Measurement device

Abstract

This measurement device 1 measures at least one type of parameter indicating the state of a fluid, and is provided with: a detection unit 10 for acquiring, as data, at least one type of sensor value necessary for the calculation of the parameter; and a calculation unit (30) that calculates a parameter on the basis of the data acquired by the detection unit (10), the calculation unit (30) acquiring a result of a regression process and / or a discrimination process relating to the parameter, the regression process and / or the discrimination process being executed using a learning model pre-constructed on the basis of the data when the mixture is mixed into the fluid and the acquired data.

Description

[0001] Cross-references between related applications

[0002] This application claims priority from Japanese Patent Application No. 2022-207533 filed in Japan on December 23, 2022, the disclosure of which is incorporated herein by reference in its entirety. Technical Field

[0003] The present invention relates to a measuring device. Background Art

[0004] Currently, technologies related to measuring devices for measuring parameters indicating the state of a fluid, including flow rate and density, are known. For example, Patent Document 1 discloses a field instrument that measures fluid state quantities and is capable of obtaining relatively accurate measurement values even when slag flow occurs.

[0005] Patent Document 1: Japanese Patent No. 6608396 Summary of the Invention

[0006] However, conventional techniques make it difficult to accurately measure parameters if impurities such as bubbles enter the fluid, potentially leading to errors in the measured parameter values. The field instrument described in Patent Document 1 addresses this problem. However, while this field instrument can be applied to fluids whose parameters remain roughly constant when no impurities are present, it is difficult to apply to fluids whose parameters vary over time due to operation, such as in a distillation process.

[0007] An object of the present invention is to provide a measuring device capable of more accurately measuring parameters even when a foreign substance is mixed into a fluid while the parameters of the fluid are changing with time.

[0008] Several embodiments involve a measuring device that measures at least one parameter representing the state of a fluid, and comprises: a detection unit for obtaining at least one sensor value required for calculating the parameter as data; and a calculation unit that calculates the parameter based on the data obtained using the detection unit, the calculation unit obtaining the result of at least one of regression processing and discrimination processing associated with the parameter performed using a learning model pre-constructed based on the data when an impurity is mixed into the fluid and the obtained data.

[0009] Thus, even when a foreign substance enters a fluid while its parameters are changing over time, the measuring device can more accurately measure the parameters. The measuring device obtains the results of at least one of a regression process and a discrimination process associated with the parameters, using a learning model pre-built based on data from the time the foreign substance entered the fluid and the acquired data. This facilitates application of the measuring device to fluids whose parameters change over time, such as during distillation processes. Even when foreign substances such as bubbles enter such fluids, the measuring device can output accurate parameter values through regression processing.

[0010] In one embodiment of the measurement device, the computing unit can perform at least one of the regression processing and the discrimination processing to obtain the result. The measurement device itself can perform at least one of the regression processing and the discrimination processing. This allows the measurement device to internally complete various processes, from the learning phase for constructing a learning model to the estimation phase using the trained learning model.

[0011] In one embodiment of the measurement device, the computing unit can acquire the results of at least one of the regression processing and the discrimination processing performed externally to the measurement device via communication. This eliminates the need for the measurement device to perform computations related to at least one of the regression processing and the discrimination processing. Consequently, the measurement device can reduce the computational load associated with its operation.

[0012] In one embodiment of the measuring device, the computing unit can construct the learning model in advance through supervised learning or unsupervised learning using the data and labels of the time when the impurity enters the fluid. This enables higher-precision learning processing based on labels. The measuring device can utilize the learning model constructed through such learning processing to perform at least one of regression processing and discrimination processing with higher precision.

[0013] In one embodiment of the measurement device, the computing unit can obtain the results of executing the regression process and the discrimination process using different learning models. The measurement device utilizes independent learning models for the regression process and the discrimination process. By utilizing multiple learning models, the measurement device can achieve higher-precision output in the regression process and the discrimination process.

[0014] Regarding a measuring device of one embodiment, the discrimination processing may include a process of discriminating whether there is mixing of impurities into the fluid, and the operation unit determines a first threshold value using the learning model, and the first threshold value is compared with the average value of a moving window of a specified time interval set for at least one of the parameters and the sensor value, and is the first threshold value used to discriminate whether there is mixing of impurities.

[0015] This allows the measurement device to more appropriately determine the first threshold. The measurement device can objectively and accurately determine the first threshold, which is arbitrarily set by the user based on their current experience, for example. This allows the measurement device to more accurately perform processing to determine the presence of contaminants using the first threshold.

[0016] In one embodiment, the measuring device may be configured such that, upon determining that a foreign substance has been introduced into the fluid, the computing unit performs the regression processing using the learning model and outputs the parameter. Thus, even if a parameter has an abnormal value during normal processing due to the influence of the foreign substance introduced into the fluid, the measuring device can regress the parameter and output a more accurate measurement value.

[0017] In one embodiment of the measurement device, the computing unit can correct the output parameter based on the difference between the output parameter and the parameter when no contaminant has been introduced into the fluid. Thus, even if the operating method and parameters at the start of the process vary from run to run, the measurement device can detect and correct errors in the regression-processed parameter resulting from such variations. Even if conditions, including the type and amount of small bubbles such as microbubbles, and the duration of prolonged bubble infiltration, vary from run to run, the measurement device can accurately correct the regression-processed parameter.

[0018] Regarding a measuring device of one embodiment, the discrimination processing may include a process of discriminating whether the operation of the measuring device has stopped, and the operation unit determines a second threshold value using the learning model, and the second threshold value is set for at least one of the parameters, and is the second threshold value used to establish a flag of being in operation or stopped.

[0019] This allows the measurement device to more appropriately determine the second threshold. The measurement device can objectively and accurately determine the second threshold, which has been arbitrarily set by the user, based on the user's current experience, etc. This allows the measurement device to more accurately perform the stop determination process using the second threshold to reduce false detections of shutdowns.

[0020] In one embodiment of the measuring device, the parameter may include at least one of the density, volume flow rate, mass flow rate, and bubble volume fraction of the fluid. This allows the measuring device to perform parameter measurement and regression processing required to function as a Coriolis flowmeter, for example. The measuring device can function as a Coriolis flowmeter.

[0021] Effects of the Invention

[0022] According to the present invention, it is possible to provide a measuring device capable of more accurately measuring parameters even when a foreign substance is mixed into a fluid while the parameters of the fluid are changing with time. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a block diagram showing a schematic configuration of a measuring device according to the first embodiment of the present invention.

[0024] Figure 2 It is used for Figure 1 A schematic diagram illustrating an overview of the regression processing and discrimination processing of the calculation unit.

[0025] Figure 3 This is a schematic diagram for explaining an example of regression processing performed using the trained first learning model.

[0026] Figure 4 This is a schematic diagram for explaining an example of the discrimination process performed using the trained second learning model.

[0027] Figure 5 It is used for Figure 1 A flowchart illustrating a first example of the operation of the measuring device.

[0028] Figure 6 It is used for Figure 1 A flowchart illustrating a second example of the operation of the measuring device.

[0029] Figure 7 It is used for Figure 1 A schematic diagram illustrating a third example of the operation of the measuring device.

[0030] Figure 8 It is used for Figure 1 A flowchart illustrating a fourth example of the operation of the measuring device.

[0031] Figure 9 This is a block diagram showing a schematic configuration of a measuring device according to a second embodiment of the present invention. DETAILED DESCRIPTION

[0032] The background and problems of the prior art are described in more detail.

[0033] If impurities such as bubbles are mixed into the fluid being measured, measuring devices such as field instruments can make it difficult to accurately measure parameters representing the state of the fluid, including flow rate and density. For example, errors may occur in the measured parameter values.

[0034] For example, Patent Document 1 shows a Coriolis flowmeter as an example of a field instrument. In a Coriolis flowmeter, for example, if bubbles as impurities are mixed into a fluid, the measured values of parameters including the density of the fluid are affected by the mixing of bubbles. In this case, as in Patent Document 1, Figure 4 As shown in FIG. 1 , the measured value of the parameter can be lower than the actual value when no bubbles are mixed in. As a result, it is expected that the error between the measured value and the actual value increases.

[0035] To address this issue, the field instrument described in Patent Document 1 uses the drive current value of the Coriolis flowmeter, known as the Drive Current, as a threshold for determining the presence of slag flow, etc. Hysteresis is typically applied to this threshold. As described above, this method uses a method that determines whether bubbles have entered the fluid and corrects the measured value when bubbles are present based on the normal output value when bubbles are not present.

[0036] However, the field instrument described in Patent Document 1 can be applied to fluids whose parameters are generally constant when no impurities are present. It is difficult to apply to fluids whose parameters vary over time due to operation, such as in a distillation process. Accurate calibration is difficult due to various factors, including the type and amount of small bubbles such as microbubbles, and the duration of long-term bubble contamination.

[0037] In order to solve the above problems, an object of the present invention is to provide a measuring device that can measure the parameters more accurately even when impurities are mixed into the fluid while the parameters of the fluid change over time.

[0038] Hereinafter, one embodiment of the present invention will be mainly described with reference to the drawings.

[0039] (First embodiment)

[0040] Figure 1 1 is a block diagram showing a schematic configuration of a measuring device 1 according to a first embodiment of the present invention. Figure 1 The configuration and functions of the measurement device 1 according to the first embodiment will be mainly described.

[0041] The measuring device 1 is configured as, for example, a field instrument 1. In the present invention, a "field instrument" includes any instrument that performs measurement processing on a physical quantity being measured and obtains a measured value. In the present invention, "physical quantities" include, for example, the temperature, pressure, flow rate, and pH of fluids, including gases and liquids, generated in the plant equipment in which the field instrument is installed, as well as the corrosion level and vibration level of the plant equipment. Without limitation, physical quantities also include state parameters, including temperature and pressure, associated with actuators such as valves, motors, and relays.

[0042] In the present invention, "plant equipment" encompasses, for example, not only chemical and other industrial plants, but also plants that manage and control wellheads and their surrounding areas, including those in gas and oil fields. Furthermore, plant equipment may include plants that manage and control hydropower, thermal power, and nuclear power generation, plants that manage and control environmental energy generation such as solar and wind power, and plants that manage and control water supply and drainage systems, dams, and the like.

[0043] For example, the measuring device 1 includes any device that measures at least one parameter representing the state of a fluid. In the present invention, "fluid" includes, for example, liquid. "Parameter" includes, for example, at least one of the density, volume flow rate, mass flow rate, and bubble volume fraction of the fluid. "Bubble volume fraction" refers to, for example, the proportion of gas components contained in the fluid. The proportion of gas components is based on a weight basis or a volume basis. The measuring device 1 measures, for example, the flow rate of a fluid flowing in a pipe. The measuring device 1 includes, for example, a Coriolis flowmeter. A Coriolis flowmeter is an example of a field instrument. The measuring device 1 has a detection unit 10, a processing unit 20, and a calculation unit 30 included in the processing unit 20.

[0044] The detection unit 10 vibrates a measuring tube 11 through which a fluid to be measured flows, and detects the vibration upstream and downstream and the temperature of the measuring tube 11. The detection unit 10 includes, in addition to the measuring tube 11, an exciter 12, an upstream sensor 13, a downstream sensor 14, and a temperature sensor 15.

[0045] The measuring tube 11 has, for example, a straight tube structure with both ends fixedly supported by support members, but the measuring tube 11 is not limited thereto and may have other shapes such as a U-shaped tube structure.

[0046] The vibrator 12 includes any vibration module that mechanically vibrates the measuring tube 11 in the vertical direction. The vibrator 12 is disposed around the measuring tube 11 through which the fluid flows. For example, the vibrator 12 is disposed near the center of the measuring tube 11. The vibrator 12 is electrically connected to the processing unit 20.

[0047] The upstream sensor 13 includes any sensor capable of detecting the vibration of the measuring tube 11 caused by the vibrator 12. The upstream sensor 13 is fixed to the side where the fluid flows into the measuring tube 11, that is, upstream of the vibrator 12 located near the center of the measuring tube 11.

[0048] The downstream sensor 14 includes any sensor capable of detecting the vibration of the measuring tube 11 caused by the vibrator 12. The downstream sensor 14 is fixed to the side where the fluid flows out of the measuring tube 11, that is, downstream of the vibrator 12 located near the center of the measuring tube 11.

[0049] The upstream sensor 13 and the downstream sensor 14 are each electrically connected to the calculation unit 30 of the processing unit 20 .

[0050] The temperature sensor 15 includes any sensor capable of measuring the surface temperature of the measuring tube 11. For example, the temperature sensor 15 is fixed to the surface of the measuring tube 11, downstream of the vibrator 12 located near the center of the measuring tube 11. The temperature sensor 15 is fixed near the downstream sensor 14. The temperature sensor 15 is electrically connected to the computing unit 30 of the processing unit 20. The temperature sensor 15 is used to reduce parameter measurement errors caused by temperature fluctuations.

[0051] The operation of the detection unit 10 configured as described above will be mainly described.

[0052] The vibrator 12 vibrates the measuring tube 11 in a predetermined vibration mode in response to the drive current IR output from the processing unit 20. For example, the vibrator 12 vibrates the measuring tube 11 in a primary vibration mode in which vibration nodes appear only at both ends of the measuring tube 11 fixedly supported by a support member.

[0053] When the measuring tube 11 is vibrated in the primary vibration mode by the vibrator 12 and the fluid being measured flows through the measuring tube 11, the measuring tube 11 vibrates in the secondary vibration mode, with vibration nodes occurring at both ends and the center of the measuring tube 11, which is fixedly supported by the support member. In practice, the measuring tube 11 can vibrate in a vibration mode that is a superposition of the primary and secondary vibration modes.

[0054] The upstream sensor 13 measures the displacement of the upstream side of the measuring tube 11, which is vibrating in the aforementioned vibration mode. The upstream sensor 13 outputs the measured displacement as a displacement signal SA to the computing unit 30 of the processing unit 20. The downstream sensor 14 measures the displacement of the downstream side of the measuring tube 11, which is vibrating in the aforementioned vibration mode. The downstream sensor 14 outputs the measured displacement as a displacement signal SB to the computing unit 30 of the processing unit 20.

[0055] The temperature sensor 15 measures the surface temperature of the measuring tube 11 at the surface located on the downstream side of the measuring tube 11. The temperature sensor 15 outputs the measured surface temperature of the measuring tube 11 to the calculation unit 30 of the processing unit 20 as a temperature signal ST.

[0056] The processing unit 20 includes an excitation circuit 21, an output unit 22, and a storage unit 23 in addition to the calculation unit 30. The calculation unit 30 includes a density calculation unit 31, a mass flow calculation unit 32, a volume flow calculation unit 33, and a learning unit 34. The learning unit 34 also includes a determination unit 35 and a regression unit 36. The determination unit 35 includes a bubble determination unit 351, a stop determination unit 352, and a phase state determination unit 353. The regression unit 36 includes a density calculation unit 361, a mass flow calculation unit 362, a volume flow calculation unit 363, and a bubble volume fraction calculation unit 364.

[0057] The excitation circuit 21 is connected to the exciter 12. The excitation circuit 21 outputs a drive current IR corresponding to the displacement signal SA to the exciter 12, thereby driving the exciter 12 in a sinusoidal manner, for example. The excitation circuit 21 may output a drive current IR corresponding to the displacement signal SB to the exciter 12 instead of the displacement signal SA, thereby driving the exciter 12 in a sinusoidal manner, for example.

[0058] The output unit 22 includes an arbitrary output interface for outputting information to the user of the measurement device 1. Examples of output interfaces include a display that outputs information as images and a speaker that outputs information as audio. Examples of displays include liquid crystal displays and organic EL (Electro Euminescent) displays.

[0059] The storage unit 23 includes any storage module, including an HDD (Hard Disk Drive), an SSD (Solid State Drive), an EEPROM (Electrically Erasable Programmable Read-Only Memory), a ROM (Read-Only Memory), and a RAM (Random Access Memory). The storage unit 23 functions as, for example, a main storage device, an auxiliary storage device, or a buffer. The storage unit 23 is not limited to being built into the measurement device 1 and may also include an external storage module connected via an electronic input / output port such as a USB (Universal Serial Bus). The storage unit 23 stores any information used for the operation of the measurement device 1.

[0060] The calculation unit 30 includes one or more processors. In the present invention, the term "processor" may be, for example, a general-purpose processor or a dedicated processor that performs specific processing, but is not limited thereto. The calculation unit 30 performs various calculations required for the operation of the measurement device 1.

[0061] The calculation unit 30 measures the vibration frequency of the measuring tube 11 based on at least one of the displacement signals SA and SB output from the detection unit 10. The density calculation unit 31 of the calculation unit 30 calculates the mass, or density, of the fluid being measured based on the measured vibration frequency. The mass flow calculation unit 32 of the calculation unit 30 calculates the mass flow rate of the fluid flowing through the measuring tube 11 based on the phase difference between the displacement signals SA and SB output from the detection unit 10. The volume flow calculation unit 33 of the calculation unit 30 calculates the volume flow rate of the fluid flowing through the measuring tube 11 by dividing the mass flow rate calculated by the mass flow calculation unit 32 by the density calculated by the density calculation unit 31.

[0062] The detection unit 10 is used to obtain at least one sensor value as data required for calculation of the above-mentioned parameters by the calculation unit 30. The calculation unit 30 calculates the parameters based on the data obtained by the detection unit 10. In the present invention, "data" includes, for example, at least one of phase difference data, vibration frequency data, fluid temperature data, drive current data, fluid pressure data, and total volume data during operation.

[0063] Phase difference data is data that uses the phase difference between the displacement signals SA and SB outputted from the upstream sensor 13 and the downstream sensor 14 of the detection unit 10 as a sensor value. Vibration frequency data is data that uses the vibration frequency of the measurement tube 11, measured based on at least one of the displacement signals SA and SB outputted from the upstream sensor 13 and the downstream sensor 14 of the detection unit 10, as a sensor value.

[0064] The fluid temperature data is data that uses the temperature of the measuring tube 11, measured based on the temperature signal ST output by the temperature sensor 15 of the detection unit 10, as a sensor value. The drive current data is data that uses the drive current IR output by the excitation circuit 21 to operate the oscillator 12 of the detection unit 10 as a sensor value.

[0065] Fluid pressure data is data obtained by inputting a measurement value from a pressure gauge installed separately from the measuring device 1, or by setting a fixed value within the calculation unit 30 as a sensor value. Total volume data during operation is data obtained by measuring the total volume of fluid flowing through the measuring tube 11 during operation of the measuring device 1 based on the displacement signals SA and SB outputted, respectively, from the upstream sensor 13 and the downstream sensor 14 of the detection unit 10 as a sensor value.

[0066] The learning unit 34 of the computing unit 30 obtains at least one of the results of regression processing and discrimination processing associated with parameters, performed using a pre-built learning model based on data obtained from the detection unit 10 and data obtained from the detection unit 10. The computing unit 30 obtains this result by performing at least one of the regression processing and discrimination processing. When the mixing of a foreign substance into the fluid is detected based on the pre-built, trained learning model, the learning unit 34 regresses the desired value as the measured value of the parameter and outputs the parameter.

[0067] In the present invention, "impurities" include, for example, air bubbles. While impurities are described below as air bubbles, this is not a limitation. Impurities may include, for example, any foreign matter that affects the measured value of a parameter obtained by the measuring device 1. The same description below applies to foreign matter.

[0068] In the present invention, "regression processing" related to parameters includes, for example, regressing the parameter's measured value to a desired value based on a pre-built, trained learning model when bubbles are mixed into the fluid and the parameter's measured value shows an abnormality. "Discrimination processing" related to parameters includes, for example, determining whether bubbles have been mixed into the fluid, determining whether the operation of the measurement device 1 has stopped, and determining the phase state of the fluid being measured.

[0069] In the present invention, "phase state" refers to the total number of gas and liquid phases in a multiphase flow of a fluid containing two or more mixed components, including gas and liquid. For example, when bubbles are mixed into a multiphase flow containing two components separated by the liquid phase, the phase state becomes three.

[0070] Figure 2 It is used for Figure 1 Schematic diagram for explaining an outline of the regression processing and discrimination processing of the calculation unit 30. Figure 2 This section summarizes the processing performed using the trained learning model.

[0071] The learning model is constructed in the regression process using models that output regression results, such as multiple regression analysis, neural networks, support vector regression, Gaussian process regression, regression trees, logistic regression, and autoregressive models. In the discrimination process, the learning model is constructed using models that output classification results, such as logistic regression, neural networks, support vector machines, classification trees, change point detection, k-nearest neighbor method, and k-means method. In the present invention, "neural networks" include, for example, self-organizing maps, convolutional neural networks, recursive neural networks, and long-short memory neural networks.

[0072] The input layer utilizes at least one type of sensor value obtained within the measuring device 1, which is a Coriolis flowmeter. For example, the sensor value used in the input layer is included in the aforementioned data. More specifically, the sensor value includes phase difference data, vibration frequency data, fluid temperature data, drive current data, fluid pressure data, and total volume data during operation.

[0073] The output layer utilizes at least one of estimated data related to parameters output from the measuring device 1, which is a Coriolis flowmeter, and estimated data related to the determination process. For example, the estimated data used in the output layer relates to volume flow rate, fluid density, mass flow rate, and bubble volume fraction. For example, the estimated data used in the output layer relates to determining whether bubbles have been incorporated into the fluid, determining the phase state of the fluid being measured, and determining whether the operation of the measuring device 1 has been stopped.

[0074] The learning unit 34 of the operation unit 30 uses the data and labels of the case when bubbles as admixtures are mixed into the fluid to construct a learning model in advance through supervised learning. The teacher data used as labels for constructing the learning model includes, for example, time series data obtained from at least one parameter when bubbles are mixed into the fluid. Time series data is continuous data that also includes the time when the bubbles are mixed. For example, the teacher data is set as time series data in which the fluid situation is recorded as a photo or video using a camera or other imaging device and the label of the time when the bubbles are mixed is continuous. The learning unit 34 constructs the learning model using the data obtained by the detection unit 10 of the measuring device 1 and the accumulated amount over time obtained based on its feature quantity as feature quantities.

[0075] This training data is obtained using, for example, measuring instruments such as level meters, weight meters, and densitometers, as well as visual inspection methods. Generally, the relationship: volume flow = mass flow / density holds true. Therefore, if training data is prepared for at least two of volume flow, mass flow, and density, training data for the remaining one can also be obtained indirectly. Based on the obtained training data, the learning unit 34 performs regression evaluation of data related to various sensor values.

[0076] Figure 3 This is a schematic diagram for explaining an example of regression processing performed using the trained first learning model L1. Figure 3 As an example, a first learning model L1 of a neural network during regression processing is shown. Figure 4 This is a schematic diagram for explaining an example of the discrimination process performed using the trained second learning model L2. Figure 4 As an example, a second learning model L2 of a neural network in the case of discrimination processing with two classes is shown.

[0077] The learning unit 34 of the computing unit 30 obtains the results of regression processing and discrimination processing performed using different learning models. The learning unit 34 uses independent learning models for the regression processing and discrimination processing. As an example, the learning unit 34 uses the same first learning model L1 for different types of parameters in the regression processing. As an example, the learning unit 34 uses the same second learning model L2 for different types of parameters in the regression processing.

[0078] Learning unit 34 can set any number of layers and neurons in the intermediate layers. It sets common nonlinear functions such as ramp, sigmoid, and hyperbolic tangent functions as activation functions. Based on the backpropagation error method, learning unit 34 sets a Softmax function for the output layer when performing multi-value classification.

[0079] The regression unit 36 of the learning unit 34 performs regression processing of the density, mass flow rate, volume flow rate, and bubble volume fraction of the fluid based on the first learning model L1 and the obtained sensor values. Figure 3 , the regression processing of the volume flow rate by the volume flow rate calculation unit 363 is shown as an example.

[0080] The density calculation unit 361 performs a regression process of the fluid density. For example, the density calculation unit 361 utilizes a logistic regression model using the total volume during operation and the fluid temperature among the sensor values in the regression process of the fluid density.

[0081] The mass flow rate calculation unit 362 performs mass flow rate regression processing. For example, the mass flow rate calculation unit 362 utilizes a linear regression model using the phase difference in the sensor value, the fluid temperature, the drive current, the density of the fluid before the regression processing, and the total volume during operation.

[0082] The volume flow rate calculation unit 363 performs volume flow rate regression processing. For example, the volume flow rate calculation unit 363 uses a linear regression model using the phase difference in the sensor value, the fluid temperature, the drive current, and the density of the fluid before the regression processing.

[0083] The bubble volume fraction calculation unit 364 performs regression processing of the bubble volume fraction. For example, the bubble volume fraction calculation unit 364 utilizes a linear regression model using the phase difference in the sensor values, the fluid temperature, the drive current, the density of the fluid before the regression processing, and the total volume during operation.

[0084] The determination unit 35 of the learning unit 34 performs determination processing such as determination of whether there is mixed air bubbles, determination of operation stoppage, and determination of phase state based on the obtained sensor value based on the second learning model L2. Figure 4 , the determination process of whether or not there is mixing of bubbles by the bubble determination unit 351 is shown as an example.

[0085] The bubble determination unit 351 performs a process of determining whether bubbles are mixed in. For example, the bubble determination unit 351 utilizes change point detection using a phase difference in a sensor value in the process of determining whether bubbles are mixed in.

[0086] The stop determination unit 352 performs a stop determination process. For example, the stop determination unit 352 utilizes the phase difference in the sensor values, the fluid temperature, the drive current, the density of the fluid before the regression process, and the total volume during operation.

[0087] The phase state determination unit 353 performs phase state determination processing. For example, the phase state determination unit 353 utilizes the phase difference in the sensor values, the fluid temperature, the drive current, the density of the fluid before the regression process, and the total volume during operation.

[0088] Figure 5 It is used for Figure 1 A flowchart illustrating a first example of the operation of the measuring device 1. Figure 5 The outline of the main operations implemented by the measurement device 1 is summarized.

[0089] In step S100 , the calculation unit 30 stores at least one type of sensor value necessary for parameter calculation acquired by the detection unit 10 as data in the storage unit 23 .

[0090] In step S101, the calculation unit 30 determines whether the number of samples sufficient to calculate the average value of the moving window within the predetermined time interval set for the sensor values stored in step S100 has been obtained. If the calculation unit 30 determines that the number of samples has been obtained, the calculation unit 30 executes the process of step S102. If the calculation unit 30 determines that the number of samples has not been obtained, the calculation unit 30 executes the process of step S105.

[0091] In step S102 , if it is determined in step S101 that the sampling number has been obtained, the calculation unit 30 calculates the average value of the moving window of the predetermined time interval set for the sensor value stored in step S100 .

[0092] In step S103, the calculation unit 30 determines whether the incorporation of bubbles into the fluid is detected based on the average value calculated in step S102. If the incorporation of bubbles is detected, the calculation unit 30 executes the process of step S104. If the incorporation of bubbles is not detected, the calculation unit 30 executes the process of step S105.

[0093] The calculation unit 30 calculates at least one of the parameters and sensor values, Figure 5 As an example in the flowchart, the average value of a moving window over a predetermined time interval set for sensor values is compared with a first threshold value to determine whether bubble contamination has been detected. If the average value exceeds the first threshold value, the calculation unit 30 determines that bubble contamination has been detected. The calculation unit 30 uses the second learning model L2 to determine the first threshold value to be compared with the average value. This first threshold value is used to determine whether bubble contamination has occurred.

[0094] In step S104, if it is determined in step S103 that the mixing of bubbles has been detected, the calculation unit 30 determines whether the operation stop of the measurement device 1 has been detected. If it is determined that the operation stop of the measurement device 1 has been detected, the calculation unit 30 executes the process of step S105. If it is determined that the operation stop of the measurement device 1 has not been detected, the calculation unit 30 executes the process of step S106.

[0095] In step S105, if it is determined in step S104 that the operation stop has been detected, the calculation unit 30 executes normal processing for when the operation of the measurement device 1 is stopped. If it is determined in step S101 that the sampling number has not been obtained, the calculation unit 30 executes normal processing related to parameter measurement of the measurement device 1. If it is determined in step S103 that the mixing of bubbles has not been detected, the calculation unit 30 executes normal processing related to parameter measurement of the measurement device 1.

[0096] In step S106, if it is determined in step S104 that the operation has not stopped, then the calculation unit 30 detected the incorporation of bubbles in step S103 and, therefore, uses the pre-built learning model and the data acquired in step S100 to perform regression processing associated with the parameters. Thus, if it is determined that bubbles have been incorporated into the fluid, the calculation unit 30 performs regression processing using the first learning model L1 and outputs the parameters.

[0097] As described above, the operation unit 30 uses the following change point detection, that is, using the average value of a moving window in an arbitrary time interval. When it is desired to detect the mixing of bubbles, the operation unit 30, for example, always monitors at least one sensor value used for regression processing and discrimination processing. When the monitored sensor value exceeds the set first threshold, the operation unit 30 determines that the mixing of bubbles has been detected. In order to reduce false detections due to factors other than the mixing of bubbles, such as noise, the operation unit 30 calculates the average value of the moving window in a specified time interval and compares it with the first threshold. The first threshold is determined based on the second learning model L2 pre-constructed by supervised learning using teacher data.

[0098] Figure 6 It is used for Figure 1 A flowchart illustrating a second example of the operation of the measuring device 1. Figure 6 More specifically, Figure 5 The determination process of stopping determination is performed in step S104.

[0099] In step S200 , the calculation unit 30 stores the calculated value of at least one parameter in the storage unit 23 .

[0100] In step S201, the calculation unit 30 determines whether the number of samples sufficient to calculate the average value of the moving window within the predetermined time interval set for the parameter value stored in step S200 is available. If the calculation unit 30 determines that the number of samples is available, the calculation unit 30 executes the process of step S202. If the calculation unit 30 determines that the number of samples is not available, the calculation unit 30 executes the process of step S209.

[0101] In step S202 , if it is determined in step S201 that the number of samples has been obtained, the calculation unit 30 calculates the average value of the moving window of the predetermined time interval set for the parameter value stored in step S200 .

[0102] In step S203, the calculation unit 30 determines whether the average value calculated in step S202 is less than or equal to the second threshold. If it is determined to be less than or equal to the second threshold, the calculation unit 30 executes the process of step S204. If it is determined to be greater than the second threshold, the calculation unit 30 executes the process of step S205. The calculation unit 30 uses the second learning model L2 to determine the second threshold value set for at least one of the parameters, and this second threshold value is used to establish the operating or stopped flag.

[0103] In step S204 , if it is determined in step S203 that the value is less than or equal to the second threshold value, the calculation unit 30 sets flag 1 and stores the flag in the storage unit 23 as information.

[0104] In step S205 , if it is determined in step S203 that the value is larger than the second threshold value, the calculation unit 30 sets a flag of 0 and stores the flag in the storage unit 23 as information.

[0105] In step S206, the calculation unit 30 determines whether the number of samples sufficient to calculate the average value of the moving window for the predetermined time interval set for the flag value stored in steps S204 and S205 has been obtained. If the calculation unit 30 determines that the number of samples has been obtained, the calculation unit 30 performs the process of step S207. If the calculation unit 30 determines that the number of samples has not been obtained, the calculation unit 30 performs the process of step S209.

[0106] In step S207, if the sampling number is determined to have been obtained in step S206, the calculation unit 30 calculates the average value of the flag value. For example, the calculation unit 30 calculates the average value by setting a moving window of a predetermined time interval for the flag values stored in steps S204 and S205. The calculation unit 30 determines whether the calculated average value is greater than or equal to 0.5. If the average value is greater than or equal to 0.5, the calculation unit 30 executes the process of step S208. If the average value is less than 0.5, the calculation unit 30 executes the process of step S209.

[0107] In step S208 , if it is determined in step S207 that the average value is greater than or equal to 0.5, the calculation unit 30 determines that the operation of the measurement device 1 is stopped.

[0108] In step S209, if it is determined in step S207 that the average value is less than 0.5, the calculation unit 30 determines that the measurement device 1 is in operation and performs normal processing related to parameter measurement of the measurement device 1. If it is determined in step S201 that the sampling number has not been obtained, the calculation unit 30 performs normal processing related to parameter measurement of the measurement device 1. If it is determined in step S206 that the sampling number has not been obtained, the calculation unit 30 performs normal processing related to parameter measurement of the measurement device 1.

[0109] about Figure 6 The stop determination shown is used, for example, to determine between the first and second distillations, or between the second and third distillations, when two or three distillations are performed per day. In general, field devices may not have sufficient hardware memory to limit power consumption and cost.

[0110] Setting a wide range for parameter average values can easily reduce false detections related to shutdowns, but the memory for the array used for averaging calculations is limited. Field instruments often calculate parameter values with a cycle time of milliseconds. For example, to calculate the average value for a 1-second period with a 10-ms sampling period, an array of 100 values is required. In reality, the risk of false detection is extremely high if sensor and parameter values fluctuate rapidly within a 1-second period due to noise and other factors. Therefore, an array of approximately 60 seconds is required.

[0111] Therefore, a 60-second array is prepared, using the values of each of the 100 cycles of 10ms as representative values. Therefore, while an array of 6,000 values would have been required to calculate the 60-second average, an array containing only 60 representative values per 1 second is sufficient.

[0112] During the stop determination, the calculation unit 30 does not immediately determine that the system is stopped if the average value of the parameter over 60 seconds is less than or equal to the second threshold. Instead, it sets a flag of 1 for the time point when the parameter is less than or equal to the second threshold and stores this flag in the 60-second array. The calculation unit 30 also sets a flag of 0 for the time point during operation and calculates the average value of these flags. This determines that the system is stopped when the value is greater than or equal to 0.5, for example. This stop determination significantly reduces the risk of false detection of a stop during operation of the measuring device 1.

[0113] The method of calculating the average value, the process of determining the stop, and the process of determining the operation vary depending on the capability and application of the measuring device 1 , and can be appropriately changed according to the specifications and application of the measuring device 1 .

[0114] Figure 7 It is used for Figure 1 A schematic diagram illustrating a third example of the operation of the measuring device 1. Figure 7 The left half of the diagram contains a set of graphs showing the changes in volume flow and density over time, indicating that the utilization Figure 6 The situation of the judgment process of stopping judgment described above. Figure 7 The other set of graphs included in the right half of the diagram showing the changes in volume flow and density over time indicates the execution of the Figure 6 The situation when the discrimination process of stopping discrimination is described. Figure 7 In FIG. 1 , the solid line represents the value of the parameter output by the normal process related to parameter measurement by the measuring device 1 , and the dotted line represents the value of the parameter output by the regression process by the measuring device 1 .

[0115] For example, about Figure 7 The upper left graph does not execute the Figure 6The stop determination process described above determines that the measuring device 1 has stopped because the volume flow rate exceeds the second threshold and temporarily decreases in the shaded area. The learning model is constructed only while the measuring device 1 is operating. Therefore, it is difficult for the measuring device 1 to output accurate parameters through regression processing when it is stopped.

[0116] If Figure 7 As shown in the upper left graph of , if the parameter drops in an instant and the measuring device 1 is mistakenly detected as being stopped, the total volume in operation used in the density regression process is reset to zero. Figure 7 As shown in the lower left graph of , the output value of the density based on the regression processing also returns to the initial value.

[0117] To reduce this inaccurate output, perform the exploit Figure 6 The processing for determining a stop is described above. This processing allows the calculation unit 30 to determine that even when a parameter decreases due to momentary noise, it is considered noise and to perform processing less susceptible to the noise. The calculation unit 30 does not determine that the measuring device 1 has stopped because the volume flow rate value temporarily decreases in the shaded area exceeding the second threshold. However, the calculation unit 30 determines that the measuring device 1 is stopped because the volume flow rate value continues to exceed the second threshold.

[0118] Figure 8 It is used for Figure 1 A flowchart illustrating a fourth example of the operation of the measuring device 1. Figure 8 The following shows an overview of the calibration process of the parameter value output by the regression process of the measuring device 1 .

[0119] Consider the case where fluid parameters such as density change over time during operation, as in a distillation process. When regression processing is performed on such a fluid, the output value significantly depends on the fluid parameters at the start of the process. Therefore, if the parameters at the start of the process vary from run to run due to differences in user operating methods, the parameter values output by the regression processing may vary.

[0120] The calculation unit 30 corrects the output parameter based on the difference between the parameter output by the regression process and the parameter when no bubbles are mixed into the fluid. More specifically, the calculation unit 30 stores the difference between the parameter under normal conditions, when no bubbles are mixed into the fluid, and the parameter based on the regression process when bubbles are detected in the storage unit 23. The calculation unit 30 performs parameter correction processing by adding the correction value to the parameter when bubbles are detected. The calculation unit 30 stores the difference used for the correction process in an array in the storage unit 23 and, after eliminating outliers, calculates the average of the difference and sets it as the correction value.

[0121] In step S300, the operation unit 30 performs the following operations, for example, by Figure 5 The calculation unit 30 determines whether the mixing of bubbles into the fluid is detected in the same manner as in step S103. If it is determined that the mixing of bubbles is detected, the calculation unit 30 executes the process of step S301. If it is determined that the mixing of bubbles is not detected, the calculation unit 30 executes the process of step S307.

[0122] In step S301 , if it is determined in step S300 that the mixing of bubbles is detected, the calculation unit 30 stores in the storage unit 23 the difference between the normal parameter when there is no mixing of bubbles into the fluid and the parameter based on the regression processing when bubbles are detected.

[0123] In step S302, the calculation unit 30 determines whether the number of samples sufficient to calculate the average value of the difference values stored in step S301 has been obtained. If the number of samples has been obtained, the calculation unit 30 performs the process of step S303. If the number of samples has not been obtained, the calculation unit 30 performs the process of step S306.

[0124] In step S303 , if it is determined in step S302 that the number of samples has been obtained, the calculation unit 30 excludes the values that are outliers from the difference values stored in step S301 .

[0125] In step S304 , after eliminating the offset value in step S303 , the calculation unit 30 calculates the average value of the difference values stored in step S301 as a correction value.

[0126] In step S305 , the calculation unit 30 adds the correction value calculated in step S304 to the parameter value obtained by the regression process when the bubbles are detected, and outputs the corrected parameter value to, for example, the output unit 22 .

[0127] In step S306 , if it is determined in step S302 that the sampling number has not been obtained, the calculation unit 30 outputs the value of the parameter obtained by the regression process to, for example, the output unit 22 .

[0128] In step S307 , if it is determined in step S300 that the mixing of bubbles is not detected, the calculation unit 30 outputs the parameter values in normal processing without performing either regression processing or correction processing, for example, to the output unit 22 .

[0129] Next, the algorithm used in the measurement device 1 will be described in detail as an example.

[0130] When the measuring device 1 starts operating, the internal timing of the measuring device 1 is started. The computing unit 30 of the measuring device 1 stores at least one of the parameter and sensor value outputted by the normal processing for determining the operation stop in the storage unit 23 each time the timing reaches 100 cycles.

[0131] In order to determine whether there is air bubble mixing, the operation unit 30 needs to obtain sample values greater than or equal to the minimum prepared array. For example, in the case of an array prepared for 60 seconds, the operation unit 30 calculates the average value after 60 seconds. Therefore, during the period when the average value is not calculated, even if air bubbles are mixed into the fluid, the determination process related to whether there is air bubble mixing does not work. Therefore, there is a trade-off between the size of the array as a countermeasure for reducing false detection and the loading time related to the determination of whether there is air bubble mixing. However, the timing starts after the power of the measuring device 1 is turned on, so it is only executed once. Therefore, the above situation usually does not become a major problem.

[0132] When the timer is equal to or greater than the set number, the calculation unit 30 always simultaneously executes the determination process regarding the presence of bubbles and the determination process regarding the shutdown. The first and second threshold values obtained by machine learning based on the second learning model L2 are installed in the measuring device 1.

[0133] Regarding the determination process regarding the presence of bubbles, as described above, the calculation unit 30 compares the average value of at least one sensor value with the first threshold value. Regarding the determination process regarding the shutdown, as described above, the calculation unit 30 compares the value of the parameter used for the application with the second threshold value.

[0134] If it is determined that bubbles are mixed into the fluid, the computing unit 30 performs regression processing using the pre-built, trained first learning model L1 and outputs the parameter values. If it is determined that the measuring device 1 is stopped, the computing unit 30 determines that the operation of an application such as distillation has stopped and sets the arrays of timer and parameter correction values and corresponding difference values to zero.

[0135] As described above, the timer is reset, and when the application process such as distillation is restarted, the parameters are stored in the array again. In this way, the calculation unit 30 repeatedly executes the determination process regarding the presence of mixed bubbles and the determination process regarding the stop of operation.

[0136] According to the measuring device 1 according to the first embodiment described above, even when a foreign substance enters a fluid while its parameters are changing over time, the parameters can be measured relatively accurately. The measuring device 1 obtains the results of at least one of a regression process and a discrimination process associated with the parameters, using a learning model pre-constructed based on data from the time the foreign substance entered the fluid and the acquired data. Consequently, the measuring device 1 can be easily applied to fluids whose parameters change over time, such as during distillation processes. Even when foreign substances such as bubbles enter such fluids, the measuring device 1 can output accurate parameter measurements through regression processing.

[0137] The measurement device 1 obtains results by performing at least one of regression processing and discrimination processing. The measurement device 1 itself performs at least one of the regression processing and discrimination processing. This allows the measurement device 1 to internally complete various processes, from the learning phase for building a learning model to the estimation phase using the trained learning model.

[0138] The measuring device 1 uses data and labels from the time of contaminants entering the fluid to pre-build a learning model through supervised learning, thereby enabling the performance of label-based learning with higher accuracy. The measuring device 1 can utilize the learning model built through this learning process to perform at least one of regression processing and discrimination processing with higher accuracy.

[0139] The measurement device 1 obtains the results of performing regression and discrimination processing using different learning models. The measurement device 1 utilizes independent learning models for the regression and discrimination processes. By utilizing multiple learning models, the measurement device 1 can achieve higher-precision output in the regression and discrimination processes.

[0140] By using a learning model to determine the first threshold for determining the presence of contaminants, the measurement device 1 can more appropriately determine the first threshold. The measurement device 1 can objectively and accurately determine the first threshold, which is arbitrarily set by the user, based on the user's current experience. Consequently, the measurement device 1 can perform the process of determining the presence of contaminants using the first threshold with greater accuracy.

[0141] If it is determined that a foreign substance has entered the fluid, the measuring device 1 performs regression processing using the learning model and outputs the parameters. Thus, even if the foreign substance enters the fluid and causes the parameters to have abnormal values during normal processing, the measuring device 1 can regress the parameters and output more accurate measurement values.

[0142] The measuring device 1 corrects the output parameters based on the difference between the parameters output by the regression process and the parameters obtained when no contaminants have entered the fluid. Thus, even if the operating method and parameters at the start of the process vary from run to run, the measuring device 1 can detect and correct errors in the regression-processed parameters resulting from these variations. Even if conditions, including the type and amount of small bubbles such as microbubbles, and the duration of prolonged bubble infiltration, vary from run to run, the measuring device 1 can accurately correct the regression-processed parameters.

[0143] By using a learning model to determine the second threshold value used to establish the operating or stopped flag, the measurement device 1 can more appropriately determine the second threshold value. The measurement device 1 can objectively and accurately determine the second threshold value arbitrarily set by the user based on the user's current experience, etc. This allows the measurement device 1 to more accurately perform the stop determination process using the second threshold value to reduce false detections of stopped operation.

[0144] The parameters include at least one of fluid density, volume flow rate, mass flow rate, and bubble volume fraction, and thus the measuring device 1 can perform parameter measurement and regression processing required to function as a Coriolis flowmeter.

[0145] In the first embodiment described above, the measuring device 1 pre-builds a learning model using supervised learning using data and labels from the time when the impurity enters the fluid. However, this is not limiting. The measuring device 1 can pre-build a learning model for regression processing using unsupervised learning or semi-supervised learning. The measuring device 1 can pre-build a learning model for discrimination processing using unsupervised learning or semi-supervised learning.

[0146] In the first embodiment described above, the measurement device 1 obtains results when performing regression processing and discrimination processing using different learning models. However, this is not limiting. The measurement device 1 may utilize a common, identical learning model for both the regression processing and the discrimination processing, rather than separate learning models. Conversely, the measurement device 1 may utilize different first learning models L1 for different types of parameters in the regression processing.

[0147] For example, the first learning model L1 may not necessarily match the volume flow rate and the density. The measurement device 1 may use different second learning models L2 for different types of discrimination in the discrimination process.

[0148] In the first embodiment described above, the measurement device 1 uses a learning model to determine the first threshold value for determining the presence of a foreign substance. However, this is not limiting. The measurement device 1 does not need to perform such determination processing using a learning model. The measurement device 1 can determine the first threshold value to a value appropriately set by the user.

[0149] In the first embodiment, when it is determined that a foreign substance has been mixed into the fluid, the measuring device 1 performs regression processing using the learning model and outputs parameters. However, the present invention is not limited thereto. The measuring device 1 may not perform such regression processing.

[0150] In the first embodiment described above, the measurement device 1 calibrates the parameters outputted by the regression process, but the present invention is not limited thereto. The measurement device 1 does not need to perform such calibration processing.

[0151] In the first embodiment described above, the measurement device 1 uses a learning model to determine the second threshold value for establishing the operating or stopped flag. However, this is not limiting. The measurement device 1 does not need to perform such determination processing using a learning model. The measurement device 1 can determine the second threshold value to a value appropriately set by the user.

[0152] In the first embodiment described above, the parameter includes at least one of fluid density, volume flow rate, mass flow rate, and bubble volume fraction, but this is not limited to this. The parameter may include any other physical quantity that can be measured by a field instrument. Accordingly, the measuring device 1 is not limited to a Coriolis flowmeter. The measuring device 1 may include other field instruments.

[0153] In the first embodiment, the fluid is described as containing liquid, but the present invention is not limited thereto. The fluid may contain gas.

[0154] (Second embodiment)

[0155] Figure 9 : is a block diagram showing a schematic configuration of a measuring device 1 according to a second embodiment of the present invention. Figure 9 The configuration and function of the measurement device 1 according to the second embodiment will be mainly described. The measurement device 1 according to the second embodiment differs from the first embodiment in that it does not include the learning unit 34 .

[0156] Other structures, functions, effects, and modifications are the same as those of the first embodiment, and the corresponding descriptions also apply to the measurement device 1 according to the second embodiment. Hereinafter, components identical to those of the first embodiment are denoted by the same reference numerals, and their descriptions are omitted. The following description focuses on the differences from the first embodiment.

[0157] In the first embodiment, the computing unit 30 performs at least one of the regression process and the discrimination process to obtain the result, but this is not limiting. The computing unit 30 itself does not have to perform at least one of the regression process and the discrimination process. The computing unit 30 can obtain the result of at least one of the regression process and the discrimination process performed externally to the measurement device 1 through communication. For example, the computing unit 30 can obtain the result of at least one of the regression process and the discrimination process performed by the external device through communication. The learning unit 34, which includes both the discrimination unit 35 and the regression unit 36, can be provided external to the measurement device 1.

[0158] The external device including the learning unit 34 includes, for example, one or more server devices capable of communicating with each other, such as those used in edge computing and cloud systems. The external device is not limited thereto and may include any general-purpose electronic device such as a mobile phone, smartphone, or tablet terminal, or a PC (Personal Computer), or may include other electronic devices specifically designed for the measurement device 1.

[0159] The computing unit 30 of the measurement device 1 further includes a data communication unit 37. The data communication unit 37 includes a communication module compatible with any wireless or wired communication standard. Examples of communication standards include wireless LAN (E-LAN) standards, near-field communication standards, mobile communication standards such as 4G (4th Generation) and 5G (5th Generation), and Internet standards. The measurement device 1 is communicatively connected to these external devices via the data communication unit 37.

[0160] The measurement device 1 performs calculations related to at least one of the regression and discrimination processes without acquiring the results of at least one of the regression and discrimination processes performed outside the measurement device 1 through communication.

[0161] In the second embodiment described above, both the discrimination unit 35 and the regression unit 36 are provided outside the measurement device 1. However, this is not limiting. Alternatively, one of the discrimination unit 35 and the regression unit 36 may be provided inside the measurement device 1, while the other may be provided outside the measurement device 1.

[0162] In the second embodiment described above, various processes from the learning phase for constructing a learning model to the estimation phase using the trained learning model are executed externally to the measurement device 1. However, this is not limiting. Alternatively, either the various processes in the learning phase or the various processes in the estimation phase may be executed by the measurement device 1, with the other executed by an external device.

[0163] While the present invention has been described based on the accompanying drawings and embodiments, it should be noted that those skilled in the art will readily be able to make various modifications and variations based on the present invention. Therefore, it should be noted that such modifications and variations are within the scope of the present invention. For example, the functions included in each structure or step can be rearranged in a logically consistent manner, and multiple structures or steps can be combined into one or divided.

[0164] For example, the present invention can also be implemented as a program describing the processing contents that realize each function of the above-mentioned measuring device 1 or a storage medium recording the program. It should be understood that the above-mentioned embodiment is also included in the scope of the present invention.

[0165] The following examples illustrate some embodiments of the present invention, but it should be noted that the embodiments of the present invention are not limited thereto.

[0166] [Note 1]

[0167] A measuring device for measuring at least one parameter indicating a state of a fluid, wherein:

[0168] The measuring device comprises:

[0169] a detection unit configured to acquire at least one sensor value required for calculating the parameter as data; and

[0170] a calculation unit that calculates the parameter based on the data acquired by the detection unit,

[0171] The calculation unit obtains a result of at least one of regression processing and discrimination processing associated with the parameter, performed using a learning model constructed in advance based on the data when the impurity is mixed into the fluid and the obtained data.

[0172] [Note 2]

[0173] The measuring device according to Supplementary Note 1, wherein

[0174] The calculation unit obtains the result by executing at least one of the regression process and the discrimination process.

[0175] [Note 3]

[0176] The measuring device according to Supplementary Note 1, wherein

[0177] The computing unit acquires, through communication, a result of at least one of the regression process and the discrimination process executed outside the measurement device.

[0178] [Note 4]

[0179] The measuring device according to any one of Supplementary Notes 1 to 3, wherein

[0180] The computing unit constructs the learning model in advance through supervised learning or unsupervised learning using the data and labels when the impurity is mixed into the fluid.

[0181] [Note 5]

[0182] The measuring device according to any one of Supplementary Notes 1 to 4, wherein

[0183] The calculation unit obtains the result when the regression process and the discrimination process are executed using the different learning models.

[0184] [Note 6]

[0185] The measuring device according to any one of Supplementary Notes 1 to 5, wherein

[0186] The determination process includes a process of determining whether or not a foreign substance has been mixed into the fluid.

[0187] The calculation unit uses the learning model to determine the first threshold, which is compared with the average value of the moving window of a specified time interval set for at least one of the parameters and the sensor value, and is the first threshold used to determine whether there is any mixing of impurities.

[0188] [Note 7]

[0189] The measuring device according to Supplementary Note 6, wherein:

[0190] If it is determined that the mixed matter has been mixed into the fluid, the calculation unit executes the regression process using the learning model and outputs the parameter.

[0191] [Note 8]

[0192] The measuring device according to Supplementary Note 7, wherein:

[0193] The calculation unit corrects the output parameter based on a difference between the output parameter and the parameter when no mixed matter is mixed into the fluid.

[0194] [Note 9]

[0195] The measuring device according to any one of Supplementary Notes 1 to 8, wherein

[0196] The determination process includes a process of determining whether the operation of the measuring device has stopped.

[0197] The calculation unit determines a second threshold value using the learning model. The second threshold value is set for at least one of the parameters and is used to set a flag indicating whether the system is in operation or stopped.

[0198] [Note 10]

[0199] The measuring device according to any one of Supplementary Notes 1 to 9, wherein

[0200] The parameters include at least one of density, volume flow rate, mass flow rate, and bubble volume fraction of the fluid.

[0201] Description of the label

[0202] 1. Measurement device

[0203] 10. Testing Department

[0204] 11. Assay tube

[0205] 12 vibrator

[0206] 13 Upstream sensor

[0207] 14 Downstream sensor

[0208] 15 Temperature sensor

[0209] 20 Processing Department

[0210] 21 Excitation circuit

[0211] 22 Output

[0212] 23 Storage

[0213] 30 Operation unit

[0214] 31 Density calculation unit

[0215] 32 Mass flow calculation unit

[0216] 33 Volume flow calculation unit

[0217] 34 study units

[0218] 35 discrimination units

[0219] 351 Bubble Identification Unit

[0220] 352 Stop judgment unit

[0221] 353 phase state determination unit

[0222] 36 regression units

[0223] 361 Density Calculation Unit

[0224] 362 Mass flow calculation unit

[0225] 363 Volume flow calculation unit

[0226] 364 Bubble Volume Fraction Calculation Unit

[0227] 37 Data Communications Department

[0228] IR drive current

[0229] L1 1st learning model

[0230] L2 Second Learning Model

[0231] SA displacement signal

[0232] SB displacement signal

[0233] ST temperature signal

Claims

1. A measuring device for measuring at least one parameter indicating a state of a fluid, wherein: The measuring device comprises: a detection unit configured to acquire at least one sensor value required for calculating the parameter as data; as well as a calculation unit that calculates the parameter based on the data acquired by the detection unit, The calculation unit obtains a result of at least one of regression processing and discrimination processing associated with the parameter, performed using a learning model constructed in advance based on the data when the impurity is mixed into the fluid and the obtained data.

2. The measuring device according to claim 1, wherein The calculation unit obtains the result by executing at least one of the regression process and the discrimination process.

3. The measuring device according to claim 1, wherein The computing unit acquires, through communication, a result of at least one of the regression process and the discrimination process executed outside the measurement device.

4. The measuring device according to any one of claims 1 to 3, wherein The computing unit constructs the learning model in advance through supervised learning or unsupervised learning using the data and labels when the impurity is mixed into the fluid.

5. The measuring device according to any one of claims 1 to 3, wherein The calculation unit obtains the result when the regression process and the discrimination process are executed using the different learning models.

6. The measuring device according to any one of claims 1 to 3, wherein The determination process includes a process of determining whether or not a foreign substance has been mixed into the fluid. The calculation unit uses the learning model to determine the first threshold, which is compared with the average value of the moving window of a specified time interval set for at least one of the parameters and the sensor value, and is the first threshold used to determine whether there is any mixing of impurities.

7. The measuring device according to claim 6, wherein If it is determined that the mixed matter has been mixed into the fluid, the calculation unit executes the regression process using the learning model and outputs the parameter.

8. The measuring device according to claim 7, wherein The calculation unit corrects the output parameter based on a difference between the output parameter and the parameter when no mixed matter is mixed into the fluid.

9. The measuring device according to any one of claims 1 to 3, wherein The determination process includes a process of determining whether the operation of the measuring device has stopped. The calculation unit determines a second threshold value using the learning model. The second threshold value is set for at least one of the parameters and is used to set a flag indicating whether the system is in operation or stopped.

10. The measuring device according to any one of claims 1 to 3, wherein The parameters include at least one of density, volume flow rate, mass flow rate, and bubble volume fraction of the fluid.

Images