Data collection device, data collection method, and data collection program

Abstract

To enable both preservation of important data and data compression.SOLUTION: A data collection device 10 disclosed herein acquires prediction data values used for predicting a measurement value from a device. The data collection device 10 predicts time-series data of the measurement value using the prediction data values. When an actual measurement value is acquired from the device, the data collection device 10 controls recording of the actual measurement value based on the time-series data of the predicted measurement value and the actual measurement value.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present invention relates to a data collection device, a data collection method, and a data collection program. [Background technology]

[0002] There is known a recorder that records measured values ​​such as voltage, temperature, and current values ​​measured by devices such as field instruments and various sensors. The recorder records the measured values ​​measured by the device as time-series measured values ​​for each channel. However, if all the time-series measured values ​​are recorded, the data volume becomes large, and the recorder's recording medium may have to be replaced frequently.

[0003] Therefore, a technique is known in which temporarily stored data is divided into a predetermined amount, an approximation formula is calculated from the divided data, and if the deviation between the approximation formula and the actual data exceeds a predetermined threshold, the technique decides to store the data. That is, this technique compresses data by storing only data that meets a condition, such as the deviation between the approximation formula calculated from the actual data and the actual data exceeding a threshold, rather than storing all data. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 04-299478 Summary of the Invention [Problem to be solved by the invention]

[0005] However, in the above-mentioned technology, data compression is performed by performing approximation based on actual measured values ​​and deleting the data, which may result in the deletion of data that is important in terms of quality, making it difficult to achieve both data compression and the preservation of important data.

[0006] An object of the present invention is to achieve both the protection of important data and data compression. [Means for solving the problem]

[0007] In order to solve the above-mentioned problems and achieve the objective, the data collection device of the present invention comprises an acquisition unit that acquires prediction data values ​​used to predict measurement values ​​obtained from a device, a prediction unit that predicts time series data of the measurement values ​​using the prediction data values, and a recording control unit that, when an actual measurement value is acquired from the device, controls the recording of the actual measurement value based on the predicted time series data of the measurement value and the actual measurement value. Effect of the Invention

[0008] The present invention has the effect of achieving both the protection of important data and data compression. [Brief description of the drawings]

[0009] [Figure 1] FIG. 1 is a diagram illustrating a data collection device according to an embodiment of the present invention. [Diagram 2] FIG. 13 is a diagram illustrating an example of a problem in the reference technology. [Diagram 3] 1 is a diagram illustrating an example of a configuration of a data collection device according to a first embodiment. [Figure 4] FIG. 4 is a diagram showing an example of measurement data according to the first embodiment. [Diagram 5] FIG. 4 is a diagram showing an example of restored data according to the first embodiment; [Figure 6] FIG. 4 is a diagram illustrating an example of an extension function of a predicted value setting according to the first embodiment. [Figure 7] 4 is a flowchart of a compression processing condition setting procedure according to the first embodiment. [Figure 8] 4 is a flowchart of a measurement data collection procedure according to the first embodiment. [Figure 9] 4 is a flowchart illustrating a compression process procedure according to the first embodiment. [Figure 10] 4 is a flowchart of a measurement data reading procedure according to the first embodiment. [Figure 11] 4 is a flowchart of a data restoration procedure according to the first embodiment. [Figure 12] FIG. 4 is a diagram illustrating an example of a data collection process according to the first embodiment. [Figure 13] FIG. 4 is a diagram illustrating an example of a data collection process according to the first embodiment. [Figure 14] FIG. 4 is a diagram illustrating an example of a data collection process according to the first embodiment. [Figure 15] FIG. 4 is a diagram illustrating an example of a data collection process according to the first embodiment. [Figure 16] FIG. 11 is a diagram illustrating a data collection device according to a second embodiment. [Figure 17] FIG. 2 is a hardware configuration diagram showing an example of a computer that realizes the functions of the data collection device according to the present embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] A data collection device, a data collection method, and a data collection program according to an embodiment of the present invention will be described in detail below with reference to the drawings. Note that the present invention is not limited to the embodiment described below.

[0011] <Description of data collection device> Fig. 1 is a diagram illustrating a data collection device 10 according to this embodiment. The data collection device 10 shown in Fig. 1 is an example of a recorder, computer, or the like that provides a technology for controlling the recording of actual measured values ​​obtained from a device based on time-series data predicted using past measured values ​​of a device such as a field instrument, measured values ​​of similar devices, simulation values, and the like.

[0012] First, the problems of the reference technology will be described. FIG. 2 is a diagram showing an example of the problems of the reference technology. Generally, as a device for recording actual measured values, a recorder is used that receives actual measured values ​​((1) in FIG. 2) input from a device such as a field instrument that operates constantly to monitor the voltage value, temperature, current value, etc. of an object, and records them in a chronological order. If the recorder that records the measured values ​​acquired by the device does not have a compression function, as shown in (2) in FIG. 2, all data such as the measurement start date and time, the measurement period indicating the period during which the measured values ​​are acquired, the unit of the measured values, multiple channels (CH) that accept inputs from multiple devices, and the input measured values ​​are recorded as they are. However, if all the input measured values ​​in a chronological order are recorded, the data volume to be recorded becomes enormous, which causes the storage capacity of the storage medium of the recorder to become tight and causes frequent replacement of the storage medium.

[0013] For this reason, a reference technique is known in which actual measured values ​​that deviate little from a predicted value obtained from an approximation formula based on temporarily stored actual measured values ​​are determined to be normal values ​​and deleted, and actual measured values ​​that deviate greatly are recorded as abnormal values. That is, in a measurement environment in which values ​​are usually normal but occasionally contain abnormal values, the reference technique compresses data by storing only abnormal values ​​in which the approximation formula calculated from the measured values ​​deviates from the actual measured values ​​by more than a threshold value, rather than storing all measured values.

[0014] However, in the reference technology, in situations where the measurement environment continuously deteriorates, such as when the number of abnormal values ​​(e.g., (3) in Figure 2) becomes greater than the number of normal values ​​(e.g., (4) in Figure 2), data that is important in terms of quality (abnormal values) may be deleted, making it difficult to achieve both the preservation of important data and data compression.

[0015] In addition, since the recording device according to the reference technology temporarily stores the measurement values ​​and calculates the approximation equation, it is not possible to compress the data in real time, and there is a problem that the measurement value compression process may be inefficient.

[0016] Therefore, the data collection device 10 according to this embodiment acquires prediction data values ​​used to predict measurement values ​​obtained from a device. The data collection device 10 predicts time-series data of measurement values ​​(hereinafter, may be simply referred to as "predicted values") using the acquired prediction data values. Then, when the data collection device 10 acquires actual measurement values ​​from a device, it controls recording of the actual measurement values ​​based on the predicted time-series data of measurement values ​​and the actual measurement values.

[0017] Returning now to Fig. 1, the processing of the data collection device 10 according to this embodiment will be specifically described. As shown in Fig. 1, the data collection device 10 is connected to a plurality of devices 90 (90a, 90b, 90c, ...). The data collection device 10 receives, as input data, actual measurement values ​​in time series acquired by the devices from the plurality of devices 90 ((1) in Fig. 1).

[0018] As shown in FIG. 1 (2-1), the data collection device 10 determines whether the input actual measurement value is to be recorded or not based on a predicted value predicted using the prediction data value, and records only the actual measurement values ​​to be recorded (FIG. 1 (2-2)).

[0019] For example, the data collection device 10 predicts a predicted value using prediction data values ​​such as a function that returns a constant with respect to time, a straight line connecting data values ​​at specific times, a first-order lag function, a transition pattern of measured values ​​(hereinafter, sometimes simply referred to as a "transition pattern"), etc. Then, the data collection device 10 sets upper and lower limit values ​​that indicate an allowable range for the predicted value based on the user's settings, past history, etc.

[0020] When an actual measurement value is input from each device, the data collection device 10 judges whether the actual measurement value is within the range from the upper limit to the lower limit of the predicted value, and records the actual measurement value in the storage unit if it is outside the range, and does not record the actual measurement value if it is within the range. The data collection device 10 can restore the recorded actual measurement value and predicted value and output them to the user in the form of a time series graph, data table, etc. ((3) in FIG. 1).

[0021] In other words, the data collection device 10 enables data compression by not recording predictable actual measured values ​​that do not need to be recorded. In addition, the data collection device 10 records actual measured values ​​that are difficult to predict and fall outside of an expected range, so that important data to be used for subsequent analysis and verification can be preserved. Therefore, the data collection device 10 can achieve both the preservation of important data and data compression by controlling the recording of actual measured values ​​obtained from a device based on predicted values ​​predicted from data values ​​for prediction.

[0022] <First embodiment> In the first embodiment, an example of an output is described in which actual measured values ​​that fall outside a range determined based on upper and lower limits for a predicted value predicted by the data collection device 10 are recorded, the actual measured values ​​and predicted values ​​are restored, and a time series graph is generated.

[0023] (Data collection device configuration) First, the configuration of the data collection device 10 according to the first embodiment will be described. Fig. 3 is a diagram showing an example of the configuration of the data collection device 10 according to the first embodiment. As shown in Fig. 3, the data collection device 10 has a power supply unit 11, an input / output unit 12, a communication unit 13, a reception unit 14, a display unit 15, a storage unit 20, and a control unit 30.

[0024] The power supply unit 11 supplies power to the data collection device 10. The power supply unit 11 can detect the power frequency. For example, the power supply unit 11 converts AC 100V or the like supplied from a wiring plug connector such as an outlet into DC voltage or the like. Note that the power supply unit 11 is not limited in terms of amperage, wattage, shape, mechanism, etc., as long as it has a mechanism capable of supplying power to the data collection device 10.

[0025] The input / output unit 12 is composed of an input unit that converts analog data output from a device or the like into digital data and inputs the digital data, and an output unit that converts digital data into analog data and outputs the analog data. For example, the input / output unit 12 converts measured values ​​such as voltage values, temperature values, and current values ​​input from a device into digital data and accepts the digital data based on a scanner, an ADC (Analog to Digital Converter), an input relay, and the like. The input / output unit 12 can also have multiple channels.

[0026] The communication unit 13 manages data communication with other devices based on a communication standard such as Ethernet (registered trademark). For example, the communication unit 13 performs data communication to record a time series graph output by the control unit 30 described later and actual measured values ​​or predicted values ​​to devices other than the device to which input and output are performed via the above-mentioned input / output unit 12. The communication unit 13 also accepts input of compression processing conditions, data values ​​for prediction, and the like based on the data communication. The communication unit 13 can be realized by a network interface card or the like.

[0027] The reception unit 14 controls input of various information to the data collection device 10. For example, the reception unit 14 is realized by a touch panel, a key switch, or the like, and receives operations and settings for causing the data collection device 10 to perform an operation desired by the user.

[0028] The display unit 15 is responsible for displaying various information of the data collection device 10. For example, the display unit 15 displays actual measurement values ​​obtained from a device, trends of the actual measurement values, recorded actual measurement values, predicted values, and the like, which are output by the control unit 30 described later, to a user or the like who operates the data collection device 10. The display unit 15 can be realized by a display such as a liquid crystal display.

[0029] The storage unit 20 stores various information referenced when the control unit 30 described below operates and various information acquired when the control unit 30 operates. The storage unit 20 has a compression processing condition DB 21, a predicted value DB 22, a measurement data DB 23, and a restored data DB 24. The storage unit 20 can be realized by, for example, a semiconductor memory element such as a RAM (Random Access Memory) or a flash memory, or a storage device such as a hard disk or an optical disk.

[0030] The compression processing condition DB21 is a database that stores compression processing conditions used when compression processing is performed by the compression processing unit 40 described later. Specifically, the compression processing condition DB21 stores compression execution settings, compression start conditions, compression end conditions, compression end operation, feature amounts, upper and lower limit values, etc. as compression processing conditions. Note that the "database" in this embodiment is an example of the name of a functional unit having a function of systematically storing various types of information stored electronically, and is not particularly limited. Note that all subsequent descriptions of databases are similar.

[0031] The compression implementation setting described above is information indicating whether or not compression processing is implemented, such as "compression processing ON" or "compression processing OFF". The compression start condition is a compression processing start flag, such as "start time", "start event of user operation, calculation result, etc." The compression end condition is a compression processing end flag, such as "end time", "end event of user operation, calculation result, etc." The compression end action is information that specifies the processing of the compression processing unit 40 after compression is ended, such as "end as is", "repeat compression processing", "wait for compression processing start", etc.

[0032] The feature amount is information used by the prediction unit 43 described later to predict a predicted value. Specifically, the feature amount includes a function calculated from past measurement values, simulation values, measurement values ​​of similar devices, input values ​​based on user experience, etc., a transition pattern, and a table of data values ​​at specific times among data values ​​for time series prediction (hereinafter, may be simply referred to as "break points").

[0033] As a specific example, the "function" is a function used by the prediction unit 43 described below to predict a predicted value. The "transition pattern" is a feature that represents, as a pattern, how a collection of data values ​​at each time generated using past measurement values, simulation values, measurement values ​​of similar devices, input values ​​based on user experience, etc., changes over time. The "breakpoint table" is a collection of data values ​​at the time points where fluctuations in the measurement values ​​are predicted.

[0034] The upper limit and lower limit included in the compression processing conditions indicate the allowable range of the predicted value for determining the "range" used by the compression processing unit 40 described below to determine whether or not to record the actual measurement value obtained from the device. That is, the upper limit is a plus-side allowable value for determining the range, and is a value to be added to or multiplied by the predicted value. On the other hand, the lower limit is a minus-side allowable value for determining the range, and is a value to be subtracted from or multiplied by the predicted value.

[0035] The predicted value DB 22 is a database that temporarily stores predicted values ​​predicted by a prediction unit 43 (described later) for display to a user, etc.

[0036] The measurement data DB23 is a database that stores, as measurement data, a common header in which the measurement start date and time, measurement conditions, etc. are set, and individual data including actual measurement values ​​for each channel, and a compression process start or stop flag. Specifically, the measurement data DB23 stores, as a common header, the measurement start date and time, the measurement period, and the compression process setting, and, as individual data, the unit of the measurement value, the channel number, the timestamp number, the actual measurement value associated with the timestamp number, or the compression process start flag or the compression process end flag, etc.

[0037] The measurement start date and time in the shared header described above is the date and time when the device connected to the data collecting device 10 started measurement, or the date and time when the data collecting device 10 started acquiring actual measured values ​​from the device. The measurement period is the period, such as every second, minute, or hour, at which the data collecting device 10 acquires actual measured values ​​from the device. The compression process setting is information related to the compression process conditions stored in the compression process condition DB 21 described above, or identification information such as a character string, symbol, or number that identifies the compression process condition.

[0038] The unit of individual data is the unit of the actual measurement value acquired by the device, such as voltage value (V), temperature (°C), current value (mA), etc. The channel number is a unique number that identifies the path connecting the devices. The timestamp number is a serial number that identifies the time series of the actual measurement value and flag information that identifies the start or stop of the compression process.

[0039] Here, an example of the measurement data stored in the measurement data DB 23 will be described. FIG. 4 is a diagram showing an example of the measurement data according to the first embodiment. FIG. 4 shows uncompressed measurement data ((1) in FIG. 4) and compressed measurement data ((2) in FIG. 4). In this embodiment, since compression can be switched between being performed or not performed as desired depending on the settings of the compression processing conditions, this section will describe both uncompressed and compressed measurement data. As shown in FIG. 4, the measurement data DB 23 stores a shared header common to all channels and individual data for each channel, such as CH1, CH2, CH3, etc.

[0040] For example, in the case of uncompressed measurement data shown in (1) of FIG. 4, the measurement data DB23 stores the measurement start date and time "2022 / 10 / 1 8:10", the measurement period "+0.5 s", the compression processing condition "non-compression processing", etc. as a shared header. In other words, it means that the measurement data is recorded every 0.5 s from 2022 / 10 / 1 8:10 without compression processing. In addition, the measurement data DB23 stores the unit "V", the channel "CH1", the time stamp number, and the actual measurement value "0; 1.05", "1; 1.035", "2; 1.045", "N; 1.055", etc. associated with the time stamp number. Note that "CH2" and "CH3" shown in (1) of FIG. 4 have different units and actual measurement values ​​of individual data, but the data configuration is similar, so the description will be omitted.

[0041] In addition, in the case of the measurement data at the time of compression shown in (2) of FIG. 4, the measurement data DB23 stores the measurement start date and time "2022 / 10 / 1 8:10", the measurement period "+0.5 s", the compression processing condition "compression processing", etc. as a shared header. In other words, it means that the measurement data measurement is recorded by performing compression processing every 0.5 s from 2022 / 10 / 1 8:10. In addition, the measurement data DB23 stores the unit "V", the channel "CH1", the compression processing start flag "0; 0x7fffffff" associated with the time stamp number, the actual measurement value "95; 1.05", "96; 1.035", "97; 1.045", the compression processing end flag "100; 0x7ffffffe", etc. Note that "CH2" and "CH3" shown in (2) of FIG. 4 have different units and actual measurement values ​​of individual data, but the data configuration is similar, so the description will be omitted. The compressed data is data that is recorded based on predicted time series data in contrast to the uncompressed data.

[0042] The restored data DB 24 is a database that temporarily stores actual measured values ​​and predicted values ​​restored by a restoration processing unit 50 described later in order to display them to a user as restored data. Specifically, the restored data DB 24 records the measurement start date and time, the measurement period, and the compression processing setting as a shared header, and the unit, the channel number, the timestamp number, and the restored actual measured values ​​or predicted values ​​associated with the timestamp numbers as individual data.

[0043] The time stamp numbers described above are serial numbers for identifying the time series of the actual measured values ​​recorded in (2) of Fig. 4 restored by the restoration processing unit 50 described below and the predicted values ​​in the sections where the actual measured values ​​are not recorded in (2) of Fig. 4. Note that the other items are the same as those described in the measurement data DB 23, and therefore will not be described here.

[0044] Here, an example of restored data temporarily stored by the restored data DB 24 will be described. Fig. 5 is a diagram showing an example of restored data according to the first embodiment. Fig. 5 shows restored data for each of CH1, CH2, and CH3 ((1) to (3) in Fig. 5). As shown in Fig. 5, the restored data DB 24 stores a shared header common to all channels, and individual data for each channel such as CH1, CH2, and CH3.

[0045] For example, the restored data DB24 stores the measurement start date and time "2022 / 10 / 1 8:10", the measurement period "+0.5s", the compression processing condition "compression processing", etc. as a shared header as restored data when compression processing is performed. In addition, the restored data DB24 stores the unit "V", the channel "CH1", the predicted values ​​"0; predicted value 0" to "94; predicted value 94", "98; predicted value 98" to "300; predicted value 300", the actual values ​​"95; 1.05", "96; 1.035", "97; 1.045", etc. associated with the time stamp number as restored data when compression processing is performed. Note that "CH2" and "CH3" shown in FIG. 5 have different units and actual values ​​and predicted values ​​of the restored data, but the data configuration is similar, so the description will be omitted. Note that the restored data is composed of the recorded data of (2) in FIG. 4 and the predicted value predicted using the prediction data.

[0046] Returning to Fig. 3, the explanation will be continued. The control unit 30 is responsible for controlling the entire data collection device 10. The control unit 30 has a compression processing unit 40, a restoration processing unit 50, and an output processing unit 60. Here, the control unit 30 can be realized by, for example, an electronic circuit such as a CPU (Central Processing Unit) or an MPU (Micro Processing Unit), or an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[0047] The compression processing unit 40 compresses the actual measured values ​​obtained from the device based on the compression processing conditions set in the shared header of the measured data and the predicted values. The compression processing unit 40 has an acquisition unit 41, a setting unit 42, a prediction unit 43, and a recording control unit 44.

[0048] The acquiring unit 41 acquires past measured values ​​obtained from the device, simulation values, measured values ​​of similar devices, input values ​​based on user experience, etc. The acquiring unit 41 also acquires actual measured values ​​such as voltage values, temperature values, and current values ​​measured by the device connected to the data collecting apparatus 10 via the above-mentioned input / output unit 12.

[0049] The setting unit 42 sets the compression processing conditions in the shared header portion of the measurement data using the compression processing conditions stored in the compression processing condition DB 21. Specifically, the setting unit 42 sets the compression implementation setting, compression start condition, compression end condition, compression end operation, feature, and upper or lower limit value of the time series data of the measurement value as the compression processing conditions in the shared header portion of the measurement data. For example, the setting unit 42 sets the compression implementation setting "ON", the compression start condition "start from 8:10 on 1 October 2022", the compression end condition "until a user operation is accepted", the compression end operation "end as is", the feature "function that returns a constant for time", the upper or lower limit value of the time series data of the measurement value "upper limit value: +1, lower limit value: -1", etc. as the compression processing setting "compression processing".

[0050] In addition, regarding the compression processing conditions described above, the setting unit 42 can also input predictive data values ​​(data values ​​such as past measurement values ​​of devices such as field equipment, measurement values ​​of similar devices, simulation values, etc.) into a trained machine learning model that takes data values ​​as input and outputs feature values, and set the output feature values ​​as compression processing conditions.

[0051] Furthermore, the setting unit 42 sets a "range" used by the compression processing unit 40 described later to determine whether or not to record an actual measurement value obtained from a device, using a predicted value predicted by the prediction unit 43 described later and an upper limit value and a lower limit value included in the compression processing conditions. The setting unit 42 may set a range based on compression processing conditions such as an upper limit value and a lower limit value input by a user or the like, or may set a range to be output based on a trained machine learning model that inputs past data values ​​and outputs a range or the like.

[0052] When the data collection device 10 predicts a predicted value based on the above-mentioned transition pattern and sets a compression process, it can modify an existing transition pattern to create a new transition pattern. Here, the modification of the transition pattern will be described. Figure 6 is a diagram showing an example of an extension function of the predicted value setting according to the first embodiment.

[0053] For example, as shown in Fig. 6(1), when setting the compression process, the data collection device 10 displays a graph of a transition pattern showing times such as t1, t2, t3, t4 on the X-axis and units of values ​​included in the transition pattern such as y1, y2 on the Y-axis to the user, etc. At this time, the data collection device 10 can display dashed lines of the predicted value, upper limit value, and lower limit value as shown in Fig. 6(1).

[0054] The data collection device 10 can change the displayed transition pattern based on the operation of the user or the like. For example, as shown in FIG. 6 (2), the data collection device 10 can change the data on the X-axis to "t1→t1'", "t2→t2'", "t3→t3'", "t4→t4'". Also, as shown in FIG. 6 (3), the data collection device 10 can change the data on the X-axis to "t1→t1'', ​​"t2→t2'', "t3→t3'', "t4→t4''". Also, although not shown in FIG. 6, the data collection device 10 can change the transition pattern on the Y-axis based on the operation of the user or the like, similar to the above-mentioned example of the X-axis.

[0055] The prediction unit 43 predicts a predicted value based on the feature amounts included in the compression processing conditions (including the feature amounts output from the model). Details of the prediction process using each feature amount will be explained in a later section, and a specific example of the prediction process by the prediction unit 43 will be described below.

[0056] As a first example of a feature, the prediction unit 43 predicts a predicted value based on a data value for prediction and a function that returns a constant for a time associated with the prediction data (hereinafter, may be referred to as a "constant conversion function"). For example, the prediction unit 43 predicts a predicted value "A1, A2, A3,..." from times "t1, t2, t3,..." based on a function that returns a constant value "A" when time "t" is input. The constant conversion function described above is a function that returns a constant number for an input of time, and is therefore used in cases where the measured value obtained by the device is predicted to be constant or fall within a certain range.

[0057] As a second example of the feature, the prediction unit 43 predicts a predicted value based on a line segment connecting break points in a time series. For example, the prediction unit 43 predicts, as a predicted value, a line segment connecting break points included in a break point table "(t1, v1), (t2, v2), ... (tn, vn)" in which a break point "v" is associated with time "t". The break point table described above represents one or more fluctuation points in the case of a control pattern based on a recipe. Note that the line predicted as a predicted value may be represented by a line connecting break points, or may be represented by a mathematical formula including parameters such as the slope of a line connecting multiple break points.

[0058] As a third example of the feature quantity, the prediction unit 43 predicts a predicted value based on a line segment connecting break points in the time series and a first-order lag function with a time constant set as a parameter. For example, the prediction unit 43 predicts a first-order lag function based on a line segment connecting break points included in the break point table "(t1, v1), (t2, v2), ... (tn, vn)" and a set time constant as a predicted value. The above-mentioned time constant is an amount representing the speed of change from a transient state to a steady state for a value that changes in a time series, and the smaller the time constant, the faster the change speed, and the larger the time constant, the slower the change speed. Note that the curve predicted as a predicted value may be represented by a line segment expressed by a first-order lag function connecting break points, or may be represented by a mathematical expression including a parameter such as the slope of a line segment connecting multiple break points.

[0059] As a fourth example of the feature, the prediction unit 43 predicts a prediction value based on a time-series transition pattern of the prediction data values. For example, the prediction unit 43 generates a set of prediction data values ​​at each time associated with the prediction data, and predicts a transition pattern "v1, v2, ... vn" as a prediction value based on the time-series transition pattern of the generated set. In addition, the prediction unit 43 predicts a transition pattern "v'1, v'2, ... v'n" as the above-mentioned transition pattern using actual measured values ​​acquired by a device similar to or related to the target device. The above-mentioned transition pattern is designed to reproduce a pattern of fluctuations in past measured values, simulation values, etc., or is designed by applying past measured values, simulation values, etc. directly to the transition pattern. In addition, the transition pattern is designed based on the measured values ​​of similar devices obtained from other similar devices, so as to reproduce a pattern of fluctuations in the measured values ​​of similar devices, or is designed by applying the measured values ​​of similar devices directly to the transition pattern.

[0060] The prediction unit 43 can also acquire measurements using a function that references data of a specified similar device, and design a transition pattern based on the measurements of the similar device. The prediction unit 43 can also design a transition pattern based on the measurements of the similar device, using the data of the specified similar device acquired in advance as is.

[0061] The prediction unit 43 can predict a predicted value based on features obtained by inputting data values ​​such as past measurement values ​​of devices such as field equipment, measurement values ​​of similar devices, and simulation values ​​into a trained machine learning model such as a supervised learning model, an unsupervised learning model, or a reinforcement learning model that takes data values ​​as input and outputs features.

[0062] The recording control unit 44 determines whether the acquired actual measurement value is to be recorded based on the set compression processing conditions and the predicted value, and if it is to be recorded, stores the actual measurement value in the measurement data DB 23.

[0063] The first embodiment is an embodiment in which the "outside" of the set range is recorded as described above. Therefore, the recording control unit 44 according to the first embodiment records the actual measurement value outside the range determined by adding or multiplying the upper limit value and subtracting or multiplying the lower limit value to the predicted value predicted by the prediction unit 43. For example, when the range is determined based on the predicted value "2", the upper limit value "predicted value + 1", and the lower limit value "predicted value - 1", the recording control unit 44 records the actual measurement value "1 or less, or 3 or more" as the recording target. Note that the conditions of "less than" and "more than" here can be changed to "exceed" and "less than" as appropriate.

[0064] The recording control unit 44 also sets time stamp numbers to the actual measurement values ​​to be recorded, the compression process start flag, and the compression process end flag as serial numbers for identifying a time series for each channel assigned to the device. For example, as shown in FIG. 4 (2), the recording control unit 44 sets the time stamp number "0" to the compression process start flag "0x7fffffff" for the individual data of CH1. The recording control unit 44 also sets time stamp numbers "95 to 97" to the recorded actual measurement values ​​"1.05, 1.035, 1.045", respectively. The recording control unit 44 also sets the time stamp "300" to the compression process end flag "0x7ffffffe". The recording control unit 44 can set the same time stamp when the uncompressed actual measurement value and the compression process start flag or the compression process end flag are at the same timing.

[0065] The restoration processing unit 50 performs restoration processing using actual measured values ​​stored in the measurement data DB 23 and predicted values ​​predicted based on the feature quantities. Specifically, for a timestamp section in which an actual measured value is recorded, the restoration processing unit 50 reads the actual measured value from the measurement data DB 23 and restores the restored data. In addition, for a timestamp section in which an actual measured value is not recorded, the restoration processing unit 50 restores a predicted value based on the feature quantities set in the compression processing conditions. An example of prediction processing in a non-recorded section by the restoration processing unit 50 will be described below.

[0066] The restoration processing unit 50 restores the predicted value based on a function that returns a constant for the time associated with the prediction data in the section where the actual measurement value is not recorded. For example, the restoration processing unit 50 converts the time "t1, t2, t3,..." of the timestamp section where the actual measurement value is not recorded to a constant "A1, A2, A3,..." based on the above-mentioned function to restore the predicted value. Note that the above-mentioned "time" may use the timestamp number, or may use the measurement time calculated based on the measurement start date and time, the measurement period, and the timestamp number.

[0067] The restoration processing unit 50 restores the predicted value based on the line segment connecting the break points of the time series in the section where the actual measurement value is not recorded. For example, the restoration processing unit 50 restores the predicted value corresponding to the time stamp section where the actual measurement value is not recorded based on the above-mentioned break point table "(t1, v1), (t2, v2), ... (tn, vn)".

[0068] The restoration processing unit 50 restores the predicted value in the section where the actual measurement value is not recorded, based on the line segment connecting the break points of the time series and the first-order lag function in which the time constant is set as a parameter. For example, the restoration processing unit 50 restores the predicted value corresponding to the time stamp section where the actual measurement value is not recorded, based on the above-mentioned break point table "(t1,v1),(t2,v2),...(tn,vn)" and the first-order lag function.

[0069] The restoration processing unit 50 generates a set of prediction data values ​​at each time associated with the prediction data in a section where no actual measurement value is recorded, and restores a predicted value based on a time series transition pattern of the generated set. For example, the restoration processing unit 50 restores a predicted value corresponding to a time stamp section where no actual measurement value is recorded, based on the above-mentioned transition pattern "v1, v2, ... vn."

[0070] The output processing unit 60 outputs a time-series graph using the restored data restored by the restoration processing unit 50. Specifically, the output processing unit 60 outputs a time-series graph, a data table, or the like created using the recorded actual measurement values ​​and time-series data of the measurement values ​​restored using the feature amounts for the sections where the actual measurement values ​​are not recorded.

[0071] Furthermore, the output processing unit 60 outputs a compression ratio calculated based on the ratio of the number of recorded actual measurement values ​​to the total number of actual measurement values ​​that can be acquired by the device. As an example, when there are 100 uncompressed actual measurement values ​​and 20 recorded actual measurement values, the output processing unit 60 can output the compression ratio as "80%" or the like in the form of numbers, text, images, graphs, etc.

[0072] (Procedure for setting compression conditions) From here, the flow of various processes performed by the data collection device 10 according to the first embodiment will be described. First, a detailed procedure for "setting compression processing conditions" will be described. Fig. 7 is a flowchart of the compression processing condition setting procedure according to the first embodiment.

[0073] 7, the data collection device 10 accepts a command to start or stop the compression process (S101). If compression is not to be performed (S102; No), the data collection device 10 ends the process of setting compression process conditions. On the other hand, if compression is to be performed (S102; Yes), the data collection device 10 performs the following process.

[0074] Here, when a feature for predicting a predicted value is generated based on a machine learning model (S103; Yes), the setting unit 42 inputs the data value for prediction to a trained machine learning model that inputs a data value and outputs a feature, and generates a feature (S104). Then, the receiving unit 14 receives the feature generated based on the machine learning model (S105).

[0075] On the other hand, if the feature is not generated based on the machine learning model (S103; No), the reception unit 14 receives the feature input by the user or the like (S105). Next, the reception unit 14 receives compression execution conditions for performing the compression process (S106).

[0076] The prediction unit 43 predicts a predicted value based on the received feature amount (S107). Then, the display unit 15 displays the predicted value (S108). Note that the display unit 15 may display the feature amount used for the prediction without displaying the predicted value, in cases such as when referring to actual measured values ​​of a similar device for which the predicted value is displayed after measurement.

[0077] Here, if the range used as a criterion for determining whether or not to record the actual measured value is to be generated based on a machine learning model (S109; Yes), the setting unit 42 generates a range based on a trained machine learning model that inputs past data values, etc. and outputs a range (S110). Then, the receiving unit 14 receives the range generated based on the machine learning model (S111). On the other hand, if the range is not to be generated based on the above-mentioned machine learning model (S109; No), the receiving unit 14 receives a range input by a user, etc. (S111). Then, the data collection device 10 ends the process.

[0078] (Measurement data recording: Measurement data collection procedure) Next, a detailed procedure of "measurement data recording: measurement data collection" will be described. Fig. 8 is a flowchart of the measurement data collection procedure according to the first embodiment.

[0079] 8, the setting unit 42 reads out the compression processing conditions (S201). The setting unit 42 sets the read out compression processing conditions in the header of the measurement data (S202).

[0080] Next, the setting unit 42 initializes the internal processing state (S203). For example, when the compression process is ON, the setting unit 42 sets the compression function to "enabled" and the compression status to "not compressing". When the compression process is OFF, the setting unit 42 sets the compression function to "disabled" and the compression status to "not compressing".

[0081] Here, the data collecting device 10 waits for the start of the measurement period (S204; No). If the measurement period starts (S204; Yes), the compression processing unit 40 reads out the measurement data from the input / output unit 12, etc. (S205). Then, the compression processing unit 40 performs the compression process (S206).

[0082] Here, the compression processing unit 40 continues the process if the data collection is not to be ended (S207; No). On the other hand, if the data collection is to be ended (S207; Yes), the data collection device 10 ends the process.

[0083] (Measurement data recording: compression processing procedure) Next, a more detailed procedure of "Performing compression processing" explained in S206 of Fig. 8 will be explained. Fig. 9 is a flowchart of the compression processing procedure according to the first embodiment. Note that after the steps in the flowchart of Fig. 9 are completed, the steps from S207 onwards in Fig. 8 will continue.

[0084] 9, the compression processing unit 40 checks the internal processing state (S301). If the internal processing state indicates that the compression function is "disabled" (S302; No), the compression processing unit 40 writes out the actual measurement value and stores it in the measurement data DB 23 (S303). Then, the data collection device 10 ends the process.

[0085] On the other hand, if the internal processing state is that the compression function is "enabled" (S302; Yes), the compression status is "not compressing" and the compression start condition is met (S304; Yes), the compression processing unit 40 updates the internal processing state to "compressing" (S305). Next, the compression processing unit 40 initializes the prediction algorithm for the predicted value by returning the reference position of the break point table and transition pattern to the beginning and resetting the time parameter of the function to 0 (S306). Then, the compression processing unit 40 writes out compression start flag data and stores it in the measurement data DB 23 (S307).

[0086] On the other hand, if the compression status is "not being compressed" and the compression start condition is not met (S304; No), the compression processing unit 40 skips the processes from S305 to S307.

[0087] If the internal processing state indicates that the compression is in progress (S308; Yes), the compression processing unit 40 predicts the predicted value (S309). On the other hand, if the internal processing state indicates that the compression is not in progress (S308; No), the compression processing unit 40 writes out the actual measured value and stores it in the measurement data DB 23 (S303). Then, the data collection device 10 ends the process.

[0088] If the measured value is within the range of the predicted values ​​(S310; Yes), the compression processing unit 40 does not write out the measured value. On the other hand, if the measured value is not within the range of the predicted values ​​(S310; No), the compression processing unit 40 writes out the measured value and stores it in the measurement data DB 23 (S311). The compression processing unit 40 can calculate the compression rate up to this point.

[0089] The compression processing unit 40 prepares to predict the next predicted value (S312). If the compression end condition is satisfied (S313; Yes), the compression end flag data is written out and stored in the measurement data DB 23 (S314). On the other hand, if the compression end condition is not satisfied (S313; No), the data collection device 10 ends the process.

[0090] The compression processing unit 40 performs a predetermined compression end operation (S315). Specifically, when waiting for a start condition (S315; (1)), the compression processing unit 40 updates the compression status as an internal processing state to "not compressing" (S316). Then, the data collection device 10 ends the process.

[0091] If the compression is completed (S315; (2)), the compression processing unit 40 updates the internal processing state to "not compressing" and the compression function to "disabled" (S317). Then, the data collection device 10 ends the process.

[0092] Furthermore, when the compression process is repeated (S315; (3)), the compression processing unit 40 initializes the prediction value calculation algorithm (S318). Then, the compression start flag data is written out and stored in the measurement data DB 23 (S319). Then, the data collection device 10 ends the process.

[0093] (Measurement data reading procedure) Next, a detailed procedure for "reading recorded measurement data" will be described. Fig. 10 is a flowchart showing the measurement data reading procedure according to the first embodiment.

[0094] 10, the restoration processing unit 50 updates the compression status as an internal processing state to "not compressing" (S401). Next, the restoration processing unit 50 reads out the timestamp range information and compression processing settings (compression processing ON / OFF application conditions, feature amount, range, etc.) included in the measurement data (S402).

[0095] The restoration processing unit 50 initializes the time stamp number (S403). The restoration processing unit 50 performs data restoration processing (S404). The restoration processing unit 50 performs processing such as displaying data and converting data as the time stamp data (S405).

[0096] If the data to be restored is not the end data (S406; No), the restoration processing unit 50 increments the time stamp number (S407). Then, the restoration processing unit 50 returns to the previous step and continues the process.

[0097] On the other hand, if the target of the restoration process is end data (S406; Yes), the data collection device 10 ends the process.

[0098] (Data recovery procedure) Next, a detailed procedure will be described for "Performing data restoration processing" explained in S404 of Fig. 10. Fig. 11 is a flowchart of the data restoration procedure according to the first embodiment. Note that after the steps in the flowchart of Fig. 11 are completed, the steps from S405 onwards in Fig. 10 will continue.

[0099] 11, the restoration processing unit 50 reads out data of a timestamp number from the measurement data (S501). For example, the restoration processing unit 50 reads out all data of the same timestamp for uncompressed actual values, compression start flag data, compression end flag data, etc. included in the measurement data.

[0100] If the read data has a compression process start flag recorded therein (S502; Yes), the restoration processing unit 50 updates the internal processing status of the compression to "compressing" (S503).Then, the restoration processing unit 50 initializes the prediction algorithm for the predicted value (S504).On the other hand, if the read data has no compression process start flag recorded therein (S502; No), the restoration processing unit 50 skips the processes of S503 and S504.

[0101] If the read data contains uncompressed actual values ​​(S505; Yes), the restoration processing unit 50 uses the uncompressed actual values ​​as restored data (S506). On the other hand, if the read data does not contain uncompressed actual values ​​(S505; No) and the compression status is "compressing" (S507; Yes), the restoration processing unit 50 calculates and determines the range of predicted values ​​from the set value as the data of the timestamp (S508). If the read data does not contain uncompressed actual values ​​(S505; No) and the compression status is "not compressing" (S507; No), the restoration processing unit 50 determines an error (S509) and ends the process.

[0102] If the compression process end flag is recorded (S510; Yes), the restoration processing unit 50 updates the compression status as the internal processing status to "not compressing" (S511). Then, the data collection device 10 ends the process.

[0103] On the other hand, if the compression process end flag is not recorded (S510; No) and compression is in progress (S512; Yes), the restoration processing unit 50 prepares to predict the next predicted value (S513). Then, the data collection device 10 ends the process.

[0104] On the other hand, if the compression process end flag has not been recorded (S510; No) and compression is not in progress (S512; No), the restoration processing unit 50 skips the step of S513 and ends the process.

[0105] (Specific examples) Hereinafter, a specific example of the compression process, restoration process, and output process of the actual measurement values ​​obtained from the device by the data collection device 10 will be described. Fig. 12 to Fig. 15 are diagrams showing an example of the data collection process according to the first embodiment.

[0106] (Example of data compression based on constant conversion functions) A specific example using a constant conversion function, which is an example of a feature, will be described with reference to Fig. 12. As shown in Fig. 12, in a case where the actual measurement value is constant or falls within a certain range, the data collection device 10 performs a compression process of the actual measurement value based on a predicted value predicted using the constant conversion function described above, a restoration process of the actual measurement value and the predicted value, and a process of outputting a time series graph.

[0107] FIG. 12 shows a step of determining whether or not actual measured values ​​obtained from a device shown in FIG. 12(1) are to be recorded, a step of recording the actual measured values ​​to be recorded shown in FIG. 12(2), and a step of restoring the actual measured values ​​and predicted values ​​and outputting a time series graph using the restored data shown in FIG. 12(3).

[0108] First, in step (1) of FIG. 12, the data collecting device 10 acquires actual measurement values ​​and the like shown in (1-1) to (1-3) of FIG. 12 in chronological order.

[0109] The data collection device 10 predicts a straight line of constant values ​​as predicted values, as shown in FIG. 12(1), based on the data values ​​for prediction and the constant conversion function.

[0110] As shown in (2-1) of Fig. 12, the data collection device 10 sets a range having a certain width in both directions of the Y axis based on the predicted value displayed as a straight line of the above-mentioned constant value. For example, the data collection device 10 adds an upper limit value "1" to the predicted value displayed as a straight line of the above-mentioned constant value and subtracts a lower limit value "1" to set a range having a certain width in both directions of the Y axis based on the predicted value.

[0111] The data collection device 10 determines that actual measurement values ​​indicated by a "circle" outside the range shown in Fig. 12(1) and error values ​​indicated by an "x within a circle" are to be recorded. On the other hand, the data collection device 10 determines that actual measurement values ​​indicated by a "black circle" that are included in the range shown in Fig. 12(1) are not to be recorded. As a result, the data collection device 10 records only actual measurement values ​​to be recorded, as shown in Fig. 12(2).

[0112] The data collection device 10 reads out and restores the actual measurement values ​​indicated by "○" in (3-1) etc. of Fig. 12 and the error values ​​indicated by "× in a circle" in (3-2) etc. of Fig. 12 as recorded actual measurement values ​​from the storage unit 20. Then, the data collection device 10 plots the restored actual measurement values ​​shown in (3-1), (3-2) etc. of Fig. 12 on a time-series graph.

[0113] On the other hand, for a section of a timestamp number where no actual measurement value is recorded, the data collection device 10 restores a predicted value by using the corresponding timestamp number and the above-mentioned constant conversion function so as to fill the section where no actual measurement value is recorded.Then, the data collection device 10 plots the restored predicted value shown in (3-3) of FIG. 12 on a time series graph so as to fill the section of the timestamp number where no actual measurement value is recorded by connecting the predicted value to the restored actual measurement value.

[0114] The data collection device 10 connects the restored actual values ​​and the predicted values ​​with a line segment, and outputs a time series graph showing the range of the error bars on both sides of the Y axis as shown in (3) of Fig. 12. Note that the actual values ​​shown in (3-2) of Fig. 12 and the like are not included in the depiction of the line segment because they are treated as error values.

[0115] In this way, the data collection device 10 can set a predicted value of a constant and a range corresponding to the predicted value, compress the actual measured value based on the range, restore the actual measured value and the predicted value, and output a time series graph.

[0116] (Example of data compression based on breakpoint tables) A specific example of using the break point table will be described with reference to Fig. 13. As shown in Fig. 13, in a case where the transition of the time series of the actual measurement values ​​is predicted to vary for each time period, the data collection device 10 performs compression processing of the actual measurement values ​​based on the predicted value, restoration processing of the actual measurement values ​​and the predicted value, and output processing of a time series graph, using a line segment connecting the data values ​​included in the break point table as a predicted value.

[0117] FIG. 13 shows a step of determining whether or not actual measured values ​​obtained from a device shown in FIG. 13(1) are to be recorded, a step of recording the actual measured values ​​to be recorded shown in FIG. 13(2), and a step of restoring the actual measured values ​​and predicted values ​​and outputting a time series graph using the restored data shown in FIG. 13(3).

[0118] First, in step (1) of FIG. 13, the data collecting device 10 acquires the actual measurement values ​​and the like shown in (1-1) and (1-2) of FIG. 13 in chronological order.

[0119] Based on the line segments connecting the data values ​​included in the break point table, the data collection device 10 predicts a straight line connecting each point at specific times (t1, y1), (t2, y2), (t3, y2), and (t4, y1) as a predicted value, as shown in Figure 13 (1).

[0120] The data collection device 10 sets a range having a certain width in both directions of the Y axis based on the predicted value displayed as a line segment connecting the above-mentioned break points, as shown in (2-1) of Fig. 13. For example, the data collection device 10 adds an upper limit value "1" to the predicted value displayed as a line segment connecting the above-mentioned break points, and subtracts a lower limit value "1" to set a range having a certain width in both directions of the Y axis based on the predicted value.

[0121] The data collection device 10 determines that the actual measurement values ​​indicated by "○" outside the range shown in Fig. 13(1) are to be recorded. On the other hand, the data collection device 10 determines that the actual measurement values ​​indicated by "●" included within the range shown in Fig. 13(1) are to be recorded. As a result, the data collection device 10 records only the actual measurement values ​​to be recorded, as shown in Fig. 13(2).

[0122] The data collection device 10 reads out and restores the actual measurement values ​​indicated by "◯" in (3-1) of Fig. 13 from the storage unit 20 as recorded actual measurement values. Then, the data collection device 10 plots the restored actual measurement values ​​shown in (3-1) of Fig. 13 on a time-series graph.

[0123] On the other hand, for a section of a timestamp number where no actual measurement value is recorded, the data collection device 10 restores a predicted value by using a breakpoint table corresponding to the corresponding timestamp number so as to fill in the section where no actual measurement value is recorded.Then, the data collection device 10 plots the restored predicted value shown in (3-2) of FIG. 13 on a time series graph so as to fill in the section of the timestamp number where no actual measurement value is recorded by connecting the predicted value to the restored actual measurement value.

[0124] The data collection device 10 connects the restored actual values ​​and predicted values ​​with a line segment, and outputs a time-series graph showing the range of the data on the Y-axis with error bars in both directions, as shown in FIG. 13(3).

[0125] In this way, the data collection device 10 can set a predicted value represented by a line segment connecting the break points and a range corresponding to the predicted value, and perform compression processing of actual measured values ​​based on the range, restoration processing of actual measured values ​​and predicted values, and output processing of a time series graph.

[0126] (Example of data compression based on breakpoint table and first-order lag function) A specific example using a break point table and a first-order lag function with a time constant as a parameter will be described with reference to Fig. 14. As shown in Fig. 14, in a case where the transition of the actual measurement value is predicted to show the behavior of a first-order lag system, the data collecting device 10 performs a compression process of the actual measurement value based on the predicted value, a restoration process of the actual measurement value and the predicted value, and an output process of a time series graph, using a line segment represented based on the break point table and the first-order lag function as a predicted value.

[0127] FIG. 14 shows a step of determining whether or not actual measured values ​​obtained from a device shown in FIG. 14(1) are to be recorded, a step of recording the actual measured values ​​to be recorded shown in FIG. 14(2), and a step of restoring the actual measured values ​​and predicted values ​​and outputting a time series graph using the restored data shown in FIG. 14(3).

[0128] First, in step (1) of FIG. 14, the data collecting device 10 acquires the actual measurement values ​​and the like shown in (1-1) and (1-2) of FIG. 14 in chronological order.

[0129] The data collection device 10 predicts the predicted value based on the breakpoint table and the first-order lag function. For example, as shown in (1) of Fig. 14, the breakpoint table is depicted as fluctuating from y1 to y2 (steady state) at time "t1" based on the time constant. However, the transition of the predicted actual measurement value does not show a linear fluctuation at t1, but shows a behavior of gradually rising in a curve with t1 as the boundary and fluctuating from y1 to y2, which is the steady state.

[0130] Therefore, in this example, the data collecting device 10 predicts the line segment connecting the break points as a line segment of a first-order lag function with a time constant as a parameter based on a known method. For example, as shown in Fig. 14(1), the data collecting device 10 predicts a line segment that starts at time "t1", gradually rises from y1 to y2, and reaches y2, which is a steady state, at time "t1'".

[0131] The data collection device 10 sets a range having a certain width in both directions of the Y axis based on the predicted value displayed as a first-order lag function with a time constant applied to the line segment connecting the above-mentioned break points. For example, the data collection device 10 adds an upper limit value "1" to the predicted value displayed as a first-order lag function with a time constant applied to the line segment connecting the above-mentioned break points, and subtracts a lower limit value "1" to set a range having a certain width in both directions of the Y axis based on the predicted value.

[0132] The data collection device 10 determines that the actual measurement values ​​indicated by "○" outside the range shown in Fig. 14(1) are to be recorded. On the other hand, the data collection device 10 determines that the actual measurement values ​​indicated by "●" included within the range shown in Fig. 14(1) are to be recorded. As a result, the data collection device 10 records only the actual measurement values ​​to be recorded, as shown in Fig. 14(2).

[0133] The data collection device 10 reads out and restores the actual measurement values ​​indicated by "○" in (3-1) of Fig. 14 from the storage unit 20 as recorded actual measurement values. Then, the data collection device 10 plots the restored actual measurement values ​​shown in (3-1) of Fig. 14 on a time-series graph.

[0134] On the other hand, for a section of a timestamp number where no actual measurement value is recorded, the data collecting device 10 restores a predicted value based on the corresponding timestamp number, the line segment connecting the above-mentioned break points, and the time constant so as to fill the section where no actual measurement value is recorded.Then, the data collecting device 10 plots the restored predicted value shown in (3-2) of FIG. 14 on a time series graph so as to fill the section of the timestamp number where no actual measurement value is recorded by connecting the predicted value to the restored actual measurement value.

[0135] The data collection device 10 connects the restored actual values ​​and predicted values ​​with a line segment, and outputs a time-series graph showing the range of the data on the Y-axis with error bars in both directions, as shown in FIG. 14(3).

[0136] In this way, the data collection device 10 can set a predicted value expressed by a first-order lag function with a time constant applied to the line segment connecting the break points, and a range corresponding to the predicted value, and can perform compression processing of the actual measured value based on the range, data restoration processing, and output processing of a time series graph.

[0137] (Example of data compression based on transition patterns) A specific example using a transition pattern designed based on past measurement values, simulation values, measurement values ​​of similar devices, etc. will be described with reference to Fig. 15. As shown in Fig. 15, in a case where the transition of past measurement values, simulation values, measurement values ​​of similar devices, etc. is predicted to be similar to the transition of actual measurement values, the data collection device 10 predicts a predicted value using the transition pattern, performs compression processing of the actual measurement value based on the predicted predicted value, restoration processing of the actual measurement value and the predicted value, and output processing of a time series graph.

[0138] FIG. 15 shows a step of acquiring actual measured values ​​from a device shown in FIG. 15(1), a step of determining whether or not the actual measured values ​​are to be recorded shown in FIG. 15(2), a step of recording the actual measured values ​​to be recorded shown in FIG. 15(3), and a step of restoring the recorded actual measured values ​​and predicted values ​​and outputting a time series graph using the restored data shown in FIG. 15(4).

[0139] First, in step (1) of FIG. 15, the data collecting apparatus 10 acquires actual measurement values ​​shown in (1-1) of FIG. 15 and the like from the device.

[0140] The data collection device 10 predicts a predicted value by reading values ​​included in the designed transition pattern and plotting them in a time series, as shown in (2) of Fig. 15. Then, as shown in (2) of Fig. 15, the data collection device 10 judges whether the acquired actual measurement value is a measurement value to be recorded, as shown in (2-1) of Fig. 15, or a measurement value not to be recorded, as shown in (2-2) of Fig. 15, for example.

[0141] The data collection device 10 sets the range displayed as error bars in both directions of the Y axis based on the above-mentioned transition pattern as the range, as shown in (2-3) of Fig. 15. For example, the data collection device 10 sets the range displayed as error bars in both directions of the Y axis as the range by adding an upper limit value "1" and subtracting a lower limit value "1" based on the above-mentioned transition pattern.

[0142] The data collection device 10 records the actual measurement values ​​indicated by "○" outside the range shown in (3-1) of Fig. 15 as actual measurement values ​​to be recorded. On the other hand, the data collection device 10 does not record the actual measurement values ​​indicated by "●" included within the range shown in (3) of Fig. 15 as actual measurement values ​​not to be recorded.

[0143] When reading or displaying, the data collection device 10 reads out and restores the actual measurement values ​​indicated by "◯" in (4-1) etc. of Fig. 15 as recorded actual measurement values ​​from the storage unit 20. Then, the data collection device 10 plots the restored actual measurement values ​​shown in (4-1) etc. of Fig. 15 on a time-series graph.

[0144] On the other hand, for a section of a timestamp number where no actual measurement value is recorded, the data collection device 10 restores a predicted value by using the corresponding timestamp number and the above-mentioned transition pattern so as to fill the section where no actual measurement value is recorded.Then, the data collection device 10 plots the restored predicted value shown in (4-2) of FIG. 15 on a time series graph so as to fill the section of the timestamp number where no actual measurement value is recorded by connecting the predicted value to the restored actual measurement value.

[0145] The data collection device 10 connects the restored actual values ​​and predicted values ​​with a line segment, and outputs a time series graph showing the range of the error bars on both sides of the Y axis, as shown in Figure 15 (4-2) etc.

[0146] In this way, the data collection device 10 can set a predicted value and a range corresponding to the predicted value, compress the actual measured value based on the range, restore the actual measured value and the predicted value, and output a time series graph.

[0147] (effect) As described above, the acquisition unit 41 of the data collection device 10 according to the first embodiment acquires prediction data values ​​used to predict measurement values ​​obtained from a device. The prediction unit 43 of the data collection device 10 predicts time-series data of the measurement values ​​using the calculated prediction data values. Then, when an actual measurement value is acquired from the device, the recording control unit 44 of the data collection device 10 controls the recording of the actual measurement value based on the predicted time-series data of the measurement values ​​and the actual measurement value. Therefore, the first embodiment has the following effects.

[0148] The data collection device 10 can compress data efficiently by recording only actual measured values ​​that are not included in a range set based on a predicted value. For example, in a system that monitors and ensures that the actual measured values ​​to be recorded are within a certain range, the data collection device 10 can provide high compression efficiency by recording only the minority actual measured values ​​that are not within the certain range. Therefore, the data collection device 10 can achieve effects such as reducing the number of work steps and costs by reducing the frequency of replacing the recording medium by extending the storage time per recording medium.

[0149] Moreover, the predicted value in the first embodiment is predicted based on past measured values, simulation values, input values ​​based on user experience, and the like. Therefore, it is considered that a normal measured value falls within a range set based on the predicted value. On the other hand, a measured value that exists outside the range is an actual measured value that is difficult to predict, such as an error value, a saturated value, or an abnormal value caused by an abnormality in equipment, etc., and is important data for managing facilities and equipment. Therefore, the data collection device 10 according to this embodiment can achieve both the preservation of the above-mentioned important data and data compression by recording an actual measured value that is outside the range set based on the predicted value.

[0150] Furthermore, the data collection device 10 controls the recording of the actual measurement value based on whether or not the actual measurement value is included within the range. Therefore, the data collection device 10 can temporarily store the actual measurement value and record important data that falls outside the range in real time without determining whether or not to record it.

[0151] The data collection device 10 also controls the recording of the actual measured values ​​based on the feature values ​​obtained based on the data values ​​for prediction, such as past measured values, simulation values, and input values ​​based on the user's experience. In other words, the predicted values ​​predicted based on the feature values ​​can be said to be normal data transitions that can be predicted in advance. Therefore, the data collection device 10 can measure the control quality of devices with fixed control methods by predicting the predicted values ​​using the feature values. For example, in the case of a device that monitors the temperature of a drug warehouse, the data collection device 10 records only the actual measured values ​​obtained from the device that fall outside the range set based on the predicted values. Then, the data collection device 10 can output information on the quality, such as the rate of temperature fluctuation in the drug warehouse and the status of stored items, from the rate of recorded actual measured values.

[0152] Furthermore, the data collection device 10 displays the compression ratio based on the ratio of the number of recorded actual measurement values. For example, when there are many recorded actual measurement values ​​and the compression ratio is low, the data collection device 10 can output information that can be used as an index for quality control, such as that there are many abnormal actual measurement values ​​outside the range.

[0153] Furthermore, the data collection device 10 performs compression processing based on set compression processing conditions, such as compression start time, compression end time, compression start conditions, compression end conditions, etc. For example, the data collection device 10 realizes the implementation of compression processing under various conditions, such as "compression processing starts after a certain time has elapsed since the start of measurement, or when a triggering event occurs", "compression processing starts when the actual measured value of a certain channel reaches a specified value", and "compression processing is implemented in batches". This allows the data collection device 10 to be flexibly operated, for example, by putting the compression processing on hold until the environment of the recording site becomes stable, which has the effect of making subsequent data analysis, etc., more efficient.

[0154] Furthermore, the data collection device 10 can combine multiple compression processes when performing compression processing. For example, the data collection device 10 can combine feature quantities such as "break point table x transition pattern" and appropriately change the order and number of compression processes, thereby both preserving important data and compressing the data even when the transition of the actual measured values ​​is expected to be complex.

[0155] Furthermore, when predicting a predicted value based on a transition pattern and setting a compression process, the data collection device 10 creates a new transition pattern by modifying an existing transition pattern. This allows the data collection device 10 to provide an effect of enabling a user or the like to efficiently create a transition pattern.

[0156] Furthermore, the data collection device 10 realizes real-time compression processing that achieves both the preservation of important data and data compression through the compression processing described above. This enables the data collection device 10 to perform compression processing without temporarily storing actual measured values, determining whether the temporarily stored actual measured values ​​are important, and deleting them, thereby realizing high-speed processing by a computer.

[0157] <Second embodiment> In the above first embodiment, an example was described in which an actual measurement value outside the range set based on a predicted value was recorded as the actual measurement value to be recorded. In the second embodiment, an example is described in which the data collection device 10 records an actual measurement value within the above-mentioned range as the actual measurement value to be recorded.

[0158] Here, a data collection device 10 according to a second embodiment will be described. The data collection device 10 according to the second embodiment acquires prediction data values ​​used to predict measurement values ​​obtained from a device. Next, the data collection device 10 predicts a predicted value using the acquired prediction data values. Then, the data collection device 10 records an actual measurement value that falls within a range determined based on an upper limit value and a lower limit value of the predicted value.

[0159] Here, the processing of the data collection device 10 according to the second embodiment will be specifically described. FIG. 16 is a diagram for explaining the data collection device 10 according to the second embodiment. As shown in FIG. 16, the data collection device 10 is connected to a plurality of devices 90 (90a, 90b, 90c, ...). The data collection device 10 receives actual measurement values ​​acquired by the devices from these plurality of devices 90 in chronological order as input data ((1) in FIG. 16).

[0160] As shown in (2-1) of Fig. 16, the data collection device 10 judges whether the actual measurement values ​​input in time series are "to be recorded" or "not to be recorded" based on the set compression processing conditions, and records only the actual measurement values ​​to be recorded. For example, the data collection device 10 predicts the predicted value using a method similar to that of the first embodiment, and sets upper and lower limit values ​​that indicate the allowable range of the predicted value.

[0161] Thereafter, when the data collection device 10 receives an actual measured value from each device, it determines whether the actual measured value is within the range between the upper limit and lower limit of the predicted value, and if it is within the range, it records the actual measured value in the memory unit 20, and if it is outside the range, it does not record the actual measured value.

[0162] That is, since data within a range of measured values ​​may be abnormal (important data), the data collection device 10 records only actual measured values ​​within an expected range in order to preserve important data to be used in subsequent analysis and verification, and eliminates other unnecessary data, thereby compressing the data. Therefore, the data collection device 10 can achieve both the preservation of important data and data compression.

[0163] (Data collection device configuration) Next, the functions different from those of the first embodiment will be specifically described. Since the data collection device 10 according to the second embodiment has a common functional unit with the data collection device 10 according to the first embodiment, only the recording control unit 44, which is different, will be described in this section, and the rest will be omitted as appropriate.

[0164] The recording control unit 44 according to the second embodiment records an actual measurement value that falls within a range determined by adding or multiplying an upper limit value and subtracting or multiplying a lower limit value to a predicted value predicted by the prediction unit 43. For example, when a range is determined based on a predicted value of "2", an upper limit value of "predicted value + 1", and a lower limit value of "predicted value - 1", the recording control unit 44 records an actual measurement value such as "1 or more, or 3 or less" as a recording target. Note that the conditions of "less than" and "more than" can be changed to "exceed" and "less than" as appropriate.

[0165] (effect) As described above, the data collecting device 10 according to the second embodiment will be described. The data collecting device 10 according to the second embodiment acquires prediction data values ​​used to predict measurement values ​​obtained from a device. Next, the data collecting device 10 predicts a prediction value using the acquired prediction data values. Then, the data collecting device 10 records an actual measurement value that falls within a range determined based on an upper limit value and a lower limit value of the time series data of the measurement value.

[0166] The data collection device 10 can compress data efficiently by recording only actual measurement values ​​that fall within a range set based on the predicted value. For example, in the case of a measurement in which the actual measurement values ​​to be recorded do not fall within a certain range and the majority of the actual measurement values ​​are outside the range, the data collection device 10 can record only actual measurement values ​​within the range that are considered to be abnormal, thereby preserving normal actual measurement values, which are important data, while compressing the data. In this way, the user can eliminate unnecessary data by having the data collection device 10 record only abnormal actual measurement values, which are important data, and improve the work efficiency and work accuracy of the subsequent analysis process.

[0167] <Modification> The following describes modified examples realized by the data collection device 10 according to the first and second embodiments. The modified examples may be implemented independently or in combination with each other as long as no contradiction is present.

[0168] (Numbers, etc.) The measurement data, restored data, compression processing conditions, various time series graphs, the number of devices or computers, users, names of functional parts of the data collection device 10, steps, names of steps, etc. used in the description of the above embodiment are merely examples and can be changed as desired. For example, the steps of the flowcharts shown in Figures 7 to 11 may be executed in a different order as long as there is no contradiction, and some steps may be omitted.

[0169] (Configuration of measurement data and restored data) The data collection device 10 is not limited to the data configuration of the measurement data and the restored data described above, and can store information included in the shared header and the individual data. For example, the data collection device 10 can store the measurement start date and time including units of seconds and milliseconds. The data collection device 10 can also store the measurement period including every millisecond, every minute, and every hour. The data collection device 10 can also store other units such as pressure, volumetric flow rate, and the remaining amount of a substance in a container. The data collection device 10 is not limited to the number of CHs "three" shown in this embodiment, and can store individual data for three or less or four or more CHs.

[0170] (internal state variables) The data collection device 10 can use variables set arbitrarily by a user, etc., other than "compress_available_f", which indicates whether the compression function is enabled or disabled, and "compressing_f", which indicates whether compression is in progress, as the above-mentioned internal state variables.

[0171] (Derivation of the first-order lag function and time constant) The data collection device 10 can derive the above-mentioned first-order lag function based on a known method for deriving a first-order lag system using a time constant. The time constant may be any value set by the user or a value calculated from past measured values ​​using the least squares method or the like.

[0172] (Combination of features used in compression processing) In the present embodiment, the data collection device 10 predicts the predicted value based on the feature amount, but the feature amount to be used is not limited to one, and may be combined as long as there is no contradiction. For example, the data collection device 10 may predict the predicted value using a plurality of feature amounts such as "constant conversion function x breakpoint table x transition pattern".

[0173] (Change the storage format depending on the compression ratio) The data collecting device 10 according to this embodiment can change the format of the measurement data based on the compression rate and save the data. For example, when the compression rate is below a predetermined threshold, the data collecting device 10 can save the measurement data in a format in which the measurement data is saved as a common timestamp for each channel. When the data collecting device 10 switches the above-mentioned format during measurement, the data collecting device 10 can do so, for example, in units of data blocks (units of measurement data files, etc.).

[0174] (An example of the internal processing state) In this embodiment, the data collection device 10 uses a predetermined variable to control whether the compression function is enabled or disabled, and whether the compression status is in compression or not. For example, the data collection device 10 can use variables such as "compress_available_f=1" which means that the compression function is enabled, "compress_available_f=0" which means that the compression function is disabled, "compressing_f=1" which means that compression is in progress, and "compressing_f=0" which means that compression is not in progress. However, the above-mentioned variables are merely examples, and the data collection device 10 can use any variable that can express ON / OFF of the compression function and whether compression is in progress without any particular limitations.

[0175] (system) The information including the processing procedures, control procedures, specific names, various data and parameters shown in the above documents and drawings can be changed arbitrarily unless otherwise specified.

[0176] In addition, each component of each device shown in the figure is a functional concept, and does not necessarily have to be physically configured as shown in the figure. In other words, the specific form of distribution and integration of each device is not limited to that shown in the figure. In other words, all or part of them can be functionally or physically distributed and integrated in any unit according to various loads, usage conditions, etc.

[0177] Furthermore, each processing function performed by each device may be realized, in whole or in part, by a CPU (Central Processing Unit) and a program analyzed and executed by the CPU, or may be realized as hardware using wired logic.

[0178] <Hardware configuration> Next, an example of a hardware configuration of the data collection device 10 will be described. Fig. 17 is a hardware configuration diagram showing an example of a computer that realizes the functions of the data collection device 10 according to this embodiment. As shown in Fig. 17, the data collection device 10 has an input / output signal receiving device 110, an input device 120, a display device 130, a communication device 140, a power supply 150, a HDD (Hard Disk Drive) 160, a processor 170, and a memory 180. Moreover, each unit shown in Fig. 17 is connected to each other by a bus or the like.

[0179] The input / output signal receiving device 110 is composed of an ADC, an input relay, etc., and is connected to external equipment that performs input / output using analog signals such as devices, converting input analog data into digital data and output digital data into analog data.

[0180] The input device 120 is a mouse, a keyboard, etc., and accepts input of information based on an operation by a user, etc. The display device 130 is a display, a printer, etc., and outputs information analyzed by the data collection device 10 to the user, etc.

[0181] The communication device 140 is a network interface card or the like, and communicates with other servers. The power supply 150 is a connector that supplies power to the data collection device 10, and is composed of an insertion plug and an insertion socket. The HDD 160 stores programs and DBs that operate the functions shown in FIG.

[0182] The processor 170 reads out a program that executes the same processes as the respective processing units shown in Fig. 17 from the HDD 160 or the like and loads it in the memory 180, thereby operating a process that executes each function described in Fig. 17 or the like. For example, this process executes a function similar to that of each processing unit of the data collection device 10. Specifically, the processor 170 executes a process that executes the same processes as the compression processing unit 40, the restoration processing unit 50, the output processing unit 60, and the like.

[0183] In this way, the data collection device 10 operates as an information processing device that executes an analysis method by reading and executing a program. The data collection device 10 can also realize functions similar to those of the above-mentioned embodiment by reading the program from a recording medium using a medium reading device and executing the read program. Note that the program in this other embodiment is not limited to being executed by the data collection device 10. For example, the present invention can be similarly applied to cases where another computer or server executes a program, or where these execute a program in cooperation with each other.

[0184] This program can be distributed via a network such as the Internet. This program can also be recorded on a computer-readable recording medium such as a hard disk, a flexible disk (FD), a CD-ROM, a magneto-optical disk (MO), or a digital versatile disk (DVD), and can be executed by being read out from the recording medium by a computer.

[0185] <Other> Some examples of combinations of the disclosed technical features are set out below.

[0186] (1) A data collection apparatus comprising: an acquisition unit that acquires prediction data values ​​used to predict measurement values ​​obtained from a device; a prediction unit that predicts time series data of the measurement values ​​using the prediction data values; and a recording control unit that, when an actual measurement value is acquired from the device, controls the recording of the actual measurement value based on the predicted time series data of the measurement value and the actual measurement value.

[0187] (2) The data collection device described in (1), wherein the prediction unit predicts the time series data of the measurement value based on the data value for prediction and a function that returns a constant for a time associated with the prediction data.

[0188] (3) The data collection device according to (1) or (2), wherein the prediction unit predicts the time series data of the measurement values ​​based on a line segment connecting data values ​​for prediction at a specific time among the time series data values.

[0189] (4) A data collection device described in any one of (1) to (3), wherein the prediction unit predicts the time series data of the measurement values ​​based on a line segment connecting data values ​​at a specific time among the data values ​​for prediction in the time series and a first-order lag function having a time constant set as a parameter.

[0190] (5) A data collection device described in any one of (1) to (4), wherein the prediction unit uses the time series of the prediction data values ​​to generate a collection of the prediction data values ​​at each time linked to the prediction data, and predicts the time series data of the measurement value based on a time series trend pattern of the generated collection.

[0191] (6) A data collection device described in any one of (1) to (5), wherein the prediction unit predicts the time series data of the measurement values ​​by inputting the data values ​​for prediction into a trained machine learning model that inputs data values ​​and outputs features, and using the features output by the trained machine learning model.

[0192] (7) A data collection device described in any one of (1) to (5), wherein the recording control unit records the actual measurement value that falls outside a range determined based on upper and lower limit values ​​of the time series data of the measurement value.

[0193] (8) A data collection device described in any one of (1) to (5), wherein the recording control unit records the actual measurement value that falls within a range determined based on upper and lower limit values ​​of the time series data of the measurement value.

[0194] (9) A data collection device according to any one of (1) to (8), further comprising a restoration processing unit that restores the recorded actual measurement values ​​and time series data of the measurement values ​​in sections in which the actual measurement values ​​are not recorded, and an output processing unit that outputs a time series graph created using the recorded actual measurement values ​​and the restored time series data of the measurement values ​​in sections in which the actual measurement values ​​are not recorded.

[0195] (10) The data collection device described in (9), wherein the restoration processing unit restores the time series data of the measurement values ​​based on a function that returns a constant for a time associated with the prediction data in a section in which the actual measurement values ​​are not recorded.

[0196] (11) The data collection device according to (9) or (10), wherein the restoration processing unit restores the time series data of the measurement values ​​based on a line segment connecting data values ​​for prediction in a time series at a specific time in a section in which the actual measurement values ​​are not recorded.

[0197] (12) A data collection device described in any one of (9) to (11), wherein the restoration processing unit restores the time series data of the measurement values ​​in a section in which the actual measurement values ​​are not recorded, based on a line segment connecting data values ​​for prediction in a time series at a specific time and a first-order lag function having a time constant set as a parameter.

[0198] (13) The data collection device according to any one of (9) to (12), wherein the restoration processing unit generates a collection of the prediction data values ​​at each time linked to the prediction data in a section in which the actual measured values ​​are not recorded, and restores the time series data of the measured values ​​based on a time series trend pattern of the generated collection.

[0199] (14) A data collection method in which a computer executes a process of acquiring prediction data values ​​used to predict measurement values ​​obtained from a device, predicting time series data of the measurement values ​​using the prediction data values, and when an actual measurement value is acquired from the device, controlling the recording of the actual measurement value based on the predicted time series data of the measurement values ​​and the actual measurement value.

[0200] (15) A data collection program that causes a computer to execute a process of acquiring prediction data values ​​used to predict measurement values ​​obtained from a device, predicting time series data of the measurement values ​​using the prediction data values, and when an actual measurement value is acquired from the device, controlling the recording of the actual measurement value based on the predicted time series data of the measurement values ​​and the actual measurement value. [Explanation of symbols]

[0201] 10 Data Collection Equipment 11 Power supply section 12 Input / output section 13. Communications Department 14 Reception Department 15 Display section 20 Memory section 21 Compression processing conditions DB 22 Prediction Value DB 23 Measurement data DB 24 Restore Data DB 30 Control section 40 Compression Processing Unit 41 Acquisition Department 42 Setting section 43 Prediction Department 44 Recording control section 50 Restoration processing section 60 Output Processing Section

Claims

1. an acquisition unit for acquiring a prediction data value used for predicting a measurement value obtained from the device; a prediction unit that predicts time-series data of the measurement values ​​by using the data values ​​for prediction; a recording control unit that, when an actual measurement value is acquired from the device, controls recording of the actual measurement value based on time series data of the predicted measurement value and the actual measurement value; A data collection device comprising:

2. The prediction unit is predicting the time series data of the measurement values ​​based on the prediction data values ​​and a function that returns a constant for a time associated with the prediction data; The data collection device of claim 1 .

3. The prediction unit is predicting the time series data of the measurement values ​​based on a line segment connecting data values ​​for prediction at a specific time among the time series data values ​​for prediction; The data collection device of claim 1 .

4. The prediction unit is predicting the time series data of the measurement values ​​based on a line segment connecting data values ​​at specific times among the time series data values ​​for prediction and a first-order lag function having a time constant set as a parameter; The data collection device of claim 1 .

5. The prediction unit is generating a collection of the data values ​​for prediction at each time associated with the prediction data using the time series of the data values ​​for prediction, and predicting the time series data of the measurement values ​​based on a time series transition pattern of the generated collection; The data collection device of claim 1 .

6. The prediction unit is A trained machine learning model receives data values ​​as input and outputs feature quantities, and predicts the time series data of the measurement values ​​using the feature quantities output by inputting the prediction data values.

6. A data collection device according to any one of claims 1 to 5.

7. The recording control unit recording the actual measurement value that falls outside a range determined based on an upper limit value and a lower limit value of the time series data of the measurement value; 6. A data collection device according to any one of claims 1 to 5.

8. The recording control unit recording the actual measurement value within a range determined based on an upper limit value and a lower limit value of the time series data of the measurement value; 6. A data collection device according to any one of claims 1 to 5.

9. a restoration processing unit that restores the recorded actual measurement values ​​and time-series data of the measurement values ​​in a section in which the actual measurement values ​​are not recorded; and an output processing unit that outputs a time series graph created using the recorded actual measurement values ​​and the time series data of the measurement values ​​restored in the section in which the actual measurement values ​​are not recorded. The data collection device of claim 1 .

10. The restoration processing unit: In a section in which the actual measurement values ​​are not recorded, the time series data of the measurement values ​​is restored based on a function that returns a constant for a time associated with the prediction data.

10. The data collection device of claim 9.

11. The restoration processing unit: restoring the time series data of the measurement values ​​based on a line segment connecting data values ​​for prediction at specific times among the time series data values ​​in a section in which the actual measurement values ​​are not recorded; 10. The data collection device of claim 9.

12. The restoration processing unit: In a section where the actual measurement values ​​are not recorded, the time series data of the measurement values ​​is restored based on a line segment connecting data values ​​for prediction at a specific time among the time series data values, and a first-order lag function having a time constant set as a parameter.

10. The data collection device of claim 9.

13. The restoration processing unit: generating a collection of the prediction data values ​​at each time associated with the prediction data in a section in which the actual measurement values ​​are not recorded, and restoring the time series data of the measurement values ​​based on a time series transition pattern of the generated collection; 10. The data collection device of claim 9.

14. The computer Obtaining prediction data values ​​to be used in predicting the measurement values ​​obtained from the device; predicting the time series data of the measurement values ​​using the prediction data values; When an actual measurement value is acquired from the device, recording of the actual measurement value is controlled based on the time series data of the predicted measurement value and the actual measurement value. Data collection methods to carry out processing.

15. On the computer, Obtaining prediction data values ​​to be used in predicting the measurement values ​​obtained from the device; predicting the time series data of the measurement values ​​using the prediction data values; When an actual measurement value is acquired from the device, recording of the actual measurement value is controlled based on the time series data of the predicted measurement value and the actual measurement value. The data collection program that runs the process.

Images