Centralized management system and method

Abstract

A centralized management system (SYS) comprising a data processor (101) that collects information regarding at least one electric motor (5) and a management device (200) that communicates with the data processor (101) via a communication network (400). The data processor (101) stores current data which corresponds to a certain period of time and is obtained by measuring a current (Iuv) flowing through the electric motor in a time-division manner and transmits current data (Iu_lim[n]) in a data amount divided in correspondence to the current frequency of the electric motor to the management device (200) via the communication network (400). The data processor (101) transmits deterioration estimation information (West[n]) to the management device together with the current data, the deterioration estimation information (West[n]) being used for identifying a cause of an abnormality in the electric motor.

Description

Centralized management system, learning device, and centralized management method

[0001] The present disclosure relates to a centralized management system, a learning device, and a centralized management method.

[0002] When driving a compressor equipped with an electric motor having bearings, most failures are caused by deterioration or damage to the motor's bearings. If the compressor is used when the motor's bearings are worn and deteriorated, the deterioration can cause the motor to stop, resulting in significant downtime (stoppage time) for the compressor's operation and reducing the compressor's availability. Therefore, it is preferable to detect the deterioration of the motor's bearings before the motor stops due to bearing wear and take appropriate measures, such as repairing the deteriorated motor's bearings. This can reduce or shorten the compressor's downtime and improve the compressor's availability.

[0003] International Publication No. 2022 / 030516 (Patent Document 1) discloses an abnormality determination device including an abnormality determination unit that determines whether a mechanical element of a compressor driven by an electric motor has an abnormality based on an electrical characteristic of the compressor. Here, the electrical characteristic is a specific frequency component of a signal representing a physical quantity correlated with the current of the electric motor that drives the compressor, and the specific frequency component is a component that is a positive integer multiple of the rotation frequency of the compressor. The abnormality determination unit determines whether a mechanical element of the compressor has an abnormality by taking into account fluctuations in the relationship between the abnormality of the mechanical element and the electrical characteristic of the compressor due to changes in the rotation frequency of the compressor.

[0004] International Publication No. 2022 / 030516

[0005] Users, such as building owners, have a need to centrally monitor the deterioration of motor bearings in the multiple properties they manage. However, the technology disclosed in International Publication No. 2022 / 030516 (Patent Document 1) does not anticipate using information from multiple compressors to improve the accuracy of abnormality determination. Therefore, high-precision analysis is not possible, making it difficult to improve services for users.

[0006] The present disclosure discloses a centralized management system that solves the above-mentioned problems, and its purpose is to collect a large amount of data to perform highly accurate analysis and provide better services to users.

[0007] The present disclosure relates to a centralized control system. The centralized control system includes a data processor that collects information about at least one electric motor and a management device that communicates with the data processor via a communication network. The data processor measures the current flowing through the electric motor in a time-sharing manner, stores the current data for a certain period of time, and transmits the current data to the management device via the communication network after dividing the data into data amounts corresponding to the current frequency of the electric motor. The data processor transmits to the management device, together with the current data, degradation estimation information that is information used to identify the cause of an abnormality in the electric motor.

[0008] According to the centralized management system of the present disclosure, highly accurate analysis can be performed from a large amount of data, and services can be provided to users.

[0009] 1 is a block diagram for explaining an example of the configuration of a centralized control system according to embodiment 1. FIG. 1 is a cross-sectional view showing an example of the internal structure of a compressor. FIG. 1 is a cross-sectional view showing an example of a bearing when the main shaft and main bearing are in a lubricated state and driving normally. FIG. 1 is a cross-sectional view showing a bearing when the main shaft and main bearing are not in a lubricated state and driving abnormally. FIG. 2 is a control block diagram of a management device according to embodiment 1. FIG. 3 is a block diagram for explaining an example of processing in the centralized control system according to embodiment 1. FIG. 4 is a flowchart for explaining an example of processing in a data processor in the control device. FIG. 5 is a diagram for explaining data processing of current data. FIG. 6 is a diagram for explaining an example of deterioration allocation processing in a data processor in the control device. FIG. 7 is a flowchart for explaining output timing to a network. FIG. 8 is a diagram for explaining a data storage state in the management device according to embodiment 1. FIG. 9 is a flowchart for explaining an example of processing in the management device. FIG. 10 is a control block diagram of a management device according to embodiment 2. FIG. 11 is a diagram for explaining an example of processing in a centralized control system according to embodiment 3. FIG. 12 is a diagram for explaining a data storage state in the management device according to embodiment 3. FIG. 13 is a flowchart for explaining output timing to a communication network according to embodiment 4.

[0010] A centralized control system, learning device, and centralized control method according to embodiments will be described below with reference to the drawings. In each drawing, components designated by the same reference numerals are identical or equivalent components, and are common to all of the embodiments described below. Furthermore, the configurations of the components shown in all of the embodiments described below are merely examples and are not limited to the configurations described below. In particular, the combinations of components are not limited to the combinations in each embodiment, and components described in other embodiments can be applied to other embodiments. Furthermore, the level of parameters such as pressure and temperature is not determined in relation to absolute values, but is determined relatively in terms of the state and operation of the device, etc.

[0011] First Embodiment A centralized control system SYS will be described with reference to the drawings. Fig. 1 is a block diagram for explaining an example of the configuration of a centralized control system according to the first embodiment. The centralized control system SYS shown in Fig. 1 includes a management device 200 and a data processor 101.

[0012] The target of fault diagnosis in the centralized control system SYS is an electric motor (not shown) included in the compressor 5. A plurality of outdoor units 1000, 2000, 3000, ... are connected to the management device 200 via a communication network 400. A data processor 101 is mounted in the control device 100 included in each outdoor unit.

[0013] The configuration of the outdoor unit 1000 will be described below as a representative example, but the outdoor units 2000, 3000, . . . have a similar configuration.

[0014] The configuration that drives the compressor 5 includes an AC power supply 1, a rectifier circuit 2, an electrolytic capacitor 3, and an inverter 4. The rectifier circuit 2 converts three-phase (UVW phase) AC power from the AC power supply 1 into DC power. The electrolytic capacitor 3 smoothes the DC power from the rectifier circuit 2. The inverter 4 converts the DC power from the rectifier circuit 2 into three-phase AC power and outputs the three-phase AC power to the compressor 5. The current sensor 6 is provided midway along a wiring 7 that runs from the inverter 4 to the compressor 5, and detects currents Iuv (e.g., currents Iu and Iv) of two phases of the three-phase AC that flows from the inverter 4 to the compressor 5.

[0015] The configuration for driving the compressor 5 further includes a control device 100 that controls the driving of the compressor 5. The control device 100 outputs gate pulses GP to the inverter 4 based on the results detected by the current sensor 6 and the bus voltage 8, thereby controlling and driving the compressor 5. In addition, the management device 200 collects data from the control devices in the multiple outdoor units via a communication network 400. The management device 200 then analyzes the collected data and outputs the cost required for maintenance to the notification device 300 via a network 500. The management device 200 is, for example, a server connected to the notification device 300 and the outdoor unit 1000 via the communication network 400.

[0016] The control device 100 includes a data processor 101, a CPU (Central Processing Unit) 102, a memory 103 (a ROM (Read Only Memory) and a RAM (Random Access Memory)), an input / output buffer (not shown), and the like. The data processor 101 and the CPU 102 load and execute programs stored in the ROM into the RAM or the like. The programs stored in the ROM contain processing procedures for the control device 100. The control device 100 controls each device in the refrigeration cycle apparatus in accordance with these programs. The control device 100 can be realized by hardware such as a computer that executes various programs. For example, an FPGA (Field-Programmable Gate Array) can be used instead of the data processor 101 or the CPU 102 as the computing entity that executes various processes. This control is not limited to software processing, and can also be processed by dedicated hardware (electronic circuits).

[0017] The control device 100 performs signal processing in a data processor 101 inside the control device 100 and transmits the processed signal to a communication network 400. The management device 200 receives the signal from the communication network 400, calculates a maintenance cost from the signal, and outputs the calculated cost to a notification device 300. The notification device 300 notifies a user or a maintenance team of the maintenance cost calculated by the management device 200. For example, the notification device 300 is installed on a monitor of an indoor unit or on a management monitor of a maintenance team. The communication network 400 is, for example, a LAN, and the network 500 is, for example, the Internet.

[0018] Fig. 2 is a cross-sectional view showing an example of the configuration of the compressor 5 in Fig. 1. As shown in Fig. 2, the compressor 5 includes a suction pipe 51, a main shaft 52, an electric motor 53, an oil pump 55, an auxiliary bearing 56, a main bearing 57, a compression mechanism 58, and a discharge pipe 59. These components are housed in a housing of the compressor 5. Lubricating oil 54 is stored in the bottom of the housing.

[0019] The compressor 5, which constitutes part of the air conditioning equipment, is connected to other elements by refrigerant piping to form a refrigeration cycle. As will be described later, refrigerant is drawn into the compressor 5 through a suction pipe 51 and discharged from a discharge pipe 59.

[0020] The suction pipe 51 is a pipe for drawing low-temperature, low-pressure refrigerant into the compressor 5. A pressure sensor, a temperature sensor, a humidity sensor, etc. may be attached to the suction pipe 51 to measure the pressure, temperature, humidity, etc. of the refrigerant flowing through the pipe. Furthermore, these sensors may be attached to piping within the air conditioning equipment to estimate the pressure, temperature, humidity, etc. of the refrigerant flowing through the suction pipe 51.

[0021] Although not shown, the electric motor 53 is connected to a three-phase AC power line and is driven in response to the voltage applied from the inverter 4. The main shaft 52 is connected to the electric motor 53 and transmits rotational energy to the compression mechanism 58. The lubricating oil 54 is collected at the bottom of the housing of the compressor 5 and is supplied to the auxiliary bearing 56 and the main shaft 52 by the oil pump 55 to lubricate the auxiliary bearing 56 and the main shaft 52. Note that in order to check the amount of the lubricating oil 54, a liquid level sensor that can detect the oil level of the lubricating oil 54 may be attached and the amount of the lubricating oil 54 may be measured.

[0022] The discharge pipe 59 is a pipe for discharging the high-temperature, high-pressure refrigerant compressed by the compression mechanism 58 to the outside of the compressor 5. A pressure sensor, a temperature sensor, a humidity sensor, etc. may be attached to the discharge pipe 59 to measure the pressure, temperature, humidity, etc. of the refrigerant flowing inside the pipe. Furthermore, these sensors may be attached to piping inside the air conditioning equipment to estimate the pressure, temperature, humidity, etc. of the refrigerant flowing inside the discharge pipe 59.

[0023] Furthermore, the state of the main shaft 52 and the main bearing 57 when the compressor 5 is driven will be described using Figures 3 and 4. Figure 3 is a cross-sectional view of the main shaft 52 and the main bearing 57 when they are in a lubricated state and driving normally. Figure 4 is a cross-sectional view of the main shaft 52 and the main bearing 57 when they are not in a lubricated state and driving abnormally. Note that the main bearing 57 shown is a plain bearing.

[0024] As shown in Figure 3 , during normal operation, sufficient lubricating oil 54 fills the gap between the main shaft 52 and the main bearing 57, ensuring a certain clearance between the main shaft 52 and the main bearing 57 and allowing the main shaft 52 to rotate smoothly. On the other hand, as shown in Figure 4 , during abnormal operation, the viscosity of the lubricating oil 54 decreases due to factors such as temperature and aging, making it impossible to ensure an oil film between the main shaft 52 and the main bearing 57. This causes partial contact between the main shaft 52 and the main bearing 57, resulting in wear of the main bearing 57 at this contact point. If the main shaft 52 continues to rotate while the main bearing 57 is already worn, the wear of the main bearing 57 will worsen, resulting in further deterioration. This may ultimately result in a shutdown of the compressor 5 (downtime), which could reduce the operating rate of the compressor 5. At the same time, whirling occurs with each mechanical angle cycle, accelerating deterioration and reducing the operating rate of the compressor 5.

[0025] Next, a description will be given of the control of the management device 200. Fig. 5 is a control block diagram of the management device according to embodiment 1. The management device 200 includes a database (DB) 201, a learning unit 202, an inference unit 203, and a maintenance cost calculation unit 204.

[0026] The database (DB) 201, details of which will be described in FIG. 6 , processes data from the communication network 400 and outputs data DWA associated with deterioration to the learning unit 202. The learning unit 202, details of which will be described in FIG. 12 , performs machine learning using the data DWA associated with deterioration and outputs a pattern generation function PGF to the inference unit 203. The inference unit 203 uses the limit current data Iu_lim[n] from the communication network 400 and the pattern generation function PGF to output a deterioration level West[n] of the compressor to the maintenance cost calculation unit 204. The maintenance cost calculation unit 204, details of which will be described later, calculates the cost of maintenance from the deterioration level.

[0027] Next, details of the internal processing of the data processor 101, the communication network 400, and the management device 200 will be described using FIG. 6. FIG. 6 is a block diagram for explaining an example of processing in the centralized management system according to the first embodiment. The data processor 101 is configured to include a limit number calculation processing unit 1011, a data limit processing unit 1012, a drive time count processing unit 1013, and a deterioration allocation processing unit 1014. The data processor 101 operates as these processing units using loaded software. Details of the control executed by these processing units will be described below.

[0028] The limit number calculation processing unit 1011 calculates the limit number N_lim[n] based on the U-phase current Iu. Details of the limit number calculation processing unit 1011 will be described with reference to FIG. 7 . FIG. 7 is a flowchart illustrating an example of processing by the data processor in the control device. In FIG. 7 , the data processor 101 executes processing to read the current Iu in step S1011-1. Note that while FIG. 7 illustrates the U-phase current Iu being read, the W-phase current Iw may be read, or a V-phase current calculated from the U-phase current Iu and the W-phase current Iw may be read. Furthermore, it is preferable to measure the U-phase current and the W-phase current during stable operation of the electric motor 53. That is, it is preferable to measure the currents at a time when there is little load fluctuation or speed fluctuation.

[0029] For example, if speed control is being implemented, current can be measured at a time when there is little speed fluctuation by controlling the speed command value to be constant for a certain period of time. Similarly, if torque control is being implemented, current can be measured at a time when there is little load fluctuation by controlling the torque command value to be constant for a certain period of time. Adjusting the timing in this way makes it possible to collect data with little current fluctuation during stable motor operation, enabling highly accurate analysis and providing better service to air conditioner users.

[0030] In step S1011-2, the data processor 101 performs moving average processing on the U-phase current Iu read in step S1011-1 in order to mitigate the influence of switching noise from the inverter 4. Note that this moving average processing is performed to prevent erroneous detection due to chattering when the U-phase current Iu value is near 0 in the processing of step S1011-3 described below.

[0031] In step S1011-3, the data processor 101 detects the timing (zero-cross timing) at which the value of the U-phase current Iu subjected to the moving average processing in step S1011-3 crosses zero. This will be further explained using FIG. 8. FIG. 8 is a diagram for explaining data processing of current data. In FIG. 8, the horizontal axis represents elapsed time, and the vertical axis represents compressor current. As shown in FIG. 8, in this zero-cross detection processing, the times at which the elapsed time reaches 0 are detected in the order of T[0], T[1], T[2], and so on, in ascending order. Here, this zero-cross timing is designated as T[n] (n is an integer equal to or greater than 0).

[0032] Returning to FIG. 7 , in step S1011-4, the data processor 101 calculates a frequency F[n] (n is an integer equal to or greater than 0) corresponding to one mechanical angle cycle. For example, the frequency F[n] is calculated using the timing T[n] (n is an integer equal to or greater than 0) detected in step S1011-3, as shown in equation (1). Note that P represents the number of pole pairs. When P = 1, the electrical angle and the mechanical angle are the same. However, when the number of pole pairs P is a positive integer other than 1, the frequency F[n] corresponding to the mechanical angle can be calculated. F[n] = 1 / (T[2nP + 2P] - T[2nP]) (1) Next, in step S1011-5, the data processor 101 calculates a limit number N_lim[n] using the frequency F[n] calculated using equation (1) and the sampling time Tsam, as shown in equation (2). This allows the number of data corresponding to one mechanical angle cycle to be sequentially calculated. Here, the sampling time Tsam is the reciprocal of the carrier frequency and is a fixed value. N_lim[n]=(1 / F[n]) / Tsam (2) According to equation (2), when the rotation speed of the motor is increasing (accelerating), N_lim[n] decreases as n increases, and when the rotation speed of the motor is decreasing (decelerating), N_lim[n] increases as n increases. Furthermore, when the motor is operating at a constant speed, N_lim[n] does not change.

[0033] Returning to FIG. 6 , the data restriction processing unit 1012 outputs the restriction current Iu_lim[n] to the communication network 400 based on the U-phase current Iu and the restriction number N_lim[n] calculated by the restriction number calculation processing unit 1011 .

[0034] 8, the control of the limit number calculation processing unit 1011 will be described in detail. The limit number calculation processing unit 1011 extracts data of the limit number N_lim[n] from the stored compressor current data at regular intervals and sets the data as the limit current Iu_lim[n]. As a result, the amount of data corresponding to one mechanical angle cycle is sequentially output to the communication network 400.

[0035] 6 , the drive time counting processor 1013 has an internal memory, counts the drive time of the compressor 5 from the U-phase current Iu based on the limit number N_lim[n], and outputs the operating time Twork[n] to the deterioration allocation processor 1014. Specifically, the drive time counting processor 1013 executes processing to start counting when the compressor 5 starts and the U-phase current Iu is flowing, and to stop counting when the compressor 5 stops. Furthermore, when the compressor 5 starts again, the drive time counting processor 1013 executes processing to start counting from the value at the time when counting was stopped the previous time.

[0036] The deterioration assignment processing unit 1014 executes a process of assigning a deterioration level West[n] according to the state of the operating time Twork[n] from the drive time count processing unit 1013. The process executed by the deterioration assignment processing unit 1014 will be specifically explained using Fig. 9. Fig. 9 is a diagram for explaining an example of deterioration assignment processing in a data processor in a control device. The horizontal axis of Fig. 9 represents operating time, and the vertical axis represents deterioration level.

[0037] For the operating time, operating times WT1 and WT2 are set in advance as threshold values. The operating time WT1 is the time during which the compressor 5 transitions from normal to slight degradation (light degradation in FIG. 9 ). The operating time WT2 is the time during which the compressor 5 transitions from slight degradation to severe degradation (severe degradation in FIG. 9 ). In other words, the operating times are set as an operating time interval during which the compressor 5 is normal (WT0 to WT1), an operating time interval during which the compressor 5 is slightly deteriorated (WT1 to WT2), and an operating drive interval during which the compressor 5 is severely deteriorated (WT2 and thereafter).

[0038] Therefore, the degradation assignment processing unit 1014 assigns the degradation state of the compressor 5 as normal, slightly deteriorated, or severely deteriorated based on the operating time Twork[n], and outputs the degradation level West[n] to the communication network 400. The thresholds WT1, WT2, etc. are calculated by calculating the skewness, kurtosis, and imbalance of the current for each individual compressor from past current data Iu_lim[n], and setting the thresholds WT1, WT2, etc. for each individual compressor based on these. Furthermore, as will be described later, in order to exclude the influence of an initial defect of the compressor 5, a determination value WTS for the operating time may be set in advance, and the degradation level West[n] may be output to the communication network 400 using the operating time Twork[n] after this determination value WTS.

[0039] Obtaining the degree of deterioration from the current waveform allows use of existing sensor information (current sensor), which reduces the amount of data and costs compared to obtaining the degree of deterioration using a separate sensor.

[0040] Next, the timing of data output to the communication network 400 will be described with reference to Fig. 10. Fig. 10 is a flowchart for explaining the timing of output to the network. First, in step S101-1, the data processor 101 calculates the operating time Twork[n] of the compressor 5 as processing of the drive time count processing unit 1013 in Fig. 6.

[0041] Next, in step S101-2, the data processor 101 determines whether the operating time Twork[n] calculated in step S101-1 is equal to or greater than the determination value WTS (see FIG. 9). If the operating time Twork[n] is equal to or greater than the determination value WTS (Yes in S101-2), processing proceeds to steps S101-3 and S101-5. On the other hand, if the operating time Twork[n] is less than the determination value WTS (No in S101-2), step S101-1 is executed again from the return. From the above, the current data can be linked to the operating time, enabling highly accurate analysis and providing better service to air conditioner users.

[0042] In step S101-3, the data processor 101 calculates a limit number N_lim[n] of the U-phase current Iu for one mechanical angle cycle as the limit number calculation processing unit 1011 in Fig. 6. Next, in step S101-4, the data processor 101 divides (Iu_lim[n]) the buffered data of the U-phase current Iu as the data limit processing unit 1012 in Fig. 6 using the limit number N_lim[n] calculated in step S101-3.

[0043] The data processor 101 performs the processes of steps S101-5 and S101-6 as the degradation allocation processing unit 1014 of FIG.

[0044] In step S101-5, the data processor 101 assigns the deterioration level West[n] according to the operating time Twork[n]. In step S101-6, the data processor 101 synchronizes the timing of the limited current Iu_lim[n] processed in step S101-4 with the deterioration level West[n] and outputs them to the communication network 400.

[0045] Although the processes of steps S101-3 and S101-4 and the process of step S101-5 are performed simultaneously in parallel, they may be performed sequentially. In this case, the order of execution of the processes of steps S101-3 and S101-4 and the process of step S101-5 does not matter.

[0046] Furthermore, although the above description is given for one outdoor unit, data can be output sequentially to the communication network 400 for a plurality of outdoor units by executing similar processing.

[0047] Returning to FIG. 6 , the processing of the database (DB) 201 in the management device 200 will be described. The database (DB) 201 includes a frequency conversion processing unit 2011 and a normalization processing unit 2012. The database (DB) 201 generates data DWA associated with degradation from the limit current Iu_lim[n] from the communication network 400 and the degradation degree West[n]. The frequency conversion processing unit 2011 converts the limit current Iu_lim[n] into the frequency domain to generate a spectrum SPE[n] corresponding to the frequency. Next, the normalization processing unit 2012 performs normalization processing to divide the spectrum SPE[n] by the fundamental component. For example, if the fundamental component is first-order, by dividing each order spectrum of the spectrum SPE[n] by the first-order spectrum, the value of the first-order component becomes 1, and the values ​​of the other order components become numbers indicating how many times the first-order spectrum they are.

[0048] Next, the data DWA associated with degradation will be described with reference to Fig. 11. Fig. 11 is a diagram for explaining the state of data storage in the management device according to the first embodiment. Fig. 11 shows the state of storage in the database (DB) 201 of Fig. 6. In Fig. 11, the horizontal columns represent orders representing the frequency domain, and the vertical rows represent spectral values ​​normalized for every n (n is an integer equal to or greater than 0).

[0049] For example, when n=0, the zeroth order A0 indicates the DC component, the first order "1" indicates the spectrum of the fundamental wave (first order), the second order A2 indicates the spectrum of the second harmonic, and the third order A3 indicates the spectrum of the third harmonic.

[0050] Since database (DB) 201 performs normalization processing using the fundamental wave (first order), as described above, the value of the first order is "1", and values ​​of the other orders A2, A3, etc. are accumulated and saved as values ​​less than 1. Furthermore, database (DB) 201 constructs a database by arranging the spectrum SPE[n] in the leftmost column and the deterioration degree West[n] in the rightmost column for each row [n]. As described above, database (DB) 201 accumulates and saves the spectrum SPE[n] and deterioration degree over time. Because sliding bearings cannot automatically repair deterioration, the longer the driving time, that is, the larger n, the worse the deterioration degree becomes.

[0051] Next, an example of the operation of the management device 200 according to the first embodiment will be described with reference to the flowchart of FIG. 12 , which illustrates the processing procedure. FIG. 12 is a flowchart illustrating an example of processing in the management device. The processing procedure described below is an example of the learning method of the present disclosure. Therefore, each process may be modified as much as possible, and processes may be omitted, replaced, or added as appropriate depending on the embodiment.

[0052] FIG. 12 shows the processes executed by the learning unit 202, the inference unit 203, and the maintenance cost calculation unit 204 shown in FIG. 5 in order.

[0053] First, the learning unit 202 performs machine learning to create a trained model. The inference unit 203 executes inference processing using the trained model that has undergone machine learning. Since the processing content differs between machine learning and inference processing, the processing procedure for machine learning will be described first.

[0054] In step S202-1, the learning unit 202 refers to the data DWA associated with degradation and acquires a spectrum SPE[n] from the data DWA as input data. The "data DWA associated with degradation" is, for example, the table in FIG. 11. The spectrum SPE[n] is the data (A0, 1, A2, etc.) other than the degradation level in the table in FIG. 11.

[0055] In step S202-2, the learning unit 202 acquires the degradation level West[n] as a label from the data DWA associated with degradation.

[0056] In step S202-3, the learning unit 202 acquires a set of data (referred to as a teacher data set) consisting of the input data acquired in step S202-1 and the label acquired in step S202-2, and performs learning with the teacher data. From the above, the learning unit 202 constructs a trained model by performing learning with the teacher data based on the teacher data set. This makes it possible to link the degree of deterioration of the compressor with the order of the current.

[0057] Here, the "trained model" specifically refers to a pattern generation function PGF, which is a function that receives a current spectrum as input and outputs a degradation level.

[0058] The machine learning targeted at the compressor in this embodiment 1 is learning with teacher data using a neural network configured by combining perceptrons. Specifically, a teacher data set consisting of input data indicating the state of current and a label corresponding to the deterioration state of the compressor is provided to the neural network, and learning is repeated while changing the weighting of each perceptron so that the output of the neural network is the same as the label.

[0059] In the learning assumption, a process called back-propagation (also called backpropagation) is repeated to adjust the weighting values ​​so as to reduce the error in the output of each perceptron.

[0060] In this way, the learning unit 202 learns the features of the training data set and inductively acquires a trained model to estimate the result from the input. That is, as described above, training with training data is performed by adjusting the weighting values ​​so that the error between the label and the output data is eliminated.

[0061] In this way, in the training with teacher data, a trained model for determining the degree of deterioration from the order information of the current is obtained as a training result. The trained model is then stored as the pattern generation function PGF of the next stage.

[0062] Note that the neural network used by the learning unit 202 for learning may be multi-layered, so that learning can be performed by so-called deep learning (also called deep learning).

[0063] In step S202-4, the trained model obtained by training with teacher data is saved as a pattern generation function PGF. The pattern generation function PGF may be updated by periodically executing the processes of steps S202-1 to S202-4. The pattern generation function PGF may also be updated by executing the processes of steps S202-1 to S202-4 each time the degradation state changes.

[0064] The pattern generation function PGF may generate the degradation level from current information. For example, a trained model obtained from at least one of current information such as current likelihood, skewness, and harmonics may be generated and saved as the pattern generation function PGF.

[0065] The pattern generation function PGF is output when the inference unit 203 performs processing, which will be described later. The inference unit 203 then determines the degree of degradation based on the normalized spectrum SPE[n].

[0066] Next, we will explain the processing performed by the inference unit 203 using the trained model after the machine learning has been performed as described above. First, in step S203-1, the inference unit 203 acquires the pattern generation function PGF, which is the trained model saved in step S202-4.

[0067] Next, in step S203-2, the inference unit 203 acquires, as input data, the normalized spectrum SPE[n] from the data DWA associated with the deterioration. Then, in step S203-3, the inference unit 203 calculates the degree of deterioration from the current information that is the target of deterioration diagnosis from the multiple outdoor units acquired from the communication network 400.

[0068] Although FIG. 9 shows an example in which the deterioration level is in three stages, the deterioration level is a state (normal, slightly deteriorated, heavily deteriorated) set by the designer for each level of maintenance cost, so the deterioration level does not necessarily have to be in three stages.

[0069] The maintenance cost calculation unit 204 calculates the cost required for maintenance based on the degree of deterioration calculated in step S203-3. The cost required for maintenance will be explained using Fig. 13. Fig. 13 is a diagram for explaining an example of processing by the management device according to the first embodiment. The horizontal axis represents the degree of deterioration, and the vertical axis represents the maintenance cost.

[0070] As shown in FIG. 13, the greater the degree of deterioration, the higher the maintenance cost. Maintenance costs include the cost of replacing parts and the cost of service personnel. The longer the service personnel's work time, the higher the maintenance cost. Therefore, the user can be notified before being charged a large fee, thereby improving user satisfaction. In this embodiment, the maintenance cost is notified to the notification device 300, but notifying the degree of deterioration may also be useful information for the user's facility management. In FIG. 13, the degree of deterioration increases with operating time, and the maintenance cost increases accordingly.

[0071] From the above, by performing the processing of the inference unit 203 described above, the management device 200 reads out a pattern generation function, which is a trained model, from the current information and determines the degree of deterioration of the compressor installed in the outdoor unit connected to the communication network 400. As a result, a large amount of versatile actual unit data necessary for deterioration diagnosis is collected, enabling highly accurate analysis and providing better service to air conditioner users.

[0072] Embodiment 2. Fig. 14 is a control block diagram of a management device 200 according to embodiment 2. The management system according to embodiment 2 differs from the management system according to embodiment 1 in the processing within the management device 200. Specifically, in embodiment 2, a parts selection unit 205 is provided instead of the maintenance cost calculation unit 204. The other components and processing are the same. This parts selection unit 205 determines the parts or the number of service personnel required for maintenance according to the degree of deterioration from the inference unit 203, and outputs the result to the notification device 300.

[0073] As a result, the number of parts and service personnel required for maintenance can be determined in advance, allowing for an estimate of the time required to procure maintenance parts and personnel, enabling better service to be provided to air conditioner users.

[0074] 15 is a diagram for explaining an example of processing in a centralized management system according to embodiment 3. In the management system according to embodiment 3, processing within management device 200 differs from that of embodiment 1. Specifically, embodiment 3 differs from embodiment 1 in the internal processing of database (DB) 201 within management device 200.

[0075] The database (DB) 201 of the third embodiment newly includes an image processing unit 2013 instead of the frequency conversion processing unit 2011 and the normalization processing unit 2021 of the first embodiment. The image processing unit 2013 has a function of processing data of the limit current Iu_lim[n] from the communication network 400 into an image. The other components and processes are similar to those of the first and second embodiments, and therefore description thereof will not be repeated.

[0076] Next, data DWA (data associated with deterioration) generated from the database (DB) 201 will be described using FIG. 16 . FIG. 16 is a diagram for explaining the state of data storage in the management device according to the third embodiment. In the database (DB) 201, data processed by the image processing unit 2013 is associated with the deterioration level West[n]. Specifically, as shown in FIG. 16 , images and deterioration levels are accumulated and saved according to the number n. In FIG. 16 , for example, if the number of pole pairs of the motor is three, the mechanical angle is three times the electrical angle, so images are saved every three electrical angle cycles. Furthermore, the deterioration level West[n] is associated and saved corresponding to these images. Note that these processes may also be applied to multiple outdoor units.

[0077] 16 shows a change in which normality continues for a while, followed by mild deterioration for two cycles, and then severe deterioration. For convenience of explanation, the mild deterioration is described as continuing for only two cycles, but the period of mild deterioration does not have to be two cycles.

[0078] The database accumulated and saved as shown in Figure 16 undergoes the learning and inference processes described in the first embodiment, and ultimately notifies the notification device 300 of the maintenance cost. In this case, the input data for steps S202-1 and S203-2 described in Figure 12 is image data. By applying machine learning to this image data, it becomes possible to estimate deterioration from the waveform shape of the current, enabling highly accurate analysis and providing better service to air conditioner users.

[0079] Fourth Embodiment In a management system according to the fourth embodiment, the timing of output to the communication network 400 is different from that in the management system according to the first embodiment. Fig. 17 is a flowchart for explaining the timing of output to the communication network 400 according to the fourth embodiment. Specifically, in the fourth embodiment, compared to the flowchart of Fig. 10 described in the first embodiment, a new process of step S101-7 is added before the process of step S101-1, and the process of step S101-8 is executed instead of the process of step S101-2.

[0080] In step S101-7, the data processor 101 acquires time information for each outdoor unit. In step S101-8, the data processor 101 determines whether the time acquired in step S101-7 is the set time. If it is the set time (YES in step S101-8), the data processor 101 proceeds to steps S101-3 and S101-5. On the other hand, if it is not the set time (NO in step S101-8), the data processor 101 returns to step S101-7.

[0081] For example, if you set a time such as once a day at 3:00 a.m., you can collect data periodically, which will enable highly accurate analysis and provide better service to air conditioner users.You can also set it to once a week depending on the frequency of use by the user.

[0082] (Summary) Finally, the present embodiment will be summarized with reference to the drawings again.

[0083] (Section 1) The present disclosure relates to a centralized control system (SYS). The centralized control system (SYS) shown in FIG. 1 includes a data processor (101) that collects information about at least one electric motor (53) and a management device (200) that communicates with the data processor (101) via a communication network (400). The data processor (101) stores current data obtained by measuring the current (Iu) flowing through the electric motor (53) in a time-sharing manner for a certain period of time, divides the data into data amounts corresponding to the electric motor's current frequency, and transmits the resulting current data (Iu_lim[n] in FIG. 6) to the management device (200) via the communication network (S101-6 in FIG. 10). The data processor (101) also transmits degradation estimation information (degradation level (West[n] in FIG. 6), which is used to identify the cause of an abnormality in the electric motor, along with the current data to the management device (200).

[0084] (Section 2) In the centralized control system described in Section 1, the amount of data corresponding to the current frequency is the amount of data measured over one mechanical angle cycle of the electric motor 53 (Figure 8: T[0] to T[2]).

[0085] (3) In the centralized management system described in paragraph 1 or paragraph 2, the deterioration estimation information (Figures 6 and 9: deterioration level West[n]) is the operating time of the motor (Figure 9: operating time) calculated by integrating the period during which current flows.

[0086] (4) In the centralized management system described in 3, the data processor 101 is configured to update the current data and send it to the management device 200 as the operating time increases (FIG. 10: YES in S101-2, execute S101-6).

[0087] (Item 5) In the centralized management system described in item 4, the data processor 101 updates the current data at most once a day.

[0088] (Item 6) In the centralized control system described in any one of items 1 to 5, the data processor 101 is configured to acquire current data when the electric motor 53 is operating stably.

[0089] (Section 7) In the centralized management system described in any one of Sections 1 to 6, the management device 200 is configured to calculate the degree of deterioration of the electric motor 53 from the acquired current data and deterioration estimation information, calculate the maintenance cost based on the degree of deterioration (Fig. 12: 204, Fig. 13), and notify the maintenance cost.

[0090] (Section 8) In the centralized management system described in Section 7, as shown in Figure 6, the management device 200 is equipped with a frequency conversion processing unit 2011 that performs a process of converting current data (limit current Iu_lim[n]) into the frequency domain and generates a spectrum SPE[n] corresponding to the frequency, and a normalization processing unit 2012 that performs a normalization process of dividing the spectrum SPE[n] by the fundamental wave component.

[0091] (Item 9) In the centralized management system described in item 7, the management device 200 is configured to notify whether or not a maintenance part is required depending on the degree of deterioration (FIG. 14: 205, 300).

[0092] (Section 10) In the centralized management system described in Section 1, the management device 200 is configured to convert the current data into image data (Figure 15: 2013, Figure 16) and link it to deterioration estimation information (Figure 15: deterioration level West[n]).

[0093] (Section 11) In another aspect, the present disclosure relates to a learning device ( FIG. 12 : 202) that optimizes an estimation model used to predict deterioration of at least one electric motor 53. The learning device includes a data processor 101 that collects information related to the electric motor 53 and a management device 200 that communicates with the data processor 101 via a communication network 400. The management device 200 acquires current data obtained by dividing the current (Iuv) flowing through the electric motor 53 calculated by the data processor 101 into fixed time intervals, acquires deterioration estimation information that is information used to identify the cause of an abnormality in the electric motor 53 ( FIG. 12 : S202-2), and constructs a learning dataset including the current data and the deterioration estimation information. The management device 200 inputs current data extracted from the learning dataset into the estimation model and optimizes the estimation model so that the output estimation result approaches the deterioration estimation information labeled in the learning dataset.

[0094] (Section 12) In yet another aspect, the present disclosure relates to a centralized management method executed by using a data processor 101 that collects information on at least one electric motor 53 and a management device 200 that communicates with the data processor 101 via a communication network 400. The centralized management method includes a step in which the data processor 101 stores current data (Iuv) measured in a time-division manner on a current flowing through the electric motor 53 for a certain period of time, a step in which the data processor 101 divides the current data into data amounts corresponding to the current frequency of the electric motor 53 and transmits the divided data (Iu_lim[n] in FIGS. 6 and 8) to the management device 200 via the communication network (S101-3, S101-4 in FIG. 10), and a step in which the data processor 101 transmits degradation estimation information (degradation degree West[n] in FIG. 6), which is information used to identify the cause of an abnormality in the electric motor, together with the current data to the management device 200 (S101-6 in FIG. 10).

[0095] The embodiments disclosed herein should be considered to be illustrative in all respects and not restrictive. The scope of the present disclosure is indicated by the claims, not by the description of the above embodiments, and is intended to include all modifications within the meaning and scope of the claims. Furthermore, the inventions described in the embodiments and each modification are intended to be practiced, as far as possible, either alone or in combination.

[0096] REFERENCE SIGNS LIST 1 AC power supply, 2 Rectifier circuit, 3 Electrolytic capacitor, 4 Inverter, 5 Compressor, 6 Current sensor, 7 Wiring, 8 Bus voltage, 51 Suction pipe, 52 Main shaft, 53 Electric motor, 54 Lubricating oil, 55 Oil pump, 56 Sub-bearing, 57 Main bearing, 58 Compression mechanism, 59 Discharge pipe, 100 Control device, 101 Data processor, 102 CPU, 103 Memory, 200 Management device, 202 Learning unit, 203 Inference unit, 204 Maintenance cost calculation unit, 205 Parts selection unit, 300 Notification device, 400 Communication network, 500 Network, 1000, 2000, 3000 Outdoor unit, 1011 Limit number calculation processing unit, 1012 Data limitation processing unit, 1013 Drive time count processing unit, 1014 Deterioration allocation processing unit, 2011 Frequency conversion processing unit, 2012, 2021 Normalization processing unit, 2013 Image processing unit, SYS Centralized management system.

Claims

1. a data processor that collects information regarding the at least one electric motor; a management device communicating with said data processor via a communications network; The data processor is configured to store current data of a current flowing through the electric motor measured in a time-division manner for a certain period of time, and to transmit the current data to the management device via the communication network by dividing the data into data amounts corresponding to a current frequency of the electric motor; The data processor is configured to transmit deterioration estimation information, which is information used to identify a cause of an abnormality in the electric motor, together with the current data to the management device; A centralized control system in which the amount of data corresponding to the current frequency is calculated based on a carrier frequency.

2. A data processor that collects information regarding at least one electric motor; a management device communicating with said data processor via a communications network; The data processor is configured to store current data obtained by measuring a current flowing through the electric motor for a certain period of time, and to transmit the current data to the management device via the communication network after dividing the data into data amounts corresponding to a current frequency of the electric motor; The data processor is configured to transmit deterioration estimation information, which is information used to identify a cause of an abnormality in the electric motor, together with the current data to the management device; A centralized management system, wherein the amount of data corresponding to the current frequency is the amount of data measured during one mechanical angle period of the motor.

3. 3. The centralized control system according to claim 1, wherein the deterioration estimation information is an operating time of the electric motor obtained by integrating a period during which the current flows.

4. The centralized management system according to claim 3 , wherein the data processor is configured to update the current data as the operating time increases and send the updated current data to the management device.

5. The centralized control system of claim 4 , wherein the data processor updates the current data at most once a day.

6. The centralized control system according to claim 1 or 2, wherein the data processor is configured to obtain the current data when the electric motor is in stable operation.

7. The centralized management system of claim 1 or 2, wherein the management device is configured to calculate a degree of deterioration of the motor from the acquired current data and the deterioration estimation information, calculate a maintenance cost based on the degree of deterioration, and notify the maintenance cost.

8. The management device includes: a frequency conversion processing unit that converts the current data into a frequency domain and generates a spectrum corresponding to the frequency; The centralized control system according to claim 7 , further comprising: a normalization processing unit that performs normalization processing of dividing the spectrum by a fundamental wave component.

9. The centralized management system according to claim 7 , wherein the management device is configured to notify whether or not a maintenance part is required depending on the degree of deterioration.

10. The centralized management system according to claim 1 , wherein the management device is configured to convert the current data into image data and associate the image data with the deterioration estimation information.

11. 1. A centralized management method implemented using a data processor that collects information about at least one electric motor, and a management device that communicates with the data processor via a communications network, comprising: a step of storing current data of a current flowing through the electric motor measured in a time-division manner by the data processor for a certain period of time; The data processor divides the current data into data amounts corresponding to a current frequency of the motor and transmits the current data to the management device via the communication network; The data processor transmits deterioration estimation information, which is information used to identify a cause of an abnormality in the electric motor, together with the current data to the management device; A centralized management method in which the amount of data corresponding to the current frequency is calculated based on a carrier frequency.

12. 1. A centralized management method implemented using a data processor that collects information about at least one electric motor, and a management device that communicates with the data processor via a communications network, comprising: a step of storing current data of a current flowing through the electric motor measured in a time-division manner by the data processor for a certain period of time; The data processor divides the current data into data amounts corresponding to a current frequency of the motor and transmits the current data to the management device via the communication network; The data processor transmits deterioration estimation information, which is information used to identify a cause of an abnormality in the electric motor, together with the current data to the management device; A centralized management method, wherein the amount of data corresponding to the current frequency is the amount of data measured during one mechanical angle period of the motor.