JP4265982B2 - Equipment diagnostic equipment, refrigeration cycle equipment, refrigeration cycle monitoring system - Google Patents

Description

  The present invention relates to a technique relating to failure diagnosis of equipment such as a compressor of a refrigeration cycle apparatus used in a refrigeration apparatus or an air conditioner, a refrigerant circuit of a refrigeration cycle apparatus, a blower, or other equipment or devices.

  A conventional compressor failure diagnosis apparatus includes an acoustic signal obtained through a filter, frequency analysis, and comparison of a main component with a normal database to detect abnormal sound. See Patent Document 1. In order to diagnose abnormalities in rotating equipment, waveform data is extracted from the vibration sensor, and frequency and temporal feature values are extracted. Both data are collated with reference data and input to a neural network to identify the presence and cause of the abnormality. There is something. See Patent Document 2. For air conditioner failure diagnosis, control data such as sensors, setting values, and abnormal signals are taken in. Further, the operation status sequence of each failure is stored in the microcomputer with operation data such as pressure and temperature, and failure diagnosis is performed. Technology has been proposed. See Patent Document 3. On the other hand, many attempts have been made to use Mahalanobis distance, which is a method of multivariate analysis, for failure diagnosis. In the old days, there are known ones that compare vibration sensor signals with normal ones, see Patent Document 4, and recently, those that try to find signs of deterioration using many types of sensors, see Patent Document 5, etc. Yes.

  Further, in the conventional refrigeration cycle apparatus described in Patent Document 6, the liquid refrigerant in the liquid reservoir and the auxiliary tank is set to the same liquid level by connecting the liquid reservoir (liquid receiving tank) and the auxiliary tank through the communication pipe. The liquid level is detected by a float type level sensor installed in the auxiliary tank, and the refrigerant leakage is detected depending on whether or not the detected liquid level is higher than a predetermined normal liquid level.

Japanese Patent Application Laid-Open No. 11-117875 (columns of FIGS. 1, 7, and 0029) Japanese Patent Laid-Open No. 10-274558 (column 0037, column 0049, FIG. 22) JP-A-2-110242 (FIGS. 4 to 11) JP 59-68643 (page 23, upper left to upper right column) JP 2000-259222 A (FIGS. 3 to 9) JP-A-10-103820

  Conventional attempts to analyze a failure from a specific signal such as sound or vibration have a problem that an extreme abnormal state can be determined, but an accurate apparatus cannot be obtained. For example, if the noise value near the compressor or vibration acceleration is measured, and if these values exceed the preset allowable limit value, an alarm signal is generated from the alarm means, the threshold value of the specific operating state amount is set. However, since it is not possible to capture the subtle and complex changes in the state quantity including the entire refrigeration cycle system, it is not possible to detect the possibility of an abnormality when a sign of failure appears. It was.

  In addition, when trying to improve accuracy, it takes too much data, and it is necessary to make judgments assuming various conditions, and it increases costs not only for sensors but also for microcomputer changes each time the target device changes and the capacity of the microcomputer increases. In addition, since the threshold for failure determination is determined by design values or specific machine tests, individual differences between actual machines cannot be taken into account, and the possibility of false detection is high.

  Even if a multivariate analysis method is used, it is not possible to determine the cause of the failure because the judgment on the threshold is insufficient or a large amount of data is required, and the cause of the failure cannot be specified. We were unable to respond quickly to maintenance.

  In addition, the conventional refrigeration cycle apparatus has a problem that a special sensor needs to be attached in order to measure the level of the liquid level in the liquid reservoir, resulting in a very expensive apparatus.

  Further, the conventional refrigeration cycle apparatus has a problem that it is difficult to install the existing refrigeration cycle apparatus because a special sensor is assembled to the apparatus.

  The present invention has been made in order to solve the above-described problems, and the object of the present invention is based on complex state quantity determination including equipment, for example, a compressor alone or the entire apparatus such as a refrigeration cycle. It is to obtain what makes it possible to detect early signs of failure. Another object of the present invention is to absorb a difference between actual machines in failure determination and to obtain a cheap and practical product that can be used anytime, anywhere. Another object of the present invention is to obtain a technology that can specify a cause of failure in failure determination and has high accuracy and high reliability.

  Another object of the present invention is to obtain an inexpensive and highly reliable apparatus or diagnosis and monitoring technology that can detect an abnormality of a single device or the entire refrigeration cycle using only information on general temperature measurement means and pressure measurement means. . Another object of the present invention is to obtain a failure diagnosis and monitoring technique that can be easily applied to existing devices or apparatuses.

  In addition, the present invention uses a correlation of a plurality of data to obtain not only a diagnosis and monitoring technology that can detect each abnormality and detect the abnormality early, but also a failure time prediction. That is the purpose.

  The device diagnostic apparatus of the present invention measures a plurality of waveform data continuously generated under the influence of operation in the operation state of the device, and the measured waveform data is an average value, a standard deviation, and the like related to each waveform. Calculates the state quantity at the time of equipment operation by combining the waveform data processing means that processes into multiple parameters for each waveform data and the multiple parameters obtained by the waveform data processing means as a combination of variables And calculating means for comparing whether or not the state quantity calculated by the calculating means is within a set range, and determining whether or not the device is in a normal operating state. Waveform data of different phenomena such as rotational torque and electric quantity, or waveform data of the same phenomenon at different positions related to the vicinity.

  The apparatus diagnostic apparatus according to the present invention includes a measuring unit that measures a measurement quantity such as a physical quantity of fluid sucked and discharged by the apparatus, a driving force that drives the apparatus, or the like, or an average value calculated from the measurement value. The second measuring means for measuring at least one type of waveform data from sound, vibration, rotational torque, electric quantity, etc. during measurement by the measuring means, and the waveform data measured from the second measuring means as the waveform Waveform data processing means for processing numerical data such as related average values, standard deviations, etc., and the waveform data processing means processes the measurement value or measurement amount obtained from the measurement means and the waveform data of the second measurement means. A calculation means for calculating a state quantity at the time of operation of the device by a calculation in which a plurality of obtained numerical data is combined as a plurality of variables and related to each other, and a state quantity calculated by the calculation means are set Device by comparing whether the range is one that was equipped with a determination unit that determines whether the normal operating condition.

  In the refrigeration cycle apparatus of the present invention, a compressor, a condenser, an expansion means, and an evaporator are connected by piping, a refrigerant is circulated therein to form a refrigeration cycle, and a flow from the discharge side of the compressor to the expansion means. State of at least one of the high pressure of the refrigerant at any position of the passage, the condensation temperature at the high pressure position, and the discharge temperature of the refrigerant at any position of the flow path from the discharge side of the compressor to the condenser The amount of refrigerant, the low pressure of the refrigerant at any position in the flow path from the expansion means to the suction side of the compressor, the evaporation temperature of the refrigerant at the low pressure position, and any one from the suction side of the compressor to the evaporator Refrigerant state quantity measuring means for measuring at least one of the two state quantities of the refrigerant suction temperature at the position, and either the compressor casing, the discharge side pipe or the suction side pipe Position vibration, compressor A pulsation state quantity measuring means for measuring at least one pulsation state of the sound pressure of air at any of the surrounding positions and an electric quantity for driving the compressor, a refrigerant state quantity measuring means, and a pulsation state quantity measurement means; A plurality of numerical values obtained from a plurality of state quantities to be measured as a plurality of variables, and a calculation means for calculating a set associated with each other and calculating a calculation result, and from the calculation result of each calculation means, This is to judge the driving state.

  The refrigeration cycle apparatus of the present invention includes any one of a refrigeration cycle in which a compressor, a condenser, an expansion means, and an evaporator are connected by piping and a refrigerant is circulated therein, and a flow path from the discharge side of the compressor to the expansion means. At least one state quantity among the high pressure of the refrigerant at the position, the condensation temperature of the refrigerant at the high pressure position, the discharge temperature of the refrigerant at any position of the flow path from the discharge side of the compressor to the condenser, And the low pressure of the refrigerant at any position in the flow path from the expansion means to the suction side of the compressor, the evaporation temperature at the low pressure, the refrigerant at any position in the flow path from the suction side of the compressor to the evaporator Refrigerant state quantity measuring means for measuring at least one of two state quantities, and a compressor casing, a compressor discharge side pipe, or a compressor suction side pipe Measure vibration at any position of First pulsating state quantity measuring means, second pulsating state quantity measuring means for measuring the sound pressure of air at any position around the compressor, and third pulsating state for measuring the current driving the compressor And at least two pulsating state quantity measuring means of the quantity measuring means, and a refrigerant state quantity obtained from the refrigerant state quantity measuring means and a state quantity obtained from at least two pulsating state quantity measuring means The operation state of the refrigeration cycle is determined by a calculation that is associated with each other.

  The device diagnosis method of the present invention measures a plurality of waveform data generated continuously under the influence of operation in the operation state of the device, and the measured waveform data is averaged, standard deviation, etc. related to each waveform. A waveform data processing step that sets a plurality of numerical values for each waveform data, a calculation step that calculates a calculation result by calculating an aggregate related to each other by combining a plurality of numerical values as a plurality of variables, and within a threshold value where the calculation result is set And a determination step for determining whether or not the device is in a normal operation state.

  This invention diagnoses whether the equipment is normal or abnormal from different types of waveform data such as noise and vibration of the equipment, and diagnoses the operating state of the refrigeration cycle device. Simple and reliable diagnosis, abnormality detection, Failure time prediction is possible. In addition, the present invention provides a diagnostic technique that is highly accurate, practical, and capable of absorbing individual machine differences and identifying the cause of failure. In the present invention, the refrigeration cycle is reliably monitored.

Embodiment 1 FIG.
The configuration of the first embodiment of the present invention will be described with reference to FIGS. FIG. 1 is an overall conceptual diagram of the present invention, in which 1 is a refrigeration cycle apparatus such as a refrigerator or an air conditioner, 2 is an operation state quantity of the refrigeration cycle apparatus 1, and a detection result is calculated, stored, and displayed. Alternatively, a board or microcomputer incorporating a device for transmitting / receiving data to / from the warning lamp and the like, 3 is a means for communicating with the outside such as a telephone line or a LAN line, and wireless, 4 is a remote of the refrigeration cycle apparatus 1 A remote monitoring room for centralized management such as monitoring and control, 5 is a computer having a display and calculation function for transmitting and receiving data to and from the refrigeration cycle apparatus 1 installed in the remote monitoring room 4, and 6 is a refrigeration cycle apparatus 1. A display device such as a liquid crystal display provided on the touch panel, 7 is an input device such as a touch panel or buttons, 8 is a warning lamp for notifying the occurrence of abnormality, and 9 is a different device. Is a speaker that generates a sound for notifying the occurrence. The refrigeration cycle apparatus 1 such as a refrigerator or an air conditioner is an air conditioner installed in a building, a refrigerator or an air conditioning system installed in a large store such as a supermarket, a refrigeration / air conditioner such as a small store, or an air conditioner in an apartment house. The remote monitoring room may monitor the plurality of facilities, or may monitor individual facilities. Or you may connect to the computer for monitoring from each house.

  FIG. 2 is a configuration diagram illustrating details of the refrigeration cycle apparatus 1 of FIG. 1. 11 is a compressor, 12 is a condenser, 13 is an expansion valve, 14 is an evaporator, 15 is a microphone that detects sound pressure in the vicinity of the compressor, or a detection element that directly contacts the compressor to detect vibrations of the compressor. A pulsating state quantity measuring means for measuring one or a plurality of waveforms such as 16, and 16 is a refrigerant state quantity measuring means for measuring a refrigerant state such as pressure and temperature of the refrigeration cycle 1 as numerical values such as an execution value, an average value, and a peak value. , 17 is a signal processing means for performing signal processing such as amplification and A / D conversion on the sound pressure signal or the like which is the measured continuous data, and 18 is the measurement data of the pulsation state quantity measuring means 15 and the refrigerant state quantity measuring means 16. Calculation means for performing various calculations based on the above, 19 is a storage means for storing past calculation results, reference values, etc., 20 is a comparison means for comparing the calculation results with the stored contents, and 21 is a determination for making a determination based on the comparison results. Means 22 This is an output means for outputting the determination result to the display device or the remote monitoring means 5. In addition, the condenser 12 and the evaporator 14 of FIG. 2 are provided with an air cooling fan. When the configuration of the present invention is a diagnostic device for equipment such as a compressor, the measurement means, calculation means, storage means, comparison means, etc. of the configuration of FIG. 2 are provided on a board attached to the compressor, and in some cases output By attaching an alarm device or the like that gives an alarm based on the received signal to the compressor, not only the device can be operated, but also a self-contained device that can self-diagnose and predict and predict the failure prediction that will be described later is obtained. For such a configuration, a simpler measuring device such as vibration and power supply current supplied to the compressor driving motor than the sound as the pulsation state quantity measuring means is more compactly collected. The sound pressure may be a flat characteristic that is not corrected for each frequency, or may be a pulsation state quantity signal that has been subjected to auditory correction such as the A characteristic or the C characteristic. The sound pressure may be measured with a microphone or a sound level meter.

  FIG. 3 is a detailed view of the pulsation state quantity detection means showing details of the pulsation state quantity measurement means 15 and the signal processing means 17, and shows the configuration of a system for detecting the sound pressure in the vicinity of the compressor 11. A filter and an amplifier, 33 is an A / D converter, and 34 is an FFT calculator.

  FIG. 4 shows a control block diagram of the compressor abnormality determination. In the figure, 41 is a state quantity A such as the pressure and temperature of each part of the refrigeration cycle, 42 is a sound pressure (state quantity B) around the compressor, and the composite variable calculation process in the calculation means 18 based on the state quantities A and B. I do. Then, storage means 19 storing past data and setting thresholds, comparison means 20 for comparing the stored data with the current value, determination means 21 for making a comprehensive determination based on the comparison result, and output for outputting the determination result The means 22 and the output determination result are displayed on the display means 23 or transmitted to the remote monitoring means 5 for monitoring the driving state at a remote place. In the description of FIGS. 1 and 2, a refrigerant circuit that circulates the refrigerant to perform air conditioning such as heating and cooling and refrigeration such as a refrigerator and a freezer warehouse, a sensor that detects the operating state of the refrigerant circuit, and a control of calculation, etc. Necessary microcomputers and substrates are housed in the refrigeration cycle apparatus, and the operation state is measured, calculated, compared, evaluated, and judged in this apparatus. However, in the vicinity of the refrigeration cycle, the sensors 15 and 16 may be provided up to the measurement 41 and 42, or the signal processing 17, 32, 33, and 34 may be provided, and the computation 18 and later may be provided in the remote monitoring room 4.

  Next, details of the compressor abnormality determination will be described. First, the operation of the refrigeration cycle apparatus will be described with reference to FIG. A refrigerant is sealed in the refrigerant circuit of the refrigeration cycle apparatus 1, the refrigerant is compressed and pressurized by the compressor 11, and the high-temperature and high-pressure refrigerant is cooled by a liquid cooling system such as an air cooling fan or water cooling (see FIG. The refrigerant is cooled and liquefied by an expansion valve 13 to become a low-temperature and low-pressure refrigerant and evaporated by the evaporator 14 by heat exchange with an air cooling fan or a liquid heat medium (not shown) such as water. It is heated and vaporized. The vaporized refrigerant returns to the suction side of the compressor 11 and moves again to the compression and pressurization step. At this time, the air or liquid exchanged with the refrigerant in the condenser 12 is heated at a high temperature and used as a heating heat source or exchanged with the outside air, and the air or liquid exchanged with the refrigerant in the evaporator 14 has a low temperature. It is cooled and used as a cooling or refrigeration / freezing heat source or exchanges heat with the outside air. Refrigerants used include natural refrigerants such as carbon dioxide, hydrocarbons, and helium, refrigerants that do not contain chlorine, such as alternative refrigerants such as HFC410A and HFC407c, or fluorocarbon refrigerants such as R22 and R134a that are used in existing products. The fluid equipment such as a compressor that circulates the refrigerant is of various types such as a reciprocating, a rotary, and a scroll. The abnormality determination of the present invention can be realized not only for new products but also for existing products that are already installed by adding a deficient sensor later.

  Next, the operation of detecting the pulsation state quantity will be described with reference to FIG. The pulsating sound pressure measured by the pulsating state quantity measuring means 15 such as a microphone or an acceleration element is converted into an electric signal, and only the sound pressure in the frequency range necessary for determination is extracted by a 32 low-pass filter or band-pass filter. The signal is amplified by an amplifier. Then, the A / D converter 33 converts the analog signal into a digital signal. Here, the conversion result signal is branched into two, one time-series sound pressure data is directly sent to the computing means 18, and the other is the FFT computing unit. After the time-series sound pressure data is converted into sound pressure data for each frequency by 34, the data is transmitted to the calculation means 18. Here, FFT is an abbreviation for Fast Fourier Transform, and is an analysis technique for converting a signal waveform from a time function to a frequency function at high speed. Here, only the sound pressure signal of noise will be described as pulsating flow detection. However, as will be described later, sound pressure signals may be picked up from a plurality of locations, or various pulsating flow detections such as vibration, voltage, and current may be performed. Alternatively, a plurality of different types of pulsating flows may be detected. The term “vibration” as used herein refers to not only the vibration in the audible region but also acoustic emission (AE), which is an elastic wave in the ultrasonic region that is generated in a pulsed manner as the material cracks and propagates. What was measured with the sensor etc. (PZT, lead zirconate titanate, etc.) may be used. With this sensor, the number of pulses, the magnitude of the pulse, and two or more sensors can be used to check for material distortion caused by cracks in the pressure relief valve in the compressor, rubbing between the rotor and stator of the rotating device, and sudden torque changes. The pulse generation position and the like can be measured by calculating from the difference between the pulse arrival times.

  The configuration of the signal processing means 17 to the output means 22 shown in FIG. 2 and FIG. 3 is a description of a system in which each set of means is built in the refrigeration cycle apparatus 1 as a substrate. A function from the means 18 to the output means 22 may be provided in the computer 5 provided in the remote monitoring room 4 of FIG. 3 may be a method in which the FFT calculation is performed by the computer 5 or the calculation means 18, and the function of each means may be arranged either in the refrigeration cycle apparatus 1 main body or in the remote monitoring room 4. What is necessary is just to satisfy the function. In addition, although it demonstrates as the computer 5 provided in the remote monitoring room 4, since this is convenient for centralized monitoring of several apparatuses, when it targets a specific apparatus, it is for movement like a mobile. Of course, it is possible to use a monitoring device so that the service person can constantly monitor while moving.

  Next, the abnormality determination operation of the compressor which is an example of the present invention will be described based on the control block diagram of FIG. 4 and FIG. The state quantity A, that is, the state quantity 41 of the refrigeration cycle is the driving force such as the pressure of each part of the refrigerant flowing through the refrigerant circuit, the temperature state quantity, or the compressor current value, power consumption, etc. necessary for grasping the operating state of the refrigeration cycle. The numerical value as the state quantity is measured by the refrigerant state quantity measuring means 16 in FIG. In order to grasp the operating state of the refrigeration cycle, in FIG. 2, the refrigerant pressure, that is, the high pressure, the high pressure saturation temperature, that is, the condensation at any position in the flow path from the discharge side of the compressor 11 to the expansion valve 13. Temperature, at least one state quantity of refrigerant temperature at any position from the discharge side of the compressor 11 to the condenser 12, that is, the discharge temperature, or a flow path from the expansion valve 13 to the suction side of the compressor 11 At least one of the refrigerant pressure, i.e., the low pressure, the low-pressure saturation temperature, i.e., the evaporation temperature, and the refrigerant temperature, i.e., the intake temperature, at any position in the flow path from the suction side of the compressor 11 to the evaporator 14. What is necessary is just to measure at least one of the state quantities. In the case of a chiller that uses cold / hot water as the heat source, if there is almost no change in the heat source temperature throughout the year, the influence is neglected or there is almost no change in the conditions on the high-pressure side or low-pressure side. Since it is handled as a fixed input value without performing measurement, it is sufficient to measure one state quantity. However, if it is difficult to treat as a fixed input, both state quantities may be measured. The pressure is measured using a pressure converter that converts the refrigerant pressure into an electrical signal, and the temperature is measured using temperature detection means such as a thermistor or a thermocouple. The pressure and temperature measurement positions are configured so that the refrigeration cycle operation status can be grasped more accurately by changing the position and adding measurement positions according to the configuration and operating characteristics of the target refrigeration cycle. Also good. The measurement of the state quantity A is performed at a certain fixed interval, for example, 1 minute interval, and the information is transmitted to the calculation means 17. When a simple abnormality determination is performed locally, the state quantity A may be input manually by reading from a pressure meter, a thermometer, or the like attached to the device. Also, in some devices and equipment where the values of state quantities such as high pressure and low pressure do not fluctuate, set the state quantities that do not fluctuate such as pressure and temperature as fixed values in advance or separate from the measured values. Manual input may be used, and these numerical values and measured values may be combined. That is, instead of the state quantity measured by the refrigerant state quantity measuring means, it is also possible to use a preset numerical value or a numerical value read or estimated from the operating state of the device. Measuring drive torque and current will be described later.

  The state quantity B, for example, the sound pressure 42 around the compressor, is information necessary for grasping the operating state of the compressor, and the state quantity is measured by the pulsating state quantity measuring means 15 in FIG. Here, the measurement position of the compressor peripheral sound pressure is in the vicinity of the position of either the casing of the compressor 11, the compressor discharge side pipe, or the compressor suction side pipe. The information amount B is pulsating time-series waveform data, and for example, data as shown in FIGS. 5 and 6 is obtained. 5 and 6 are explanatory diagrams showing a time-series waveform of sound pressure (upper diagram) and an amplitude spectrum (lower diagram) for each frequency after FFT processing. Detailed description of FIGS. 5 and 6 will be described later. Since the information amount B is waveform data, the sampling period is as short as 40 kHz, for example, and the information amount becomes enormous. For this reason, the waveform data is sent to the calculation means 18 and then processed by the calculation means 18 as a feature parameter representing the feature of the waveform, and is taken out as a representative numerical value for a certain period.

  The state quantity A and the state quantity B are measured in a mutually related state, and data measured in the same time zone or related time zones are used. The feature quantity parameters of the waveform data of the state quantity B are roughly classified into two types: feature quantity parameters based on time waveform data and feature quantity parameters based on frequency domain data. Hereinafter, each parameter will be described. Examples of the feature parameter based on the entire time waveform of the time waveform data include the following. Equation (1) is the average value, Equation (2) is the standard deviation, Equation (3) is the skewness, and Equation (4) is the kurtosis. In addition, although it demonstrates individually by the combination of state quantity A and B, you may measure several state quantity A and may summarize a correlation only with state quantity A.

  Here, X represents time series data, the suffix i represents the time series number of the measurement time waveform, and N represents the total number of waveform digital data. In the time domain analysis, all parameters from i = 1 to N are subjected to averaging calculation processing such as moving average processing for noise removal, and then each parameter is calculated. Further, the skewness represents asymmetry with respect to the time axis with respect to the average value of the waveform, and the kurtosis is a feature amount representing the impact level of the waveform. Subsequently, as the feature amount parameters based on the frequency domain data using the FFT calculation result, there are, for example, those having the intersection frequency and the extreme value frequency as described below.

  Here, fi is the frequency, P (fi) is the power spectrum, i is the FFT operation result spectrum number (low frequency side is the starting point), and N is the FFT frame size. The crossing frequency is a parameter that represents how many times the waveform has crossed the zero point per unit time, and the extreme value frequency is a parameter that represents how many times the peak (extreme value) of the waveform exists per unit time.

  As described above, the state quantity B is handled as the feature parameter of the sound pressure waveform data by the computing means 18, so that a huge amount of waveform data features, that is, feature quantities that are easy to handle in both normal and abnormal states, It is possible to perform calculation processing as an average value, a standard deviation, a skewness, a kurtosis, a crossing frequency, an extreme value frequency, and the like as in the equations 1) to (6). Further, parameters representing waveform features other than the above-described feature amount parameters can be added as necessary, and feature amounts that do not appear so much during normal operation and tend to appear when equipment is damaged or deteriorated.

  Note that the state quantity A and the state quantity B are processed in accordance with the measurement interval of the state quantity A in order to handle them as data in the same column, and the values of the state quantity A and the state quantity B are set at regular time intervals. The arithmetic processing is performed so that For example, if A is an interval of 1 minute, B is processed using all waveform data for 1 minute, or the amount of data is reduced to reduce calculation load, for example, only the last 10 seconds of a waveform of 1 minute are used. In addition, a value obtained from the same time zone or a related time zone is adopted so as to obtain a measurement amount related to the interval of the state quantity A.

  FIG. 5 shows an example of a sound pressure waveform in a normal state, and FIG. 6 shows an example of a sound pressure waveform in an abnormal state (refrigerant liquid back). The upper diagrams in FIGS. 5 and 6 mean time-series waveform data, where the horizontal axis indicates time in seconds and the vertical axis indicates the waveform change value of the electrical signal. The lower diagram shows time-series data converted into data for each frequency, with the horizontal axis representing frequency and the vertical axis representing amplitude spectrum. Conversion from time series data to frequency data is performed by an FFT computing unit. Time waveform data of a specific time, for example, several seconds is acquired, and the obtained time waveform is converted into frequency data. As a result, the time waveform data and frequency data in the same time zone can be compared. As can be seen from these comparisons, the characteristics of the waveform are clearly different such that the amplitude of the sound pressure time waveform increases and the value of the frequency peak increases when an abnormality occurs. The feature amount parameter described above replaces such a waveform feature with a numerical index. By comparing and analyzing these feature amount parameters, it is possible to determine a compressor abnormality. The state quantity A represents the operating state of the refrigeration cycle, and the state quantity A changes when an abnormality occurs. Therefore, by determining the state quantity A and the state quantity B in a comprehensive manner, a more accurate compressor abnormality determination can be performed. In the above description, the time waveform is FFT processed and converted to frequency data. However, other methods such as octave analysis and 1/3 octave analysis used in noise analysis may be used, and wavelet transform may be used. Such processing may be performed, and the feature amount of the waveform may be calculated using data obtained by performing computation or filtering on the time waveform.

  Next, a method for combining the state quantity A and the state quantity B into a composite variable and a method for detecting a compressor abnormality using the composite variable will be described. As an example of a method of processing a plurality of state quantities, there is a generally known Mahalanobis distance. , Mahalanobis distance, for example, is described in “Multivariate analysis that can be easily understood” issued by Tokyo Book Co., Ltd. on October 26, 1992, and is a technique used in the field of multivariate analysis. . Hereinafter, a method for detecting an abnormality of the compressor using the Mahalanobis distance will be described. Deterioration and failure are more complicated in the initial stage than in the final stage, such as breakage or insulation short-circuits, and the phenomenon that appears on the surface, data, and surface, especially in the initial stage. This is a combination of complex factors such as data, and the complex structure may be simplified by grasping them in a multiple rather than a unified way, and a technique called multivariate analysis is adopted. ing. However, simply using multivariate analysis cannot find the desired result, for example, an early stage defect. In the present invention, a practical diagnostic technique can be obtained from the correlation between variables.

  The total number of state quantities A representing the refrigeration cycle operating state and state quantities B (feature quantity parameters, etc.) obtained from the sound pressure waveform is m, and each state quantity is assigned to a variable X. Define the operating state quantity of. Next, reference data is collected for a total of n (two or more) combinations of the operating state quantities X1 to Xm in the reference normal operating state.

  And each average value mi and standard deviation (sigma) i (reference | standard variation degree) of X1-Xm are calculated | required by the following (7) Formula and (8) Formula. Note that i is the number of items (the number of parameters), and here, it is set to 1 to m and indicates a value corresponding to X1 to Xm. The standard deviation here is the square root of the expected value of the square of the difference between the variable and its average value.

  Next, normalization is performed in which the original X1 to Xm are converted into X1 to Xm by the following equation (9) using the average value mi and the standard deviation σi. That is, the variable is converted into a random variable having an average of 0 and a standard deviation of 1. In the following equation (9), j takes any value from 1 to n and corresponds to each of n measured values.

Next, in order to perform analysis using data with the variables normalized to mean 0 and variance 1, the correlation between X1 to Xm as the variance-covariance matrix, that is, the correlation matrix R indicating the relationship between the variables and the inverse matrix of the correlation matrix R ⁻¹ is defined by the following formula (10). In the following equation (10), k is the number of items (the number of parameters), and here it is m. Moreover, i and p show the value in each item, and take the value of 1-m here.

  After such calculation processing, the Mahalanobis distance is obtained based on the following equation (11). In the equation (11), j takes any value from 1 to n and corresponds to each of n measured values. Further, k is the number of items (number of parameters), which is m here. Further, a11 to akk are coefficients of the inverse matrix of the correlation matrix of the above equation (10), and the Mahalanobis distance is about 1 in the reference data, that is, in the normal operation state, and is within 4 or less. When an abnormality occurs, the numerical value increases, and the distance increases according to the degree of abnormality (degree of departure from normal). Although the dissimilarity required for cluster analysis, that is, Mahalanobis distance, is used here, multivariate analysis methods such as standardized Euclidean distance, Minkowski distance, and other shortest distance methods and longest distance methods may be used. .

  Here, the concept of Mahalanobis distance and the calculation flow will be described with reference to FIGS. FIG. 7 illustrates the relationship between Mahalanobis distance and its appearance rate. As shown in the figure, it is possible to determine the positional relationship of the calculated Mahalanobis distance with respect to the reference data group regardless of the number of parameters, and to confirm the failure state of the compressor. In the reference data group, the Mahalanobis distance has an average value of about 1 and is 4 or less even when variation is considered.

FIG. 8 is a flowchart for calculating the Mahalanobis distance. First, the average value of the reference data, the standard deviation, the inverse matrix of the correlation matrix, and the number of items are set (ST1), and the refrigeration cycle operation state quantity is acquired (ST2). Next, the acquired data is standardized based on the above equation (9) (ST3), and then the Mahalanobis distance is set to 0 as the initial value, and the counters i and j are set to the initial value 1 (ST4). ). Then, the counters i and j are changed until the number of items k is reached, and the calculation of the equation (11) is performed by repeating the calculation of ST5 to ST7 and dividing the integrated value obtained in ST8 by the number of items k. The Mahalanobis distance D ² can be determined.

  Next, a compressor abnormality detection method using Mahalanobis distance in an actual machine will be described with reference to a flowchart shown in FIG. FIG. 9 assumes operation immediately after installation of equipment such as a refrigeration cycle apparatus and a blower at the installation site. First, in ST21, the presence or absence of initial learning is confirmed. If initial learning is necessary, ST22 is performed. Thus, normal operation is learned as a reference space, and forced abnormal operation is performed at ST23 to learn as an abnormal space. Note that the step of ST23 is not necessarily required, and it can be substituted by grasping the characteristics at the time of abnormality by a factory test or simulation before shipping and correcting using the state of normal operation measured by an actual machine. The details of the normal operation reference space learning method and the abnormal space creation method will be described later. In ST24, the Mahalanobis distance between the reference space and each abnormal space is calculated to obtain a threshold value for each abnormality.

Next, an abnormality detection method after the end of initial learning after ST25 will be described. In ST25, the operation state quantity is measured, and in ST26, the operation state quantity data is normalized. In ST27, the square root of the Mahalanobis distance D ² for each of the reference space and each abnormal space is calculated by the following equation (12), and the occurrence probability of each abnormality is calculated by the equation (13) in ST28. Evaluate and estimate the cause of failure from the probability of occurrence of each abnormality. Here, the reason why the Mahalanobis distance D ² is raised to the 1/2 power in the equation (12) is that the distance D ² is a square value, and therefore the value increases quadratically as the distance increases. This is because by using the square root distance D, the distance linearly increases in accordance with the degree of abnormality, so that the increase in distance and the increase in degree of abnormality are proportional and easy to handle sensuously. In Expression (13), “Initial D” is the Mahalanobis distance when the abnormal space is applied to the initial normal state data. In the initial normal state, it represents the distance to normal based on the abnormality. “Current D” represents a distance when an anomalous space is applied to current measurement data. “Current D” takes a large value in the initial normal state (because the difference between the abnormal state and the normal state is large), but “Current D” becomes a small value as the degree of abnormality progresses (gradually from normal to abnormal) The probability of occurrence of anomaly approaches 100%.

  If the result of evaluation is normal in ST30, the process proceeds to normal reference space supplement processing ST31 (details will be described later). . Then, the service person who has been notified of the failure performs repairs such as failure repair and overhaul, and the equipment is restored to a normal state.

  Each process in the flowchart of FIG. 9 described above is performed by the calculation means 18, the storage means 19, the comparison means 20, the determination means 21, and the output means 22 of FIG. In ST21, whether or not the initial learning is present is determined by the determining means 21, ST22 and ST23 and ST24, and the learning related processing is calculated by the calculating means 18 and stored in the storage means 19. The calculation process of the Mahalanobis distance in ST25 to ST28 is performed based on the data of the reference space and the abnormal space stored in the storage unit 19 in the calculation unit 18, and the failure determination in ST29 to ST30 is performed by the comparison unit 20 and the determination. This is performed by the means 21, and the output of ST32 is performed by the output means 22.

In the above description, the learning operation of learning the reference space for the normal state of ST21, ST22, ST23, ST24 or the abnormal space for each abnormal state is performed by measuring the reference value necessary for calculating the Mahalanobis distance. Represents the operation of calculating and storing as a reference value, specifically, the average value m of equation (7), the standard deviation σ of equation (8), and the inverse matrix of the correlation matrix of equation (9) Indicates that R ⁻¹ is calculated.

  Each abnormal space stores an average value and standard deviation of each parameter and a correlation coefficient of each parameter. Using the average value of each parameter of each abnormal space, the distance between the reference space and each abnormal space can be obtained by calculating the Mahalanobis distance from the normal reference space. In ST24, this is set as a threshold value. ing.

  ST25 to ST29 are portions representing an operation of performing data measurement in actual machine operation and determining whether or not there is a failure. The distance (square root of Mahalanobis distance) between each abnormal space obtained in ST24 and the normal reference space is set as initial D1 and initial D2. Then, the distance D0 between the measured current driving state quantity data and the normal reference space, and the distances D1 and D2 between the abnormal spaces are obtained. Note that D0 takes a value of 2 or less in the initial state. Then, the degree of approach to each abnormal space is calculated from equation (13), and the occurrence probability of each abnormality is obtained. Then, in ST30, the abnormality occurrence probabilities are compared to determine the cause of the failure. In the above description, the case where the number of target abnormal spaces is two has been described. However, a plurality of number of abnormal spaces can be prepared in accordance with the characteristics of the target refrigeration cycle apparatus.

  As described above, by defining the normal reference space and the anomalous space and determining the probability of occurrence for each anomaly, grasp the degree of abnormality by increasing the distance to the normal reference space (Mahalanobis distance or square root of Mahalanobis distance) It is possible to identify the cause of the abnormality by reducing the distance (Mahalanobis distance or the square root of the Mahalanobis distance) to each anomalous space.

  The concept of the Mahalanobis distance between the abnormal space and the normal space is shown in FIG. In FIG. 10, the normal reference space is an image diagram in which each abnormal space exists at a position away from the origin at the coordinate center. Since the Mahalanobis distance is actually a multidimensional space, FIG. 10 is an image diagram representing this two-dimensionally. The normal reference space and the abnormal space each have a region with variations, and it is possible to determine whether the current operating state is normal or abnormal by determining which space it belongs to It becomes possible. The distance between each abnormal space and the normal space can be calculated by obtaining the Mahalanobis distance between the normal reference space and the representative data (average value data) of the abnormal space. For example, if this distance is 1000, the current refrigeration cycle operating state quantity is calculated using the normal reference space, and the distance is 1000. If the distance from the abnormal space is close to zero, this may be abnormal. Is expensive. As for the threshold for each abnormality, the distance between the normal reference space in each abnormality and the Mahalanobis between each abnormality space is calculated as described above. For example, if it is desired to detect the abnormality early, 1/10 is set as the threshold for the abnormality. The threshold is set in advance as described above. In addition, if only the abnormal space is learned and the initial state is normal, a method of determining abnormality from the degree of approach to the abnormality, or a method of learning only the normal space and determining the degree of abnormality from the degree of separation from the normal space, etc. It is possible to discriminate between normal and abnormal and the degree of each. That is, it is not always necessary to learn both normal and abnormal spaces.

  In addition, in the failure simulation test at the installation site, the test cannot be performed in extremely bad operating conditions that lead to compressor failure, so the failure condition is divided into several levels and the abnormal space is learned according to each level. It may be. In FIG. 10, an abnormal space 1 represents an example. In this example, the abnormal space 1 is divided into an abnormal level 1 to an abnormal level 3 according to the degree of abnormality, and the level 1 and level 2 abnormal spaces are learned in the installation site test. Do. Level 3 is a level that actually causes the compressor to break, and is an abnormal space in which measurement is performed in advance in the test room and learning is performed.

  In this way, by classifying abnormalities according to the degree of abnormality, it is possible to create an actual space in combination with the actual machine for an area with a low degree of abnormality that can be simulated by the actual machine. It is possible to detect early abnormalities in line with.

  In addition, by classifying the level of abnormality and creating an abnormal space for each abnormal level, accurate failure prediction is possible even when the abnormal level is low, and it is easy to distinguish from other abnormalities, It is possible to predict failure and identify the cause of failure at an early stage before an abnormality occurs and the refrigeration cycle apparatus breaks down.

  Next, the contents of reference space learning will be described. The reference space learning is a normal reference space replenishment process which is ST31 in FIG. 9, and FIG. 11 shows a flowchart of the normal reference space replenishment process. The present invention deals with continuous waveform data which is a pulsation state quantity. In this case, data for a certain period is taken and the feature amount within that period is calculated. Therefore, when the control for changing the state of the expansion valve or load of the refrigeration cycle is performed, the operating range naturally changes and the measurement conditions differ. In normal operation, operation is performed over various operation ranges as shown in the table shown in ST41 depending on the high-pressure and low-pressure operation ranges of the refrigeration cycle. For this reason, in order to increase the determination accuracy of failure diagnosis, it is necessary to create normal reference spaces according to the high and low pressure operation states of the refrigeration cycle. ST41 is a table of high and low pressure operating ranges (Pd1 indicates a pressure range such as 1 MPa to 1.2 MPa, Ps1 indicates a pressure range of 0.1 MPa to 0.2 MPa, for example). The unlearned range is indicated by a mark. In light of this table, it is determined in ST41 and ST42 whether the high and low pressure range in the current operating state has already been learned. If the learning has not been completed, the operation state quantity is stored in the storage means in ST43, and when the number of times of operation in the same high and low pressure range reaches N times (N is 100 times, for example), ST45. A new reference space is created using the data stored in the past at ST46, the new reference space created at ST46 is stored, and the Mahalanobis distance between the reference space and each abnormal space is calculated at ST47 to calculate each abnormality. Find the threshold for. Note that the initial learning performed in the initial stage of operation corresponds to the normal reference space creation process from the state where the past normal learning has never been performed in the flowchart of FIG.

  As described above, by creating a reference space according to the high-pressure and low-pressure operating ranges of the actual refrigeration cycle, it becomes possible to create a normal reference space corresponding to various operating conditions and complete absorption of individual device differences. .

  Next, anomalous space learning will be described. There are two types of abnormal space: a method of learning with an actual machine after installation of the equipment at the installation site and a method of creating an abnormal space using data obtained by simulating the failure state of the same model in the test room in advance. There is. As for the former, a failure state that can simulate a failure state at an installation site is targeted, for example, a refrigerant liquid bag, a refrigerator oil depletion, and the like. For these failures, simulate the refrigerant liquid back state by opening the expansion valve of the refrigeration cycle, or temporarily evacuating oil from the bottom of the compressor. Create an anomalous space from operating conditions. The created abnormal space is stored in the storage means, and is used to determine an abnormal state.

  The latter method of performing a failure simulation test in advance in the test room is intended for failures that are difficult to simulate at the installation site, such as bearing abnormalities, motor abnormalities, and tooth contact. For these failures, a compressor that can simulate an abnormal condition is created, a refrigeration cycle device equipped with this compressor is tested in a test room, and abnormal operation state data is collected. Create an anomalous space. The abnormal space prepared in advance is stored in the storage unit in advance when the refrigeration cycle apparatus is shipped, so that it can be applied to an actual machine. Also, a part of the fault simulation test can be substituted by simulation.

  In addition, as another learning method for anomalous space, when parameters that show signs when a target failure occurs are clear in advance, after normal reference space learning, the data of each parameter used for the normal reference space On the other hand, a method is conceivable in which only the value of a parameter in which a sign appears remarkably when an abnormality occurs is forcibly changed to a value estimated when a failure occurs, and abnormal operation state quantity data is newly created. The number of parameters to be forcibly changed may be determined according to the phenomenon appearing at the time of abnormality, and it may be only one or plural. For example, when only the variation in the waveform increases, the standard deviation of the feature parameter created based on the waveform may be increased. Then, a part of values in the normal reference space is replaced with the newly created abnormal operation state quantity data to define the abnormal space. When this abnormal space learning method is used, it is necessary to create an abnormal space corresponding to each normal reference space. Therefore, every time the normal reference space supplement processing of FIG. 11 is performed to learn a new normal reference space. In addition, the abnormal space is learned by the above method.

  This makes it possible to create an anomaly space based on the normal state of the actual machine, and to completely absorb individual differences due to variations in the actual machine, if the parameters that show signs in the event of an abnormality are clear in advance. It becomes possible to do.

  When the operation of the refrigeration cycle apparatus is continued, an unexpected failure that cannot be covered in the initially predicted abnormal space may occur. As a countermeasure for such a case, there is a new abnormality learning function, and its concept is shown in the flowchart of FIG. In the figure, ST51 is the detection of the occurrence of an abnormality, and the cause of the failure cannot be specified in the failure cause evaluation determination flow, but the Mahalanobis distance is increased and it can be determined that an abnormality has occurred in the refrigeration cycle apparatus. In such a state, first, a time zone in which a corresponding abnormality has occurred is selected from the past time zones displayed on the display means 6 in FIG. 1 by an operation using the input device 7 in FIG. Note that the data for the past several days is always stored in the storage means, and in ST52, an arbitrary location is selected from this data. In ST53, the abnormal space is learned using the operation data (abnormal data) in the selected time zone. In ST54, the learned abnormal space is stored in the storage means as a new abnormal space. In the failure cause evaluation after the new abnormal space is stored, the new abnormal space is also determined.

  In addition, although the said description demonstrated learning operation in the operating device of the input means of an actual refrigeration cycle apparatus, the same learning operation by information terminals, such as a remote personal computer, in a remote monitoring means is also possible. Alternatively, the input means does not need to be permanently installed in the refrigeration cycle apparatus, and in the event of an abnormality, a maintenance tool can be installed so that a service person can download data from the refrigeration cycle apparatus, analyze it, and write information to the refrigeration cycle apparatus. You may be allowed to go to maintenance with a personal computer.

  As described above, by providing an additional learning function even for a new failure, it is possible to accurately deal with failure determination by post-processing even for a failure that cannot be predicted at the beginning of design. In addition, the learned information on the new anomalous space is stored in the device diagnostic device and the remote monitoring means. By using such information, it can be added to the storage means of the same model to be shipped or a similar different model. It is also possible to deploy to many models.

  FIG. 13 is a graph in which time is plotted on the horizontal axis and D value (square root of Mahalanobis distance) is plotted on the vertical axis, and shows the transition of the D value over time from a normal state when a certain abnormality occurs. It is. The D value is a value of 2 or less in a normal state, and the D value gradually changes to a larger value with the transition of time with respect to a certain abnormality as shown in the figure. Therefore, it is possible to estimate the time until failure from the relationship between the increasing tendency of the D value and the failure threshold, and it is possible to prevent the apparatus from being abnormally stopped by performing accurate maintenance before the estimated failure time. It becomes possible to prevent. For example, if it takes one month from the initial normal state until the D value reaches half of the threshold value, it can be predicted that it will take another month before the D value reaches the threshold value and falls into a failure state. Further, when the D value change method is not proportional, for example, when the increase rate of the D value in the last week has increased, the failure time is predicted using the change rate of the D value in the last week. This makes it possible to predict failure more accurately.

  In the above description, the Mahalanobis distance is used as the abnormality determination means, and the method of performing abnormality determination by converting a multi-item parameter into one index has been described. If possible, pay attention to specific items such as standard deviation and skewness, and determine whether an abnormality is detected based on whether or not this item exceeds a threshold. A method of performing threshold determination by paying attention to harmonic components may be used. When the above method is applied to determine the abnormality of a compressor installed in a refrigeration cycle device such as an air conditioner / refrigerator, the compressor bearing section or the like may be used depending on the compression ratio of the compressor (ratio of high pressure to low pressure). Since such torque conditions change greatly, the threshold value must be changed accordingly. As this method, learning of the normal operation state is effective, and the content thereof is almost the same as the flowchart of the normal reference space supplement processing described with reference to FIG. The content will be described below with reference to FIG. 11, taking as an example the case where attention is paid to the harmonic component of the specific frequency. In ST41 and ST42, it is determined whether or not the high / low pressure range of the current operating state has already been learned. If the learning has not been completed, the operating state is stored in the storage means in ST43. When the number of operations in the same high and low pressure range reaches N (N is 100, for example) or more, the data stored in the past in ST45 is used to show a remarkable abnormality in the normal state. The average value of the sound pressure in the wave component, for example, the Z-order harmonic of the bearing frequency (Z-order harmonic = 50 * Z (Hz) if the bearing frequency is 50 Hz) is stored, and this is used as the reference value in the normal state. To do. Next, in ST46, this reference value is stored as a function relating to high pressure and low pressure. The threshold is set, for example, α times the reference state, and the threshold is calculated in ST47. As described above, by setting reference values according to the high and low pressure operating ranges of the actual refrigeration cycle and thresholds based on these, it is possible to perform complete absorption of individual device differences and threshold determination corresponding to various operating states. It becomes possible. In the above example, the number of harmonic component items has been described as one, but a plurality of items may be used depending on the target abnormal state. Moreover, although the reference value is the sound pressure value at the specific harmonic, processing such as non-dimensionalization as a ratio to the sound pressure overall value may be added.

  In the above description, the state quantity B has been described on the assumption of sound pressure. However, the same effect can be obtained by using a vibration or current waveform instead. In the case of vibration, a vibration measuring element such as a contact-type or non-contact-type acceleration measuring means or a displacement measuring means for measuring the vibration at any position of the compressor casing or the discharge side piping or suction side piping of the compressor In the case of current, the current waveform may be measured by using a current measuring means such as a coil provided on the power supply line to the compressor. In the case where the current is difficult to measure from the power supply of the compressor, the same effect can be obtained by measuring the current from the power line of the refrigeration cycle apparatus. In addition, since a magnetic flux is generated around the current line, a non-contact type current sensor that can generate an eddy current from the magnetic flux and measure the current may be provided near the current line. The above description has explained the technique of diagnosing with the combination of the state quantity A and the state quantity B. As for the state quantity A, a measured quantity such as a current that is unrelated to the instantaneous value is obtained by measuring a physical quantity or an execution value related to a refrigerant having a large change time delay. On the other hand, the state quantity B is waveform data that captures a relatively short time change and captures data as a tendency for a certain period of time. By combining many variables obtained from both, mechanical, electrical, Diagnosis of failure as a whole, including effects from other sources not related to accidents. In particular, the compressor used in the refrigeration cycle discharges, sucks, and circulates the refrigerant flowing through the refrigeration cycle, and it is effective for practical diagnosis to use variables including the physical quantity of the refrigerant. The same applies to fluid machinery such as blowers related to physical quantities of wind flow and pumps related to water, food, and chemical liquids, and devices that move objects such as fax machines, printers, and production lines. It can also be used with other drive devices. In particular, in the case of a blower used in a refrigeration cycle, it is clear from the fact that the performance and characteristics of the refrigeration cycle change that the physical quantity of the refrigerant other than the flow of wind as the fluid may be measured as described above. As described above, the cause of the abnormality can be accurately grasped by picking up and using data characteristic of various abnormal spaces shown in FIG.

  Next, these waveform data are convenient for grasping characteristics if they are a plurality of mutually related waveform data that are continuously generated under the influence of operation in the operation state of the equipment. For example, noise and vibration are It is easy to generate feature data that are related to each other at the time of failure such as contact that leads to mechanical damage, and the feature clearly appears in both normal and abnormal situations. In addition, instantaneous values such as noise / vibration and rotational torque, electric power, and current are closely related to rotating motors such as steel production lines, fans, compressors, etc., and their drive motors. Is good. Of course, other electric quantities such as electromagnetic force generated between the stator and the rotor of the motor also cause instantaneous speed fluctuations. Furthermore, for motors, waveform data of different phenomena such as ground current, noise radio waves leaking to the surroundings, and shaft voltage are not only electrically related to each other, but also to distinguish them from accidents such as mechanical systems. A combination that is effective for obtaining a characteristic change such as a combination of vibration and leakage current is also acceptable. Or, even if the sound is measured, the same phenomenon at different positions, such as near each other, is similar to different data in the mechanical contact situation such as the circumference of the compressor bearing. It is convenient to collect the data from the data and make it a variable as characteristic data based on this data, because the characteristics clearly appear in both normal and abnormal conditions.

  When driving data such as current or waveform data of driving torque is used as waveform data, it will include many effects such as compressor and blower rotation fluctuations, inverter control fluctuations, etc. Failure alone cannot be determined by pressure alone or vibration alone. For example, in the case of a combined device even if a spectrum of harmonics is analyzed over a wide spectrum, it is necessary to separate other influences, and if this is examined one by one, a very practical diagnostic device cannot be obtained. The present invention can easily solve such problems. When an induction motor and a DC brushless motor are used for the motor, the way of generating the harmonics is different, and it is necessary to sort the data by learning the operation range. In addition, when there are a plurality of compression chambers of the compressor, there is torque pulsation equivalent to the compression chamber during one rotation, so it is necessary to exclude this amount. Since the torque pulsation due to the pulsation of the refrigerant is quite large, it is necessary to separate it from the effects of the failure of the compressor and other equipment, and the torque varies greatly depending on the ratio of high pressure to low pressure, which is the compression ratio of the compressor. It is not possible to determine the failure by using only the harmonic components. It is necessary to measure the high and low pressures of the refrigerant with the refrigerant state quantity measuring means that is measuring the torque and current waveform, and to check the normal and abnormal data in the operation range divided by the compression ratio. For several tens of minutes after starting the compressor, the high and low pressures of the refrigeration cycle change without stability. The signal due to the torque of the compressor or a failure signal such as a tooth contact affected by the torque changes during that time. By using this property, it is possible to distinguish between signals that are affected by torque and signals that are not affected, such as failure of an electric system such as a capacitor. Control of equipment on the load side of the refrigeration cycle, for example, in the case of a showcase that is a refrigeration system, there is one heat exchanger on the heat source side, that is, a condenser, and there are multiple load side heat exchangers, that is, evaporators. Since the temperature is adjusted by adjusting the flow rate of the refrigerant to the evaporator, even if the frequency of the compressor does not change due to opening and closing of this adjusting solenoid valve, the state quantity of the high-pressure and low-pressure refrigeration cycle changes. As a result, the torque of the compressor varies. In order to take this fluctuation into account, the reference state is memorized in relation to the torque and compression ratio, etc., and the operation range is meticulously studied, the average over a certain period of time is taken, and the number of waveform averaging is increased It is necessary to increase accuracy. Considering the failure diagnosis, it is necessary to judge from the state quantity of the refrigerant so that the special state of the refrigeration cycle is not included in the failure. For example, the pressure of the refrigerant in the compressor is different between the liquid back operation that should be viewed as a kind of abnormality and the operation that is superheated, that is, the operation that should be registered as the operation range in the normal state. In other words, the compressor is distinguished from normal and abnormal in the operating range in which the superheat is on.

  Next, the notification operation after the compressor abnormality is detected will be described. When an abnormality or a sign of the compressor is detected, there are two cases: the case where the abnormality is notified at the installation site and the case where the information is notified in the remote monitoring room. When notifying the abnormality at the installation site, either a method of notifying the abnormality with the warning lamp 8 or the speaker 9 in FIG. 1 and a method of displaying the abnormality content on the display device 6 such as a liquid crystal display or both of them are used. Is possible. When the abnormal situation is urgent and serious, the combined use of the warning lamp 8, the speaker 9 and the display device 6 is effective. In the stage where the abnormality is small or the prediction stage, only the display device 6 is used for notification, and a serviceman is used during maintenance. If it is configured so that the abnormal tendency can be confirmed, it is possible to grasp an appropriate maintenance time. The display device 6 displays information related to compressor abnormality detection. As this information, for example, installation location name, address, contact information, equipment name, compressor model name, specifications, etc., device specific items, compressor normal / abnormal judgment result character display, abnormality degree numerical display and history display by graph, Mahalanobis distance time transition graph and numerical data history, abnormality type judgment result text display, learning status such as Mahalanobis distance learning presence, equipment operating state such as pressure temperature of each part, measurement waveform graph such as sound vibration and FFT calculation It is possible to display a history image of a measured amount, a calculation result, a determination result, and an operation state such as a result graph, a feature parameter calculation result and a history. These displays may be in a plurality of screen formats that can be arbitrarily selected as necessary.

As for notification to the remote monitoring room, the contents and degree of abnormality are notified to the remote monitoring room by communication means such as a telephone line, LAN, and radio. In the remote monitoring room, a service person is dispatched according to the state of the abnormality, but if the cause of the abnormality can be grasped remotely, the necessary parts can be prepared to deal with the abnormality before going to the site, Rapid maintenance can be performed. In addition, it is also possible to notify the information directly to the information receiving means such as the mobile phone of the service person at the same time as informing the remote monitoring room. As explained above, the composite variable is calculated from the pressure and temperature of the refrigerant of the refrigeration cycle apparatus and the state quantity of sound, vibration or current, etc., and comparison is made between the normal space and the abnormal space where learning is performed. By determining the abnormality, it is possible to detect an early sign of failure, absorb individual machine differences, and identify the cause of failure.

  A configuration for detecting a plurality of state quantities B according to the present invention will be described with reference to FIG. FIG. 14 is basically the same as the refrigeration cycle apparatus of the first embodiment shown in FIG. 2 except that there are both means for detecting sound pressure and vibration near the compressor. In the configuration of FIG. 14, the refrigerant circulates through the compressor 11, the condenser 12, the expansion valve 13, and the evaporator 14, and 16 is a refrigerant state quantity detection unit that detects the refrigerant state such as the pressure and temperature of the refrigeration cycle 1. , 51 is a first pulsation state quantity detecting means for detecting vibration at any position of the compressor casing, the compressor discharge side pipe or the compressor suction side pipe, and 52 is the first pulsation state quantity detection. Installed in the vicinity of the means 51, for example, at a position about 1 to 3 cm away from the first pulsation state quantity detection means 51, and detects the sound pressure near the vibration measurement position detected by the first pulsation state quantity detection means 51. Second pulsation state quantity detection means. Reference numeral 17 denotes signal processing means for performing signal processing such as amplification and A / D conversion on the output signals of the first pulsation state quantity detection means 51 and the second pulsation state quantity detection means 52, and 18 denotes refrigerant state quantity detection means 16. And a calculation means for performing various calculations based on the detection results of the first pulsation state quantity detection means 51 and the second pulsation state quantity detection means 52, 19 a storage means for storing past calculation results, reference values and the like, 20 Is a comparison means for comparing the operation result with the stored content, 21 is a determination means for making a determination based on the comparison result, and 22 is an output means for outputting the determination result. The overall conceptual diagram and other symbols in this example are the same as those in FIGS.

  FIG. 15 is a diagram showing details of the first pulsation state quantity detection means 51 and the second pulsation state quantity detection means 52, and shows a system configuration for detecting vibration and sound pressure in the vicinity of the compressor 11. is there. The first pulsation state quantity detection means 51 is a means for converting sound pressure such as a microphone into an electric signal, and the second pulsation state quantity detection means 52 is an electric signal for a vibration state quantity such as an acceleration pickup or a displacement sensor. , 32 is a filter and amplifier, 33 is an A / D converter, and 34 is an FFT calculator. In this example, sound pressure and vibration are detected at the same time, but the detection method is the same as that in FIG. 3, and this example is different in that it corresponds to 2ch.

  Next, the signal processing operation in the computing means 18 after simultaneous detection of sound pressure and vibration, which is a feature of this example, will be described. Using the two time-series waveform signals of sound pressure and vibration obtained from the first pulsation state quantity detection means 51 and the second pulsation state quantity detection means 52, a correlation function is obtained by Expression 14.

  Here, x (t) and y (t) are time-series signal waveforms related to time t, τ is a time shift, and T is a time interval. When the compressor becomes abnormal, vibration is generated and sound is generated, so the sound pressure waveform and vibration waveform should basically be the same waveform signal, but the surrounding background noise is superimposed on the sound pressure waveform, Since the vibration at the time of starting the compressor unrelated to the vibration due to the abnormality of the compressor appears in the vibration waveform, the change in the waveform due to the abnormality and the noise may not be distinguished from only one of the signals. Therefore, by obtaining a correlation function of two time-series waveforms of sound pressure and vibration, only the common component remains in the two signals, the noise component is removed, and a more accurate pulsation state quantity can be extracted. By using this correlation function to obtain the feature parameter based on the time waveform data already described, it is possible to perform an analysis that eliminates the influence of noise. Note that the noise removal method using the correlation function can achieve the same effect not only with a sound pressure waveform and a vibration waveform but also with a plurality of combinations of other waveform data such as a sound pressure waveform and a current waveform, or a vibration and a current waveform. Naturally, the current waveform measurement may be performed not only at the motor terminal but also at the middle of the connected power line or at the power terminal.

  The refrigerant state quantity calculation processing method, learning method, and the like are the same as those described above, and in this example, the abnormality determination processing is the same as the content described using FIGS. 4 to 13 as described above. This is possible by performing the determination process.

  In this example, the correlation function is obtained from the sound pressure waveform and the vibration waveform and the cross spectrum processing is performed as an example. However, the cross result can also be obtained by multiplying the FFT result of the sound pressure waveform and the FFT result of the vibration waveform. A spectrum can be obtained. Then, by performing inverse FFT processing on the cross spectrum calculation result, it is possible to return to the time domain waveform, and using the cross spectrum calculation result and the time domain waveform calculation result by the inverse FFT, the same processing such as the feature amount parameter is performed. Even if it does, it will become possible to analyze the influence of noise similarly. The cross spectrum processing is not limited to the sound pressure waveform and the vibration waveform, but may be any combination of different pulsation waveforms measured in the same time zone such as the sound pressure waveform and the current waveform, and the current waveform and the vibration waveform.

  In this example, the correlation function is obtained from the sound pressure waveform and the vibration waveform. However, two microphones are used and one is installed at a position where the sound pressure change due to the compressor abnormality is large. The other may be configured to be installed at a position where the change in sound pressure due to the compressor abnormality is small. Then, by taking the difference between the measured waveforms of the two microphones, the background noise that appears in the same manner in both waveforms can be eliminated, and a waveform in which only the influence of the abnormality remains can be obtained. Even if comprised in this way, the same abnormality determination process is possible and there exists the same effect.

  At this time, the number of microphones is not limited to two, and a larger number may be used. FIG. 16 shows an installation example in which a multipoint microphone is applied around the compressor. Since the motor and the compression rotor are rotating inside the compressor, if the bearing is damaged, an abnormal sound corresponding to the rotation angle is generated. Therefore, as shown in FIG. 16, when the microphone 53 is placed so as to surround the compressor, the correlation between the abnormal phenomenon and the plurality of microphones is obtained, and abnormality determination is performed based on the correlation, one Abnormalities that could not be grasped by the microphone can be surely grasped.

  In the description of the present invention, in order to capture subtle changes in the refrigeration cycle apparatus, in addition to the sound pressure or vibration information, the pressure and temperature information of the refrigeration cycle apparatus is also used in combination to diagnose the compressor failure. However, without using the pressure and temperature information of the refrigeration cycle apparatus, the failure diagnosis of the compressor can be performed by applying the signal processing method described above also based on the sound pressure and vibration near the compressor, or only the sound pressure or only the vibration information. Is possible.

  As described above, the composite variable is calculated from the pressure and temperature of the refrigeration cycle apparatus and the state quantities of vibration and the sound pressure in the near field. By determining the abnormality, it is possible to remove a noise component included in vibration or sound pressure, detect an early sign of failure with high accuracy, absorb individual machine differences, and identify the cause of the failure. In the refrigeration cycle apparatus and compressor according to the present invention, a refrigerant state quantity detection means comprising the pressure and temperature of the refrigeration cycle apparatus, and a pulsation state quantity for detecting at least one state quantity of sound or vibration around the compressor or current And a detecting means for calculating a composite variable from these state quantities and determining an abnormality of the compressor using the composite variable.

  Further, the refrigeration cycle apparatus of the present invention is a refrigeration cycle apparatus in which a compressor, a condenser, an expansion means, and an evaporator are connected by piping and a refrigerant is circulated therein to constitute a refrigeration cycle. Refrigerant pressure or high pressure at any position in the flow path to the expansion means, high pressure saturation temperature or condensation temperature, refrigerant temperature or discharge temperature at any position in the flow path from the discharge side of the compressor to the condenser And at least one state quantity, and the pressure or low pressure of the refrigerant at any position in the flow path from the expansion means to the suction side of the compressor, the low pressure saturation temperature or evaporation temperature, and the suction side of the compressor to the evaporator A refrigerant state quantity detecting means for detecting at least one state quantity from the refrigerant temperature at any position in the flow path, that is, a suction temperature, and a compressor casing or a compressor discharge side pipe Or a pulsation that detects at least one state quantity from vibration at any position on the suction side of the compressor, sound pressure of air at any position around the compressor, and current flowing through the power line of the compressor A state quantity detection means, a calculation means for calculating two or more state quantities detected by the refrigerant state quantity detection means and the pulsation state quantity detection means, and obtaining a composite variable, and a detection value by each state quantity detection means or A storage means for storing a calculated value such as a composite variable calculated therefrom and a determination means for determining a compressor abnormality from the composite variable obtained by the calculating means are provided.

  Further, the refrigeration cycle apparatus of the present invention is a refrigeration cycle apparatus in which a compressor, a condenser, an expansion means, and an evaporator are connected by piping and a refrigerant is circulated therein to constitute a refrigeration cycle. Refrigerant pressure or high pressure at any position in the flow path to the expansion means, high pressure saturation temperature or condensation temperature, refrigerant temperature or discharge temperature at any position in the flow path from the discharge side of the compressor to the condenser And at least one state quantity, and the pressure or low pressure of the refrigerant at any position in the flow path from the expansion means to the suction side of the compressor, the low pressure saturation temperature or evaporation temperature, and the suction side of the compressor to the evaporator A refrigerant state quantity detecting means for detecting at least one state quantity from the refrigerant temperature at any position in the flow path, that is, a suction temperature, and a compressor casing or a compressor discharge side pipe Or a first pulsation state quantity detection means for detecting vibration at any position of the pipe on the suction side of the compressor, and a second pulsation for detecting sound pressure of air at any position around the compressor. It is possible to ensure a highly accurate state by providing the state quantity detection means and installing the second pulsation state quantity detection means in the vicinity of the first pulsation state quantity detection means. Moreover, you may provide the 3rd pulsating flow measurement means which measures the electric current which drives a compressor. In that case, the first, second, and third pulsating flow measuring means may be combined in any way, and the accuracy can be improved by calculating using all of them.

  Further, the refrigeration cycle apparatus of the present invention calculates a relational variable from the first pulsation state quantity detection means and the second pulsation state quantity detection means, and is detected by the refrigerant state quantity detection means and the pulsation state quantity detection means 2 An arithmetic means for calculating a composite variable by calculating two or more state quantities or related variables, a storage means for storing a detection value by each state quantity detection means or an arithmetic value such as a composite variable calculated therefrom, and an arithmetic means Since it has a judging means for judging a compressor abnormality from the obtained composite variable, the abnormality detection is simple and reliable.

  The refrigeration cycle apparatus of the present invention is a refrigeration cycle apparatus in which a compressor, a condenser, an expansion means, and an evaporator are connected by piping and a refrigerant is circulated therein to constitute a refrigeration cycle. First pulsation state quantity detection means for detecting vibration at any position of the discharge pipe or compressor suction pipe, and the sound pressure of air at any position around the compressor Since the second pulsation state quantity detection means is provided and the second pulsation state quantity detection means is installed in the vicinity of the first pulsation state quantity detection means, a highly reliable device can be obtained.

  Further, the refrigeration cycle apparatus of the present invention calculates a relational variable from the first pulsation state quantity detection means and the second pulsation state quantity detection means, and two or more state quantities detected by the pulsation state quantity detection means or Computation means for computing a composite variable by computing the relational variable, storage means for storing a computation value such as a detection value by the state quantity detection means or a composite variable computed from them, and compression from the composite variable obtained by the computation means And a determination means for determining a machine abnormality.

  Further, the refrigeration cycle apparatus of the present invention is a refrigeration cycle apparatus in which a compressor, a condenser, an expansion means, and an evaporator are connected by piping and a refrigerant is circulated therein to constitute a refrigeration cycle. It is provided with a pulsation state quantity detection means for detecting the sound pressure of the air at either the compressor discharge side pipe or the compressor suction side pipe, and the pulsation state quantity detection means is installed in the vicinity of the measurement target. .

  Further, the refrigeration cycle apparatus according to the present invention includes a calculation means for calculating a relational variable from the pulsation state quantity detection means and calculating two or more state quantities or relation variables detected by the pulsation state quantity detection means to obtain a composite variable. , A storage means for storing a detection value by the state quantity detection means or a calculation value such as a composite variable calculated therefrom; and a determination means for determining a compressor abnormality from the composite variable obtained by the calculation means. .

  The refrigeration cycle apparatus diagnosis method of the present invention extracts the state in which the refrigeration cycle apparatus is operating normally from the detection values by the state quantity detection means stored in the storage means or the state feature values calculated from them. And learning. In the refrigeration cycle apparatus diagnosis method of the present invention, any one of a learned value detected by each state quantity detecting means during normal operation or a state feature value calculated from them is forced to another value. And a step of newly calculating a composite variable after the conversion, and a step of setting the newly calculated composite variable as a threshold value when the determination unit determines that the compressor is abnormal.

  In addition, by combining the learning methods of FIGS. 9, 11, and 12 of the present invention, measurement data, that is, refrigerant state quantity and pulsation state, for an existing machine or for a device whose specification at a remote location is not clear. Learning can be started first by receiving a quantity of measured data continuously and directly or through communication. Initially, the diagnosis is inaccurate, but it starts with recording of operation data, and then the configuration and operation of the present invention are obtained by obtaining information on equipment specifications and installation status, checking the operating history during maintenance, checking records for abnormal operation, and performing simulation tests. Thus, it is possible to obtain an accurate diagnosis in a relatively short period of time. As a result, work to diagnose failures of refrigeration cycles and equipment using communications such as the Internet should be performed separately from work used for refrigeration cycles and equipment operations, work for manufacturing refrigeration cycles and equipment, or maintenance work. I can do it. In this case, information necessary for failure diagnosis may be exchanged only with the currently operating equipment, the type of equipment, the specifications, and the like, as well as the type of measurement equipment and data for the equipment and equipment. Of course, if you want to include failure prediction, you will also need operation records, failure history, maintenance records, and so on.

  The diagnosis method of the refrigeration cycle apparatus according to the present invention is based on the value of the composite variable in the normal state and the calculated value of the current composite variable by the calculation means and the threshold value or the threshold value and elapsed time set by the user in advance. A step of calculating a time until the degree reaches a threshold value, that is, a step of predicting a failure.

  In the refrigeration cycle apparatus of the present invention, the judging means for judging the compressor abnormality includes refrigerant liquid back to the compressor, refrigeration oil depletion / deterioration inside the compressor, compressor bearing abnormality, compressor swinging abnormality, It is a judgment means that can discriminate any or all of compressor related failure events such as compressor motor abnormality, compressor valve / scroll, rotor compression chamber components such as rotor damage, compressor tooth contact, etc. And Of course, the failure can be similarly determined not only for the compressor but also for other devices and devices such as a blower driven by a motor. The composite variable described above is characterized by the Mahalanobis distance.

  Next, a method of detecting refrigerant leakage as a refrigeration cycle abnormality will be considered. When there is a refrigerant leak, the refrigerant flowing into the supercooling means provided between the condenser 12 and the expansion means 13 is a two-phase refrigerant. The cooling capacity is reduced, and the subcooling (supercooling degree) of the refrigerant at the inlet of the expansion means 13 becomes smaller than the state without refrigerant leakage or the initial stage of refrigerant leakage. Therefore, if the change in the subcool is captured, the refrigerant leakage can be specified. Further, if the refrigerant pressure pulsation waveform due to the fluctuation of the refrigerant pressure is measured, it is possible to catch the bubbles that are the flash gas in the liquid refrigerant due to the decrease in subcooling, which is a sign of refrigerant leakage. Further, if the sound pressure around the expansion valve during the refrigeration cycle is measured, it is possible to detect the flash gas mixture due to the subcool reduction by the change in the sound pressure. Therefore, it is only necessary to capture and combine changes in sound and vibrations indicating changes in the physical state of the refrigerant.

  In the refrigeration apparatus, the heat exchange amount in the condenser 12 differs when the outside air temperature is different. Further, the ambient air temperature of the evaporator 14 built in the load side device such as a showcase or a refrigerator is constantly controlled by the opening degree of the expansion means 13. Further, the compressor 11 performs capacity control, number control, or ON / OFF control so that the refrigeration cycle operates normally. In the refrigeration system, since the refrigeration cycle is formed by circulating refrigerant in the piping, the state quantities of the refrigeration cycle change in correlation with each other. Each state quantity of the refrigeration cycle such as subcool changes.

  That is, the subcool of the refrigeration cycle changes depending on any of the heat exchange amount in the condenser 12, the control state of the expansion means 13, the control state of the compressor 11, and the refrigerant leakage amount. Similarly, the state quantity of the other refrigeration cycle is also determined by any of the heat exchange amount in the condenser 12, the control state of the flow path opening / closing means 36 and the expansion means 13, the control state of the compressor 11, and the refrigerant leakage amount. It also changes depending on. Therefore, even if only the change in the subcool of the refrigeration cycle is measured, it cannot be specified whether the change in the subcool is due to refrigerant leakage or the change in the operating state of the refrigeration cycle.

  However, since the change factors other than the refrigerant leakage are generated in the normal operation of the refrigeration apparatus, a plurality of state quantities including the subcool of the refrigeration cycle are measured in the operation state in which the refrigerant leakage does not occur, and these are mutually connected. If it can be handled as an assembly having a correlation, it will be separated from the assembly when a refrigerant leak occurs, so that the refrigerant leakage can be specified. As described above, as a method of capturing a plurality of state quantities as an aggregate, there is a method of using the Mahalanobis distance already described.

When the Mahalanobis method is used for detecting refrigerant leakage in the refrigeration cycle, as a result of examination, it has been found that the characteristic quantities of refrigerant leakage in the refrigeration apparatus are high pressure, low pressure, and subcool. The feature amount is a state amount in which a change appears when the phenomenon occurs. Now, the high pressure of the refrigeration cycle is X ₁ , the low pressure is X ₂ , the subcool is X _3, and X _{2 to} X _{3 are} changed in a state where there is no refrigerant leakage to make a total of n (2 or more) combinations, X _{1 to} X ₃ are measured. The measured value is used as reference data. The average values and standard deviations (data variation degrees) of X _{1 to} X ₃ have already been described in Expressions 1 and 2. Then converted original _X 1 to X ₃ in _x 1 ~x ₃ are normalized as Equation 9 using them. Note that j takes any value from 1 to n and corresponds to each of n measured values. As shown in Expression 10, a correlation matrix R indicating a correlation between x _{1 to} x ₃ and an inverse matrix R ⁻¹ of the correlation matrix are obtained.

  Data can be handled as an aggregate having a certain distribution by the matrix indicating the average value, standard deviation, and correlation. This collection of data is called a unit space. A unit space with respect to a normal state as a base for judgment, here, a state without refrigerant leakage, is referred to as a reference space. Data constituting this reference space is referred to as reference data.

The Mahalanobis distance D ² is defined by Equation 11. Note that j in the equation takes any value from 1 to n and corresponds to each of the n measured values. K is the number of items (number of parameters), which is 3 here. Further, a _{11 to} a _kk are coefficients of the inverse matrix of the correlation matrix, and the Mahalanobis distance is about 1 in the reference space, that is, when there is no refrigerant leakage. Then, the high pressure X ₁ , the low pressure X ₂ and the subcool X ₃ corresponding to the refrigerant leakage amount to be detected are measured, and the Mahalanobis distance in the refrigerant leakage state is obtained as described above, and this is stored as a threshold value. At this time, the inverse matrix of the correlation matrix is obtained in a state where there is no refrigerant leakage as a reference.

  The concept of Mahalanobis distance is shown in FIG. FIG. 17 shows the correlation between two parameters, high pressure and subcool. That is, as the high pressure increases, the subcool increases. Although there is variation, there is a correlation between the high pressure and the subcool, and in a state where there is no refrigerant leakage, it falls within a certain range, and these are used as reference data to create a reference space. The other state quantities also have a correlation like the high pressure and the subcool. Then, with respect to the reference space (reference data), whether the data to be determined is normal or abnormal is determined based on the Mahalanobis distance.

  Further, as already explained in FIG. 7, the Mahalanobis distance and its appearance rate are normal or abnormal depending on the positional relationship of the calculated Mahalanobis distance with respect to the reference space, regardless of the number of parameters. Judgment can be made. In the reference space, the Mahalanobis distance has an average of about 1, and even when variation is taken into consideration, the distance is 4 or less. And in an actual machine, the measurement means which measures each state quantity of a freezing apparatus is provided, and these measured values are processed with a previous formula, and the Mahalanobis distance is calculated | required. Then, the magnitude of the Mahalanobis distance corresponds to the refrigerant leakage amount, and the refrigerant leakage can be known from the magnitude of the Mahalanobis distance. Note that the Mahalanobis distance is normally 4 or less in the reference space (normal state). Therefore, when it exceeds this, it is considered abnormal. However, in practice, there is also a problem of detection error, so the threshold value for judging refrigerant leakage is set to an appropriate value larger than 4, for example, 50. Note that the threshold value is set to a value corresponding to the refrigerant amount at a stage where refrigerant leakage occurs before the refrigeration cycle reaches uncooled.

  In addition, although the explanation has been given here on the assumption that the refrigerant leakage is estimated based on the three state quantities of the high pressure, the low pressure, and the subcool of the refrigeration cycle, the present invention is not limited to this. The condensation temperature (condenser saturation temperature) may be used instead of the high pressure, and the evaporation temperature (evaporator saturation temperature) may be used instead of the low pressure. Further, the Mahalanobis distance may be obtained using a larger amount of state, which improves detection accuracy. Further, the liquid pipe temperature detecting means for obtaining the subcool may be installed in the outlet pipe of the supercooling means, but is not limited to this, and may be installed anywhere as long as it is a liquid pipe. Play. However, it is more preferable that the subcooling at the position where the liquid pipe temperature detection means is installed is as large as possible because the refrigerant leakage detection accuracy is high, and it is more preferable to install it on the high pressure side and as close as possible to the expansion means. These state quantities may be sound, vibration, or pressure pulsation.

  In addition, here, a refrigeration apparatus having a liquid reservoir has been described as an example. However, other equipment such as an air conditioner having a liquid reservoir has a liquid reservoir and stores excess refrigerant in the liquid reservoir. Needless to say, the same principle produces the same effect. In addition, if the apparatus is configured to store excess refrigerant in the liquid reservoir, the same can be said even if other equipment configurations are different. For example, in a refrigeration apparatus having a liquid reservoir and an accumulator, excess refrigerant is stored in the liquid reservoir. Therefore, the same effect is obtained by the same principle. Furthermore, in an apparatus without a liquid reservoir, excess refrigerant accumulates inside the condenser, and the amount of state such as high pressure and low pressure of the refrigeration cycle changes depending on the amount of refrigerant in the circuit, and judgment including this overall change is necessary. Become. The refrigerant state quantity measuring means in the refrigeration cycle at this time may measure high pressure, low pressure, subcool, superheat, or discharge temperature. The subcool is obtained from the liquid tube temperature-condensation temperature, and the superheat is obtained from the suction temperature-evaporation temperature.

  Further, the Mahalanobis distance may be output as it is as the refrigerant leakage amount. The square root of the Mahalanobis distance is called the D value, and a D value corresponding to the limit refrigerant leakage amount is obtained, and this is corresponded to the maximum output voltage, for example, 5 V, no refrigerant leakage, leakage amount, small amount, medium amount A method is also conceivable in which the output means 22 outputs the D value and the voltage in correspondence from the large leakage amount to the limit refrigerant leakage amount. The Mahalanobis distance is a value that is proportional to the square of the deviation of each state quantity, but the D value is a value that is proportional to the deviation of each state quantity, and is a value that can be easily handled to correspond to a voltage or the like.

  It is also possible to estimate the refrigerant leak rate from the state of Mahalanobis distance or D value changing, and to predict the time when the limit refrigerant leak amount is reached. That is, as shown in FIG. 13 described above, when the Mahalanobis distance or the D value continues to increase, it is expected that some modulation continues even if the failure has not yet occurred. Once the refrigerant leakage occurs, the refrigerant leakage does not stop unless the refrigerant leakage portion is closed or refilled, so the Mahalanobis distance and D value continue to increase. Therefore, if the Mahalanobis distance or D value continues to increase, it can be said that the possibility of refrigerant leakage is high. Even if the Mahalanobis distance or D value does not reach the threshold, it can be determined that the refrigerant leaks. In addition, the time to reach the threshold, that is, the time to reach the limit amount of the refrigerant leakage can be predicted from the change speed of the distance. Since the state quantity of the refrigeration cycle is constantly changing, the Mahalanobis distance and the D value change even if the refrigerant leakage amount does not change. Therefore, the increasing trend here does not necessarily have to be a monotonous increase, but means that it tends to increase as a whole except for a slight increase or decrease. Then, based on the prediction of the time until the refrigerant leakage reaches the limit amount, the time when the refrigerant leakage amount reaches the limit refrigerant leakage amount may be output from the output means as a voltage. For example, the time may be set in proportion to the distance so that it is within one day for 5V, within one week for 3V, within one month for 1V, and no refrigerant leakage for 0V.

  Although the data used here is described as if it were a constant value, even if the data is changing, if the average value of the data for a certain period of time is taken, it can be handled in the same way, and the same effect can be achieved. Needless to say. In addition, here, as an example of using a Mahalanobis distance as a method of capturing a plurality of state quantities as an aggregate, other methods may be used.

  As described above, the subcool (or liquid tube temperature) in the normal state is the high pressure (or condensation temperature) and low pressure (or evaporation temperature) or the difference between the high pressure and low pressure (or the difference between the condensation temperature and evaporation temperature). It is possible to detect refrigerant leakage by learning and storing the relationship and observing the change.

In any method, any refrigerant may flow in the refrigeration cycle of the refrigeration apparatus. For example, a single component refrigerant such as R22 or R32, a three-component mixed refrigerant such as R407C, Thus, a mixed refrigerant composed of two components, an HC refrigerant such as propane, a natural refrigerant such as CO _2, or the like can be used. Refrigerants that have a negative impact on global environmental protection can be replaced if leakage begins even a little. In addition, the leakage of the flammable refrigerant can be processed in advance before the problem occurs if the safety limit value determined by the standard or the like is displayed. Furthermore, in a refrigeration system using a flammable refrigerant or a refrigerant containing a considerable amount of flammable components, such as propane, R32 or R410A, as a refrigerant, the refrigerant leakage is extremely dangerous from the viewpoint of safety. Is detected and output as an electric signal such as a voltage or a communication code, it is necessary to give priority to the output of another refrigeration apparatus. And it can notify a refrigerant | coolant leak early by making an output means into voltage output or electric current output, connecting to an alarm device, and issuing a warning by sound or light.

  As described above, the abnormality such as the compressor includes not only a failure of the device but also a time-dependent change such as a deterioration of the device, and any object whose operation state changes can be detected. For example, deterioration due to the life of the compressor 11 or liquid back, dirt or breakage of the condenser 12 or the evaporator 14, deterioration or failure of the blower of the condenser 12 or the blower of the evaporator, or provision in the piping of the refrigerant circulation circuit Clogging of dust collection strainer and refrigerant drying dryer, deterioration of refrigeration oil used in compressor 11 (detected by clogging of piping, poor lubrication of compressor, change in heat transfer, etc.) Can be detected and discriminated by configuration.

  In addition, the unit space for calculation includes an average value, a standard deviation, and a correlation coefficient of each feature amount, and these are stored in a memory on a substrate that is also received as a control device or an electrical product. In order to learn all or a part of the actual machine, it must be stored in a rewritable memory.

  In addition, in the case where the unit spaces are different as shown in FIG. 10, the data includes five data of high pressure, low pressure, discharge temperature, superheat, and subcool except for vibration, sound pressure, driving force, etc. in different combinations. Although explained, it is not limited to this. In some refrigeration cycle apparatuses, if the high pressure is too low, it is not preferable in terms of the reliability of the equipment, and therefore, there is a refrigeration cycle apparatus provided with a high pressure maintaining means. In this case, whether or not the high-pressure maintaining means works is different in the summer when the high pressure is high and in the winter when the high pressure is low, and the operation of the refrigeration cycle is different. Therefore, if the same reference space and anomalous space are used throughout the year, anomaly discrimination accuracy may deteriorate. In such a case, as shown in FIG. 18, it is preferable to have a plurality of reference spaces and anomalous spaces every year and use them according to the season. It should be noted that the proper use of this season may be performed according to the outside air temperature, but the actual machine often does not have an outside air temperature detecting means, and in that case, it is properly used depending on the high pressure range. In FIG. 18, the vertical axis represents the outside air temperature and the horizontal axis represents the passage of time throughout the year. The reference space when installed in winter is 1 and the reference space when the outdoor temperature is high in summer is 4. Thus, an explanation is provided in which a plurality of reference spaces are provided according to changes in the outside air temperature.

  Moreover, since the present invention is configured as described above, it is possible to remotely monitor the abnormality (failure and deterioration) of the device, so that the abnormality of the device can be found without going to the site. Can be detected early. And in the past, it was necessary to go to the site first to understand the cause of the abnormality and then take countermeasures later. Therefore, it is possible to prepare in advance and go to the work site, and to shorten the time to recovery. For example, when a compressor failure occurs, it can be detected remotely, so a spare compressor can be prepared and dispatched to the site.

  The degree of abnormality of the equipment and the refrigeration cycle is judged from the value calculated by the calculation means of the present invention, and the limit time when stable operation cannot be continued is predicted, so that safe operation can be performed. For example, it predicts a limit time that can maintain a pre-stored cooling capacity, and includes output means for outputting the predicted limit time as an electric signal such as the magnitude of voltage or current. The signal is a voltage output or a current output according to the degree of abnormality with the maximum amount of limit abnormality that can maintain a predetermined cooling capacity as a maximum value, and anyone can know the state of abnormality and maintenance is easy.

  Also, by defining multiple abnormal states according to the degree of abnormality of the device for one abnormality cause, and estimating the degree of abnormality of the device from the change in distance between the current operating state of the device and the plurality of abnormal states An easy-to-use diagnostic device such as continuation of driving in various states can be obtained. Furthermore, it has a means for extracting and learning the normal state of the equipment from the actual operation data, so that a reliable judgment can be obtained. Further, the value or distance which is a calculation result for determining the abnormality of the device is a numerical value obtained by processing the Mahalanobis distance or the Mahalanobis distance, and can be determined by accurate data.

  The device monitoring system of the present invention measures a plurality of waveform data continuously generated under the influence of operation in the operation state of the device, and the measured waveform data is averaged, standard deviation, etc. related to each waveform. A waveform data processing means for processing a plurality of parameters for each waveform data, and a plurality of parameters obtained by the waveform data processing means are combined as a plurality of variables to calculate a state quantity during operation of the equipment. Comparing the calculation means with the determination means for comparing whether the state quantity calculated by the calculation means is within the set range, whether the equipment is in a normal operating state or the degree of abnormality, and the information of the information through the equipment and the communication means A remote monitoring device for exchanging information, the information being measured waveform data, the processing result processed by the waveform data processing means, the state quantity calculated by the computing means, and the judgment Determination and Te in the range of at least one of the determination results, can be reliably equipment monitoring and becomes the user can perform driving with peace of mind.

  The device monitoring system of the present invention measures waveform data continuously generated under the influence of operation in the operation state of the device, and the measured waveform data is converted into a plurality of parameters such as an average value and a standard deviation related to the waveform. A waveform data processing means to be processed, a calculation means for calculating a state quantity at the time of operation of the device by a combination of a plurality of parameters obtained by the waveform data processing means as a plurality of variables, and a calculation means; The calculation result in the current operation state of the device and the calculation result when the device is operating in the normal state, or the calculation when the device is operating in the abnormal state or when it is estimated to be abnormal operation The distance between the results is obtained, and the time or the degree of abnormality until the obtained distance reaches the preset device operation limit value with respect to the elapsed operation time of the device is estimated. And a remote monitoring device connected to the device via the communication means and monitoring the operating state of the device. The time estimated by the failure prediction estimating device is transmitted to the remote monitoring device and this is transmitted to the remote monitoring device. Since the estimated time is displayed, the administrator who manages the device can continue the operation without difficulty and can grasp the maintenance schedule.

  The remote monitoring system of the present invention is configured to transmit measurement data or calculation values to a remote device via a network or a public line, so that any problem can be easily dealt with and effective in continuing operation. is there. In addition, safe operation is possible by connecting an output means for outputting an abnormality as an electrical signal or communicating with others as a communication code to a remote device via a network or a public line.

  As described above, the present invention has means for extracting and learning the state in which a device or apparatus is normally operated from the measured values stored in the storage means or the calculated values calculated from them. Therefore, stable data can always be obtained. Further, a step of forcibly converting any one of the measured values stored in the storage means or the calculated values calculated from them into another value, and after the conversion, newly adding the composite variable Since there is a step of calculating and a step of setting the newly calculated composite variable as a threshold value when the determining means determines fluid leakage, a threshold value for easily determining the range of abnormality can be set. It is also possible to determine this threshold value from data obtained by experimenting in advance.

  The present invention relates to a plurality of means for storing a state quantity or a calculated value from the state quantity when the device is operating normally, and a calculated value from the state quantity or the state amount in an abnormal state where the device has an abnormality. Means for estimating the device or means for reproducing the abnormal state of the device, means for calculating the distance between the normal state, the abnormal state, and the current operating state of the device, and the distance or abnormality between the current operating state of the device and the normal state Means for estimating the normal state, abnormal state, degree of abnormality, or cause of abnormality of a device from the change in distance from the state, provided in the vicinity of the refrigeration cycle apparatus or remotely via a network or public line, and via the network or public line Since it is configured to transmit measurement data or calculation values, it is possible to operate the equipment with peace of mind. Further, by providing output means for outputting the time when the predicted abnormality limit is reached by an electric signal such as a voltage or current magnitude, the detected abnormality can be transmitted at an early stage.

It is a whole conceptual diagram of Embodiment 1 of this invention. It is a block diagram of Embodiment 1 of this invention. It is detail drawing of the pulsation state amount detection means which detects the sound pressure of Embodiment 1 of this invention. It is a control block diagram of the compressor abnormality determination of Embodiment 1 of this invention. It is sound pressure waveform data explanatory drawing at the time of normal of Embodiment 1 of this invention. It is sound pressure waveform data explanatory drawing at the time of abnormality of Embodiment 1 of this invention. It is explanatory drawing explaining the relationship between the distance of Mahalanobis of Embodiment 1 of this invention, and its appearance rate. It is a calculation flowchart of the Mahalanobis distance of Embodiment 1 of the present invention. It is a flowchart showing the compressor abnormality detection method using the Mahalanobis distance in the real machine of Embodiment 1 of the present invention. It is explanatory drawing showing the concept of the Mahalanobis distance of the abnormal space and normal space of Embodiment 1 of this invention. It is a flowchart showing the normal reference space replenishment processing content of Embodiment 1 of this invention. It is a flowchart showing the novel abnormality learning function content of Embodiment 1 of this invention. It is explanatory drawing showing the time transition of the Mahalanobis distance of Embodiment 1 of this invention. It is another block diagram of Embodiment 1 of this invention. It is a pulsation state quantity detection means detail drawing which detects the sound pressure and vibration of Embodiment 1 of this invention. It is a pulsation state quantity detection means detail drawing which detects the sound pressure of Embodiment 1 of this invention in multiple points. It is a figure which shows the concept of the Mahalanobis distance of Embodiment 1 of this invention. It is a figure which shows the division | segmentation method of the reference space in the year of Embodiment 1 of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Refrigeration cycle apparatus, 2 Microcomputer, 3 Telephone line or LAN, 4 Remote monitoring room, 5 Computer, 6 Display apparatus, 7 Input apparatus, 8 Warning lamp, 9 Speaker, 11 Compressor, 12 Condenser, 13 Expansion valve, 14 Evaporator, 15 pulsation state quantity detection means, 16 refrigerant state quantity detection means, 17 signal processing means, 18 calculation means, 19 storage means, 20 comparison means, 21 judgment means, 22 output means, 32 filter amplifier, 33 A / D Converter, 34 FFT, 51 1st pulsation state quantity detection means, 52 2nd pulsation state quantity detection means, 53 Microphone.

Claims (14)

A first operating state quantity is measured from a physical quantity of a fluid such as a refrigerant having a time delay in a change in suction and discharge by the equipment during operation of the equipment and a physical quantity such as a current effective value of a driving equipment that drives the equipment. Pulsating pulsation including a short time change such as an instantaneous value during operation of the device, directly from or in the vicinity of the measuring device and the fluid-related device such as the pipe connected to the device or the device A second measuring means for measuring a state quantity and a plurality of measured quantities measured by the first measuring means are each subjected to arithmetic processing such as an average value and a standard deviation to obtain a first state quantity. The measurement amount measured by the measurement means is subjected to arithmetic processing such as skewness, kurtosis, and crossing frequency as time waveform data or frequency data to obtain the second state quantity, and the combination aggregate using the calculated arithmetic values as variables. Summarized in the plurality of Computation means for computing a state quantity having surveying characteristics; whether or not the device is in a normal operating state by comparing the state quantities having the characteristics of the plurality of measurement quantities computed by the computation means with the stored content stored in advance Determination means for determining the first state quantity and the first state at regular intervals by performing measurement and calculation processing of the second measurement means in accordance with a measurement interval measured by the first measurement means. A device diagnostic apparatus characterized in that the value of the state quantity of 2 is made uniform. A plurality of operating state quantities are determined at different points from the physical quantity of a fluid such as a refrigerant having a time delay in the change in suction and discharge of the equipment during operation of the equipment and the physical quantity such as the current effective value of the driving equipment that drives the equipment. A first measuring means for measuring;
A plurality of pulsating state quantities that pulsate including a short time change such as an instantaneous value during operation of the device, directly or in the vicinity of the device such as a pipe connected to the device or a pipe connected to the device Second measuring means for measuring at a relevant position;
The plurality of measurement quantities measured by the first measurement means are each subjected to arithmetic processing such as an average value and standard deviation, and the plurality of measurement quantities measured by the second measurement means are converted to time waveform data or frequency. An arithmetic means for performing arithmetic processing such as skewness, kurtosis, and intersection frequency as data, and calculating each state value as a variable in a combined set and calculating a state quantity having the characteristics of the plurality of measurement quantities;
The state quantity calculated by the calculating means from the plurality of measured quantities measured when the operation of the equipment is determined to be normal is stored as a state quantity in the normal state of the equipment, and the equipment is The calculation is performed from a plurality of measurement amounts corresponding to a plurality of different abnormal states, which are measured when determined to be different from each other, or set according to different abnormal states so as to obtain different abnormal states. State quantity storage means for storing a plurality of different state quantities calculated by the means as state quantities of a plurality of abnormal states of the equipment, and a plurality of different quantities measured by the measuring means during operation of the equipment. A plurality of state quantities during operation calculated by the calculation means using a plurality of measurement quantities corresponding to the abnormal state as a variable and the state quantity in the normal state or the compound quantity in the normal state. Device diagnostic apparatus characterized by comprising: a determination means for determining the cause of the abnormality from a plurality of abnormal conditions which differ respectively comparing the difference between the state quantity of the abnormal state of the. A state quantity calculated by the calculation means in a state where the device is normally operated is learned and stored as a normal operation state, or a state calculated by the calculation means in a state where the device is abnormally operated The apparatus diagnosis apparatus according to claim 1, wherein the device diagnosis apparatus learns and stores the quantity as an abnormal driving state. 2. The pulsation state quantity measured by the second measuring means is calculated by converting from a function of time to a function of frequency, performing octave analysis, or wavelet transform. Or the apparatus diagnostic apparatus of 2. Whether the device is in a normal operating state or an abnormal operating state is determined by whether the operating state of a fluid device such as a compressor, a pump, or a blower that handles a flammable fluid or a fluid harmful to the human body, or a driving device of this fluid device. 4. The apparatus diagnosis apparatus according to claim 1, wherein the apparatus diagnosis apparatus determines whether it is normal or abnormal. The comparison of the state quantities calculated during operation of the device is based on each of the operating quantities calculated by the calculating means using a plurality of measured quantities corresponding to a plurality of different abnormal states measured by the measuring means as variables. It is a comparison between a plurality of state quantities and state quantities of the plurality of abnormal states, and by this comparison, the determination that the device is abnormal is based on the lack of lubricating oil in the device, the deterioration of lubricating oil, etc. Judgment of equipment bearing part abnormality, equipment drive motor part abnormality, contact between fixed part and rotating part of the equipment, breakage of rotating part or fixing part of the equipment, swiveling of the rotating part, and liquid back The device diagnosis apparatus according to claim 1, wherein the device diagnosis apparatus determines whether or not any of these abnormalities is included. The amount of electricity measured during operation of the device is a numerical value or a waveform selected from drive current, electromagnetic force, radio wave, leakage current, shaft voltage, and the like, according to any one of claims 1 to 6. Equipment diagnostic device. When calculating the state quantity during operation of the device by calculation related to each other as a combination of multiple variables, forcibly convert one or more of the numerical values obtained by measurement or calculation processing to another value The device diagnosis apparatus according to claim 1, wherein a threshold value for determining an abnormal state is obtained from a state quantity obtained by using the converted value as a composite variable. The determination means determines whether or not the device is in a normal operating state based on whether or not the calculation result of the calculation means is normal, and the calculation at the time of measurement within the range determined to be normal 9. The device diagnosis apparatus according to claim 1, wherein the failure time of the device is estimated based on a relationship between a result and a preset threshold value. A refrigeration cycle in which a compressor, a condenser, an expansion means, and an evaporator are connected by piping and a refrigerant is circulated therein;
High pressure of refrigerant at any position of flow path from discharge side of compressor to expansion means, condensation temperature of refrigerant at high pressure position, flow path from discharge side of compressor to condenser At least one of the discharge temperature of the refrigerant at any of the positions, and the low pressure of the refrigerant at any position of the flow path from the expansion means to the suction side of the compressor, the low pressure Measure at least one of two measured quantities: the refrigerant evaporation temperature at the position, and at least one measured quantity of the refrigerant suction temperature at any position in the flow path from the suction side of the compressor to the evaporator. Refrigerant state quantity measuring means,
Vibration at any position of the compressor casing, the discharge side pipe of the compressor, or the suction side pipe of the compressor, the sound pressure of air at any position around the compressor, the compression A pulsation state quantity measuring means for measuring at least one pulsation state quantity of the electric quantity for driving the machine;
A plurality of measured quantities measured by the refrigerant state quantity measuring means are each subjected to arithmetic processing such as an average value and standard deviation, and the pulsating state quantities measured by the pulsating state quantity measuring means are distorted as time waveform data or frequency data. Calculating means for calculating the state quantity having the characteristics of the plurality of measured quantities and pulsating state quantities by calculating the degrees, kurtosis, crossing frequency, etc. Prepared,
The refrigeration cycle is characterized in that an operating state of the refrigeration cycle is determined from a calculation result of the calculation means that performs measurement and calculation processing of the pulsation state quantity measurement means in accordance with a measurement interval measured by the refrigerant state quantity measurement means. apparatus. The pulsation state quantity measuring means is a first pulsation state quantity measuring means for measuring vibration at any position of the compressor casing, the compressor discharge side pipe or the compressor suction side pipe, At least two of the second pulsation state quantity measuring means for measuring the sound pressure of air at any position around the compressor, and the third pulsation state quantity measuring means for measuring the current for driving the compressor The refrigeration cycle is calculated by combining the refrigerant state quantity obtained from the refrigerant state quantity measuring means and the pulsating state quantity obtained from at least two pulsating state quantity measuring means in association with each other. The refrigeration cycle apparatus according to claim 10, wherein an operation state of the refrigeration cycle is determined. The refrigeration according to claim 10 or 11, wherein a numerical value set in advance or a numerical value read or estimated from an operating state of the device is used instead of the measurement amount measured by the refrigerant state quantity measuring means. Cycle equipment. A refrigeration cycle in which a compressor, a condenser, an expansion means, and an evaporator are connected by piping and a refrigerant is circulated therein;
High pressure of refrigerant at any position of flow path from discharge side of compressor to expansion means, condensation temperature of refrigerant at high pressure position, flow path from discharge side of compressor to condenser At least one of the discharge temperature of the refrigerant at any of the positions, and the low pressure of the refrigerant at any position of the flow path from the expansion means to the suction side of the compressor, the low pressure Measure at least one of two measured quantities: the refrigerant evaporation temperature at the position, and at least one measured quantity of the refrigerant suction temperature at any position in the flow path from the suction side of the compressor to the evaporator. Refrigerant state quantity measuring means,
Vibration at any position of the compressor casing, the discharge side pipe of the compressor, or the suction side pipe of the compressor, the sound pressure of air at any position around the compressor, the compression A pulsation state quantity measuring means for measuring at least one pulsation state quantity of the electric quantity for driving the machine;
A plurality of measured quantities measured by the refrigerant state quantity measuring means are each subjected to arithmetic processing such as an average value and standard deviation, and the pulsating state quantities measured by the pulsating state quantity measuring means are distorted as time waveform data or frequency data. Computing means for computing degrees, kurtosis, intersection frequency, etc., and computing the state quantities having the characteristics of the plurality of measured quantities and pulsation state quantities by combining the computed values as variables into a combined set;
When the operation of the refrigeration cycle is determined to be normal, the calculation means calculates the plurality of measurement quantities measured by the refrigerant state quantity measurement means and the pulsation state quantities measured by the pulsation state quantity measurement means. The calculated state quantity is stored as a state quantity of the normal state of the device, and each is measured so that the refrigeration cycle is determined to be different from each other, or different from each other so that different abnormal states can be obtained. A plurality of measured amounts corresponding to a plurality of different abnormal states set according to the abnormal state and a plurality of different state amounts calculated by the calculating means from the pulsating state amount are set to a plurality of abnormalities of the device. State quantity storage means for storing the state quantity as a state;
During operation of the refrigeration cycle, the calculation unit calculates a plurality of measurement amounts corresponding to a plurality of different abnormal states measured by the refrigerant state amount measurement unit and the pulsation state amount measurement unit during operation of the refrigeration cycle. A judgment means for comparing the difference between each of the plurality of state quantities and the state quantity of the normal state or the state quantities of the plurality of abnormal states and judging the cause of the abnormality from a plurality of different abnormal states, and
The distance between each of the plurality of state quantities during operation of the refrigeration cycle and the state quantity in the normal state or the state quantities in the abnormal state is obtained, and the obtained distance is a preset threshold. The cause of the abnormality is transmitted to the remote monitoring device via the communication means together with at least one of the result of judging whether the refrigeration cycle apparatus is in a normal operation state and the degree of abnormality by comparing whether or not Refrigeration cycle monitoring system. Between the calculation result when the refrigeration cycle is calculated in the current operating state and the calculation result when the refrigeration cycle is operating in a normal state or the calculation result when the refrigeration cycle is operating in an abnormal state A failure prediction estimating means for estimating a time until the determined distance reaches an operation limit value set in advance with respect to the operation elapsed time of the refrigeration cycle,
14. The refrigeration cycle monitoring system according to claim 13, wherein the time estimated by the failure prediction estimation unit is displayed on a remote monitoring device that is connected via a communication unit and monitors an operation state of the device.

Images