Information processing apparatus and set pressure calculation method

Abstract

The information processing device and the set pressure calculation method of the present application can calculate the set pressure of a fluid machine that ensures a necessary pressure when an end pressure fluctuates while the fluid machine is operating. The information processing device includes a learning model construction section that constructs a learning model that takes a supply pressure, an end pressure, and a pressure loss characteristic quantity at a first time as inputs and outputs an end pressure at a second time; an end state prediction section that inputs the supply pressure, the end pressure, and the pressure loss characteristic quantity at a prescribed time to the learning model and predicts an end pressure at a next time; a maximum pressure loss exploration section that explores a maximum pressure loss based on the supply pressure, the end pressure, and the pressure loss characteristic quantity while the fluid machine is operating; an optimal supply pressure exploration section that explores an optimal supply pressure based on the necessary pressure, the end pressure, and the maximum pressure loss; and a set pressure calculation section that calculates the set pressure based on the optimal supply pressure and the set pressure while operating.

Description

Technical Field

[0001] This invention relates to a technique for setting pressure in computational fluid dynamics machinery. Background Technology

[0002] Several solutions have been proposed for the above-mentioned technologies in the past.

[0003] For example, Patent Document 1 describes a compressor device comprising: a motor speed of the compressor controlled based on a detected value of the compressor's compressed gas discharge pressure; a speed-controlled compressor for capacity control to make the detected value consistent with a preset discharge pressure target value; a base load compressor for capacity control by repeatedly performing load operation and no-load operation; and a control device for controlling each of the above compressors. The control device is characterized in that it has an arithmetic unit that automatically adjusts the compressor's discharge pressure target value to keep the terminal pressure of the compressed gas user equipment constant, regardless of the magnitude of the pressure loss between the compressor's discharge outlet and the compressed gas user equipment.

[0004] The invention describes a compressor device according to Patent Document 1, which automatically adjusts the discharge pressure to a set target value based on the end pressure of the compressed gas application device, thereby providing a compressor device and control method that can operate stably without response delay and prevent unnecessary power consumption.

[0005] Existing technical documents

[0006] Patent documents

[0007] Patent Document 1: Japanese Patent Application Publication No. 2009-13961 Summary of the Invention

[0008] The technical problem that the invention aims to solve

[0009] In the manufacturing industry, efforts are underway to reduce energy consumption. In typical factories, fluid machinery, such as air compressors, accounts for 20% to 30% of the total electricity consumption, and energy-efficient operation of fluid machinery is one of the technical challenges.

[0010] The purpose of this invention is to provide a technology for calculating the set pressure of fluid machinery that can ensure the required pressure (hereinafter referred to as necessary pressure) of the end device even when the pressure of the end device (hereinafter referred to as end pressure) changes during the operation of fluid machinery.

[0011] Technical solutions to the problem

[0012] A first aspect of the present invention is an information processing apparatus for calculating the set pressure of a fluid machine, comprising: a learning model building unit for building a learning model, wherein the learning model takes as input the pressure of the fluid supplied from the fluid machine to the piping at a first moment (i.e., supply pressure), the pressure of the fluid at the end of the piping (i.e., end pressure), and a pressure loss characteristic quantity, and outputs the end pressure at a second moment after a predetermined time has elapsed since the first moment; an end state prediction unit that takes the supply pressure, end pressure, and pressure loss characteristic quantity at the predetermined time as input to the learning model built by the learning model building unit, and predicts the end pressure at the next moment; a maximum pressure loss exploration unit that explores the maximum value of pressure loss (i.e., maximum pressure loss) based on the supply pressure, end pressure, and pressure loss characteristic quantity during the operation of the fluid machine; an optimal supply pressure exploration unit that explores the optimal supply pressure that makes the end pressure higher than the necessary pressure based on the necessary pressure at the end of the piping, the end state prediction unit, and the maximum pressure loss explored by the maximum pressure loss exploration unit; and a set pressure calculation unit that calculates the set pressure of the fluid machine based on the optimal supply pressure explored by the optimal supply pressure exploration unit and the set pressure during the operation of the fluid machine.

[0013] Invention Effects

[0014] According to the present invention, even under conditions of fluctuating terminal pressure, the set pressure of the fluid machinery that ensures the necessary pressure can be calculated during operation. This contributes to energy conservation in factories using fluid machinery. Attached Figure Description

[0015] Figure 1 This is a schematic structural diagram of the air compressor system of the application information processing device in Embodiment 1 of the present invention.

[0016] Figure 2 This is a hardware structure diagram of the information processing device according to Embodiment 1 of the present invention.

[0017] Figure 3 This is a functional block diagram of the information processing device according to Embodiment 1 of the present invention.

[0018] Figure 4 This is an example of data measured in order to construct a learning model in Embodiment 1 of the present invention.

[0019] Figure 5 This is a flowchart illustrating the learning model construction process of Embodiment 1 of the present invention.

[0020] Figure 6 This is a diagram illustrating an example of the learning model of Embodiment 1 of the present invention.

[0021] Figure 7This is an example of the output screen when constructing the learning model in Embodiment 1 of the present invention.

[0022] Figure 8 This is a flowchart illustrating the setting pressure calculation process of Embodiment 1 of the present invention.

[0023] Figure 9 This is a flowchart illustrating the maximum pressure loss exploration process of Embodiment 1 of the present invention.

[0024] Figure 10 This is a flowchart illustrating the optimal supply pressure exploration process of Embodiment 1 of the present invention.

[0025] Figure 11 This is a flowchart illustrating the setting pressure calculation process of Embodiment 1 of the present invention.

[0026] Figure 12 This is an example of the screen output when calculating and setting the pressure in Embodiment 1 of the present invention.

[0027] Figure 13 This is a schematic structural diagram of the air compressor system of the application information processing device in Embodiment 2 of the present invention.

[0028] Figure 14 This is an example of data measured in order to construct a learning model in Embodiment 2 of the present invention.

[0029] Figure 15 This is a flowchart illustrating the learning model construction process of Embodiment 2 of the present invention.

[0030] Figure 16 This is a flowchart illustrating the maximum pressure loss exploration process of Embodiment 2 of the present invention. Detailed Implementation

[0031] <Example 1>

[0032] For the information processing apparatus of Embodiment 1 of the present invention, refer to Figures 1 to 12 This will be explained. Furthermore, in this specification and the accompanying drawings, components that have essentially the same functional structure are labeled with the same drawing number, thus omitting redundant explanations.

[0033] Figure 1 The schematic structure of the air compressor system 100 using the information processing device in Embodiment 1 is shown. Furthermore, in Embodiment 1, an air compressor will be used as an example of a fluid machinery.

[0034] like Figure 1As shown, the compressed air system 100 of Embodiment 1 includes an air compressor 101, an air tank 103, piping 102 and 104, and terminal devices 105 and 106. The air compressor 101 is connected to the air tank 103 via piping 102. The air tank 103 is connected to the terminal devices 105 and 106 via piping 104. A pressure gauge 107 is installed on piping 104. A pressure gauge 108 and a flow meter 109 are installed in the terminal device 105. A pressure gauge 110 and a flow meter 111 are installed in the terminal device 106.

[0035] Compressed air discharged from air compressor 101 is stored in air tank 103 via piping 102. The compressed air stored in air tank 103 is delivered to end devices 105 and 106 via piping 104. The delivered compressed air is used by end devices 105 and 106.

[0036] The pressure of the compressed air discharged from the air compressor 101 is measured by pressure gauge 107. From this, the pressure of the compressed air supplied from the air compressor 101 to the piping 104, i.e., the supply pressure, is determined. The pressure of the compressed air used by the terminal devices 105 and 106 is measured by pressure gauges 108 and 110. From this, the terminal pressure in the terminal devices 105 and 106 is determined.

[0037] The flow rate of compressed air used by terminal devices 105 and 106 is measured by flow meters 109 and 111. The flow rate of compressed air in terminal devices 105 and 106 (hereinafter referred to as terminal flow rate) is thus determined.

[0038] Additionally, in Example 1, only one air compressor was described, but multiple air compressors may be used. In this case, compressed air discharged from multiple air compressors is temporarily stored in an air tank via piping. The pressure of the compressed air stored in the air tank is measured as the supply pressure.

[0039] In addition, although Example 1 describes two terminal devices, there can also be one or more terminal devices.

[0040] Furthermore, in Embodiment 1, pressure gauges and flow meters are installed in each terminal device, but they may also be installed only in a portion of the multiple terminal devices. When pressure gauges and flow meters are installed only in a portion of the multiple terminal devices, it is preferable to install them sequentially starting from the terminal device where pressure is to be maintained. Alternatively, if there is no terminal device where pressure is to be maintained, it is preferable to install the pressure gauge and flow meter in the terminal device located furthest from the air compressor.

[0041] Furthermore, in Example 1, piping 104 includes one branch and two elbows, but the piping structure is arbitrary. The length and diameter of the piping vary depending on the factory where it is applied. Circulating piping may also be used.

[0042] Figure 2 This indicates the hardware structure of the information processing device 200.

[0043] like Figure 2 As shown, the information processing device 200 includes a control device 201, a main storage device 202, an auxiliary storage device 203, an input device 204, and an output device 205.

[0044] The control device 201 comprises one or more computing devices or combinations thereof. Examples of computing devices include a CPU (Central Processing Unit), an MPU (Micro Processor Unit), a GPU (Graphics Processing Unit), a DSP (Digital Signal Processor), and a FPGA (Field-Programmable Gate Array). The control device 201 reads the program stored in the auxiliary storage device 203 and, in cooperation with the main storage device 202, performs computing operations.

[0045] The main storage device 202 is composed of temporary storage devices or combinations thereof. Examples of main storage devices 202 include SRAM (Static Random Access Memory), DRAM (Dynamic Random Access Memory), cache, and flash memory. The main storage device 202 temporarily stores data required by the control device 201 for performing computational processing.

[0046] The auxiliary storage device 203 is composed of a non-volatile storage device or a combination thereof. Examples of the auxiliary storage device 203 include HDD (Hard Disc Drive) and SSD (Solid State Drive). The auxiliary storage device 203 stores the programs and data required for computational processing by the storage control device 201.

[0047] The input device 204 may be, for example, a keyboard, a pointing device, a touch panel, or a combination thereof. The input device 204 accepts inputs of supply pressure, end pressure, and pressure loss characteristics.

[0048] The output device 205 may be, for example, a liquid crystal display (LCD), an organic EL display, a touch panel, a projector, or a combination thereof. Prescribed information is displayed on the output device 205 through computational processing by the control device 201. The output device 205 outputs the end pressure calculated using the learning model and the end-state prediction unit 201b.

[0049] Figure 3 This indicates the functions of the information processing device 200. The functions are explained below.

[0050] The control device 201 includes a learning model construction unit 201a, an end state prediction unit 201b, a maximum pressure loss exploration unit 201c, an optimal supply pressure exploration unit 201d, and a set pressure calculation unit 201e.

[0051] The learning model construction unit 201a constructs a learning model that takes the supply pressure, terminal pressure, and pressure loss characteristic quantity (hereinafter referred to as pressure loss characteristic quantity) at a first moment as input and outputs the terminal pressure at a second moment after a predetermined time from the first moment. In the construction of the learning model in Example 1, the supply pressure measured by pressure gauge 107, the terminal pressure measured by pressure gauges 108 and 110, and the flow rate measured by flow meters 109 and 111 (which serve as pressure loss characteristic quantities) are used. The constructed learning model is stored in the learning model storage unit 203b of the auxiliary storage device 203, described later.

[0052] The end-point state prediction unit 201b inputs the supply pressure, end-point pressure, and pressure loss characteristic quantity at a specified time into the learning model, and predicts the end-point pressure at a specified time elapsed from the specified time (hereinafter referred to as the "next time"). In the end-point state prediction unit 201b of Embodiment 1, the supply pressure measured by pressure gauge 107, the end-point pressure measured by pressure gauges 108 and 110, and the flow rate measured by flow meters 109 and 111 are used for prediction.

[0053] The maximum pressure loss exploration unit 201c explores the maximum pressure loss during the operation of the air compressor 101. In Example 1, this exploration uses the flow rate measured by flow meters 109 and 111.

[0054] The optimal supply pressure exploration unit 201d explores the minimum supply pressure (hereinafter referred to as the optimal supply pressure) that satisfies the necessary pressure of the terminal devices 105 and 106, the maximum pressure loss explored in the maximum pressure loss exploration unit 201c, and the terminal state prediction unit 201b.

[0055] The set pressure calculation unit 201e calculates the set pressure based on the optimal supply pressure explored in the optimal supply pressure exploration unit 201d and the set pressure during the operation of the air compressor 101.

[0056] The main storage device 202 has various data storage units 202a. The various data storage units 202a temporarily store data required by the control device 201 to perform various processes.

[0057] The auxiliary storage device 203 includes a measurement data storage unit 203a and a learning model storage unit 203b. The measurement data storage unit 203a stores the supply pressure measured by pressure gauge 107, the terminal pressure measured by pressure gauges 108 and 110, and the flow rate measured by flow meters 109 and 111. The learning model storage unit 203b stores the learning model constructed by the learning model construction unit 201a.

[0058] The input device 204 includes a data input unit 204a for learning model construction, a runtime data input unit 204b, and various parameter input units 204c.

[0059] The output device 205 includes a data output unit 205a for learning model construction, a screen output unit 205b for learning model construction, and a runtime data output unit 205c.

[0060] The following is for reference. Figures 4 to 12 The process of calculating the set pressure of the air compressor 101 using the information processing device 200 of Embodiment 1 is described. This process consists of a learning model construction process and a set pressure calculation process.

[0061] refer to Figures 4 to 7 This explains the learning model construction process.

[0062] The learning model construction process involves building a learning model used in the pressure calculation process. It is assumed that the air compressor system 100 is constructed in a variety of factories. Therefore, when building the learning model, the characteristics of the air compressor system 100 need to be known beforehand. This is achieved by measuring pressure and flow data using pressure gauges 107, 108, and 110 and flow meters 109 and 111.

[0063] The data measurement operation is performed by the user before building the learning model. The user is, for example, a service personnel from the manufacturer of the air compressor 101. The user inputs the measured data using the input device 204 in the information processing unit 200. The input measurement data is stored in the measurement data storage unit 203a of the auxiliary storage unit 203.

[0064] Figure 4 This is an example of data measured to build a learning model.

[0065] Measurement data 400, for example, consists of time, supply pressure, terminal pressure, and terminal flow rate.

[0066] Column 401 indicates the measurement time. Column 402 indicates the pressure measured by pressure gauge 107. Column 403 indicates the pressure measured by pressure gauge 108. Column 404 indicates the flow rate measured by flow meter 109. Column 405 indicates the pressure measured by pressure gauge 110. Column 406 indicates the flow rate measured by flow meter 111. The sampling time for the measurement is arbitrary. However, it must be an interval that can measure the pressure variation caused by changes in the amount of compressed air used.

[0067] Figure 5 This represents the process of building and processing learning models.

[0068] Before building a learning model, users need to set the data to be used for building the learning model, the parameters of the learning model, and the data range.

[0069] The data used to build the learning model includes Figure 4 The measurement data shown is used by the user to set the data for building the learning model from the measurement data stored in the measurement data storage unit 203a using the input device 204.

[0070] The data interval represents the interval in the measurement data used for learning the model. The user uses input device 204 to set the data interval and the parameters of the learning model.

[0071] The data used by the user to construct the learning model, the parameters of the learning model, and the information of the data range set by the user using the input device 204 are stored in the main storage device 202 or the auxiliary storage device 203.

[0072] After the user completes the setting of the data, parameters, and data range for building the learning model, the learning model is constructed in the control device 201. The following refers to... Figure 5 Detailed explanation.

[0073] After processing begins (S501), the learning model construction unit 201a of the control device 201 calls the learning model construction data, learning model parameters, and data range information stored in the main storage device 202 or the auxiliary storage device 203. In S502, the learning model construction unit 201a constructs the learning model (f(p)). supply [k],p end [k],V end [k])). More specifically, a learning model is constructed by solving the optimization problem described in (S502) based on the measurement data within the defined data interval. Here, k represents the time in the data. p supply [k] represents the supply pressure at time k. end [k] represents the pressure of the terminal device at time k. V end [k] represents the flow rate in the terminal device.

[0074] The measured values ​​of supply pressure, terminal pressure, and terminal flow are input into the learning model constructed in S502, thereby outputting the terminal pressure at the next moment.

[0075] The function f representing the structure of the constructed learning model can be constructed using, for example, a linear regression model or a neural network. However, the model constructed here needs to represent the pressure change response, such as the corresponding change in pressure and flow rate at the terminal device. Furthermore, even when inputting certain values ​​as supply pressure, terminal pressure, and terminal flow rate to the learning model, and continuously calculating future terminal pressure, it needs to have the characteristic that the terminal pressure converges to a predetermined value (without diverging). In the case of a linear regression model, this model characteristic is achieved by determining the model parameters so that the absolute values ​​of the model's eigenvalues ​​are less than 1.

[0076] The parameters of the learning model constructed in S502 are stored in the learning model storage unit 203b of the main storage device 202 or the auxiliary storage device 203 (S503). Alternatively, the model itself can also be stored.

[0077] Figure 6 This represents an example of a saved learning model. Column 601 represents the model ID used to identify the model. Column 602 represents the name of the end device that is the object of the model. Columns 603 to 606 represent the parameters of the model calculated by the learning model construction unit 201a.

[0078] Here, for reference Figure 7 This section details how the user specifies the data used for model building, sets the model parameters, and defines the data range. Figure 7 The image shown is an example of the screen output by the output device 205 when constructing the learning model.

[0079] Set the data for building the learning model in region 701.

[0080] In zone 702, the user specifies which end devices among multiple end devices will have the learning model built for. Additionally, the user sets the ID of the built learning model in zone 702.

[0081] Set the data range in region 703.

[0082] The curve 704 represents the supply pressure and terminal pressure data in the data set in region 701 and the data range set in region 703.

[0083] The curve 705 represents the end flow data in the data range specified in region 703 within the data set in region 701.

[0084] Curve 706 represents curve 704 and the estimated value of the terminal pressure. The estimated value of the terminal pressure is calculated based on the data used to build the learning model and the terminal state prediction unit 201b.

[0085] Curve 707 shows the error between the data of the end pressure shown in curve 706 and the estimated value of the end pressure.

[0086] In any of the graphs 704, 705, 706, and 707, the horizontal axis represents time. The vertical axis of graphs 704, 706, and 707 represents pressure values, while the vertical axis of graph 705 represents flow rate values.

[0087] Region 708 represents the average prediction error shown in curve 707. Region 709 represents the maximum prediction error shown in curve 707.

[0088] The user confirms curves 706 and 707, and regions 708 and 709, thereby confirming the error level used to construct the learning model. If the values ​​displayed in regions 708 and 709 are greater than the user's predicted values, the user checks whether noise exists in the data set in region 701. If noise is present, the noisy data intervals are excluded in region 703.

[0089] Next, refer to Figures 8 to 12 This indicates the setting of pressure calculation processing.

[0090] Figure 8 This indicates the setting of pressure calculation processing.

[0091] Before calculating the set pressure, the user enters the operating data, the required pressure, and the set pressure of the air compressor as the operating setting value.

[0092] The runtime data consists of data measured by various sensors during the operation of the air compressor 101. The runtime data is related to... Figure 4 The learning model shown is constructed using the same measurement data. However, unlike the data measured to build the learning model, it is necessary to measure the characteristics of the air compressor during operation. Therefore, the operating data is preferably data measured continuously for at least one week. Furthermore, a data range is set when inputting the operating data.

[0093] The required pressure is the pressure specified by the terminal devices 105 and 106 used. Additionally, when inputting the required pressure, the number of terminal devices N should also be input.

[0094] The setpoint of the air compressor 101 during operation is the set pressure of the air compressor 101 when the running data is acquired.

[0095] These processes performed by the user are executed using the input device 204 in the information processing device 200. The runtime data, necessary pressure, and set pressure of the running air compressor input by the user using the input device 204 are stored in the main storage device 202 or the auxiliary storage device 203.

[0096] After the user completes the input of runtime data, the input of necessary pressure, and the input of the set pressure of the air compressor during operation, the set pressure is calculated in the control device 201 after these processes are completed. See below for reference. Figures 8 to 11 Detailed explanation of the pressure calculation process.

[0097] After processing begins (S801), in S802, the control device 201 calls to retrieve the runtime data, necessary pressure, and set pressure of the running air compressor stored in the main storage device 202 or auxiliary storage device 203.

[0098] In S803, the maximum pressure loss exploration unit 201c of the control device 201 explores the maximum pressure loss of the terminal devices 105 and 106 based on the running data, and extracts the maximum pressure loss characteristic quantity.

[0099] Here, for reference Figure 9 Provide a detailed explanation of the steps in S803.

[0100] Figure 9 This indicates the maximum pressure loss during the exploration process.

[0101] After the processing starts (S901), in S902, the maximum pressure loss exploration unit 201c of the control device 201 obtains the number of terminal devices N, running data and usage data range stored in the main storage device 202 or the auxiliary storage device 203.

[0102] The number of terminal devices N is the number of terminal devices equipped with pressure gauges and flow meters. In Example 1, N=2. The number of terminal devices N is set when the necessary pressure is input to the terminal devices.

[0103] The data range is set when inputting runtime data and is used in the maximum pressure loss exploration process.

[0104] In S903, one end device is specified as the object from the number of end devices N.

[0105] In S904, the characteristic quantity of the maximum pressure loss during operation is obtained using the runtime data from the usage data range of one terminal device. Here, the characteristic quantity of the maximum pressure loss in Example 1 is the maximum flow rate measured by the flow meter installed in the terminal device.

[0106] Here, the estimated flow rate obtained from the input supply pressure and terminal pressure of the learning model constructed in the learning model construction process can also be used as the feature quantity of the maximum pressure loss.

[0107] The ability to use inferred flow occurs when the function f representing the structure of the constructed learning model is a linear regression model or a neural network, etc. By using inferred flow, the set pressure, including the error of the learning model, can be calculated.

[0108] In S905, the characteristic value of the maximum pressure loss acquired is stored. The characteristic value of the maximum pressure loss is stored in the main storage device 202 or the auxiliary storage device 203. The processing of S904 and S905 is repeated a number of times equivalent to the number of end devices acquired in S902 (S906 and S908).

[0109] return Figure 8 In S804, the optimal supply pressure exploration unit 201d uses the characteristic quantity of the maximum pressure loss obtained in S803 and the end state prediction unit 201b to calculate the optimal supply pressure.

[0110] Here, for reference Figure 10 The steps of S804 are explained in detail.

[0111] Figure 10 This indicates the optimal supply pressure exploration process.

[0112] After processing begins (S1001), in S1002, the optimal supply pressure exploration unit 201d of the control device 201 acquires the number of end devices N, the number of steps Nt, and the convergence judgment value S stored in the main storage device 202 and the auxiliary storage device 203. The number of steps Nt and the convergence judgment value S are input simultaneously when the set pressure of the air compressor during operation is input.

[0113] In S1003, initialize the array of optimal supply pressures for each terminal device.

[0114] In S1004, the necessary pressure and the maximum pressure loss characteristic obtained in S803 are obtained.

[0115] In S1005 and S1006, the optimal supply pressure exploration unit 201d sets appropriate initial values ​​for the terminal pressure and the supply pressure.

[0116] In S1007, the array P that stores the pressure estimation values ​​obtained from the learning model is initialized. j pred .

[0117] In S1008, the optimal supply pressure exploration unit 201d uses the terminal state prediction unit 201b to input the supply pressure, terminal pressure, and maximum pressure loss characteristic quantities into the learning model. From this, the terminal pressure at the next moment is calculated.

[0118] In S1009, the calculated end pressure is added to and stored in the array P of the pressure calculation value. j pred The array P, which is calculated and appended to the calculated pressure values, is used to deduce and append these values. j pred The terminal pressure is stored in the main storage device 202 or the auxiliary storage device 203.

[0119] The processes in S1008 and S1009 are executed repeatedly for the number of steps Nt set in S1002 (S1010 and S1016).

[0120] Upon completion of steps S1010 and S1016, a judgment is made in S1011. In S1011, it is determined whether the minimum value of the calculated terminal pressure stored in the pressure calculation array is greater than the necessary pressure of the terminal equipment (whether the value obtained by subtracting the necessary pressure of the terminal equipment from the minimum calculated terminal pressure is greater than 0). If the minimum calculated terminal pressure is less than the necessary pressure, the optimal supply pressure exploration unit 201d corrects the set supply pressure and executes the processes from S1007 to S1010. Conversely, if the minimum calculated terminal pressure is greater than the necessary pressure, it is determined whether the difference between the minimum calculated terminal pressure and the necessary pressure of the terminal equipment is greater than the convergence judgment value (S1012).

[0121] In S1012, if the difference between the calculated minimum terminal pressure and the required pressure of the terminal equipment is greater than the convergence judgment value, the optimal supply pressure exploration unit 201d corrects the set supply pressure and executes the processing in S1007 to S1010. If the difference between the calculated minimum terminal pressure and the required pressure of the terminal equipment is less than the convergence judgment value, the supply pressure set at this time is added and stored in the optimal supply pressure array O (S1013).

[0122] Steps S1004 to S1013 are repeated an amount of time equivalent to the number of end devices N. Thus, the optimal supply pressure is calculated for each target end device.

[0123] return Figure 8 In S805, the setting pressure calculation unit 201e uses the optimal supply pressure calculated in S804 to calculate the setting pressure.

[0124] Here, for reference Figure 11 The steps of S805 are explained in detail.

[0125] Figure 11 This indicates the setting of pressure calculation processing.

[0126] In S1102, the set pressure calculation unit 201e of the control device 201 obtains the current number of air compressors and their set pressures stored in the main storage device 202 or the auxiliary storage device 203. The current number of air compressors is input simultaneously with the set pressure of the air compressors currently in operation.

[0127] In S1103, the minimum value is obtained from the acquired set pressure.

[0128] In S1104, the maximum value is obtained from the optimal supply pressure array calculated in step S804. The difference p between the obtained maximum value and the current minimum set pressure is calculated. diff .

[0129] In S1105, the difference p is subtracted from the current set pressure of each air compressor. diff The value of is used to calculate the optimal set pressure for each air compressor.

[0130] Figure 12 This is an example of a 1200-degree output screen when calculating the set pressure.

[0131] Set the runtime data used when calculating the set pressure in zone 1201.

[0132] Enter the number of end devices for which pressure gauges and flow meters are set in zone 1202.

[0133] Enter the necessary pressure on the end device of the object in zone 1203.

[0134] Configure the object end device in zones 1204 and 1210.

[0135] In zone 1205, set the data range used to calculate the set pressure from the runtime data set in zone 1201.

[0136] The graph 1206 represents the supply pressure, terminal pressure, and flow rate of the terminal device set in region 1204 and the data range set in region 1205.

[0137] Set the number of steps Nt and the convergence criterion S in region 1207.

[0138] Enter the number of air compressors in zone 1208.

[0139] Enter the air compressor's set pressure in zone 1209. Here, "set pressure" includes both the lower limit pressure and the upper limit pressure, in addition to the set pressure. The "lower limit pressure" is the pressure used to start other compressors when the pressure is below a threshold. The "upper limit pressure" is the pressure used to stop the compressor when the pressure is above a threshold.

[0140] The curve 1211 represents the following four items. First, the maximum pressure loss characteristic of the terminal equipment set in region 1210. Second, the optimal supply pressure obtained by the optimal supply pressure exploration unit 201d. Third, the calculated terminal pressure. Fourth, the necessary pressure input in region 1203.

[0141] The optimal settings for the air compressor calculated by the setting pressure calculation unit 201e are displayed in area 1212. Furthermore, the optimal settings for the air compressor calculated by the setting pressure calculation unit 201e are stored in the main storage device 202 or the auxiliary storage device 203.

[0142] The optimal setting display area 1212 is the area that displays the current optimal setting of the air compressor calculated by the setting pressure calculation unit 201e. Furthermore, in Embodiment 1, the number of terminal devices N is N=2, and the number of compressors is 1, but the number of each unit is arbitrary.

[0143] Based on Example 1, a learning model is constructed to predict future terminal pressures based on measured values ​​of supply pressure, terminal pressure, and flow rate. Using the constructed learning model and the measured values ​​of supply pressure, terminal pressure, and flow rate during operation, the optimal set pressure for the air compressor can be calculated.

[0144] <Example 2>

[0145] For the information processing apparatus 200 of Embodiment 2 of the present invention, refer to Figures 13 to 16 The following explanation will be provided. Furthermore, in Example 2, an air compressor will be used as an example to illustrate fluid machinery.

[0146] In the information processing apparatus 200 of Embodiment 1, flow rate is used as the characteristic quantity of maximum pressure loss. However, flow meters for measuring flow rate are more expensive than pressure gauges. Therefore, it is preferable to be able to calculate the set pressure using only a pressure gauge. Thus, in the information processing apparatus 200 of Embodiment 2, the case where the terminal flow rate is not used as measurement data in the learning model construction process and the set pressure calculation process will be described.

[0147] Figure 13 This illustrates the schematic structure of the compressed air system 100 using the information processing device 200 in Embodiment 2. (Compared to...) Figure 1 Unlike others, no flow meter was installed.

[0148] In addition, the hardware structure and functions of the information processing device 200 are similar to Figure 2 and Figure 3 The same as shown.

[0149] The following is for reference. Figures 14 to 16 This describes the process of calculating the set pressure of the air compressor 101 using the information processing device 200 of Embodiment 2.

[0150] refer to Figures 14 to 16 Explain the learning model construction process.

[0151] In Example 2, pressure gauges 107, 108 and 110 were used to measure the supply pressure and the terminal pressure.

[0152] Figure 14 This is an example of data measured to build a learning model.

[0153] Measurement data 1400 consists, for example, time (column 1401), supply pressure (column 1402), and terminal pressure (columns 1403, 1404).

[0154] Figure 15 This represents the process of building and processing learning models.

[0155] Before building a learning model, users need to set the data to be used for model building, the parameters of the model, and the data range.

[0156] The data used for model building includes Figure 14 The measurement data shown is the data.

[0157] A data interval represents the interval in the measurement data used to learn the model.

[0158] These processes performed by the user are executed using the input device 204 in the information processing device 200.

[0159] After these processes are completed, a learning model is built in the control device 201. The following refers to... Figure 15 Detailed explanation.

[0160] In S1502, the learning model construction unit 201a calculates a pressure loss characteristic quantity based on the supply pressure and the terminal pressure. Here, the pressure loss characteristic quantity is the calculated value of the pressure difference between the supply pressure and the terminal pressure at each time. Alternatively, the pressure loss characteristic quantity can be the value obtained by simply subtracting the terminal pressure from the supply pressure. Alternatively, information such as the change since the previous time step can be used to express the pressure difference at the next time step from the value obtained by subtracting the terminal pressure from the supply pressure. The pressure difference at the next time step can, for example, be expressed by adding the rate of change of the pressure difference at the previous time step to the current pressure difference.

[0161] In S1503, a learning model is constructed using the learning model construction unit 201a. More specifically, the learning model is constructed by solving the optimization problem described in (S1503) based on measurement data within a defined data interval. Here, k represents the time in the data. supply [k] represents the supply pressure at time k. end [k] represents the pressure of the terminal device at time k.

[0162] The measured values ​​of supply pressure, terminal pressure, and pressure loss characteristics are input into the learning model constructed in S1503, thereby outputting the terminal pressure at the next moment.

[0163] The function f representing the structure of the constructed learning model can be built using, for example, a linear regression model or a neural network. However, the model constructed here needs to represent the pressure change response, such as the corresponding change in pressure and flow rate at the terminal device. Furthermore, even when a certain value is input to the learning model as the supply pressure, terminal pressure, and terminal flow rate, and future terminal pressure is continuously calculated, the model needs to have the characteristic that the terminal pressure converges to a predetermined value (without diverging). In the case of a linear regression model, this characteristic is achieved by determining the model parameters so that the absolute values ​​of the model's eigenvalues ​​are less than 1.

[0164] The learning model constructed in S1503 is stored in the learning model storage section 203b of the auxiliary storage device 203 (S1504).

[0165] The learning model stored in S1504 is used to... Figure 6 Save it in the same format.

[0166] The user specifies the data used for model building, sets the model parameters, and defines the data range to... Figure 7 The same format is used.

[0167] Next, refer to Figure 16 Explanation of pressure calculation and processing.

[0168] The pressure calculation process is set to be similar to that of Example 1, except for a portion thereof. Figure 8 The calculation process for the set pressure shown is the same.

[0169] The following explains the differences from the set pressure calculation process in Example 1.

[0170] Figure 16 This represents the exploration process for maximum pressure loss.

[0171] In S1602, the maximum pressure loss exploration unit 201c acquires the number of terminal devices N, runtime data, and usage data range.

[0172] In S1603, one terminal device from the number N of terminal devices is designated as the target. In S1604, the maximum pressure loss detection unit 201c calculates the pressure difference between the supply pressure and the terminal pressure, i.e., the pressure loss characteristic quantity p. loss .

[0173] In S1605, the maximum pressure loss exploration unit 201c uses runtime data from the usage data range of one terminal device to obtain a characteristic quantity of the maximum pressure loss during operation. Here, the characteristic quantity of the maximum pressure loss in Example 2 is the maximum value of the pressure loss characteristic quantity calculated in S1604.

[0174] Here, the calculated values ​​obtained by substituting the supply pressure and end pressure into the learning model can also be used as the feature quantity of the maximum pressure loss.

[0175] The possibility of using extrapolated values ​​arises when the function f representing the structure of the constructed learning model is a linear regression model or a neural network, etc. By using extrapolated values, the set pressure, including the error of the learning model, can be calculated.

[0176] In S1606, the characteristic value of the maximum pressure loss obtained is stored. The processing from S1604 to S1606 is repeated a number of times equivalent to the number of end devices obtained in S1602 (S1607 and S1608).

[0177] The optimal supply pressure exploration unit 201d uses the characteristic quantity of the maximum pressure loss explored in S1601 to S1609 and the terminal state prediction unit 201b to calculate the optimal supply pressure. The optimal supply pressure exploration processing and the setting pressure calculation processing are the same as in Example 1. Figure 10 and Figure 11 same.

[0178] Based on Example 2, a learning model is constructed to predict future terminal pressures based on measured values ​​of supply pressure and terminal pressure. Using the constructed learning model and the measured values ​​of supply pressure and terminal pressure during operation, the optimal set pressure for the air compressor can be calculated.

[0179] In Examples 1 and 2, an air compressor was used as an example of fluid machinery, but the present invention is not limited to these examples. It can also be used in other applications involving fluids, such as pumps. Furthermore, various modifications can be made without departing from the spirit of the invention.

[0180] Explanation of reference numerals in the attached figures

[0181] 100: Compressed Air System

[0182] 101: Air Compressor

[0183] 102, 104: Piping

[0184] 103: Air canister

[0185] 105, 106: Terminal devices

[0186] 107, 108, 110: Pressure gauges

[0187] 109, 111: Flow meters

[0188] 200: Information processing device

[0189] 201: Control device

[0190] 201a: Learning Model Construction Department

[0191] 201b: End-of-Terminal Prediction Unit

[0192] 201c: Maximum Pressure Loss Exploration Department

[0193] 201d: Optimal Supply Pressure Exploration Department

[0194] 201e: Setting Pressure Calculation Unit

[0195] 202: Main storage device

[0196] 203: Auxiliary storage device

[0197] 203a: Measurement Data Storage Department

[0198] 203b: Learning Model Storage Unit

[0199] 204: Input device

[0200] 204a: Data Input Section for Learning Model Construction

[0201] 204b: Runtime Data Input Section

[0202] 204c: Various parameter input sections

[0203] 205: Output device

[0204] 205a: Data Output Section for Learning Model Construction

[0205] 205b: Learning Model Construction Screen Output Department

[0206] 205c: Runtime Data Output Section

[0207] 400: Measurement data

[0208] 401, 402, 403, 404, 405, 406: Columns

[0209] 601, 602, 603, 604, 605, 606: Columns

[0210] 701, 702, 703, 708, 709: Areas

[0211] 704, 705, 706, 707: Figure

[0212] 1200: Screen

[0213] 1201, 1202, 1203, 1204, 1205, 1207, 1208, 1209, 1210: Areas

[0214] 1206, 1211: Curve graph

[0215] 1212: Optimal display area setting

[0216] 1401, 1402, 1403, 1404: Columns.

Claims

1. An information processing device for calculating the set pressure of computational fluid dynamics machinery, characterized in that, include: The learning model construction unit constructs a learning model, wherein the learning model takes the pressure of the fluid supplied from the fluid machinery to the piping at the first moment, i.e., the supply pressure, the pressure of the fluid at the end of the piping, i.e., the end pressure, and the pressure loss characteristic quantity as inputs, and outputs the end pressure at the second moment after a predetermined time has passed from the first moment. The terminal state prediction unit inputs the supply pressure, terminal pressure, and pressure loss characteristic quantity at a specified time into the learning model constructed by the learning model construction unit, and predicts the terminal pressure at the next time. The maximum pressure loss exploration unit explores the maximum value of pressure loss, i.e., the maximum pressure loss, based on the supply pressure, terminal pressure, and pressure loss characteristic quantity during the operation of the fluid machinery. The optimal supply pressure exploration unit explores the optimal supply pressure that makes the end pressure higher than the necessary pressure, based on the necessary pressure at the end of the piping, the end state prediction unit, and the maximum pressure loss explored by the maximum pressure loss exploration unit. and The set pressure calculation unit calculates the set pressure of the fluid machinery based on the optimal supply pressure explored by the optimal supply pressure exploration unit and the set pressure during the operation of the fluid machinery.

2. The information processing apparatus as described in claim 1, characterized in that: The pressure loss characteristic is the flow rate of the fluid at the end of the piping, i.e., the end flow rate or the calculated value of the pressure difference between the supply pressure and the end pressure.

3. The information processing apparatus as described in claim 1 or 2, characterized in that: It also includes input devices and output devices. The input device accepts inputs of the supply pressure, the terminal pressure, and the pressure loss characteristic quantity. The output device outputs the end pressure predicted by the learning model constructed by the learning model construction unit and the end state prediction unit.

4. The information processing apparatus as described in claim 1 or 2, characterized in that: It also includes input devices and output devices. The input device accepts the necessary pressure input from the end of the piping. The output device outputs the optimal supply pressure obtained by the optimal supply pressure exploration unit.

5. The information processing apparatus as described in claim 1 or 2, characterized in that: It also includes input devices and output devices. The input device accepts the input of a set pressure during the operation of the fluid equipment. The output device outputs the set pressure of the fluid machinery calculated by the set pressure calculation unit.

6. A method for calculating the set pressure of a computational fluid dynamics machine, characterized in that, include: The learning model construction steps for constructing a learning model, wherein the learning model takes the pressure of the fluid supplied from the fluid machinery to the piping at the first moment, i.e., the supply pressure, the pressure of the fluid at the end of the piping, i.e., the end pressure, and the pressure loss characteristic quantity as inputs, and outputs the end pressure at the second moment after a specified time from the first moment. The terminal state prediction step involves inputting the supply pressure, terminal pressure, and pressure loss characteristic at a specified time into the learning model constructed in the learning model construction step, and predicting the terminal pressure at the next time step. The maximum pressure loss exploration step explores the maximum value of pressure loss, i.e., the maximum pressure loss, based on the supply pressure, terminal pressure, and pressure loss characteristic quantities during the operation of the fluid machinery. The optimal supply pressure exploration step explores the optimal supply pressure that makes the end pressure higher than the necessary pressure, based on the necessary pressure at the end of the piping, the end state prediction step, and the maximum pressure loss explored in the maximum pressure loss exploration step. and A pressure calculation step is set up to calculate the set pressure of the fluid machinery based on the optimal supply pressure explored in the optimal supply pressure exploration step and the set pressure during the operation of the fluid machinery.

7. The method for calculating the set pressure as described in claim 6, characterized in that: The pressure loss characteristic is the flow rate of the fluid at the end of the piping, i.e., the end flow rate or the calculated value of the pressure difference between the supply pressure and the end pressure.

8. The method for calculating the set pressure as described in claim 6 or 7, characterized in that, include: The step of accepting inputs of the supply pressure, the terminal pressure, and the pressure loss characteristic; and The step of outputting the end pressure predicted by the learning model constructed in the learning model construction step and the end state prediction step.

9. The method for calculating the set pressure as described in claim 6 or 7, characterized in that, include: The steps for receiving the necessary pressure input at the end of the piping; and Output the steps for finding the optimal supply pressure in the optimal supply pressure exploration step.

10. The method for calculating the set pressure as described in claim 6 or 7, characterized in that, include: The step of accepting input of a set pressure during the operation of the fluid device; and The step of outputting the set pressure of the fluid machinery calculated in the set pressure calculation step.

Images