Vacuum pump device

Abstract

This vacuum pump device (1) comprises: a collection unit (31) that collects operation parameters of a dry vacuum pump (3) of the vacuum pump device; a model generation unit (32) that generates discrimination models to be used for discriminating whether or not the operating state is normal on the basis of the collected data of the operation parameters collected by the collection unit (31); a model registration unit (33) in which the discrimination models are registered; a registration control unit (34) that controls the operation parameter collection operation performed by the collection unit (31), the discrimination model generation operation performed by the model generation unit (32), and the discrimination model registration operation performed by the model registration unit (33); and an input unit (35) that issues a command to the registration control unit (34) so as to perform a new registration, an additional registration, or an update registration of the discrimination models. As a result, the discrimination models can be easily generated and registered.

Description

Technical Field

[0001] The present invention relates to a vacuum pump device configured of a vacuum pump such as a dry vacuum pump, and more particularly to a vacuum pump device having a predictive maintenance function that uses a discrimination model to discriminate whether the operating state of the vacuum pump is normal or not. Background Art

[0002] In recent years, vacuum pumps have increasingly been used in applications such as semiconductor device manufacturing sites, where sudden stoppage during use would cause extensive damage. For this reason, vacuum pumps used in such applications are required to be equipped with a predictive maintenance system that constantly monitors the operating state and predicts and maintains abnormalities that would cause sudden stoppage in advance.

[0003] Regarding the realization of a predictive maintenance system mounted on a vacuum pump, a method is known in which a discrimination model that serves as an indicator of whether the vacuum pump is normal or not is prepared, and abnormalities are discriminated from operating parameters representing the operating state of the vacuum pump (hereinafter referred to as "operating parameters") obtained from the vacuum pump during operation using the discrimination model (for example, Patent Document 1).

[0004] Patent Document 1 discloses a diagnostic device that predicts and diagnoses abnormalities in rotating machinery such as pump motors using a normal model as a discrimination model. The normal model used in this diagnostic device is generated based on values obtained from the motor current of a normally operating motor (paragraph 0037 of Patent Document 1). Prior Art Documents Patent Documents

[0005] Patent Document 1 Japanese Unexamined Patent Application Publication No. 2022-138234 Summary of the Invention Problems to be Solved by the Invention

[0006] Vacuum pumps are operated in a variety of installation environments and conditions, and the threshold for determining normal operation parameters detected from the vacuum pump during operation changes depending on the installation environment and conditions. Therefore, when performing predictive maintenance on vacuum pumps, it is preferable to use a discrimination model that is tailored to the installation environment and conditions, and it is necessary to generate a discrimination model each time the installation environment or conditions change. However, in conventional predictive maintenance systems installed on vacuum pumps, generating and registering discrimination models has not been easy.

[0007] In view of these points, the main objective of the present invention is to propose a vacuum pump device that can easily generate and register discrimination models. [Means for solving the problem]

[0008] [1] In order to solve the above problems, the vacuum pump device according to the present invention includes: an acquisition unit that collects operating parameters representing the operating state of the vacuum pump; a model generation unit that generates a discrimination model used to determine whether the operating state is normal or not based on the acquired data of operating parameters collected by the acquisition unit; a model registration unit where the discrimination model is registered; a registration control unit that controls the operation of collecting operating parameters by the acquisition unit, the operation of generating a discrimination model by the model generation unit, and the operation of registering the discrimination model to the model registration unit; and an input unit that commands the registration control unit to perform at least one of the registration operations of new registration, additional registration, and update registration of the discrimination model. Furthermore, the registration control unit of this vacuum pump is equipped with a registration function that performs at least one of the following registration operations for discrimination models: a new registration function, an additional registration function, and an update registration function. The new registration function is activated when a new registration command is received and causes the collection unit to collect operating parameters, the model generation unit to generate a new discrimination model based on the collected operating parameters, and the generated discrimination model to be newly registered in the model registration unit. The additional registration function is activated when an additional registration command is received and causes the collection unit to collect operating parameters, the model generation unit to generate a new discrimination model based on the collected operating parameters, and the generated discrimination model to be added to the model registration unit in addition to the already registered discrimination models registered in the model registration unit. The update registration function is activated when an update registration command is received and causes the collection unit to collect operating parameters, the model generation unit to generate a new discrimination model based on the collected operating parameters, and the generated discrimination model to be registered in the model registration unit in place of the already registered discrimination models registered in the model registration unit.

[0009] [2] In the vacuum pump device according to the present invention, the discrimination model generated by the model generation unit is preferably a normal model generated using the MT method (Mahalanobis-Taguchi system) from a group of multidimensional information that includes several different operating parameters indicating the normal operating state of the vacuum pump.

[0010] [3] In the vacuum pump device according to the present invention, the collected data used to generate the discrimination model is preferably a series of data detected at a predetermined sampling period over a predetermined time.

[0011] [4] In the vacuum pump device according to the present invention, the input unit preferably includes one or more operating members that are operated to input the command.

[0012] [5] In the vacuum pump device according to the present invention, the discrimination model includes a first discrimination model and a second discrimination model for determining whether different first and second operating states of the vacuum pump are normal. The model registration unit comprises a first registration area for registering the first discrimination model and a second registration area for registering the second discrimination model, and the input unit preferably comprises an operating member for specifying the registration area for inputting a command to the registration control unit to specify whether the registration operation instructed by the command above should be performed on the first registration area or the second registration area.

[0013] [6] Furthermore, it is preferable that the input unit is configured to also be able to instruct the registration control unit to perform the registration operation of each of the multiple discrimination models, including the first discrimination model and the second discrimination model, as a series of registration operations, by having the operating state of the vacuum pump determined based on at least one of the operating parameters, and automatically switching the discrimination model to be generated and registered. [Effects of the Invention]

[0014] According to the present invention, a vacuum pump device is provided that allows for the easy generation and registration of discrimination models. [Brief explanation of the drawing]

[0015] [Figure 1] This is a block diagram showing the configuration of the vacuum pump device of Embodiment 1. [Figure 2] This is an explanatory diagram of the MT method used in Embodiment 1. [Figure 3] This graph shows the operating status of the dry vacuum pump in Embodiment 1. [Figure 4] This diagram illustrates the differences in unit space depending on the operating state in Embodiment 1. [Figure 5] This is a block diagram showing the configuration of the predictive maintenance system of Embodiment 1. [Figure 6] This flowchart primarily illustrates the processes performed in the predictive maintenance system for the vacuum pump device of Embodiment 1. [Figure 7] It is a block diagram showing the configuration of the vacuum pump device according to Embodiment 2. [Figure 8] It is a flowchart mainly showing processing performed in the predictive maintenance system for the vacuum pump device according to Embodiment 2. [Figure 9] It is a flowchart showing each processing in the discrimination model registration step of FIG. 8. MODE FOR CARRYING OUT THE INVENTION

[0016] Hereinafter, a vacuum pump device as an embodiment to which the present invention is applied will be described with reference to the drawings. It should be noted that each drawing does not necessarily strictly reflect actual dimensions or all actual configurations.

[0017] (Embodiment 1) FIG. 1 is a block diagram showing the configuration of a vacuum pump device 1 according to Embodiment 1. As shown in FIG. 1, the vacuum pump device 1 includes, as a vacuum pump, a two-stage dry vacuum pump 3 arranged inside a device case 2. For example, the dry vacuum pump 3 is constituted of two stages: a mechanical booster pump 4 at the front stage (hereinafter simply referred to as "booster pump 4") and a screw-type vacuum pump 5 at the rear stage (hereinafter referred to as "back pump 5"). An intake port 6 of the dry vacuum pump 3 extends from the booster pump 4 and opens upward on the upper surface of the device case 2. An exhaust port 7 of the dry vacuum pump 3 extends from the back pump 5 and opens laterally on the side surface of the device case 2. The booster pump 4 and the back pump 5 are connected via an internal pipe. The dry vacuum pump 3 causes air sucked from the intake port 6 to pass through the booster pump 4 and the back pump 5 in this order, discharges the air from the exhaust port 7, and evacuates the inside of a container connected to the intake port 6 side.

[0018] Furthermore, in the vacuum pump device 1, detection elements 10 such as measurement devices or detection sensors are installed, and operating parameters representing the operating state of the dry vacuum pump 3 are detected from various perspectives by each detection element. Specifically, in the vacuum pump device 1, vibration detection elements 10a, 10b and current detection elements 10c, 10d are installed to detect the respective vibrations and currents of the booster pump 4 and the backing pump 5. Further, a temperature detection element 10e (thermocouple 10e) is installed in the vacuum pump device 1 to detect the temperature of the backing pump 5. In addition, a back pressure detection element 10f is installed in the exhaust path of the backing pump 5 in the vacuum pump device 1 to detect the back pressure of the dry vacuum pump 3.

[0019] Furthermore, the vacuum pump device 1 includes an operation control system 20 that controls the operation of the dry vacuum pump 3, and a predictive maintenance system 30 that determines whether the operating state of the dry vacuum pump 3 is normal or not.

[0020] The operation control system 20 is electrically connected to the booster pump 4 and the backing pump 5. The operation control system 20 receives an operation command from a user via an operation panel 21 or the like and a maintenance command from the predictive maintenance system 30 described later, and controls operations such as starting / stopping for each of the booster pump 4 and the backing pump 5 in accordance with the contents of the operation command and the maintenance command.

[0021] The predictive maintenance system 30 constantly monitors the operating state of the dry vacuum pump 3, grasps the operating state of the dry vacuum pump 3, determines maintenance timing, occurrence of abnormalities, and the like, notifies the user of the results, and reflects the results in the control of the operation control system 20. Here, before describing the predictive maintenance system 30 in detail, first, an outline of the predictive maintenance system 30 in the vacuum pump device 1 according to Embodiment 1 will be described.

[0022] Figure 2 is an explanatory diagram of the MT method used in Embodiment 1. Figure 3 is a graph showing the operating status of the dry vacuum pump 3 in Embodiment 1. Figure 4 is a diagram illustrating the difference in unit space depending on the operating state in Embodiment 1. In the predictive maintenance system 30 of the vacuum pump device 1, the system attempts to determine whether the operating state of the dry vacuum pump 3 is normal or not from different operating parameters (multidimensional measured values) detected by at least two types of detection elements (10a to 10f) among the detection elements 10. Therefore, in the predictive maintenance system 30, in order to distinguish the multidimensional measured values ​​on a single scale, the MT method (Mahalanobis-Taguchi system) using quality engineering and statistical techniques is used to determine whether the operating state of the dry vacuum pump 3 is normal or not using an index called the Mahalanobis distance (hereinafter referred to as "MD value").

[0023] In the MD value-based discrimination method, as shown in Figure 2, a single measuring stick is created by integrating multidimensional information from a homogeneous group (in Embodiment 1, the normal operating parameters of the dry vacuum pump 3) with respect to the objective. The range in which this homogeneous group (the group of normal operating information for the dry vacuum pump 3) is distributed is called the unit space. In the MD value-based discrimination method, the distance ("Mahalanobis distance (MD value)") from the center of the unit space to individual objects that do not belong to this homogeneous group (in Embodiment 1, the data for the dry vacuum pump 3 when it is malfunctioning) is determined using this measuring stick, and it is determined whether or not they are malfunctioning. In the vacuum pump device 1 using this discrimination method, a normal model generated using the MT method (Mahalanobis-Taguchi system) from a group of multidimensional information containing several different operating parameters that indicate the normal operating state of the dry vacuum pump 3 is used as the discrimination model. Then, in Embodiment 1, the MD value in the unit space of this discrimination model is calculated, and it is determined whether or not it is malfunctioning based on the idea that the larger the MD value, the greater the degree of malfunction. For example, whether something is normal or not can be determined by whether the exponentially weighted moving average (EWMA) of the calculated MD values ​​exceeds a threshold.

[0024] Furthermore, when the dry vacuum pump 3 performs vacuuming in the container connected to the intake port 6, it operates by repeatedly going through different operating states, such as an idle state where the motor speed is not increased, a process state where the motor speed is periodically increased, and (in some cases, other operating states), as shown in Figure 3. As a result, the pump behavior differs in each operating state, so the predictive maintenance system 30 considers it difficult to determine whether it is normal or not using only one discrimination model, and therefore uses multiple discrimination models to determine whether it is normal or not. In other words, the vacuum pump device 1 uses a first discrimination model and a second discrimination model to determine whether the dry vacuum pump 3 is normal or not in its different first operating state (e.g., idle state) and second operating state (e.g., process state). The predictive maintenance system 30 will be described in detail below.

[0025] Figure 5 is a block diagram showing the configuration of the predictive maintenance system 30 in the vacuum pump device 1 of Embodiment 1. As shown in Figure 5, the predictive maintenance system 30 is configured to include a collection unit 31, a model generation unit 32, a model registration unit 33, a registration control unit 34, an input unit 35, and a discrimination unit 36.

[0026] The data collection unit 31 is electrically connected to the detection element 10 and collects operating parameters representing the operating state of the dry vacuum pump 3 detected by the detection element 10 for use in generating a discrimination model. The data collected by the data collection unit 31 and used for generating the discrimination model is a series of data detected at a predetermined sampling period over a predetermined time. For example, the data collection unit 31 collects operating parameters related to the vibration and current of the booster pump 4, the vibration and current of the back pump 5, and the temperature and back pressure of the back pump 5 as a series of continuous data collected every second for one hour for use in generating a discrimination model. In addition, when determining whether the dry vacuum pump 3 is operating normally, the data collection unit 31 collects operating parameters representing the operating state of the dry vacuum pump 3 detected by the detection element 10. For example, the collection unit 31 collects the vibration and current of the booster pump 4, the vibration and current of the back pump 5, and the operating parameters related to the temperature and back pressure of the back pump 5 every second as a series of operating state data in 60-second units, so that the discrimination unit 36 ​​(described later) can determine every 60 seconds whether the dry vacuum pump 3 is operating normally or not.

[0027] The model generation unit 32 generates a discrimination model used to determine whether the operating state of the dry vacuum pump 3 is normal or not, based on the collected data of operating parameters collected by the collection unit 31. As described above, in Embodiment 1, it is determined whether the dry vacuum pump 3 operating in different first operating states (idle operation state) and second operating states (process operation state) is normal or not. For this reason, the model generation unit 32 can generate at least a first discrimination model and a second discrimination model corresponding to each operating state. Also, as described above, the dry vacuum pump 3 is composed of a booster pump 4 and a back pump 5, and it is determined whether each pump is normal or not. For this reason, the model generation unit 32 can generate a first discrimination model and a second discrimination model for the booster pump 4, as well as a first discrimination model and a second discrimination model for the back pump 5, corresponding to each pump. In Embodiment 1, as described above, the MT method is used to determine whether the dry vacuum pump 3 is normal or not. Therefore, the discrimination model generated by the model generation unit 32 is a normal model generated using the MT method (Mahalanobis-Taguchi system) from a group of multidimensional information that includes several different operating parameters indicating the normal operating state of the dry vacuum pump 3 (for example, the discrimination model for the back pump 5 in Embodiment 1 includes operating parameters of vibration, current, temperature, and back pressure, and the discrimination model for the booster pump 4 in Embodiment 1 includes operating parameters of vibration and current).

[0028] The model registration unit 33 is a memory having a registration area for registering data such as memory elements and storage media, and a discrimination model is registered in this registration area. In the example of Embodiment 1, as described above, a first discrimination model and a second discrimination model are used to determine whether the booster pump 4 and the back pump 5 are normal or not in their respective first operating state (e.g., idle operation state) and second operating state (e.g., process operation state). For this reason, the model registration unit 33 includes a first registration area for registering the first discrimination model and a second registration area for registering the second discrimination model for each of the booster pump 4 and the back pump 5.

[0029] The registration control unit 34 controls the operation of collecting operation parameters by the collection unit 31, the operation of generating a discrimination model by the model generation unit 32, and the operation of registering the discrimination model to the model registration unit 33. The registration control unit 34 is equipped with a registration function that performs registration operations instructed by commands from at least the input unit 35, described later, including a new registration function for discrimination models, an additional registration function, and an updated registration function.

[0030] The new registration function in the registration control unit 34 is activated when a new registration command is received, and it performs the following: collection of operation parameters by the collection unit 31, generation of a new discrimination model by the model generation unit 32 based on the collected operation parameters, and new registration of the generated discrimination model to the model registration unit 33.

[0031] Furthermore, the additional registration function in the registration control unit 34 is activated when an additional registration command is received, and is a function that collects operating parameters by the collection unit 31, generates a new discrimination model by the model generation unit 32 based on the collected operating parameters, and adds the generated discrimination model to the model registration unit 33 in addition to the already registered discrimination models registered in the model registration unit 33.

[0032] The update registration function in the registration control unit 34 is activated when an update registration command is received, and includes the collection of operation parameters by the collection unit 31, the generation of a new discrimination model by the model generation unit 32 based on the collected operation parameters, and the registration of the generated discrimination model in the model registration unit 33 in place of an already registered discrimination model registered in the model registration unit 33.

[0033] The input unit 35 includes, for example, an operating member 37 such as a push button switch, a toggle switch, or a touch panel installed on the side of the device case 2, and accepts input operations from the user via the operating member 37. The input unit 35 commands the registration control unit 34 to perform at least one of the following registration operations: new registration, additional registration, and update registration of the discrimination model. The input unit 35 is operated to input at least one of the commands for new registration, additional registration, and update registration. In Embodiment 1, as described above, multiple discrimination models such as a first discrimination model and a second discrimination model are used as discrimination models, so the input unit 35 is equipped with an operating member for specifying the registration area that inputs a command to the registration control unit 34 specifying whether the registration operation instructed by the command given by the registration control unit 34 should be performed on the first registration area or the second registration area. In Embodiment 1, the input unit 35 is configured as an operating member 37 and includes a plurality of push-button switches (37a, 37b, ...) corresponding to the first registration area and the second registration area, respectively. When one of the push-button switches (for example, 37a) is pressed, the input unit 35 commands the registration control unit 34 to generate a discrimination model and then register the discrimination model in the registration area corresponding to the push-button switch (for example, 37a).

[0034] The discrimination unit 36 ​​analyzes the operating parameters representing the operating state of the dry vacuum pump 3 during operation, collected by the collection unit 31, using the discrimination model registered in the model registration unit 33, and determines whether the operating state of the dry vacuum pump 3 is normal or not. Specifically, in Embodiment 1, the discrimination unit 36 ​​generates operating state data based on the operating parameters of the dry vacuum pump 3 during operation, collected by the collection unit 31. Next, the discrimination unit 36 ​​calculates the MD value (Mahalanobis distance) in the unit space of the discrimination model used for discrimination from among the discrimination models registered as normal models in the model registration unit 33, and determines that it is normal if the calculated MD value is smaller than the MD value of the boundary line separating the inside and outside of the unit space, and that it is potentially abnormal or abnormal if it is larger. Next, the discrimination unit 36 ​​outputs the discrimination result so as to present it to the user or so as to reflect it in the control of the operation control system 20.

[0035] In the vacuum pump device 1 configured in this way, the dry vacuum pump 3 is operated under the control of the operation control system 20. Operating parameters indicating the operating state of the dry vacuum pump 3 are detected by the detection element 10 and collected as needed by the collection unit 31 of the predictive maintenance system 30. The dry vacuum pump 3 in operation is monitored and its operating state is determined by the discrimination unit 36 ​​of the predictive maintenance system 30, using the discrimination model registered in the model registration unit 33 to determine whether or not it is operating normally. On the other hand, in order for the predictive maintenance system to make accurate judgments in the vacuum pump device, it is necessary to use a discrimination model that is appropriate to the installation environment and installation state. In the vacuum pump device 1, when the user performs an input operation from the input unit 35, the registration function in the registration control unit 34 registers a new discrimination model based on the operating parameters of the dry vacuum pump 3 in that installation environment and installation state.

[0036] Figure 6 is a flowchart showing the processes performed in the vacuum pump device 1 of Embodiment 1, particularly in the predictive maintenance system 30. Next, referring to Figure 6, the processes performed in the vacuum pump device 1, particularly in the predictive maintenance system 30, will be explained. In the vacuum pump device 1, after the power is turned on and the device starts up, and after a certain period of time has elapsed since the device started up or after it has been confirmed that the dry vacuum pump 3 has entered a stable state, the startup read step ST1 is repeated for the number of selectable discrimination models that can be registered, and then the predictive maintenance step ST2 is repeatedly executed until a stop command is received. In addition, in the predictive maintenance system 30, the discrimination model registration step ST3 or the discrimination step ST4 is executed depending on the state of the model registration unit 33 or the input unit 35. Note that in the processes performed in the predictive maintenance system 30, a model generation flag is used for conditional branching, but the model generation flag is "OFF" at startup. The procedure will be explained below.

[0037] In the startup read step ST1, first, as the model registration unit confirmation process ST11, the registration status of the discrimination model in the model registration unit 33 is checked. If the model registration unit confirmation process ST11 confirms that a discrimination model is registered in the model registration unit 33, the discrimination model read process ST12 reads out the registered discrimination model so that it can be used in the discrimination step ST4 described later. On the other hand, if the model registration unit confirmation process ST11 confirms that no discrimination model is registered in the model registration unit 33, the model generation flag is set to "ON" as the model generation flag ON process ST13. In the startup read step ST1, these processes are repeated for the number of discrimination models that can be registered in the model registration unit 33. For example, if, as in Embodiment 1, at least a first discrimination model and a second discrimination model for the booster pump 4, and a first discrimination model and a second discrimination model for the back pump 5 can be registered, these processes are repeated at least four times to read out each discrimination model.

[0038] In the predictive maintenance step ST2, first, the input status of the input unit 35 is checked as the input unit confirmation process ST21. If it is confirmed in the input unit confirmation process ST21 that there is input to the input unit 35, the model generation flag is set to "ON" as the model generation flag ON process ST22. On the other hand, if it is confirmed in the input unit confirmation process ST21 that there is no input to the input unit 35, the process proceeds to the next step. Next, as the model generation flag confirmation process ST23, it is checked whether the model generation flag is ON or OFF. If it is ON, the process proceeds to the discrimination model registration step ST3, and if it is OFF, the process proceeds to the discrimination step ST4. In other words, in the predictive maintenance system 30, if a discrimination model is not registered in the model registration unit 33 at startup, or if there is input to the input unit 35 even if a discrimination model is registered in the model registration unit 33 at startup, the discrimination model registration step ST3 is executed. Otherwise, the discrimination step ST4 is executed.

[0039] The discrimination model registration step ST3 is a process that generates and registers a discrimination model for use in predictive maintenance of the dry vacuum pump 3 from operating parameters obtained by actually operating the dry vacuum pump 3 under conditions where the operating state can be understood in order to generate a discrimination model. In the discrimination model registration step ST3, the data acquisition process ST31, discrimination model generation process ST32, and discrimination model registration process ST33 are executed sequentially by one of the registration functions of the registration control unit 34, such as the new registration function, the additional registration function, and the update registration function. In the data acquisition process ST31, the acquisition unit 31 collects the operating parameters of the dry vacuum pump 3 for a predetermined time (for example, 1 hour). In the discrimination model generation process ST32, the model generation unit 32 generates a new discrimination model based on the operating parameters collected in the data acquisition process ST31. In the discrimination model registration process ST33, the discrimination model generated in the discrimination model generation process ST32 is registered in the registration area of ​​the model registration unit 33 corresponding to the operating state specified when input to the input unit 35. After a new discrimination model is registered in the discrimination model registration process ST33, the new discrimination model is read out in the discrimination model readout process ST34 so that it can be used in the discrimination step ST4 described later, and the model generation flag is changed from ON to OFF in the model generation flag OFF process ST35.

[0040] The discrimination step ST4 is a process for determining the operating state of the dry vacuum pump 3 while it is running. In the discrimination step ST4, the data acquisition process ST41, the discrimination process ST42, and the result output process ST43 are executed sequentially. In the discrimination step ST4, for example, in the data acquisition process ST41, the acquisition unit 31 collects the operating parameters of the dry vacuum pump 3 for a predetermined time (for example, 60 seconds). Next, in the discrimination process ST42, the discrimination unit 36 ​​generates operating state data based on the operating parameters of the dry vacuum pump 3 while it is running, which were collected in the data acquisition process ST41. Next, in the discrimination process ST42, the discrimination unit 36 ​​calculates the MD value (Mahalanobis distance) in the unit space of the read discrimination model for the operating state data, and determines that it is normal if the calculated MD value is smaller than the MD value of the boundary line separating the inside and outside of the unit space, and that it is potentially abnormal or abnormal if it is larger. In the discrimination process ST42, a series of processes are performed to generate operating state data corresponding to the read discrimination model and to determine the generated operating state data. In other words, if multiple discrimination models are read, this series of processes in discrimination process ST42 will be performed multiple times, corresponding to each discrimination model, through parallel or sequential processing. Next, in result output processing ST43, the discrimination unit 36 ​​outputs the discrimination result to be presented to the user or reflected in the control of the operation control system 20. Once discrimination step ST4 is executed, if there is no input to the input unit 35, the model generation flag remains "OFF" and the process will flow again from model generation flag confirmation process ST23 to discrimination step ST4. Therefore, discrimination step ST4 will be repeatedly executed until there is input to the input unit 35 and a stop command is received, and the operation of the dry vacuum pump 3 in operation will be monitored.

[0041] The vacuum pump device 1 of Embodiment 1 includes a collection unit 31 for collecting operating parameters, a model generation unit 32 for generating a discrimination model based on the collected operating parameter data collected by the collection unit 31, a model registration unit 33 for registering the discrimination model, and a registration control unit 34 for controlling the operation of collecting operating parameters by the collection unit 31, the operation of generating a discrimination model by the model generation unit 32, and the operation of registering the discrimination model to the model registration unit 33. The registration control unit 34 controls the collection unit 31, the model generation unit 32, and the model registration unit 33, and provides functions for new registration, additional registration, and update registration of discrimination models. In other words, in the vacuum pump device 1, the registration function of the registration control unit 34 collects the operating parameters of the dry vacuum pump 3 that is in operation, generates a discrimination model from the collected operating parameters, and registers that discrimination model in the model registration unit 33, thereby registering a new discrimination model in a series of steps. Here, the vacuum pump device 1 is further equipped with an input unit 35 that commands the registration control unit 34 to perform at least one of the following registration operations: new registration, additional registration, and update registration of the discrimination model. As a result, with the vacuum pump device 1, the input to the input unit 35 is controlled by the registration control unit 34 and the registration of a new discrimination model is carried out in a series of steps, making it easy to generate and register discrimination models.

[0042] Furthermore, in the vacuum pump device 1, the input unit 35 is equipped with one or more operating members 37 that are operated to input commands for new registration, additional registration, and update registration of discrimination models. Therefore, discrimination models can be generated and registered more easily simply by operating the operating members 37.

[0043] Furthermore, in the vacuum pump device 1, the input unit 35 is configured as an operating member (37a, 37b…) for specifying the registration area, which inputs a command to the registration control unit 34 specifying whether to perform new registration, additional registration, or update registration of a discrimination model for the first registration area or the second registration area. As a result, discrimination models can be easily generated, and multiple discrimination models can be easily and accurately registered.

[0044] Furthermore, the vacuum pump device 1 is a normal model generated by the model generation unit 32 using the MT method (Mahalanobis-Taguchi system) from a group of multidimensional information containing several different operating parameters that indicate the normal operating state of the dry vacuum pump 3. The collected data used to generate the discriminative model is a series of data detected over a predetermined time period at a predetermined sampling period. As a result, the normal model can be created from a series of data detected by the dry vacuum pump 3 operating normally at a predetermined sampling period without waiting for an abnormal shutdown. Therefore, the vacuum pump device 1 allows for the timely and easy generation and registration of a discriminative model.

[0045] (Embodiment 2) Figure 7 is a block diagram showing the configuration of the vacuum pump device 101 of Embodiment 2. Figure 8 is a flowchart mainly showing the processes performed in the predictive maintenance system 130 of the vacuum pump device 101 of Embodiment 2. Figure 9 is a flowchart showing each process in the discrimination model registration step ST103 of Figure 8.

[0046] The vacuum pump device 101 according to Embodiment 2 has basically the same configuration as the vacuum pump device 1 according to Embodiment 1, but differs in the following respects. Specifically, in the vacuum pump device 101 according to Embodiment 2, as shown in Figure 7, an operating member 137 different from the operating member 37 of Embodiment 1 (see Figure 1) is installed in the input unit 135. Also, in the vacuum pump device 101 according to Embodiment 2, as shown in Figures 8 and 9, the processing performed in the predictive maintenance system 130 is different from the processing performed in the predictive maintenance system 30 of Embodiment 1 (see Figure 6). In the following, components and processing similar to those in the vacuum pump device 1 according to Embodiment 1 are denoted by the same reference numerals as in Embodiment 1 and their explanation is omitted.

[0047] As shown in Figure 7, the operating member 137 installed in the input unit 135 is configured to include a plurality of push-button switches (37a, 37b, ...) corresponding to the first registration area and the second registration area, respectively. When one of the push-button switches (e.g., 37a) is pressed, it commands the registration control unit 34 to generate a discrimination model and then register the discrimination model in the registration area corresponding to the push-button switch (e.g., 37a). The operating member 137 further includes an additional registration operating member 137a and a continuous registration operating member 137b.

[0048] The additional registration operation member 137a is, for example, a toggle switch, and it gives a command to the registration control unit 34 to perform a registration operation using the new registration function or update registration function, or to perform a registration operation using the additional registration function, after generating a discrimination model. Specifically, when the additional registration operation member 137a is ON, input from the push button switches (37a, 37b, ...) is also made, instructing the registration control unit 34 to register the discrimination model using the additional registration function. Also, when the additional registration operation member 137a is OFF, input from the push button switches (37a, 37b, ...) is also made, instructing the registration control unit 34 to register the discrimination model using the new registration function or update registration function.

[0049] The continuous registration operation member 137b is, for example, a changeover switch, and commands the registration control unit 34 to perform the registration operation of each of the multiple discrimination models as a series of registration operations, by determining the operating state of the dry vacuum pump 3 based on at least one of the operating parameters and automatically switching which discrimination model to generate and register. Specifically, when the continuous registration operation member 137b is ON, the device is also activated, and commands the registration control unit 34 to first generate and register a first discrimination model as a series of registration operations, monitor the average current value of the booster pump 4 using the registered first discrimination model, and generate and register a second discrimination model when the monitored average current value exceeds a threshold.

[0050] Next, referring to Figures 8 and 9, the processes performed in the vacuum pump device 101, particularly in the predictive maintenance system 130, will be explained. In the vacuum pump device 101, similar to the vacuum pump device 1 of Embodiment 1, when the power is turned on and the device starts up, after a certain period of time has elapsed since the device started up or after it has been confirmed that the dry vacuum pump 3 has entered a stable state, the startup read step ST101 is executed, and then the predictive maintenance step ST102 is repeatedly executed until a stop command is received. In addition, in the predictive maintenance system 130, the discrimination model registration step ST103 or the discrimination step ST4 is executed depending on the state of the model registration unit 33 or the input unit 135. Note that in the processes performed in the predictive maintenance system 130, a model generation flag is used for conditional branching, but the model generation flag is "OFF" at startup. The procedure will be explained below.

[0051] In the startup read step ST101, first, as the model registration unit confirmation process ST11, the registration status of the discrimination model in the model registration unit 33 is checked. If it is confirmed in the model registration unit confirmation process ST11 that a discrimination model is registered in the model registration unit 33, the discrimination model read process ST12 reads the registered discrimination model so that it can be used in the discrimination step ST4 described later, and proceeds to the predictive maintenance step ST102. On the other hand, if it is confirmed in the model registration unit confirmation process ST111 that no discrimination model is registered in the model registration unit 33, the input waiting process ST113 waits until there is input in the input unit 135. In the input waiting process ST113, once input to the input unit 135 is confirmed, the process proceeds to the predictive maintenance step ST102.

[0052] In the predictive maintenance step ST102, first, the input status of the input unit 135 is checked as the input unit confirmation process ST21. If it is confirmed in the input unit confirmation process ST21 that there is input to the input unit 135, the model generation flag is set to "ON" as the model generation flag ON process ST22. On the other hand, if it is confirmed in the input unit confirmation process ST21 that there is no input to the input unit 135, the process proceeds to the next step. Note that if input to the input unit 135 is confirmed in the input waiting process ST113, it is also assumed that there is input in the input unit confirmation process ST21. Next, as the model generation flag confirmation process ST23, it is checked whether the model generation flag is ON or OFF. If it is ON, the status of the additional registration operation member 137a of the input unit 135 is checked as the additional registration confirmation process ST124, and after checking the status of the continuous registration operation member 137b of the input unit 135 as the continuous registration confirmation process ST125, the process proceeds to the discrimination model registration step ST103. If it is OFF, the process proceeds to the discrimination step ST4. In other words, in the predictive maintenance system 130, if there is input to the input unit 135, the discrimination model registration step ST103 is executed with processing content selected according to the content of the input to the input unit 135, as described later, and if not, the discrimination step ST4 is executed.

[0053] The discrimination model registration step ST103 is a process that generates and registers a discrimination model for use in predictive maintenance of the dry vacuum pump 3 from the operating parameters obtained by actually operating the dry vacuum pump 3 in a state where the operating conditions can be understood in order to generate a discrimination model. In the discrimination model registration step ST103, depending on the state of the model generation flag in the model generation flag confirmation process ST23, and the operation status of the additional registration operating member 137a and the continuous registration operating member 137b confirmed in the additional registration confirmation process ST124 and the continuous registration confirmation process ST125 (ST125a, 125b), the single new registration process ST103A, the continuous new registration process ST103B, the single additional registration process ST103C, or the continuous additional registration process ST103D is executed. Specifically, after confirming that the model generation flag in the model generation flag confirmation process ST23 is ON, the registration process is selected and executed as follows. Specifically, if it is confirmed that the state of the additional registration operation member 137a is OFF and the state of the continuous registration operation member 137b is OFF, the single new registration process ST103A is executed. If it is confirmed that the state of the additional registration operation member 137a is OFF and the state of the continuous registration operation member 137b is ON, the continuous new registration process ST103B is executed. If it is confirmed that the state of the additional registration operation member 137a is ON and the state of the continuous registration operation member 137b is OFF, the single additional registration process ST103C is executed. If it is confirmed that the state of the additional registration operation member 137a is ON and the state of the continuous registration operation member 137b is ON, the continuous additional registration process ST103D is executed.

[0054] In the one-time new registration process ST103A, as shown in Figure 9(A), the data acquisition process ST31, the discrimination model generation process ST32, and the discrimination model registration process ST33 are executed sequentially. In the data acquisition process ST31, the acquisition unit 31 collects the operating parameters of the dry vacuum pump 3 for a predetermined time (for example, 1 hour). In the discrimination model generation process ST32, the model generation unit 32 generates a new discrimination model based on the operating parameters collected in the data acquisition process ST31. In the discrimination model registration process ST33, the discrimination model generated in the discrimination model generation process ST32 is newly registered or updated in the registration area of ​​the model registration unit 33 corresponding to the operating state specified when input to the input unit 35. After the new discrimination model is registered in the discrimination model registration process ST33, the new discrimination model is read out as the discrimination model reading process ST34 so that it can be used in the discrimination step ST4 described later, and the model generation flag is changed from ON to OFF as the model generation flag OFF process ST35.

[0055] In the continuous new registration process ST103B, as shown in Figure 9(B), the data acquisition process ST31a, the first discrimination model generation process ST32a, and the first discrimination model registration process ST33a are executed sequentially to first generate and register a first discrimination model for a first operating state (e.g., idle operation state). In the data acquisition process ST31a, the acquisition unit 31 collects the operating parameters of the dry vacuum pump 3 for a predetermined time (e.g., 1 hour). In the first discrimination model generation process ST32a, the model generation unit 32 generates a new discrimination model based on the operating parameters collected in the data acquisition process ST31a. In the first discrimination model registration process ST33a, the first discrimination model generated in the first discrimination model generation process ST32a is newly registered or updated in the registration area of ​​the model registration unit 33 corresponding to the first operating state. After a new discrimination model is registered in the first discrimination model registration process ST33a, the new first discrimination model is read out in the first discrimination model readout process ST34a so that it can be used in the next process and in the discrimination step ST4 described later. Next, in the continuous new registration process ST103B, in order to automatically switch and continuously generate and register a second discrimination model for a second operating state that has a different operating state from the first operating state, the operating parameter monitoring process ST36 monitors the operating parameters of the dry vacuum pump 3 until they exceed a threshold in order to determine whether the operating state of the dry vacuum pump 3 has changed. Specifically, in the operating parameter monitoring process ST36, the first discrimination model read out in the first discrimination model readout process ST34a is used to monitor the average current of the booster pump 4 until it exceeds a threshold. Next, in the continuous new registration process ST103B, when the operating parameter monitoring process ST36 confirms that the operating parameters of the dry vacuum pump 3 have exceeded a threshold, it is determined that the operating state has changed, and in order to generate and register a second discrimination model for the second operating state (e.g., process operation state), the data acquisition process ST31b, the second discrimination model generation process ST32b, and the second discrimination model registration process ST33b are executed sequentially. In the data acquisition process ST31b, the acquisition unit 31 collects the operating parameters of the dry vacuum pump 3 for a predetermined time (e.g., 1 hour).In the second discrimination model generation process ST32b, the model generation unit 32 generates a new discrimination model based on the operating parameters collected in the data acquisition process ST31b. In the second discrimination model registration process ST33b, the second discrimination model generated in the second discrimination model generation process ST32b is newly registered or updated in the registration area of ​​the model registration unit 33 corresponding to the second operating state. After the new second discrimination model is registered in the second discrimination model registration process ST33b, the new second discrimination model is read out in the second discrimination model reading process ST34b so that it can be used in the discrimination step ST4 described later, and the model generation flag is changed from ON to OFF in the model generation flag OFF process ST35.

[0056] In the one-time additional registration process ST103C, as shown in Figure 9(C), the discrimination model registration process ST33 performs the same processing as the one-time new registration process ST103A, except that the discrimination model generated in the discrimination model generation process ST32 is added rather than newly registered or updated.

[0057] In the continuous additional registration process ST103D, as shown in Figure 9(D), the first discrimination model registration process ST33a and the second discrimination model registration process ST33b perform the same processing as the continuous new registration process ST103B, except that the discrimination models generated in the first discrimination model generation process ST32a or the second discrimination model generation process ST32b are added rather than newly registered or updated.

[0058] The discrimination step ST4 is the same as in Embodiment 1 and will not be described.

[0059] Although the present invention has been described above based on the above embodiments, the present invention is not limited to the above embodiments. It can be implemented in various forms without departing from the spirit of the invention, and for example, the following modifications are also possible.

[0060] (1) The number and types of components, connection methods, number of measurements, measurement time, and flow of each process described in the above embodiments are examples and can be changed within the scope that does not impair the effects of the present invention.

[0061] (2) In the embodiments described above, a vacuum pump device 1 that performs predictive maintenance using the MT method is used as an example, but the present invention is not limited thereto. A vacuum pump device that performs predictive maintenance using other analysis methods may also be used. For this reason, the discrimination model registered in the present invention is not limited to a normal model. For example, the discrimination model may be registered as an abnormal model.

[0062] (3) In the embodiments described above, the dry vacuum pump 3 is described as being composed of two pumps, a booster pump 4 and a back pump 5, but the present invention is not limited thereto. The present invention can be applied even if the dry vacuum pump is composed of one or three or more pumps.

[0063] (4) In the embodiments described above, the discrimination model for the booster pump was described using an example generated from vibration and current operating parameters, and the discrimination model for the back pump was described using an example generated from vibration, current, temperature, and back pressure operating parameters. However, the present invention is not limited thereto. For example, the discrimination model for the booster pump may be generated from vibration, current, and temperature operating parameters. Also, for example, the discrimination model may be generated from operating parameters other than vibration, current, temperature, and back pressure.

[0064] (5) In the embodiments described above, a first discrimination model corresponding to the idle operation state of the dry vacuum pump 3 and a second discrimination model corresponding to the process operation state of the dry vacuum pump 3 were set up for discrimination. However, the present invention is not limited thereto. A third discrimination model, a fourth discrimination model, etc., may be set up to correspond to other operating states.

[0065] (6) In the embodiments described above, the input unit 35 was described as being configured to include a plurality of push-button switches (37a, 37b, ...) corresponding to the first registration area and the second registration area, respectively, as the operating member 37, but the present invention is not limited thereto. For example, as the operating member, an operating member such as a toggle switch, an operation panel, or an operation lever may be used that allows input by specifying each registration area. Alternatively, a single push-button switch may be used as the operating member, and the switching function may be realized by time or number of presses, such as long press or rapid press. Furthermore, the input unit may be configured to include a transmitter such as a communication terminal and a receiver installed in a vacuum pump device that can receive signals from the transmitter, and input may be received remotely from the transmitter.

[0066] (7) In the embodiments described above, the model registration unit 33 capable of registering (storing) discrimination models was described as being located inside the device case 2 as shown in Figure 1, etc., but the present invention is not limited thereto. For example, the model registration unit may be configured to be externally attached via connection terminals or wireless communication.

[0067] (8) In the above-described embodiment 2, a first discrimination model is generated and registered, the average current value of the booster pump 4 is monitored using the registered first discrimination model, and a second discrimination model is generated and registered when the monitored average current value exceeds a threshold. However, the present invention is not limited thereto. For example, another discrimination model may be automatically generated and registered when it is determined that the operating state has changed by monitoring the operating parameters of the back pump 5. Alternatively, another discrimination model may be automatically generated and registered when it is determined that the operating state has changed by monitoring operating parameters other than the average current value, such as vibration and temperature.

[0068] (9) In Figure 1, the predictive maintenance system 30 is depicted as being located inside the device case 2 and connected to the operation control system 20 and the detection element 10 by internal wiring or wires, but the present invention is not limited thereto. For example, the predictive maintenance system may be configured as a separate unit from the unit housed in the device case and located outside the device case. In that case, the predictive maintenance system may exchange data wirelessly with the operation control system and detection element located inside the device case.

Claims

1. A collection unit that collects operating parameters representing the operating status of the vacuum pump, A model generation unit generates a discrimination model used to determine whether the operating state is normal or not, based on the collected data of the operating parameters collected by the collection unit. A model registration unit where the aforementioned discrimination model is registered, A registration control unit controls the operation of collecting the operation parameters by the collection unit, the operation of generating the discrimination model by the model generation unit, and the operation of registering the discrimination model to the model registration unit. The registration control unit is provided with an input unit that commands it to perform at least one of the following registration operations: new registration, additional registration, and update registration of the discrimination model. It is equipped with, The registration control unit includes a registration function that performs at least the registration operation instructed by the command, among the new registration function, additional registration function and update registration function of the discrimination model. The aforementioned new registration function is activated upon receiving a new registration command and causes the collection unit to collect the operation parameters, the model generation unit to generate a new discrimination model based on the collected operation parameters, and the model registration unit to register the generated discrimination model. The aforementioned additional registration function is activated upon receiving an additional registration command and includes the following functions: collection of the operation parameters by the collection unit, generation of a new discrimination model by the model generation unit based on the collected operation parameters, and additional registration of the generated discrimination model in the model registration unit, in addition to the already registered discrimination models registered in the model registration unit. The update registration function is activated upon receiving an update registration command and includes the following functions: the collection unit collects the operation parameters; the model generation unit generates a new discrimination model based on the collected operation parameters; and the model registration unit registers the generated discrimination model in place of an already registered discrimination model. The aforementioned discrimination model includes a first discrimination model and a second discrimination model for determining whether the different first and second operating states of the vacuum pump are normal or not. The input unit instructs the registration control unit to perform the registration operation for each of the multiple discrimination models, including the first discrimination model and the second discrimination model, as a series of registration operations, while automatically switching the discrimination model to be generated and registered based on at least one of the operation parameters, which determines the operating state of the vacuum pump. A vacuum pump device characterized by the following features.

2. The discrimination model generated by the model generation unit is The vacuum pump apparatus according to claim 1, which is a normal model generated using the MT method (Mahalanobis-Taguchi system) from a group of multidimensional information including several different operating parameters that indicate the normal operating state of the vacuum pump.

3. The collected data used to generate the aforementioned discrimination model is The vacuum pump device according to claim 1, which is a series of data detected over a predetermined period of time with a predetermined sampling period.

4. The vacuum pump device according to claim 1, wherein the input unit comprises one or more operating members that are operated to input the command.

5. The model registration unit comprises a first registration area for registering the first discrimination model and a second registration area for registering the second discrimination model, The vacuum pump device according to claim 1, wherein the input unit is provided with an operating member for specifying a registration area that inputs a command to the registration control unit to specify whether the registration operation instructed by the command should be performed on the first registration area or the second registration area.

Images