Malfunction prediction system

The failure prediction system for ventilation devices addresses the high cost and limited accuracy of conventional methods by using a neural network to analyze rotational speed and voltage data, providing accurate and cost-effective failure timing and factor prediction.

WO2024247178A1PCT designated stage expired Publication Date: 2024-12-05MITSUBISHI ELECTRIC CORP

Patent Information

Application Number
PCT/JP2023/020322
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-31
Publication Date
2024-12-05

AI Technical Summary

Technical Problem

Conventional failure prediction systems for ventilation devices require multiple sensors to detect various parameters, increasing costs and failing to infer non-motor failure factors, while existing machine learning-based solutions are limited in predicting failure timing and factors accurately.

Method used

A failure prediction system that includes an operation information detection means, a storage means, a learning device, and an inference device to detect and infer failure timing and factors using rotational speed and voltage data from ventilation devices, reducing the need for multiple sensors and improving prediction accuracy at a lower cost.

Benefits of technology

Enables accurate and cost-effective prediction of failure timing and factors in ventilation devices, prompting timely maintenance and reducing operational costs by using a neural network-based approach to analyze rotational speed and voltage data.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure JP2023020322_05122024_PF_FP_ABST
    Figure JP2023020322_05122024_PF_FP_ABST
Patent Text Reader

Abstract

This malfunction prediction system (50) comprises: an operation information detection means (8) for detecting operation information of a ventilation device (1); a storage means (16) for storing a time transition of the operation information; a learning device (14) for learning a malfunction time and a malfunction factor of the ventilation device (1); and an inference device (15) for inferring a predicted malfunction time and a predicted malfunction factor of the ventilation device (1).
Need to check novelty before this filing date? Find Prior Art

Description

Failure Prediction System

[0001] The present disclosure relates to a failure prediction system that predicts when a ventilation device will fail based on operating information of the ventilation device.

[0002] Conventionally, there are devices that predict when an air conditioner will fail, using past and present values ​​of outdoor temperature, room temperature, current, voltage, and operating time acquired by sensors installed in the air conditioner, such as the air conditioner described in Patent Document 1. There are also devices that predict the lifespan of an electric motor, such as the machine learning device described in Patent Document 2.

[0003] JP 2013-213669 A Japanese Patent No. 6140229 A

[0004] According to the conventional technology described above, in order to predict when a failure will occur, it is necessary to prepare multiple sensors for detecting multiple values ​​such as outside air temperature, room temperature, current, and voltage, which leads to an increase in costs. Also, the above-mentioned lifespan prediction using machine learning has the problem that, although it can predict the lifespan of the electric motor itself, it cannot infer failure factors other than those of the electric motor of the equipment.

[0005] The present disclosure has been made in consideration of the above, and aims to provide a failure prediction system that can accurately and easily predict the time and cause of failure of a ventilation device at low cost.

[0006] In order to solve the above-mentioned problems and achieve the objectives, the failure prediction system of the present disclosure is characterized by having an operation information detection means for detecting operation information of a ventilation device, a storage means for storing the time progression of the operation information, a learning device for learning the failure timing and failure factors of the ventilation device, and an inference device for inferring the failure prediction timing and failure prediction factors of the ventilation device.

[0007] The failure prediction system according to the present disclosure has the effect of being able to accurately and easily predict the timing and cause of a ventilation device failure at low cost.

[0008] 6 is a diagram showing an example of the time transition of the rotation speed of the blades and the time transition of the voltage applied to the motor for each of the plurality of failure modes shown in FIG. 6; FIG. 1 is a schematic diagram showing a scatter plot of correlation between the rotation speed of the blades and the deceleration time after a predetermined time has elapsed since the supply of drive power to the motor is stopped for a number of ventilation fans, using contour lines. FIG. 2 is a schematic diagram showing the configuration of a learning device included in the failure prediction system according to embodiment 1. FIG. 3 is a diagram showing an example of a three-layer neural network.Schematic diagram showing a scatter diagram of the correlation between the blade rotation speed in the high operation mode and the blade rotation speed in the low operation mode using contour lines. FIG. 1 shows a processor in the case where at least some of the functions of the operation information detection means and communication means of the ventilation device of the failure prediction system according to embodiment 1 are realized by the processor. FIG. 2 shows a processing circuit in the case where at least some of the functions of the operation information detection means and communication means of the ventilation device of the failure prediction system according to embodiment 1 are realized by the processing circuit.

[0009] A failure prediction system according to an embodiment will be described in detail below with reference to the accompanying drawings.

[0010] Embodiment 1. Figure 1 is a diagram showing an overall view of a failure prediction system 50 according to embodiment 1. The failure prediction system 50 includes a ventilation device 1 that is the target of failure prediction, an information processing device 2 installed in a location remote from the ventilation device 1, and an equipment management terminal 3 installed in a location remote from the ventilation device 1 and the information processing device 2. The ventilation device 1, the information processing device 2, and the equipment management terminal 3 are connected to a global network 4. The ventilation device 1 can communicate with the information processing device 2 and the equipment management terminal 3 via the global network 4. The information processing device 2 can communicate with the equipment management terminal 3 via the global network 4. The equipment management terminal 3 is operated by an equipment manager 5.

[0011] FIG. 2 is a diagram showing the configuration of a failure prediction system 50 according to the first embodiment. The ventilation device 1 includes one or more blades 6, one or more motors 7, and an operation information detection unit 8 that detects operation information of the ventilation device 1. The operation information detection unit 8 includes a rotation speed detection unit 9 that detects the rotation speed of the blades 6 and a motor applied voltage detection unit 10 that detects the voltage applied to the motor 7. The motor applied voltage detection unit 10 detects the voltage value applied to the motor 7. The operation information detection unit 8 further includes a cumulative operation time detection unit 11 that detects the cumulative operation time of the ventilation device 1. The rotation speed detection unit 9 may detect the rotation speed of the motor 7 instead of the rotation speed of the blades 6. That is, the operation information includes the rotation speed of the blades 6 or the motor 7, the voltage value applied to the motor 7, and the cumulative operation time of the ventilation device 1. The ventilation device 1 further includes a communication unit 12. The ventilation device 1 is connected to a global network 4 via the communication unit 12. The communication means 12 may be built into the ventilation device 1, or may be provided outside the housing that constitutes the ventilation device 1. The ventilation device 1 may further include functional components such as a filter and an air path damper.

[0012] The information processing device 2 has communication means 13, a learning device 14 that learns the failure timing and failure factors of the ventilation device 1, an inference device 15 that infers the failure prediction timing and failure prediction factors of the ventilation device 1, and storage means 16 that stores time progression of operating information of the ventilation device 1. Examples of the storage means 16 include electromagnetic storage means such as an SSD (Solid State Drive) or an HDD (Hard Disk Drive). Programs for realizing the functions of the learning device 14 and the inference device 15 are stored in the storage means 16. The information processing device 2 is connected to a global network 4 via the communication means 13.

[0013] The equipment management terminal 3 has a communication means 17 and a display device 18 that displays the predicted failure time and failure prediction factors of the ventilator 1. The equipment management terminal 3 is connected to the global network 4 via the communication means 17. Examples of the equipment management terminal 3 include information terminals such as a PC (Personal Computer), a smartphone, or a tablet.

[0014] 3 is a diagram showing the configuration of a motor circuit 101 according to embodiment 1. The motor circuit 101 has a detection unit 150, a rectification unit 110, a smoothing unit 120, a switching unit 130, and a motor winding unit 140. Reference numeral 102 denotes wiring.

[0015] The detection unit 150 detects whether the input from the switch 103 is received from the strong terminal or the weak terminal, and outputs the detection result to the switching control unit 131 of the switching unit 130 .

[0016] The rectifier 110 has a diode bridge circuit 111, which converts the negative voltage component of the AC input voltage into a positive voltage and performs full-wave rectification, thereby converting the AC voltage into a pulsating DC voltage.

[0017] The smoothing unit 120 has a smoothing capacitor 121, which smoothes the pulsating voltage.

[0018] The switching unit 130 switches the switching elements in sequence to sequentially switch the supply of voltage to the motor winding unit 140 of the stator, thereby applying a rotating magnetic field to the motor 7. One switching control method, for example, is to use a magnet rotor as the rotor, provide a Hall element on the stator, detect the rotation angle of the rotor using the switching control unit 131, and sequentially switch the switching elements in response to the detected rotation angle.

[0019] The switching control unit 131 of the switching unit 130 receives a signal indicating which of the strong and weak terminals has been selected from the detection unit 150 and controls the strong and weak operation. In the strong and weak operation, the switching control unit 131 controls the phase angles of the applied voltages to the three phases of the motor 7, that is, the U phase, V phase, and W phase. The switching control unit 131 performs the following control, for example. The voltage smoothed by the smoothing unit 120 is designated as Vp.

[0020] 4 and 5 are diagrams for explaining the control operation of switching control unit 131 included in switching unit 130 of motor circuit 101 according to embodiment 1. In the strong operation mode, as shown in Fig. 4, switching control unit 131 repeats the following for the voltage of each phase: +Vp for a phase angle of 120°, 0 for a phase angle of 60°, -Vp for a phase angle of 120°, and 0 for a phase angle of 60°, and forms a three-phase drive voltage by shifting the phases of the U phase, V phase, and W phase by 120°.

[0021] In the weak operation mode, as shown in FIG. 5, the switching control unit 131 repeats the following for the voltage of each phase: +Vp for a phase angle of 60°, 0 for a phase angle of 120°, −Vp for a phase angle of 60°, and 0 for a phase angle of 120°, and forms a three-phase drive voltage by shifting the phases of the U phase, V phase, and W phase by 120°.

[0022] The motor windings 140 are wound around the stator, and by switching the current in sequence, a rotating magnetic field is generated, causing the rotor to rotate.

[0023] A Hall element PH1 161 is provided on the stator of the motor 7, and detects the rotation of the rotor. A rotation speed detector NO1 162 converts the rotation detected by the Hall element PH1 161 into the rotation speed of the blades 6 and outputs it.

[0024] The input line to the motor winding section 140 is provided with a motor applied voltage detector motor applied voltage VO1·163, which outputs the applied voltage to the motor 7.

[0025] Next, using Figs. 6 to 11, the main causes of failure in the ventilation device 1 and the possibility of classifying the causes of failure based on the time transition of the operation information will be described.

[0026] FIG. 6 is a diagram illustrating an example of the main failure factors of the ventilation device 1 included in the failure prediction system 50 according to embodiment 1 and the accompanying changes in operating conditions. As shown in FIG. 6 , the main failure factors of the ventilation device 1 can be broadly divided into an increase in pressure loss due to clogging of the ventilation air passage, such as the filter or blades 6, a decrease in the voltage applied to the motor 7 due to aging of electronic components, and an increase in mechanical friction due to aging of mechanical components inside the motor 7. Naturally, multiple failure factors may progress simultaneously in parallel. The example illustrated in FIG. 6 is a well-known fact. FIG. 7 is a diagram illustrating an example of the time progression of the rotational speed of the blades 6 and the time progression of the voltage applied to the motor 7 for each of the multiple failure modes F1, F2, and F3 shown in FIG. 6 . Operating points 208 and 210 in FIG. 7 , time progressions 213, 214, and 215 of the rotational speed of the blades 6 in FIG. 7 , and time progressions 216, 217, and 218 of the voltage applied to the motor 7 will be described later using FIG. 9 , FIG. 10 , and FIG. 11 .

[0027] Changes in the operating state that occur in association with the major failure factors shown in FIG. 6 will be described using FIGS. 8 to 11 . FIG. 8 is a diagram showing the operating characteristics of the ventilation device 1 included in the failure prediction system 50 according to embodiment 1. The operating characteristics of the ventilation device 1 can be represented by a set of a graph 201, in which the horizontal axis represents the air volume of the ventilation device 1 and the vertical axis represents the static pressure of the ventilation device 1, and a graph 202, in which the horizontal axis represents the air volume of the ventilation device 1 and the vertical axis represents the rotational speed of the blades 6. The operating point of the ventilation device 1 is determined by the intersection 206 of the air path pressure loss curve 204 and the air volume-static pressure curve 203, and the rotational speed is determined as the value corresponding to the intersection 206 of the air volume-rotational speed curve 205. Changes in the operating state appear as shifts in the operating point and can be observed as changes in the rotational speed of the blades 6 and the voltage applied to the motor 7. FIG. 9 is a diagram showing an example of a time progression 211 of the rotation speed of the blades 6 and a time progression 212 of the voltage applied to the motor 7 observed in the ventilation device 1 of the failure prediction system 50 according to the first embodiment.

[0028] Assume that the ventilation air duct, such as the filter or the blades 6, becomes increasingly clogged. FIG. 10 is a diagram showing changes in the pressure loss curve of the air duct when the clogging of the ventilation air duct, such as the filter or the blades 6, progresses. Clogging of the ventilation air duct causes an increase in pressure loss in the air duct, and when clogging of the ventilation air duct occurs, the pressure loss curve of the air duct shifts to a higher-pressure curve 207, as shown in FIG. 10. In other words, the operating point 208 of the ventilator 1 shifts in the direction of an increase in static pressure, a decrease in air volume, and an increase in the rotation speed of the blades 6. FIG. 9 shows an example of a time transition 213 of the rotation speed of the blades 6 and a time transition 216 of the voltage applied to the motor 7 when the clogging of the ventilation air duct, such as the filter or the blades 6, progresses.

[0029] Next, let us consider a case where the electronic components deteriorate over time. The deterioration of the electronic components causes a decrease in the voltage applied to the motor 7, which, as shown in FIG. 11 , shifts the airflow-static pressure curve 209 toward lower airflow and static pressure, and simultaneously shifts the airflow-impeller rotation speed curve 205 toward lower rotation. In other words, the operating point 210 of the ventilator 1 shifts toward lower static pressure, lower airflow, and lower rotation speed of the impellers 6. FIG. 11 shows the change in the operating point 210 of the ventilator 1 as the electronic components deteriorate over time. FIG. 9 shows an example of the time transition 214 of the rotation speed of the impellers 6 and the time transition 217 of the voltage applied to the motor 7 when the electronic components inside the motor deteriorate over time.

[0030] Next, we will show an example of deterioration of electronic components over time. Capacitors are a typical example of electronic components that are prone to deterioration. When the smoothing capacitor 121 deteriorates, it is no longer able to store a predetermined amount of charge, and some of the charge is lost through leakage, resulting in a drop in the smoothed voltage.

[0031] Next, consider a case where the mechanical components inside the motor 7 deteriorate over time. The deterioration of the mechanical components inside the motor 7 causes an increase in mechanical friction, and as shown in FIG. 11 , the operating point of the ventilation device 1 shifts in the direction of a decrease in static pressure, a decrease in airflow, and a decrease in the rotational speed of the blades 6. However, the phenomenon of deterioration of the mechanical components inside the motor 7 differs from the phenomenon of deterioration of electronic components in that the voltage applied to the motor 7 does not decrease. FIG. 9 shows an example of a time transition 215 of the rotational speed of the blades 6 and a time transition 218 of the voltage applied to the motor 7 when the mechanical components inside the motor 7 deteriorate over time.

[0032] All of the above-mentioned causes of failure of the ventilation device 1 cause a change in the rotation speed of the blades 6 or a change in the voltage applied to the motor 7. Therefore, as shown in Fig. 9 , if the rotation speed of the blades 6 and the voltage applied to the motor 7 fall outside the appropriate rotation speed range 219 and appropriate voltage range 220 intended for the product, it can be assumed that a failure has occurred. The appropriate rotation speed range 219 and appropriate voltage range 220 may be set freely.

[0033] The time transition 211 of the rotation speed of the blades 6 and the time transition 212 of the voltage applied to the motor 7 show different trends depending on the cause of the failure, so it is possible to identify the main cause of the failure by observing these time transitions. Therefore, the time transition 211 of the rotation speed of the blades 6, the time transition 212 of the voltage applied to the motor 7, and the cumulative operating time of the ventilator 1 are used as input data to the learning device 14 and the inference device 15.

[0034] FIG. 12 is a schematic diagram showing a scatter diagram of the correlation between the rotation speed of the blades 6 and the voltage applied to the motor 7 for a number of ventilation fans, using contour lines. The inner lines indicate a higher occurrence frequency. The innermost line corresponds to the median value.

[0035] In the normal state, the rotation speed of the blades 6 and the voltage applied to the motor 7 are distributed in a normal distribution with variations around a central value.

[0036] In the F1 mode, which is an increase in pressure loss due to clogging of the ventilation air passage, such as the filter or the blades 6, the rotation speed of the blades 6 increases, but since there is no significant change in the voltage applied to the motor 7, a distribution centered on the side where the rotation speed of the blades 6 is higher than in the standard state is obtained, as shown in F1 in Figure 12.

[0037] In the F2 mode, the rotation speed of the blades 6 decreases and the voltage applied to the motor 7 also decreases due to aging degradation of electronic components. As a result, a distribution is obtained as shown in F2 in FIG. 12, where the rotation speed of the blades 6 is lower than in the standard state and the voltage applied to the motor 7 is also centered on the lower side.

[0038] In the F3 mode, where mechanical friction increases due to deterioration of the mechanical parts inside the motor 7 over time, the rotation speed of the blades 6 decreases, but since there is no significant change in the voltage applied to the motor 7, a distribution is obtained, as shown in F3 in Figure 12, with the rotation speed of the blades 6 centered on the side lower than in the standard state.

[0039] In this way, the failure mode can be determined by classifying the distribution of the rotation speed of the blades 6 and the voltage applied to the motor 7 based on multiple indexes, namely the rotation speed of the blades 6 and the voltage applied to the motor 7.

[0040] Next, the configuration of the learning device 14 will be described with reference to Fig. 13. Fig. 13 is a diagram showing the configuration of the learning device 14 included in the failure prediction system 50 according to the first embodiment. The learning device 14 includes a data acquisition unit 303 and a model generation unit 304. Fig. 13 also shows storage means 16.

[0041] The data acquisition unit 303 acquires learning data including the operating information of the ventilation device 1 and the failure time and cause of the failure of the ventilation device 1. Specifically, the data acquisition unit 303 acquires the time transition 301 of the operating information and the failure time and cause of the failure (correct answer) 302 as learning data.

[0042] The model generation unit 304 uses the learning data to generate a trained model for inferring the predicted failure timing and failure prediction factors of the ventilation device 1 from the time transition 301 of the operation information of the ventilation device 1. Specifically, the model generation unit 304 learns the failure timing and failure factors of the ventilation device 1 based on learning data created based on a combination of the time transition 301 of the operation information output from the data acquisition unit 303 and the failure timing and failure factor (correct answer) 302. The model generation unit 304 generates a trained model for inferring the optimal failure prediction timing and failure prediction factors based on the time transition 301 of the operation information of the ventilation device 1. The learning data is data in which the time transition 301 of the operation information of the ventilation device 1 is associated with the failure timing and failure factor (correct answer) 302.

[0043] The learning device 14 and the inference device 15 are used to learn and infer the timing and cause of failure of the ventilation device 1. The learning device 14 and the inference device 15 may be devices separate from the ventilation device 1, for example, connected to the ventilation device 1 via a network. The learning device 14 and the inference device 15 may be built into the ventilation device 1. The learning device 14 and the inference device 15 may exist on a cloud server.

[0044] Known algorithms such as supervised learning, unsupervised learning, or reinforcement learning can be used as the learning algorithm used by the model generation unit 304. As an example, a case where the model generation unit 304 uses a neural network will be described.

[0045] The model generation unit 304 learns the time and cause of failures by so-called supervised learning, for example, according to a neural network model. Here, supervised learning refers to a technique in which a learning device is provided with a set of data of input and result, and the learning device learns features present in learning data created based on the set of data of input and result, thereby inferring a result from the input.

[0046] A neural network is composed of an input layer made up of multiple neurons, a hidden layer made up of multiple neurons, and an output layer made up of multiple neurons. The hidden layer may be composed of one layer or multiple layers.

[0047] 14 is a diagram showing an example of a three-layer neural network. For example, in a three-layer neural network such as that shown in FIG. 14, when multiple inputs are input to input layers X1-X3, values ​​obtained by multiplying the input values ​​by weights W1 (w11-w16) are input to intermediate layers Y1-Y2, and values ​​obtained by further multiplying the results by weights W2 (w21-w26) are output from output layers Z1-Z3. The output results vary depending on the values ​​of weights W1 and W2.

[0048] The neural network learns the time of failure and the cause of failure through so-called supervised learning in accordance with learning data created based on a combination of the time progression 301 of the operating information acquired by the data acquisition unit 303 and the time of failure and the cause of failure (correct answer) 302.

[0049] That is, the neural network learns by inputting the time transition of the driving information 301 into the input layer and adjusting the weights W1 and W2 so that the results output from the output layer approach the failure time and failure cause (correct answer) 302.

[0050] The model generation unit 304 generates and outputs a trained model by performing the above-described learning.

[0051] The storage means 16 stores the trained model output from the model generation unit 304.

[0052] Next, the learning process performed by the learning device 14 will be described with reference to Fig. 15. Fig. 15 is a flowchart showing the procedure of the learning process performed by the learning device 14 included in the failure prediction system 50 according to embodiment 1. Fig. 15 also shows the procedure of the operation performed by the storage means 16.

[0053] In step S101, the data acquisition unit 303 acquires the time transition of the driving information 301 and the failure time and failure cause (correct answer) 302. In Fig. 15 , the operation of step S101 is indicated by the phrase "data acquisition." The data acquisition unit 303 may acquire the time transition of the driving information 301 and the failure time and failure cause (correct answer) 302 simultaneously, or may acquire the time transition of the driving information 301 and the failure time and failure cause (correct answer) 302 at different times.

[0054] In step S102, the model generation unit 304 learns the failure time and failure cause by so-called supervised learning in accordance with the learning data generated based on a combination of the time progression 301 of the operating information acquired by the data acquisition unit 303 and the failure time and failure cause (correct answer) 302, and generates a learned model.

[0055] In step S103, the storage means 16 stores the trained model generated by the model generation unit 304.

[0056] Next, the configuration of the storage unit 16 will be described with reference to Fig. 16. Fig. 16 is a diagram showing the configuration of the storage unit 16 included in the failure prediction system 50 according to embodiment 1. The storage unit 16 includes a data acquisition unit 402 and a storage unit 403.

[0057] The data acquisition unit 402 acquires, as operating information 401, the rotation speed of the blades 6 detected by the rotation speed detection means 9, the applied voltage to the motor 7 detected by the motor applied voltage detection means 10, and the cumulative operating time of the ventilation device 1 detected by the cumulative operating time detection means 11 at a certain time.

[0058] The data acquired by the data acquisition unit 402 is added to the end of the data stored in the storage unit 403 and stored.

[0059] Next, a process in which the storage unit 16 stores the time transition of the operation information 401 will be described with reference to Fig. 17. Fig. 17 is a flowchart showing the procedure of the storage process performed by the storage unit 16 included in the failure prediction system 50 according to the first embodiment.

[0060] In step S201, the data acquisition unit 402 acquires, as operation information 401, the rotation speed of the blades 6 detected by the rotation speed detection means 9, the voltage applied to the motor 7 detected by the motor applied voltage detection means 10, and the cumulative operation time of the ventilator 1 detected by the cumulative operation time detection means 11 at a certain time. In Fig. 17, the operation of step S201 is indicated by the phrase "acquire data."

[0061] In step S202, the data acquired by the data acquisition unit 402 is added to the end of the data stored in the storage unit 403 and stored.

[0062] The storage means 16 repeatedly executes this process at any fixed interval to create data on the time transition of the driving information 401. The repetition period may be freely set.

[0063] Next, the configuration of the inference device 15 will be described with reference to Fig. 18. Fig. 18 is a diagram showing the configuration of the inference device 15 included in the failure prediction system 50 according to the first embodiment. Fig. 18 also shows the storage means 16 and the equipment management terminal 3. The inference device 15 includes a data acquisition unit 502 and an inference unit 503.

[0064] The data acquisition unit 502 acquires time transition 501 of the operation information of the ventilation device 1 .

[0065] The inference unit 503 uses the trained model to infer the predicted failure timing and failure prediction factors of the ventilation device 1. More specifically, the inference unit 503 uses the trained model for inferring the predicted failure timing and failure prediction factors of the ventilation device 1 from the time transition 501 of the operating information, and outputs the predicted failure timing and failure prediction factors of the ventilation device 1 from the time transition 501 of the operating information acquired by the data acquisition unit 502. That is, the inference unit 503 inputs the time transition 501 of the operating information acquired by the data acquisition unit 502 into the trained model, and outputs the predicted failure timing and failure prediction factors of the ventilation device 1 inferred from the time transition 501 of the operating information.

[0066] Next, the process by which the inference device 15 obtains the failure prediction time and failure prediction factors will be described with reference to Fig. 19. Fig. 19 is a flowchart showing the procedure of the operation performed by the inference device 15 included in the failure prediction system 50 according to the first embodiment. Fig. 19 also shows the procedure of the operation performed by the equipment management terminal 3.

[0067] In step S301, the data acquisition unit 502 acquires time transition 501 of the operating information of the ventilation device 1. In Fig. 19, the operation of step S301 is indicated by the phrase "data acquisition."

[0068] In step S302, the inference unit 503 acquires a trained model from the storage means 16, inputs the time progression 501 of the operating information into the acquired trained model, and obtains the predicted failure time and failure prediction factors of the ventilation device 1.

[0069] In step S303, the inference unit 503 outputs the predicted failure time and failure prediction factors of the ventilator 1 obtained by using the trained model to the equipment management terminal 3. In Figure 19, the operation of step S303 is indicated by the phrase "data output."

[0070] In step S304, the display device 18 of the equipment management terminal 3 displays the predicted failure time and failure prediction factors of the ventilator 1. FIG. 20 is a diagram showing the configuration of the display device 18 included in the failure prediction system 50 according to embodiment 1. The display device 18 includes a communication unit 601 that communicates with the information processing device 2 via the global network 4, an acquisition unit 602 that acquires the predicted failure time and failure prediction factors of the ventilator 1 inferred by the inference device 15, and a display unit 603 that displays the predicted failure time and failure prediction factors of the ventilator 1 acquired by the acquisition unit 602. For example, the display unit 603 is realized by a liquid crystal display device. In step S304, the acquisition unit 602 acquires the predicted failure time and failure prediction factors of the ventilator 1 inferred by the inference device 15, and the display unit 603 displays the predicted failure time and failure prediction factors of the ventilator 1 acquired by the acquisition unit 601.

[0071] As described above, the failure prediction system 50 according to the first embodiment includes an operation information detection means 8 that detects operation information of the ventilation device 1, a storage means 16 that stores the time progression of the operation information, a learning device 14 that learns the failure timing and failure factors of the ventilation device 1, and an inference device 15 that infers the predicted failure timing and failure prediction factors of the ventilation device 1. Therefore, the failure prediction system 50 can accurately and easily predict the failure timing and failure factors of the ventilation device 1 at low cost. The failure prediction system 50 displays the predicted failure timing and failure prediction factors of the ventilation device 1 on the display device 18, and can therefore appropriately prompt users of the ventilation device 1 to perform maintenance on the ventilation device 1.

[0072] In the first embodiment, a case where supervised learning is applied to the learning algorithm used by the model generation unit 304 has been described, but the present invention is not limited to this. As for the learning algorithm, reinforcement learning and the like can also be applied in addition to supervised learning.

[0073] Embodiment 2 In embodiment 2, the operating information does not include the voltage value applied to the motor 7, but instead includes the deceleration time, which is the time from when the supply of power to the motor 7 is stopped until the rotation speed of the blades 6 or the motor 7 is decelerated to a predetermined value. The rest is the same as embodiment 1. In embodiment 2, differences from embodiment 1 will be mainly described.

[0074] The operating information includes the rotation speed of the blades 6 or the motor 7, the deceleration time which is the time from when the supply of power to the motor 7 is stopped until the rotation speed of the blades 6 or the motor 7 decelerates to a predetermined value, and the cumulative operating time of the ventilation device 1.

[0075] 21 is a diagram showing the configuration of the operation information detection means 8 included in the failure prediction system according to embodiment 2. The operation information detection means 8 includes a rotation speed detection means 9 that detects the rotation speed of the blades 6 or the motor 7, a deceleration time detection means 19 that detects the deceleration time that is the time from when the supply of power to the motor 7 is stopped until the rotation speed of the blades 6 or the motor 7 decelerates to a predetermined value, and a cumulative operation time detection means 11 that detects the cumulative operation time of the ventilator 1.

[0076] The deceleration time detection means 19 detects the deceleration time as follows: First, the deceleration time detection means 19 stops the supply of drive power to the motor 7. The blades 6 and motor 7 rotate due to inertia, and the rotation speed gradually decreases due to air resistance against the blades 6, and eventually the blades 6 and motor 7 stop. As described above, the deceleration time is the time from when the supply of power to the motor 7 is stopped until the rotation speed decelerates to a predetermined value. The predetermined value is, for example, a rotation speed that is 50% of the initial rotation speed.

[0077] The deceleration time is measured, for example, at night. After the measurement of the deceleration time is completed, the deceleration time detection means 19 returns the ventilator 1 to its original operating state. This makes it possible to acquire the operating state of the ventilator 1 while minimizing the impact on the user who installed the ventilator 1.

[0078] As described above, the predetermined value for measuring the deceleration time is preferably, for example, 50% of the initial rotation speed. When the rotation speed is near zero, the measurement accuracy of the rotation speed decreases, so it is preferable to measure the rotation speed before it reaches zero. Furthermore, when the motor 7 stops, the static friction coefficient is greater than the kinetic friction coefficient, so depending on the abnormality, the ventilation device 1 may not restart. However, by returning to normal operation before stopping, the risk of the ventilation device 1 not restarting can be reduced.

[0079] 22 is a diagram showing changes in the rotation speed of the blades 6 after the supply of drive power to the motor 7 is stopped. Reference numeral 232 indicates changes in the rotation speed of the blades 6 when clogging of the ventilation air passage, such as the filter or the blades 6, progresses. Reference numeral 231 indicates changes in the rotation speed of the blades 6 when deterioration of electronic components progresses over time. Reference numeral 233 indicates changes in the rotation speed of the blades 6 when deterioration of mechanical components inside the motor 7 progresses over time.

[0080] The rotation speed of the blades 6 decreases over time in the same way when the ventilation air passage, such as the filter or blades 6, becomes clogged and when the electronic components deteriorate over time. On the other hand, when the mechanical components inside the motor 7 deteriorate over time, frictional resistance is large, so the decrease in the rotation speed of the blades 6 is greater than in other cases. Regarding the change in the rotation speed of the blades 6 with respect to the cumulative operating time in Figure 9, the changes are similar in the mode of deterioration of the electronic components and the mode of deterioration of the mechanical components inside the motor 7, but by comparing the deceleration time, it is possible to identify the cause of the failure.

[0081] That is, by periodically measuring the deceleration time and comparing the time transition of the deceleration time, it becomes possible to identify the failure mode.

[0082] FIG. 23 is a schematic diagram showing, using contour lines, a scatter diagram of the correlation between the rotational speed of the blades 6 and the deceleration time after a predetermined time has elapsed since the supply of drive power to the motor 7 was stopped for a number of ventilation fans. The inner lines indicate a higher occurrence frequency. The innermost line corresponds to the median value.

[0083] In the normal state, the rotation speed and deceleration time of the blades 6 are distributed with a normal distribution having a variation around a central value.

[0084] In the F1 mode, where pressure loss increases due to clogging of the ventilation air passage such as the filter or the blades 6, the rotation speed of the blades 6 increases and the deceleration time increases slightly correspondingly, so a distribution is obtained as shown in F1 in FIG. 23, with the rotation speed of the blades 6 being higher and the deceleration time being slightly longer than in the standard state.

[0085] In the F2 mode, the voltage applied to the motor 7 decreases due to deterioration of electronic components over time, and the rotation speed of the blades 6 decreases, and the deceleration time increases slightly accordingly. As a result, a distribution is obtained as shown in F2 in FIG. 23, which is centered on the side where the rotation speed of the blades 6 is lower and the deceleration time is slightly shorter than in the standard state.

[0086] In the F3 mode, where mechanical friction increases due to deterioration over time of the mechanical parts inside the motor 7, the rotation speed of the blades 6 decreases and the deceleration time decreases significantly, so that a distribution is obtained as shown in F3 in FIG. 23, where the rotation speed of the blades 6 is lower and the deceleration time is significantly shorter than in the standard state, and which is centered on the side where the deceleration time is shorter than in the F2 mode.

[0087] In this way, by classifying the distribution of the rotation speed and deceleration time of the blades 6 based on a plurality of indexes, that is, the rotation speed and deceleration time of the blades 6, it is possible to determine the failure mode.

[0088] In the second embodiment, the operation information includes the time transition of the rotation speed of the blades 6, the time transition of the deceleration time, and the cumulative operation time of the ventilation device 1, and this operation information is used as input data to the learning device 14 and the inference device 15. Other learning methods and the like are implemented in the same way as in the first embodiment.

[0089] In addition to the time transition of the rotation speed of the blades 6, the time transition of the deceleration time, and the cumulative operating time of the ventilation device 1, the time transition of the voltage applied to the motor 7 described in the first embodiment may be added to the operating information. By adding the operating information items, the failure mode can be accurately determined.

[0090] Embodiment 3 In embodiment 3, the operating information does not include the voltage value applied to the motor 7, but includes the time transition of the rotation speed of the blades 6 during high speed operation and the time transition of the rotation speed of the blades 6 during low speed operation. Other aspects are the same as in embodiment 1. In embodiment 3, differences from embodiment 1 will be mainly described.

[0091] The operation information includes the rotation speed of the blades 6 or the motor 7 during strong operation, the rotation speed of the blades 6 or the motor 7 during weak operation, and the cumulative operation time of the ventilation device 1. Strong operation means operation of the ventilation device 1 in the strong operation mode, and weak operation means operation of the ventilation device 1 in the strong operation mode.

[0092] 24 is a diagram showing the configuration of the operation information detection means 8 included in the failure prediction system according to embodiment 3. The operation information detection means 8 includes a rotation speed detection means 9 that detects the rotation speed of the blades 6 or the motor 7, a strong / weak operation mode detection means 20 that detects whether the ventilation device 1 is operating at strong or weak mode, and a cumulative operation time detection means 11 that detects the cumulative operation time of the ventilation device 1.

[0093] The strong / weak operation is switched by the switch 103 in FIG. 3, and the detector 150 in FIG. 3 detects the strong / weak operation mode.

[0094] Measurement of the rotation speeds in each of the strong and weak operation modes is performed, for example, at night. After completing the measurement of the rotation speeds, the strong and weak operation mode detection means 20 returns the ventilator 1 to its original operating state. This makes it possible to obtain the operating state while minimizing the impact on the user who installed the ventilator 1.

[0095] Fig. 25 is a schematic diagram showing a scatter diagram of the correlation between the rotation speed of the blades 6 in the high operation mode and the rotation speed of the blades 6 in the low operation mode for a number of ventilation fans, using contour lines. The inner lines indicate a higher occurrence frequency. The innermost line corresponds to the median value.

[0096] In the normal state, the rotation speed of the blades 6 in the strong operation mode and the rotation speed of the blades 6 in the weak operation mode are correlated in proportion to each other around a central value, and a normal distribution with variations is obtained.

[0097] In the F1 mode, the pressure loss increases due to clogging of the ventilation air passage such as the filter or the blades 6. The rotation speed of the blades 6 in the strong operation mode and the rotation speed of the blades 6 in the weak operation mode both increase, and a normal distribution with variation is obtained as an extension of the correlation in the standard mode.

[0098] In the F2 mode, the applied voltage to the motor 7 decreases due to deterioration of electronic components over time. The rotation speed of the blades 6 in both the strong operation mode and the weak operation mode decreases, and a normal distribution with variation is obtained as an extension of the correlation in the standard mode.

[0099] In the F3 mode, which shows an increase in mechanical friction due to the aging of mechanical components inside the motor 7, the mechanical friction consists of a proportional component that increases in proportion to the rotation speed and a constant component that is generated at a constant rate regardless of the rotation speed. As the rotation speed decreases, the proportion of the constant component increases, and the influence of the mechanical friction becomes greater. In other words, the weak operation mode has a distribution centered on the side where the rotation speed of the blades 6 in the weak operation mode is lower than that in the strong operation mode, as shown in F3 in Figure 25, compared to F2 in Figure 25.

[0100] In this way, the failure mode can be determined by classifying the distribution of the rotation speed of the blades 6 in the strong operation mode and the rotation speed of the blades 6 in the weak operation mode based on multiple indexes, namely, the rotation speed of the blades 6 in the strong operation mode and the rotation speed of the blades 6 in the weak operation mode.

[0101] In the third embodiment, the operation information includes the time transition of the rotation speed of the blades 6 in the strong operation mode, the time transition of the rotation speed of the blades 6 in the weak operation mode, and the cumulative operation time of the ventilation device 1, and this operation information is used as input data to the learning device 14 and the inference device 15. Other learning methods and the like are implemented in the same manner as in the first embodiment.

[0102] In addition to the time transition of the rotation speed of the blades 6 in the high operation mode and the time transition of the rotation speed of the blades 6 in the low operation mode, the time transition of the voltage applied to the motor 7 described in embodiment 1 may be added to the operation information. The time transition of the deceleration time described in embodiment 2 may be added to the operation information. By adding items of operation information, the failure mode can be accurately determined.

[0103] 26 is a diagram showing a processor 91 when at least some of the functions of the operation information detection means 8 and the communication means 12 of the ventilation device 1 of the failure prediction system 50 according to embodiment 1 are realized by the processor 91. In other words, at least some of the functions of the operation information detection means 8 and the communication means 12 may be realized by the processor 91 executing a program stored in a memory 92. The processor 91 is a CPU (Central Processing Unit), a processing system, an arithmetic system, a microprocessor, or a DSP (Digital Signal Processor). FIG. 26 also shows the memory 92.

[0104] When at least some of the functions of the driving information detection means 8 and the communication means 12 are realized by the processor 91, the at least some of the functions are realized by the processor 91 and software, firmware, or a combination of software and firmware. The software or firmware is written as a program and stored in the memory 92. The processor 91 realizes at least some of the functions of the driving information detection means 8 and the communication means 12 by reading and executing the program stored in the memory 92.

[0105] When at least some of the functions of the driving information detection means 8 and the communication means 12 are realized by the processor 91, the ventilation device 1 has a memory 92 for storing a program that results in the execution of at least some of the steps executed by the driving information detection means 8 and the communication means 12. It can also be said that the program stored in the memory 92 causes a computer to execute at least some of the procedures or methods executed by the driving information detection means 8 and the communication means 12.

[0106] The memory 92 may be, for example, a non-volatile or volatile semiconductor memory such as a RAM (Random Access Memory), a ROM (Read Only Memory), a flash memory, an EPROM (Erasable Programmable Read Only Memory), an EEPROM (registered trademark) (Electrically Erasable Programmable Read-Only Memory), a magnetic disk, a flexible disk, an optical disk, a compact disk, a mini disk, or a DVD (Digital Versatile Disk).

[0107] 27 is a diagram showing a processing circuit 93 in a case where at least some of the functions of the operation information detection means 8 and the communication means 12 included in the ventilation device 1 of the failure prediction system 50 according to embodiment 1 are realized by the processing circuit 93. In other words, at least some of the functions of the operation information detection means 8 and the communication means 12 may be realized by the processing circuit 93.

[0108] The processing circuitry 93 is dedicated hardware, and may be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a combination thereof.

[0109] Some of the functions of the driving information detection means 8 and the communication means 12 may be realized by dedicated hardware that is separate from the hardware that realizes the remaining functions of the driving information detection means 8 and the communication means 12.

[0110] Some of the functions of the driving information detection means 8 and the communication means 12 may be realized by software or firmware, and the remaining functions may be realized by dedicated hardware. In this way, the functions of the driving information detection means 8 and the communication means 12 can be realized by hardware, software, firmware, or a combination thereof.

[0111] At least some of the functions of the rotation speed detection means 9, the motor applied voltage detection means 10, and the cumulative operating time detection means 11 included in the operating information detection means 8 of the first embodiment may be realized by a processor or a processing circuit. For example, the functions of the rotation speed detection means 9, the motor applied voltage detection means 10, and the cumulative operating time detection means 11 may be realized by a processor or a processing circuit separate from the processors or processing circuits that realize the functions of the other means. The processor in question is a processor equivalent to the processor 91. The processing circuit in question is a processing circuit equivalent to the processing circuit 93.

[0112] At least some of the functions of the communication means 13 of the information processing device 2 included in the failure prediction system 50 according to the first embodiment may be realized by a processor or a processing circuit. At least some of the functions of the communication means 17 of the equipment management terminal 3 included in the failure prediction system 50 according to the first embodiment may be realized by a processor or a processing circuit. The processor is equivalent to the processor 91. The processing circuit is equivalent to the processing circuit 93.

[0113] At least some of the functions of the detection unit 150 and the switching control unit 131 included in the motor circuit 101 according to the first embodiment may be implemented by a processor or a processing circuit. The processor is equivalent to the processor 91. The processing circuit is equivalent to the processing circuit 93.

[0114] At least some of the functions of the data acquisition unit 303 and the model generation unit 304 of the learning device 14 of the failure prediction system 50 according to the first embodiment may be realized by a processor or a processing circuit. The processor is a processor equivalent to the processor 91. The processing circuit is a processing circuit equivalent to the processing circuit 93.

[0115] At least a part of the functions of the data acquisition unit 402 included in the storage unit 16 of the failure prediction system 50 according to the first embodiment may be realized by a processor or a processing circuit. The processor is equivalent to the processor 91. The processing circuit is equivalent to the processing circuit 93.

[0116] At least some of the functions of the data acquisition unit 502 and the inference unit 503 included in the inference device 15 of the failure prediction system 50 according to the first embodiment may be realized by a processor or a processing circuit. The processor is equivalent to the processor 91. The processing circuit is equivalent to the processing circuit 93.

[0117] At least some of the functions of the communication means 601 and the acquisition means 602 included in the display device 18 of the failure prediction system 50 according to the first embodiment may be realized by a processor or a processing circuit. The processor is equivalent to the processor 91. The processing circuit is equivalent to the processing circuit 93. The display means 603 included in the display device 18 is realized by, for example, a liquid crystal display device.

[0118] At least some of the functions of the rotation speed detection means 9, the deceleration time detection means 19, and the cumulative driving time detection means 11 included in the driving information detection means 8 of the second embodiment may be realized by a processor or a processing circuit. The processor is a processor equivalent to the processor 91. The processing circuit is a processing circuit equivalent to the processing circuit 93.

[0119] At least some of the functions of the rotation speed detection means 9, the strong / weak operation mode detection means 20, and the cumulative operation time detection means 11 included in the operation information detection means 8 of the third embodiment may be realized by a processor or a processing circuit. The processor is a processor equivalent to the processor 91. The processing circuit is a processing circuit equivalent to the processing circuit 93.

[0120] The configurations shown in the above embodiments are merely examples, and may be combined with other known technologies, or different embodiments may be combined with each other, or part of the configuration may be omitted or modified within the scope of the gist of the invention.

[0121] 1 Ventilation device, 2 Information processing device, 3 Equipment management terminal, 4 Global network, 5 Equipment manager, 6 Blade, 7 Motor, 8 Operation information detection means, 9 Rotational speed detection means, 10 Motor applied voltage detection means, 11 Cumulative operation time detection means, 12, 13, 17, 601 Communication means, 14 Learning device, 15 Inference device, 16 Storage means, 18 Display device, 19 Deceleration time detection means, 20 Strong / weak operation mode detection means, 50 Failure prediction system, 91 Processor, 92 Memory, 93 Processing circuit, 101 Motor circuit, 102 Wiring, 103 Switch, 110 Rectification unit, 111 Diode bridge circuit, 120 Smoothing unit, 121 Smoothing capacitor, 130 Switching unit, 131 Switching control unit, 140 Motor winding unit, 150 Detection unit, 161 Hall element PH1, 162 Rotation speed detection unit NO1, 163 Motor applied voltage detection unit motor applied voltage VO1, 301 Time transition of operation information, 302 Failure time and failure cause (correct answer), 303, 402, 502 Data acquisition unit, 304 Model generation unit, 401 Operation information, 403 Memory unit, 503 Inference unit, 602 Acquisition means, 603 Display means.

Claims

1. A failure prediction system comprising: an operation information detection means for detecting operation information of a ventilation device; a storage means for storing the time progression of the operation information; a learning device for learning the failure timing and failure causes of the ventilation device; and an inference device for inferring the failure prediction timing and failure prediction causes of the ventilation device.

2. The failure prediction system according to claim 1, further comprising a ventilation device, the ventilation device having one or more blades and one or more motors.

3. The failure prediction system according to claim 2, characterized in that the operating information includes the rotation speed of the blade or the motor, the voltage value applied to the motor, and the cumulative operating time of the ventilation device.

4. The failure prediction system according to claim 3, characterized in that the operating information detection means comprises: a rotation speed detection means for detecting the rotation speed of the blades or the motor; a motor applied voltage detection means for detecting a voltage value applied to the motor; and a cumulative operating time detection means for detecting a cumulative operating time of the ventilation device.

5. The failure prediction system described in claim 2, characterized in that the operating information includes the rotational speed of the blades or the motor, the deceleration time which is the time from when the supply of power to the motor is stopped until the rotational speed decelerates to a predetermined value, and the accumulated operating time of the ventilation device.

6. The failure prediction system according to claim 5, characterized in that the operating information detection means comprises: a rotation speed detection means for detecting the rotation speed of the blades or the motor; a deceleration time detection means for detecting a deceleration time, which is the time from when the supply of power to the motor is stopped until the rotation speed decelerates to the predetermined value; and a cumulative operating time detection means for detecting a cumulative operating time of the ventilation device.

7. The failure prediction system according to claim 2, characterized in that the operating information includes the rotational speed of the blades or the motor during strong operation, the rotational speed of the blades or the motor during weak operation, and the cumulative operating time of the ventilation device.

8. The failure prediction system according to claim 7, characterized in that the operating information detection means comprises: a rotation speed detection means for detecting the rotation speed of the blades or the motor; a strong / weak operating mode detection means for detecting whether the ventilation device is operating in strong or weak operation; and a cumulative operating time detection means for detecting a cumulative operating time of the ventilation device.

9. The failure prediction system described in claim 4, 6 or 8, characterized in that the learning device has: a data acquisition unit that acquires learning data including the operating information of the ventilation device and the failure timing and failure factors of the ventilation device; and a model generation unit that uses the learning data to generate a trained model for inferring the failure prediction timing and failure prediction factors of the ventilation device from the time progression of the operating information of the ventilation device.

10. The failure prediction system described in claim 9, characterized in that the inference device has a data acquisition unit that acquires time progression of operating information of the ventilation device, and an inference unit that uses the learned model for inferring the failure prediction timing and failure prediction factors of the ventilation device from the time progression of the operating information acquired by the data acquisition unit of the inference device, and outputs the failure prediction timing and failure prediction factors of the ventilation device from the time progression of the operating information acquired by the data acquisition unit of the inference device.

11. A failure prediction system as described in claim 10, further comprising a display device for displaying the predicted failure time and failure prediction factors of the ventilation device, the display device having an acquisition means for acquiring the predicted failure time and failure prediction factors of the ventilation device inferred by the inference device, and a display means for displaying the predicted failure time and failure prediction factors of the ventilation device acquired by the acquisition means.

12. The failure prediction system described in claim 11, characterized in that the memory means, the learning device and the inference device are provided in an information processing device, and the information processing device is capable of communicating with the ventilation device and the display device via a global network.

13. The failure prediction system according to claim 12, characterized in that the ventilation device has a communication means for communicating with the information processing device via the global network.

14. The failure prediction system according to claim 13, wherein the display device further comprises a communication means for communicating with the information processing device via the global network.

Citation Information

Patent Citations

  • Mine main ventilator fault diagnosis method based on neural network

    CN113568396A

  • Fine pre-diagnosis and precise operation and maintenance device and system for ventilator

    CN114893428A

  • Apparatus diagnosing device, refrigeration cycle device, apparatus diagnosing method, apparatus monitoring system and refrigeration cycle monitoring system

    JP2005241089A

  • Apparatus management system

    JP2020135020A

  • Methods for optimizing fan efficiency and / or performance or fan placement

    JP2021512419A

Cited By

  • Fault prediction and health management method and system for ventilation system end equipment

    CN122364790A

  • Air conditioning system, inspection method, computer program, storage medium, air conditioner, and inspection terminal

    EP4793564A1