucial. Repeatability of results in measurements for predefined analyses forms an industrial specification in stating the operating conditions of the GC apparatus. Repeatability may account for deviations or errors in consistency. Recalibration in the GC apparatus may be performed to correct for deviations and errors in order to ensure the correctness of the results. However, there may be other inconsistencies or deviations from the normal behavior of the GC device that may not be detected. Moreover, generating confidence for the measured results may be difficult due to such inconsistencies or deviations. In addition, recalibration and other checks of the GC device may be performed in a scheduled manner to detect consistency deviations. However, such detection of defects or faults in the GC apparatus may be delayed, and thus errors in the measurements may be propagated until such detection and correction.

Disclosure of Invention

The present disclosure discloses a method for detecting a fault associated with a GC device in a process plant. The method is performed by at least one server operatively coupled with the GC apparatus. The GC device is further associated with at least one database configured to store at least one of real-time gas chromatograms and historical gas chromatograms. Historical gas chromatograms and real-time gas chromatograms are derived from a GC apparatus by performing one or more gas chromatograms of a gas mixture. The at least one server is configured to communicate with the at least one database to receive the real-time gas chromatogram and the historical gas chromatogram and to detect a failure.

Initially, at least one server is configured to receive a real-time gas chromatogram from at least one database. The real-time gas chromatogram can be obtained by performing a real-time gas chromatogram for the gas mixture in a GC apparatus. The real-time gas chromatogram is a graph representing values of composition of each of a plurality of compounds in a gas mixture and retention times (retentions times) associated with the corresponding compounds in a predefined sequence.

Upon receiving the real-time gas chromatogram, the at least one server is configured to detect at least one real-time symptom for the real-time gas chromatogram. At least one real-time symptom may be detected by comparing the graph to predetermined configuration data associated with the real-time gas chromatogram. The at least one real-time symptom indicates a change in the graph relative to the predetermined configuration data. The predetermined configuration data indicates an exact value and a retention time of the composition of each of the plurality of compounds. In an embodiment, the at least one real-time symptom is detected by initially detecting a plurality of peaks in the graph. The plurality of peaks is compared to predetermined configuration data to identify at least one of normal peaks from the plurality of peaks and one or more false peaks (ghost peaks) in the graph. Further, one or more image processing techniques are used to determine at least one real-time symptom associated with a peak other than a normal peak.

Upon detecting the at least one symptom, the at least one server is configured to determine one or more faults from a plurality of faults associated with the GC device. One or more faults are determined using the at least one real-time symptom and the fault signature data. Fault signature data is generated using a machine learning model trained by providing historical gas chromatograms and a plurality of faults received from at least one database. The historical gas chromatogram is obtained by performing one or more previous gas chromatograms in the GC apparatus. The fault signature data indicates an association of each of the plurality of faults with a plurality of symptoms. In an embodiment, the fault signature data is a look-up table comprising rows and columns as the plurality of faults and the plurality of symptoms. One or more faults may be determined by mapping at least one real-time symptom from the plurality of symptoms in the fault signature data with the plurality of faults of the fault signature data.

Further, a confidence score is determined for each of the one or more determined faults. The confidence score indicates the probability of occurrence of the corresponding fault in the GC device.

Drawings

FIG. 1 illustrates a block diagram of a system for detecting faults associated with GC devices in a process plant, the system including GC devices, at least one server, and at least one database, according to an embodiment of the present disclosure;

FIG. 2 shows an exemplary representation of a gas chromatogram generated by a GC apparatus, representing the composition and retention time of each of a plurality of compounds in a gas mixture, in accordance with an embodiment of the disclosure;

figure 3a illustrates an exemplary representation of predetermined configuration data for a GC apparatus, representing precise values and retention times of the composition of each of a plurality of compounds in a gas mixture, in accordance with an embodiment of the disclosure;

fig. 3b and 3c show exemplary representations of real-time gas chromatograms from a GC apparatus, in accordance with embodiments of the disclosure;

FIGS. 4a and 4b illustrate exemplary representations of at least one symptom indicative of a change in a real-time gas chromatogram relative to predetermined configuration data, in accordance with an embodiment of the present disclosure; and

FIG. 5 illustrates a flow diagram for detecting a fault associated with a GC device in a process plant according to an embodiment of the disclosure.

Detailed Description

The present disclosure discloses a method and control system for detecting faults associated with GC devices in a process plant. The control system includes at least one database and at least one server operatively coupled to the GC apparatus. At least one server is configured to perform the method steps of the present invention for detecting a failure in a GC apparatus. The at least one database may be configured to store at least one of historical gas chromatograms and real-time gas chromatograms of the GC device. The real-time gas chromatogram can be the output of a gas chromatogram performed in real-time by a GC apparatus. The historical gas chromatogram may comprise one or more gas chromatograms, which may be the output of one or more previous gas chromatograms performed by the GC apparatus. In an embodiment, at least one server may be configured to communicate with at least one database to receive real-time gas chromatograms and historical gas chromatograms and to detect faults. In real time, the real-time gas chromatogram is received and analyzed to detect at least one symptom in the real-time gas chromatogram. Further, one or more faults associated with the GC device are determined based on the at least one symptom associated with the GC device and predetermined configuration data. A confidence score indicative of the probability of occurrence of each of the determined one or more faults may also be determined by the at least one server. By the proposed invention, the accuracy of detection of a fault in a GC apparatus can be obtained. Further, the present invention proposes to perform detection of a fault in real time in a GC apparatus, thereby avoiding propagation of an error due to the fault.

The process plant may be a chemical plant, oil refinery, natural gas plant, etc. that includes a plurality of equipment. At least one device from the plurality of devices may be operable with a gas mixture comprising a plurality of compounds. In an embodiment, the gas mixture may include a predefined concentration (also referred to as composition) of each of a plurality of compounds. The GC apparatus may be coupled to at least one device that injects the gas mixture. The GC apparatus may be configured to extract the gas mixture from the at least one device to measure the composition of each of the plurality of compounds. Fig. 1 illustrates a block diagram of a factory environment including a GC apparatus 100 and a control system associated with the GC apparatus 100 for measuring faults. The control system further comprises at least one server 111 and at least one database 110. The GC apparatus 100 may include a process line 101, a probe 102, a bypass filter 103, a GC oven (oven) 104, and a GC controller 109 along with multiple inlets and multiple outlets. The GC apparatus 100 may be configured to measure the composition of a plurality of compounds. GC oven 104 further includes sample valve 105, separation column 106, detector 107, and heater system 108. A gas mixture from at least one device in a process plant may be extracted via a process line 101 and a probe 102. In an embodiment, the bypass filter 103 may be configured to filter contaminants in the gas mixture. In an embodiment, the bypass filter 103 may be placed in the sample system of the GC apparatus 100. The contaminant and residual gas mixture from the bypass filter 103 may be ejected as a gas mixture back (return) via one of a plurality of outlets in the GC apparatus 100. Further, the gas mixture is injected into the GC oven 104 while being filtered. A sample valve 105 in GC oven 104 may be configured to combine the gas mixture with a carrier gas and pass the combination through a separation column 106. The carrier gas may be injected into the sample valve 105 via one of a plurality of inlets in the GC apparatus. The carrier gas is used in the mobile phase of the gas chromatography performed in the GC apparatus 100. In an embodiment, the carrier gas may be injected manually by a positive displacement injector (not shown). The carrier gas may be one of helium, nitrogen, argon, hydrogen, air, and the like. In embodiments, any carrier gas known in the art may be used in the present invention. In addition, the gas mixture entering the separation column 106 interacts with the coated walls of the separation column 106. The processing of the gas mixture inside the separation column 106 may be referred to as the stationary phase of the gas chromatography performed by the GC apparatus 100. In an embodiment, a separation column may be associated with the heater system 108 to supply heat. The heat may facilitate volatilization of the gas mixture.

In an embodiment, the separation column 106 performs a chemical process on the stationary phase, thereby causing elution of different molecules in the gas mixture. Thereby, a separation of each of the plurality of compounds in the gas mixture may be achieved. The separation of the plurality of compounds causes each of the plurality of compounds to be injected at a different time from the separation column 106 according to a predefined sequence. In an embodiment, the separation of the gas mixture may be achieved due to differences in partition (partioning) behavior between the mobile gas phase and the stationary phase in the separation column 106. In embodiments, the separation of the plurality of compounds may be based on different strengths of interactions associated with each of the plurality of compounds in the stationary phase.

The separated gas mixture may be injected into detector 107 of GC oven 104. The detector 107 may be configured to measure the composition of each of a plurality of compounds in the gas mixture. The detector 107 in the GC oven 104 may be one of a thermal conductivity detector, a flame ionization detector, a discharge ionization detector, a pulse ionization detector, or the like. In embodiments, one or more detectors known in the art (which may be configured to measure the composition of a plurality of compounds) may be implemented as a detector in the present invention.

The output of the detector 107 may be provided to a GC controller 109, which GC controller 109 may be configured to plot the composition and retention time of each of the plurality of compounds to output a gas chromatogram. An exemplary representation of a gas chromatogram 200 is illustrated in fig. 2. The gas chromatogram 200 may be a graph representing values of the composition of each of a plurality of compounds in a gas mixture. The values of the compositions may be plotted against the retention time of the plurality of compounds. The holding time for each of the plurality of compounds may be a length of time of holding of the corresponding compound in the predefined sequence. As illustrated in fig. 2, each of the plurality of peaks in the graph corresponds to a compound from the plurality of compounds. In the examples, the height of the peaks indicates the value of the composition of the compound.

In an embodiment of the present invention, the historical gas chromatogram may be derived by performing one or more previous gas chromatograms in the GC apparatus 100. The historical gas chromatogram associated with each of the one or more previous gas chromatograms may be similar to the gas chromatogram 200 illustrated in fig. 2. The historical gas chromatogram can be stored in at least one database 110. In an embodiment, when performing one or more gas chromatographs, the GC apparatus 100 can be configured to detect errors and generate an alarm based on the errors. The evaluation of the gas mixture may not take into account the results of such one or more gas chromatographs. In the present invention, gas chromatograms from such one or more gas chromatograms can be stored as historical gas chromatograms in at least one database 110.

In real-time, the GC apparatus 100 may be configured to perform real-time gas chromatography as previously described and generate a real-time gas chromatogram for the input gas mixture. The real-time gas chromatogram may be similar to that of the gas chromatogram 200 illustrated in fig. 2. The real-time gas chromatogram can be stored in at least one database 110. At least one server 111 of the present invention may be configured to communicate with at least one database 110 to receive real-time gas chromatograms and historical gas chromatograms. In an embodiment, the at least one database 110 may include a first database and a second database (not shown in the figures). The first database may be configured to store historical gas chromatograms and the second database may be configured to store real-time gas chromatograms. In an embodiment, at least one server may be in communication with a first database to receive real-time gas chromatograms and may be in communication with a second database to receive historical gas chromatograms.

In an embodiment, the at least one server 111 may comprise a processor and a memory (not shown in the figures) for detecting a failure in the GC apparatus 100. The memory may be operatively coupled to the processor. The memory may include modules and data that, when executed, may cause the processor to perform the detection of faults as disclosed in the present invention. In an embodiment, the at least one database 110 may be an integral part of the at least one server 111. In an embodiment, at least one server 111 may be in direct communication with the GC device 100 to receive the real-time gas chromatogram and the historical gas chromatogram. In an embodiment, the at least one server 111 may be a cloud-based server that may be connected to the GC apparatus 100 via any communication network known to those skilled in the art. In an embodiment, the at least one server 111 may be a dedicated server embedded in the GC apparatus 100. In an embodiment, the at least one server 111 may be configured to communicate with a plurality of GC devices to detect a failure in each of the plurality of GC devices.

In the present invention, in real-time, at least one server 111 may be configured to receive the real-time gas chromatogram to detect at least one real-time symptom for the real-time gas chromatogram. At least one real-time symptom can be detected by comparing the real-time gas chromatogram to predetermined configuration data associated with the real-time gas chromatogram. In an embodiment, the predetermined configuration data may indicate an exact value and a holding time for the composition of each of the plurality of compounds. In an embodiment, the predetermined configuration data may also be referred to as a GC configuration. In an embodiment, the predetermined configuration data may be generated based on a priori knowledge about the gas chromatography performed by the GC apparatus 100. In an embodiment, the predetermined configuration data may be generated based on existing troubleshooting knowledge related to the GC apparatus 100.

In an embodiment, a user (which may be an operator of the GC apparatus 100, an analyst of the gas chromatogram, or anyone with said knowledge) may input predetermined configuration data to the at least one server 111. In an embodiment, the predetermined configuration data may be a numerical value of the composition of the plurality of compounds. In an embodiment, a user may provide predetermined configuration data in one or more forms known in the art to at least one server 111. In an embodiment, at least one server 111 may be configured to convert the numerical values of the components into a graphical form. In an embodiment, a user may enter a graphical form of predetermined configuration data into at least one server 111. Fig. 3a shows an exemplary representation of predetermined configuration data 300a in graphical form. Each of the plurality of peaks in the predetermined configuration data 300a corresponds to a respective compound from the plurality of compounds. In an embodiment, the height of each of the plurality of peaks indicates a value of the composition of the respective compound, and the width of each of the plurality of peaks indicates a retention time associated with the respective compound in the predefined sequence.

The at least one server 111 may be configured to detect the at least one symptom by comparing the real-time gas chromatogram to predetermined configuration data. The at least one symptom may be an abnormality or deviation or change in the gas chromatogram. In embodiments, the at least one symptom may be detected using one or more techniques known to those skilled in the art. The at least one symptom may be attributable to at least one of baseline perturbations, irregularities in the shape and size of the peak, the appearance of false peaks, loss of resolution, difficulty in quantitative calculations, wide solvent front, and the like. One or more false peaks may be additional peaks not present in the predetermined configuration data. In an embodiment, one or more ghost peaks may also be referred to as phantom peaks. In embodiments, loss of resolution may be detected when an overlap of peaks occurs in the gas chromatogram.

In an embodiment, to detect the at least one symptom, the at least one server 111 may be configured to perform peak identification in the real-time gas chromatogram. May be implemented as one or more techniques known to those skilled in the art for detecting multiple peaks. One of the techniques may include: iteratively finding the highest point in the graph of the gas chromatogram; fitting a Gaussian curve to the highest point; looking for the residue from the fit to remove peaks; and the process is repeated. Other techniques may include, but are not limited to, finding all peaks using both zero crossings in the first derivative and thresholding in the data, finding peaks using transitions from steady state to transient, finding local maxima and peaks using wavelet convolution, finding peak shaping data points using a pattern matching logical classifier. In an embodiment, an adaptive filter using a quadratic Savitsky-Golay moving window filter may be used to detect multiple peaks. The window may have a predefined scale. In an embodiment, the predefined scale may be a width of a largest peak from the plurality of peaks. Further, the plurality of peaks is compared to predetermined configuration data to identify at least one of normal peaks from the plurality of peaks and one or more false peaks in the graph. This helps identify normal behavior in the chart. In an embodiment, a matching algorithm may be implemented for identifying at least one of one or more false peaks and normal peaks. Further, at least one real-time symptom associated with a peak other than the normal peak is determined. In an embodiment, the median and mode of peaks other than normal peaks are determined to identify the reference and transition points in the peaks. This aids in further filtering of the abnormal peaks. A bell-shaped curve fit is performed on the anomalous peak to extract the variation in the anomalous peak or peak characteristics. One or more techniques may be used to detect at least one symptom. Techniques may include, but are not limited to, spike detection methods to examine higher order derivatives in the signal, use of deviations from gaussian curves, kurtosis, etc. in skew indices to detect the most irregular peaks, use of the frequency response of the signal, and noise characterization of changes along a baseline.

It is contemplated that a real-time gas chromatogram associated with the predetermined configuration data 300a illustrated in fig. 3a may be performed. Further, consider a real-time gas chromatogram 300b as will be illustrated in fig. 3 b. To detect at least one symptom, the real-time gas chromatogram 300b may be compared to predetermined configuration data 300 a. Changes between the real-time gas chromatogram 300b and the predetermined configuration data 300a may be detected as at least one symptom. For real-time gas chromatogram 300b, peak 301 and peak 302 may be detected as at least one symptom. Similarly, consider a real-time gas chromatogram 300b as will be illustrated in fig. 3 c. To detect at least one symptom, the real-time gas chromatogram 300c may be compared to predetermined configuration data 300 a. Changes between the real-time gas chromatogram 300c and the predetermined configuration data 300a may be detected as at least one symptom. For real-time gas chromatogram 300c, peak 303 may be detected as at least one symptom.

Fig. 4a and 4b show exemplary representations of a plurality of symptoms indicative of changes in a real-time gas chromatogram relative to predetermined configuration data. Fig. 4a shows a symptom that may be detected as a result of a baseline perturbation. Symptom 401.1 may occur due to spiking of values in the real-time chromatogram. Symptom 401.2 may occur due to the addition of noise in the real-time chromatogram. Symptom 401.3 may occur due to a meandering (wandering) of values in the real-time chromatogram. Symptom 401.4 may occur due to a drift in values in the real-time chromatogram. Symptom 401.5 may occur due to a shift in values in the real-time chromatogram. Symptom 401.6 may occur due to a slippage of values in the real-time chromatogram. Fig. 4b shows symptoms that can be detected due to irregularities in the peaks. Symptom 401.7 may occur due to tailing (tailing) of peaks in the real-time chromatogram. Symptom 401.8 may occur due to tongue stretching (front) of the peak in the real-time chromatogram. Symptom 401.9 may occur due to splitting of peaks in the real-time chromatogram. Symptom 401.10 may occur due to a reduction in peaks in the real-time chromatogram. Symptom 401.11 may occur due to rounding of peaks in the real-time chromatogram. Symptom 401.12 may occur due to inversion of peaks in the real-time chromatogram. One or more other deviations or anomalies known to those skilled in the art may be inferred by the at least one server 111 as a plurality of symptoms.

Upon detecting the at least one symptom, the at least one server 111 is configured to determine one or more faults from a plurality of faults associated with the GC apparatus 100.

A plurality of faults associated with the GC apparatus 100 may be associated with at least one of the separation column 106, the sample, the temperature, the detector 107, the sample valve 105, the carrier gas, the membrane, the gas mixture stream, the injector, the GC controller 109, the setup of the GC apparatus 100, etc. The one or more faults may be caused by one or more factors. One or more faults may be determined using the at least one real-time symptom and fault signature data. In an embodiment, the at least one server may be configured to perform the mapping of the at least one symptom to the failure signature data. The mapping may be performed by various methods, such as one-to-one or many-to-many mapping. In a preferred embodiment, the many-to-many mapping may be performed using a machine learning procedure.

Further, the fault signature data may be generated using a machine learning model trained by providing historical gas chromatograms and a plurality of faults received from the at least one database 110. The fault signature data indicates an association of each of a plurality of faults with a plurality of symptoms associated with the GC apparatus 100. In embodiments, the machine learning model may be one of a neural network, a decision tree, a shopping basket analysis, or the like. Using the machine learning model, each of a plurality of symptoms is appended to the induced fault. This can be characterized as signatures of different causal faults. A list of such causal faults with their data signatures is generated from the machine learning model. May be implemented as one or more techniques known to those skilled in the art for training a machine learning model.

In an embodiment, various troubleshooting manuals may be used for the gas chromatograph of the GC apparatus 100 for generating relationships between different kinds of problems and various symptoms associated with the problems.

In an embodiment, the generated fault signature data may be quantized in a look-up table. An exemplary representation of fault signature data may be as shown in table 1 below:

component part	Noise(s)	Reduction of the size of the peak	Tongue and tail extension in the peak	Overlap of peaks	Shift of hold time
						Contamination of the separation column	1	1	1	1	1
Fluctuations in the sample	0	1	0	0	0
						Fluctuations in temperature	0	1	1	1	1
Contamination of the detector	1	0	0	0	0
						Less gas mixture flow	0	0	0	0	1

TABLE 1

According to table 1, the multiple failures associated with the GC apparatus 100 may be contamination of the separation column 106, fluctuations in the sample, fluctuations in temperature, contamination of the detector 107, and a lesser flow of gas mixture. Consider that a value of "1" in the lookup table for a fault and symptom indicates that the fault is associated with the symptom, and a value of "0" in the lookup table for a fault and symptom indicates that the fault is not associated with the symptom. Thus, when the at least one symptom includes noise, a reduction in the size of the peak, a tongue and tail of the peak, an overlap of the peaks, and a shift in retention time, contamination of the separation column 106 may be determined. Fluctuations in the sample may be determined when the at least one symptom comprises a decrease in the size of the peak. Fluctuations in temperature may be detected when the at least one symptom includes a reduction in the size of the peak, an elongation and tailing of the peak, an overlap of the peaks, and a shift in retention time. When the at least one symptom includes noise, contamination of the detector 107 may be detected. When the at least one symptom includes a shift in hold time, a lesser flow of the gas mixture may be determined. Further, a confidence score is determined for each of the one or more determined faults. It is noted by those skilled in the art that while the binary value (0, 1) is currently used to describe table 1, table 1 may also include a score.

Consider that the at least one symptom detected by the at least one server 111 will be a reduction in the size of the peak, a stretching and tailing of the peak, an overlap of the peaks, and a shift in retention time. By mapping the detected at least one symptom with a plurality of faults in a look-up table, one or more faults may be determined as fluctuations in the sample, fluctuations in temperature, and a lesser flow of the gas mixture. Similarly, consider that the at least one symptom detected by the at least one server 111 will be an overlap of noise and peaks. By mapping the detected at least one symptom with a plurality of faults in a look-up table, one or more faults may be determined as contamination of the separation column 106, fluctuations in temperature and contamination of the detector 107. When the detected at least one symptom is associated with a fault, the fault may be determined to be part of one or more faults.

In an embodiment, at least one server 111 may be configured to determine a confidence score for each of the one or more determined faults. In an embodiment, the confidence score indicates the probability of occurrence of the corresponding fault in the GC apparatus 100. In an embodiment, a confidence score may be calculated based on the association of the at least one detected symptom with the determined one or more faults. According to the previous example, considering the overlap of noise and peaks as at least one symptom, one or more faults are determined as contamination of the separation column 106, fluctuations in temperature and contamination of the detector 107. A confidence score associated with each of the one or more faults, including contamination of the separation column 106, fluctuations in temperature, and contamination of the detector 107, is determined. The confidence score for contamination of the separation column 106 may be 20%, the confidence score for fluctuations in temperature may be 30%, and the confidence score for contamination of the detector 107 may be 50%. May be implemented as one or more techniques known to those skilled in the art for determining a confidence score. In embodiments, a manual check or automatic recalibration can be initiated on the GC apparatus 100 based on one or more fault and confidence scores.

Fig. 5 illustrates a flow diagram for detecting a fault associated with a GC apparatus 100 in a process plant according to an embodiment of the present disclosure.

At block 501, at least one server 111 may be configured to receive a real-time gas chromatogram from at least one database 110. The real-time gas chromatogram can be obtained by performing real-time gas chromatography for the gas mixture in the GC apparatus 100. The real-time gas chromatogram may be a graph representing values of composition of each of a plurality of compounds in a gas mixture versus retention time associated with the corresponding compounds in a predefined sequence.

At block 502, at least one server 111 may be configured to detect at least one real-time symptom for a real-time gas chromatogram. At least one real-time symptom may be detected by comparing the graph to predetermined configuration data associated with the real-time gas chromatogram. The at least one real-time symptom indicates a change in the graph relative to the predetermined configuration data.

At block 503, at least one server 111 may be configured to determine one or more faults from a plurality of faults associated with the GC device 100. The one or more faults may be determined using at least one real-time symptom and fault signature data received from the at least one database 110. The fault signature data may be generated using a machine learning model that is trained by providing a plurality of fault and historical gas chromatograms.

Embodiments of the present invention provide a systematic assessment of correctness and confidence in the results of the GC apparatus 100. This enables early corrective action that helps accurate, consistent, and reliable GC measurements.

Embodiments of the present invention provide greater confidence in the output of the GC apparatus 100.

This written description uses examples to disclose 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

。