JP2015132439A - Equipment diagnostic apparatus, equipment diagnostic method, and equipment diagnostic program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an equipment diagnostic apparatus capable of ensuring clear determination basis and minimizing overall risk.SOLUTION: An equipment diagnostic apparatus of the present invention comprises: an input data management unit that creates a first probability distribution for determining whether equipment is abnormal using normal sample data on normal equipment, and that creates a second probability distribution for determining whether the equipment is normal using abnormal sample data on abnormal equipment; a threshold determination unit that determines a threshold common to the first probability distribution and the second probability distribution so as to obtain a smallest sum of a first probability that is a probability that an actual machine as diagnostic target equipment is determined to be abnormal on the basis of the first probability distribution and a second probability that is a probability that the actual machine as the diagnostic target equipment is determined to be normal on the basis of the second probability distribution; and a diagnostic unit that determines whether the actual machine is normal on the basis of a magnitude relation between the common threshold and diagnostic target data acquired from the actual machine.

Description

  The present invention relates to a device diagnosis apparatus, a device diagnosis method, and a device diagnosis program.

  Recently, the operation state of equipment is controlled by a computer to increase the thermal efficiency. For example, regarding air-conditioning related devices, there are many examples in which a computer performs energy saving measures for a single device or an entire integrated system. Compared to general equipment, air-conditioning equipment includes many drive parts, liquid passage parts, and parts with significant temperature changes, and therefore, deterioration over time is likely to proceed. For example, as air-conditioning equipment in a large-scale office building deteriorates, the thermal efficiency naturally decreases. In addition, extreme cases such as sudden outages can cause huge damage to economic activity. Therefore, in order to predict the deterioration of the air-conditioning related equipment, it is important to carry out fine monitoring and diagnosis on a daily basis at the site or by operation from a remote place.

  The diagnostic apparatus for an absorption refrigerator of Patent Document 1 detects a state quantity of the absorption refrigerator and corrects the detected state quantity to a value in a rated operating condition. Then, a predetermined threshold is applied to the corrected state quantity to determine whether or not the absorption refrigerator is normal. The failure diagnosis apparatus of Patent Literature 2 diagnoses a failure of a showcase for a store and a refrigerator. The failure diagnosis apparatus collects in advance normal showcase state data (for example, internal temperature) and creates a probability density function indicating the distribution of the data. If the state data measured now exceeds a threshold (significant level) such as upper (or lower) 1% or 5% in the probability distribution, it is determined that the showcase and the refrigerator are abnormal. .

Japanese Patent Laying-Open No. 2004-20076 (Claim 1, FIG. 7, FIG. 8, etc.) JP 2003-172567 A (Claim 1, paragraph 0084, etc.)

  In the invention of Patent Document 1, the normal / abnormal determination is arbitrarily divided depending on how the threshold is set. Further, the invention of Patent Document 1 may make a false diagnosis that the cause of the abnormality is “B” even though the true cause of the abnormality is “A”. Although the invention of Patent Document 2 introduces the concept of statistical hypothesis testing and reduces the possibility of misdiagnosis, it uses one probability density function and a fixed significance level. That is, out of a plurality of risks that can occur at the same time, only one risk is excluded. Moreover, since the invention of Patent Document 2 does not determine whether it is normal or not according to the type, seriousness, economic influence, etc. of the abnormality, it lacks the viewpoint of overall balance. Therefore, an object of the present invention is to provide a device diagnosis apparatus that has a clear determination basis and can minimize the overall risk.

The device diagnostic apparatus of the present invention uses the normal sample data of the normal device to create a first probability distribution for determining that the device is abnormal, and uses the abnormal sample data of the abnormal device. And an input data management unit that creates a second probability distribution for determining that the device is normal, and determines that the actual machine that is the device to be diagnosed is abnormal based on the first probability distribution A first probability that is a probability to be performed and a second probability that is a probability that an actual machine that is a diagnosis target device is determined to be normal based on the second probability distribution, Whether or not the actual machine is normal based on a threshold value determination unit that determines a threshold common to the probability distribution of 1 and the second probability distribution, and the magnitude relationship between the common threshold and the diagnosis target data acquired from the actual machine A diagnostic unit for determining .
Other means will be described in the embodiment for carrying out the invention.

  According to the present invention, it is possible to provide a device diagnosis apparatus that has a clear determination basis and can minimize the overall risk.

It is a figure explaining the structure of an absorption-type cold / heat machine. It is a figure explaining the structure of an apparatus diagnostic apparatus, and the relationship between an apparatus diagnostic apparatus and an absorption-type cooling / heating machine. It is a figure which shows an example of apparatus operation information. (A), (b) and (c) is a figure which shows the probability density function of a sample value. (A) is a diagram illustrating a tilt with respect to the cooling rated capacity ratio of [Delta] T _HG in a given year, (b) are diagrams illustrating a tilt with respect to the cooling rated capacity ratio of [Delta] X _HG in a given year. It is a figure which shows the relationship of the reference value of air_conditioning | cooling capability, cold water exit temperature, cooling water exit temperature, and high temperature regenerator temperature. (A) is a flowchart of a data preparation process procedure, (b) is a flowchart of a threshold value determination process procedure. It is a flowchart of a diagnostic processing procedure. It is a flowchart (continuation) of a diagnostic processing procedure. It is a flowchart of a cooling water inlet temperature and a cooling performance coefficient estimation processing procedure. (A) And (b) is an example of the time series graph of diagnostic object data.

  Hereinafter, a mode for carrying out the present invention (referred to as “the present embodiment”) will be described in detail with reference to the drawings. The present embodiment is an example in which the present invention is applied to an “absorption chiller / heater”. Of course, the present invention can also be applied to other air conditioning related devices, and can also be applied to devices other than air conditioning related devices.

(Absorption type cooling / heating machine)
The structure of the absorption chiller / heater 42 will be described with reference to FIG. The absorption chiller / heater 42 is a kind of refrigerator. Generally, a refrigerator changes the state of a refrigerant from a liquid to a gas by controlling the pressure, and takes away heat corresponding to the heat of vaporization from the cold water. The absorption chiller / heater 42 is characterized in that water is used as a refrigerant and a low pressure is generated by causing the absorbent to absorb the refrigerant (water). As the absorbent, an aqueous lithium bromide solution is generally used. The absorption chiller / heater 42 of the present embodiment is also such a type of absorption chiller / heater.
The absorption chiller / heater 42 includes an evaporator 51, an absorber 52, a condenser 53, a low temperature regenerator 54, a high temperature regenerator 55, a low temperature heat exchanger 56 and a high temperature heat exchanger 57. In the evaporator 51, a refrigerant is dripped with respect to the coiled tube which made the heat-transfer area extremely large. Cold water passes through the tube. Since the inside of the evaporator 51 is in an ultra-low pressure state of, for example, about 1/100 atm, the dropped refrigerant (water) is easily vaporized at a low temperature of, for example, about 5 ° C. Then, the refrigerant takes the heat of vaporization from the cold water, and the temperature of the cold water decreases accordingly.

The absorber 52 and the evaporator 51 are in communication with each other, and refrigerant (water vapor) passes between them. In the absorber 52, an absorbent is dripped with respect to a coiled tube. Cooling water passes through the tube. The absorbent (aqueous lithium bromide solution) absorbs the refrigerant (water vapor). As a result, the low pressure state in the evaporator 51 is maintained. The absorbent that has absorbed the refrigerant releases heat to the cooling water in the tube while reducing its concentration, and is stored at the bottom of the absorber 52. The cooling water flows into a coiled tube in the condenser 53.
The high temperature regenerator 55 receives the absorbent flowing from the absorber 52 via the low temperature heat exchanger 56 and the high temperature heat exchanger 57, and heats the received absorbent by burning heavy oil or the like with a boiler. To do. Then, the refrigerant (water vapor) is separated from the absorbent, and the concentration of the absorbent gradually increases. The high concentration absorbent flows into the low temperature regenerator 54 via the high temperature heat exchanger 57. The separated water vapor flows into a coiled tube in the low temperature regenerator 54.

The low temperature regenerator 54 has a tube through which a refrigerant (water vapor) passes. The absorbent that has flowed into the low-temperature regenerator 54 is heated by the refrigerant, and again separates the refrigerant (water vapor). The separated refrigerant (water vapor) flows into the condenser 53, and the absorbent having a higher concentration is stored at the bottom of the low-temperature regenerator 54. The stored absorbent flows into the low-temperature heat exchanger 56. The refrigerant (water vapor) in the tube flows into the condenser 53 and is stored as water.
The condenser 53 and the low-temperature regenerator 54 communicate with each other, and refrigerant (water vapor) passes between them. The condenser 53 has a tube through which cooling water passes. The refrigerant (water vapor) flowing from the high temperature regenerator 55 via the low temperature regenerator 54 is condensed on the tube and stored in the bottom of the condenser 53. The accumulated refrigerant (water) flows into the evaporator 51.

The low-temperature heat exchanger 56 and the high-temperature heat exchanger 57 apply heat from the high-temperature absorbent to the low-temperature absorbent in the process until the absorbent flows from the absorber 52 to the high-temperature regenerator 55 (saving fuel such as heavy oil). To preheat). The role of cold water is roughly as follows.
・ Cold water absorbs heat at the air conditioning load (air conditioner, etc.).
・ Cooling water absorbs the heat of cold water and releases it to the outside (cooling tower, etc.).
-Refrigerant mediates the release of heat of cold water to cooling water by changing its state.
The absorbent creates a low pressure by absorbing the refrigerant (water vapor).
In the present embodiment, the refrigerant is water. The refrigerant is in a state of pure water not containing the absorbent (in the condenser 53 and in the evaporator 51) → a state in the absorbent (aqueous solution) (in the absorber 52) → a state of pure water vapor separated from the absorbent (The space above the high-temperature regenerator 55) → the state of pure water not containing an absorbent, and so on, repeatedly transits.

  Many sensors are arrange | positioned on each apparatus which comprises the absorption-type cooling / heating machine 42, and detects the temperature etc. of each apparatus. Many sensors are also installed on pipes (or coiled tubes) connecting each device, and detect the temperature of liquid (gas) such as cold water, the flow rate per unit time, the concentration, and the like. For example, the sensor at position Δ58 in FIG. 1 detects “cold water outlet temperature”, the sensor at position Δ59 detects “cooling water outlet temperature”, and the sensor at position Δ60 indicates “low temperature”. Regenerator drain temperature ”is detected. And although not shown in figure, auxiliary equipment, such as a pump and a pressure-reduction valve, is installed in the required position on piping.

(Equipment diagnostic equipment, etc.)
The configuration of the device diagnostic apparatus 1 and the relationship between the device diagnostic apparatus 1 and the absorption chiller / heater 42 will be described with reference to FIG. The device diagnosis apparatus 1 is operated by, for example, a service company that performs building care. The device diagnosis apparatus 1 is a general computer, and includes a central control device 11, an input device 12, an output device 13, a main storage device 14, an auxiliary storage device 15, and a communication device 16. These are connected to each other by a bus. The auxiliary storage device 15 stores device operation information 31 (details described later). The input data management unit 21, the threshold value determination unit 22, and the diagnosis unit 23 in the main storage device 14 are programs. Thereafter, when the subject is described as “XX section”, the central control device 11 reads each program from the auxiliary storage device 15 and loads it into the main storage device 14, and then the function of each program (detailed later). Shall be realized.

  The absorption chiller / heater 42 is operated by a customer of a service company. Usually, a service company has a plurality of customers, and each customer often operates a plurality of absorption chillers 42. The sensor 46 is the sensor described above. The control device 45 is a so-called microcomputer, and controls the overall operation of the absorption chiller / heater 42 while receiving a signal from the sensor 46 in accordance with a program loaded therein. The monitoring terminal device 44 also receives a signal from the sensor 46. One control device 45 and one monitoring terminal device 44 are usually installed for each absorption chiller / heater 42. The repeater 41 receives the signal of the sensor 46 from each monitoring terminal device 44. Then, the signal is transmitted to the device diagnostic apparatus 1 via the network 2. The number controller 43 is connected to the controller 45 of each absorption chiller / heater 42, and the absorption chiller / heater 42 that is actually operated according to environmental conditions such as the size of the air conditioning load and the outside air temperature. Control the number of units.

  In FIG. 2, the example in which the device diagnosis apparatus 1 of the service company remotely monitors the customer's absorption chiller / heater 42 via the (external) network 2 has been described. In addition to the service company centrally managing the absorption chiller / heater 42 of many customers from the monitoring center, the customer can also monitor the absorption chiller / heater 42 operated in-house on site. That is, an example in which the device diagnostic apparatus 1 is integrated with the repeater 41 or the monitoring terminal device 44 is also possible.

(Equipment operation information)
The apparatus operation information 31 is demonstrated along FIG. In the device operation information 31, in association with the data ID stored in the data ID column 101, the cooling / heating machine ID column 102 has the cooling / heating machine ID, the cooling performance coefficient column 103 has the cooling performance coefficient, and the high temperature regenerator. The temperature column 104 has a high temperature regenerator temperature, the high temperature regenerator concentration column 105 has a high temperature regenerator concentration, the cold water outlet temperature column 106 has a cold water outlet temperature, and the cooling water outlet temperature column 107 has a cooling water outlet temperature. The acquisition time column 108 stores an acquisition time point, and the determination column 109 stores a determination flag.

The data ID in the data ID column 101 is an identifier that uniquely identifies the record of the device operation information 31.
The chiller / heater ID in the chiller / heater ID column 102 is an identifier that uniquely identifies the absorption chiller / heater 42.
The cooling performance coefficient in the cooling performance coefficient column 103 is a value defined by the following equations 1 and 2. Note that J / second = kW holds.

The high temperature regenerator temperature in the high temperature regenerator temperature column 104 is the temperature of the high temperature regenerator 55 itself, and is equal to the temperature of the absorbent stored in the high temperature regenerator 55. Here, the unit of the high temperature regenerator temperature is Celsius temperature (° C.), but it may be absolute temperature (K).
The high temperature regenerator concentration in the high temperature regenerator concentration column 105 is the concentration of the absorbent stored in the high temperature regenerator 55. Here, the unit of concentration is “wt%”.
The cold water outlet temperature in the cold water outlet temperature column 106 is the temperature of the cold water at the position of Δ58 in FIG.
The cooling water outlet temperature in the cooling water outlet temperature column 107 is the temperature of the cooling water at the position of Δ59 in FIG.

The acquisition point in the acquisition point column 108 is the year, month, day, hour, minute, and second when the sensor 46 acquires a value such as a high temperature regenerator temperature. In the present embodiment, it is assumed that all data such as the high-temperature regenerator temperature are simultaneously detected by the sensor 46 at a certain point in time or calculated based on the detected data. Furthermore, the input data management part 21 shall create the record of the apparatus operation information 31 based on the detected or calculated data in real time.
The determination flag in the determination column 109 is a result of diagnosis for the absorption chiller / heater 42. The determination flags include “normal” indicating that the absorption chiller / heater 42 is normal, “abnormal” indicating abnormality, and a flag (details described later) indicating a specific abnormal state. “*” Will be described later.

  Focus on records 127-129. The absorption chiller / heater “M98” is, for example, a test machine (normal sample machine) in a state before the customer starts operation. The high-temperature regenerator temperature and the like of the absorption chiller / heater “M98” are acquired every 5 minutes from “20130101 10:00:00”. The user (service company) of the device diagnostic apparatus 1 obtains sample values such as a high-temperature regenerator temperature in a normal state on the assumption that the absorption chiller / heater “M98” is normal.

  Focus on records 130-132. The absorption chiller / heater “M99” is, for example, a test machine (abnormal sample machine) in a state in which aged deterioration has progressed after a customer starts operation. The high-temperature regenerator temperature and the like of the absorption chiller / heater “M99” are acquired every 5 minutes from “20130101 10:00:00”. The user of the device diagnostic apparatus 1 acquires a sample value such as a high-temperature regenerator temperature in an abnormal state on the assumption that the absorption chiller / heater “M99” is abnormal.

  That is, “normal *” in the determination column 109 indicates that the data of the record is a sample value in a normal state. Similarly, “abnormal *” indicates that the data of the record is a sample value in an abnormal state. On the other hand, the determination column 109 of the records 121 to 126 is blank. The absorption chiller / heater “M01” and the absorption chiller / heater “M02” are “real machines” being operated by customers. The device diagnostic apparatus 1 receives data such as the high temperature regenerator temperature from these actual machines and determines whether or not the actual machines are normal. When determining the threshold value for the determination, both the record storing “normal *” and the record storing “abnormal *” are used. Finally, the determination result is stored (details will be described later).

  The types of data listed in the column 103 to column 107 of the device operation information 31 are merely examples. In addition to these, for example, there may be columns such as a cold water inlet temperature, a cooling water inlet temperature, a cold water flow rate, a cooling water flow rate, and a heat input amount from a boiler. In addition, there may be a column for storing secondary data calculated by processing these primary data. For example, there may be columns of “cooling capacity” calculated based on “cold water flow rate”, “cold water outlet temperature”, and “cold water inlet temperature” which are primary data. Further, separately from the device operation information 31, association information that stores constants such as rated cooling capacity and chilled water specific heat in association with the chiller / heater ID may be stored in the auxiliary storage device 15.

FIGS. 4A, 4B, and 4C are diagrams illustrating probability density functions of sample values. In these figures, an example of “high temperature regenerator temperature” is adopted, but the idea is the same even for other types of data. First, attention is focused on FIG. The input data management unit 21 acquires, for example, a plurality of high-temperature regenerator temperatures, for example, of a certain normal sample machine in a certain period from the record of the device operation information 31. The input data management unit 21 performs statistical processing using the acquired plurality of values as a population, and calculates the average μ ₀ and variance σ ₀ ² of the population. Then, assuming that the population follows a normal distribution, a probability density function p ₀ indicating a normal distribution N (μ ₀ , σ ₀ ² ) is created.
In the present embodiment, the probability density function p ₀ has the form of Equation 3.

Thus, the probability density function p ₀ is a function having “μ ₀ ” and “σ ₀ ² ” as parameters and “x” as an input variable, and the output variable of the function is a distribution in which “x” occurs. It is. That is, when the high temperature regenerator temperature of the actual machine is “x”, the distribution in which the actual machine is normal is p ₀ (x). The probability density function p ₀ is a left-right symmetric curve having an average μ ₀ as an axis of symmetry, and has two inflection points. Each inflection point corresponds to a position of “μ ₀ −σ ₀ ” and “μ ₀ + σ ₀ ”. “Σ ₀ ” is the positive square root of the variance “σ ₀ ² ” and is the standard deviation of the population.

Next, pay attention to FIG. The input data management unit 21 acquires, for example, a plurality of high-temperature regenerator temperatures of a certain abnormal sample machine in a certain period in the record of the device operation information 31. Input data management unit 21, a plurality of values obtained performs statistical processing of the population, to calculate the average mu ₁ and variance sigma ₁ ² of that population. Then, on the premise that the population follows a normal distribution, a probability density function p ₁ indicating a normal distribution N (μ ₁ , σ ₁ ² ) is created.
In the present embodiment, the probability density function p ₁ has the form of Equation 4.

Thus, the probability density function p ₁ is a function having “μ ₁ ” and “σ ₁ ² ” as parameters and “x” as an input variable, and the output variable of the function is a distribution in which “x” occurs. It is. That is, when the high temperature regenerator temperature of the actual machine is “x”, the distribution in which the actual machine is abnormal is p ₁ (x). The probability density function p ₁ is a left-right symmetric curve having an average μ ₁ as a symmetry axis, and has two inflection points. Each inflection point corresponds to a position of “μ ₁ −σ ₁ ” and “μ ₁ + σ ₁ ”. Note that “σ ₁ ” is the positive square root of the variance “σ ₁ ² ” and is the standard deviation of the population.

and the average value mu ₀ of p _0, when comparing the average value mu ₁ of _{p 1,} not necessarily a μ ₀ <μ _1. Deterioration of air-conditioning equipment is manifested as a decrease in thermal efficiency. As a result, the value such as the cold water outlet temperature acquired from the abnormal sample machine is often larger than the value acquired from the normal sample machine (μ ₀ <μ ₁ ). Further, when the absorption chiller / heater 42 deteriorates, for example, the efficiency of the absorbent to absorb the refrigerant (water vapor) decreases, resulting in an increase in the high-temperature regenerator concentration. On the other hand, as air-conditioning related equipment that supplies hot water, steam, and the like deteriorates, the temperature of hot water, steam, etc. that can be supplied decreases (μ ₀ > μ ₁ ). Hereinafter, as a representative example, an example in which μ ₀ <μ ₁ is described. and variance sigma ₀ ² of p _0, when comparing the variance sigma ₁ ² of _{p 1,} often becomes σ ₀ ² <σ ₁ ^2.

Returning to FIG. “S ₀ ” is a significance level of “5%”. The area of the region between the horizontal axis and p ₀ (x) is always “1”. Of the areas, the area E ₀ of the right side of “s ₀ ” is “0.05”. Therefore, when the high temperature regenerator temperature “x” of the actual machine exceeds the significance level “s ₀ ”, it is natural to consider that the actual machine is abnormal.
Again, it progresses to FIG.4 (b). “S ₁ ” is a 5% significance level. The area of the region between the horizontal axis and p ₁ (x) is always “1”. Among the areas, the area E ₁ of the left part of “s ₁ ” is “0.05”. Therefore, when the value “x” of the high temperature regenerator temperature of the actual machine is less than the significance level “s ₁ ”, it is natural to consider that the actual machine is normal. The significance level may be another value, for example, “1%”.

(Hypothesis test)
The theory of statistical hypothesis testing is often applied industrially. The idea of hypothesis testing is very similar to the mathematical paradox. First, a null hypothesis is established. The null hypothesis is a hypothesis that the user of the statistical result thinks that “this hypothesis is probably false, so we want to reject it eventually”. By the way, the opposite hypothesis of the null hypothesis is called the alternative hypothesis. The alternative hypothesis is a hypothesis that the user of the statistical results thinks that “this hypothesis is probably true, so we want to adopt it eventually”. Now, assume that the null hypothesis is “the actual machine is normal”. Then, the alternative hypothesis is “the actual machine is abnormal”. In FIG. 4A, when the actual high-temperature regenerator temperature “x” exceeds “s ₀ ”, the null hypothesis is rejected and the alternative hypothesis is adopted. In the end, we can conclude that the actual machine is abnormal. Conversely, suppose that the null hypothesis is “the actual machine is abnormal”. Then, the alternative hypothesis is “the actual machine is normal”. In FIG. 4B, if the actual high-temperature regenerator temperature “x” is less than “s ₁ ”, the null hypothesis is rejected and the alternative hypothesis is adopted. In the end, we can conclude that the actual machine is normal.

The fact that the actual machine is determined to be abnormal although it is normal is called a first type of error. The probability of occurrence of the first type of error is E _{0 in} FIG. The fact that the actual machine is judged to be normal despite being abnormal is called a second type error. The probability of occurrence of the second type of error is E _{1 in} FIG. Assume that a common value “s” is used as “s ₀ ” in FIG. 4A and “s ₁ ” in FIG. In order to avoid the first type of error, the value of “s” may be increased. Conversely, in order to avoid the second type of error, the value of “s” may be reduced. The probability that the first type error will occur and the probability that the second type error will occur are in a trade-off relationship. In other words, the other cannot be reduced without increasing one.

  Then, focusing on minimizing the probability of the first type (second type) error, avoiding the situation of ignoring the probability of the second type (first type) error. It turns out that it is necessary. That is, it is necessary to determine a threshold value that minimizes the overall risk regardless of whether the first type error or the second type error occurs. FIG. 4C described below is based on this concept.

FIG. 4C is a diagram in which the probability density function p ₀ (x) and the probability density function p ₁ (x) are displayed on the same plane. Of the area of the region between the horizontal axis and p ₀ (x), the area on the right side of the threshold “s” is E ₀₀ . Of the area of the region between the horizontal axis and p ₁ (x), the area on the left side of the threshold “s” is E ₁₁ . Furthermore, the amount of damage caused by being judged abnormal even though the actual machine is normal is C _0, and is caused by judging that the actual machine is normal even though the actual machine is abnormal the amount of damage to the _{C 1.}
Damages C _0, the damages caused to the results customers who had to stop the actual believe that the actual machine is abnormal, replacement parts and which has been exchanged for not abnormal expenses, purchase of a new actual Costs and old machine disposal costs. Damages C ₁ is no longer obtained occurred in the result actual emergency inspection required is continued to be used without being carried out has stopped suddenly customer losses, forced to update trouble and extra to other devices Such as the amount of equipment purchased, the future revenue that the service company would have obtained from the customer. Then, the total damage amount C is defined as shown in Equation 5.

The total amount of loss C is determined according to changes in C ₀ , E ₀₀ , C ₁ and E ₁₁ . Now, when C ₀ and C ₁ are determined, the total damage amount C changes in accordance with the position of the threshold “s” in FIG. Expression 6 represents Expression 5 using an integral symbol. Expression 6 is a function C (s) having “s” as an input variable and C as an output variable. Therefore, “s” that minimizes C can be obtained.

In order to obtain “s” that minimizes C, first, a derivative is derived by differentiating Equation 6 by “s”. Then, “s” is determined such that the derived derivative value is “0”. That is, it is only necessary to solve Equation 7 in which “s” is an unknown. For example, when σ ₀ = σ ₁ = σ, “s” as a solution can be expressed as in Equation 8 using C ₀ or the like.

When μ ₀ > μ ₁ is satisfied, p ₀ (x) dx and p ₁ (x) dx are interchanged with each other in Equation 6 and C ₀ and C ₁ are interchanged with each other. That is, the right test is performed for the smaller average of p ₀ (x) and p ₁ (x), and the left test is performed for the other. In Equation 8, it is not assumed that μ ₀ = μ ₁ .

  In the above, an example has been described in which actual machine data to be compared with the threshold value (hereinafter referred to as “diagnosis target data”) is the value of the high-temperature regenerator temperature. However, the threshold value determination unit 22 can also determine threshold values for other types of diagnosis target data. Furthermore, it is possible to more specifically diagnose an abnormality in the actual machine according to a combination of comparison results (magnitude relationship) between each diagnosis target data and its threshold value (detailed later).

  For example, the maximum value in one season of the cooling capacity of the actual machine may be less than the threshold value, and the cooling performance coefficient may be less than the threshold value. In this case, the diagnosis unit 23 can diagnose that the actual cold water flow rate itself is excessive, and that the cooling capacity and the cooling performance coefficient are likely to be apparently small. As another example, the maximum value in one season of the cooling capacity of the actual machine may be equal to or higher than the threshold value, and the cooling performance coefficient may be lower than the threshold value. In this case, although the cooling capacity is exhibited, the thermal efficiency is low, and the diagnosis unit 23 can diagnose that the efficiency of the boiler of the high-temperature regenerator 55 is reduced. Next, an example in which diagnosis target data is acquired through a more complicated calculation process will be described.

With reference to FIG. 5A, an example will be described in which the gradient of ΔT _HG in a certain year with respect to the cooling rated capacity ratio is used as diagnosis target data for the high temperature regenerator temperature T _HG . Now, assume that v values of the high-temperature regenerator temperature of an actual machine are detected in the year A.

The first term T _HG (A, n) on the right side of Equation 9 represents the n-th (n = 1, 2,..., V) value among the v values. The second term T _HGo (Q, T _CHout , T _CDout ) on the right side is a reference value for the high temperature regenerator temperature. The reference value is “should be a temperature”. ΔT _HG (A, n) on the left side indicates the difference from the reference value of the high temperature regenerator temperature. The reference value T _HGo (Q, T _CHout , T _CDout ) is a function value having Q, T _CHout and T _CDout as input variables. Here, Q is the cooling capacity, T _CHout is the cold water outlet temperature, and T _CDout is the cooling water outlet temperature.

FIG. 6 is referred to with reference to FIG. FIG. 6 is a diagram _showing a relationship (originally a four-dimensional relationship) of the cooling capacity Q, the chilled water outlet temperature T _CHout , the chilled water outlet temperature T _CDout and the reference value T _{HGo in} a two-dimensional plane. This figure shows the following.
_{-Under the} condition that other input variables are constant, the reference value T _HGo increases as the coolant outlet temperature T _CDout increases.
Similarly, the reference value T _HGo decreases as the cold water outlet temperature T _CHout increases.
Similarly, the reference value T _HGo increases as the cooling capacity Q increases.
Incidentally, for example, it is assumed that the cooling capacity Q is 600 kW. Then, the cooling water outlet temperature _{T CDOUT} a is 32.5 ° C., coolant outlet temperature _{T CHout} is assumed to be 11.0 ° C. (regression line the bottom). At this time, the reference value T _HGo is about 120 ° C.

ΔT _HG calculated in this way can be diagnostic target data by itself. However, even if the deterioration state of a certain absorption chiller / heater 42 is the same, ΔT _HG changes depending on the cooling capacity exhibited at that time. For example, some customers often use near the cooling capacity upper limit, and some customers often use the cooling capacity with a margin. Therefore, it is necessary to introduce the concept of “cooling rated capacity ratio” as indicated by the horizontal axis in FIG.
Returning to FIG. The cooling rated capacity ratio in FIG. 5A is a value defined by Equation 10.

The input data management unit 21 indicates a combination of a certain ΔT _HG value and a cooling rated capacity ratio at the time when the T _HG (A, n) value corresponding to the value is detected, and the vertical axis is ΔT _HG . Plot on a plane where the horizontal axis is the cooling rated capacity ratio. The input data management unit 21 plots a total of v points. Next, the input data management unit 21 determines a regression line passing through the origin using the least square method. The slope of this straight line can also be diagnostic target data.

  In this way, the input data management unit 21 can determine the slope of the regression line for each year. Note that the cooling capacity itself may be used in place of the cooling rated capacity ratio on the horizontal axis in FIG. When the cooling capacity rating ratio is adopted on the horizontal axis, the distribution range of points is normalized to 0-1. Therefore, the difference in slope for each year often appears more clearly. Generally, as the aging deterioration of the absorption chiller / heater 42 progresses, the value of the slope increases.

FIG. 5B is a diagram for explaining an example in which the slope of ΔX _HG with respect to the cooling rated capacity ratio in a certain year is used as diagnosis target data. In FIG. 5B, the vertical axis _represents the difference ΔX _HG from the reference value X _HGo of the high temperature regenerator concentration X _HG as compared to FIG. The only difference is that Equation 11 is used.

Furthermore, although not shown, it is also possible to use the slope of ΔT _LGd with respect to the cooling rated capacity ratio in a certain year as diagnosis target data. Similar to Equations 9 and 11, the difference ΔT _LGd of the low temperature regenerator drain temperature T _LGd from its reference value T _LGdo is defined as Equation 12. Similarly, a slope of ΔT _LGd with _{respect to} the cooling rated capacity ratio in a certain year is also defined.

(Processing procedure)
Hereinafter, the processing procedure of this embodiment will be described. There are four processing procedures: (1) data preparation processing procedure, (2) threshold determination processing procedure, (3) diagnostic processing procedure, and (4) cooling water inlet temperature / cooling performance coefficient estimation processing procedure. Three items other than (4) are executed in the order of (1) → (2) → (3). (1) Assuming that the data preparation processing procedure is executed, it is assumed that the device operation information 31 (FIG. 3) is updated to the latest state and stored in the auxiliary storage device 15. If previous generation remote monitoring system is used, cooling water inlet temperature may not be measured. Therefore, (4) is executed when the cooling performance coefficient is further estimated after estimating the cooling water inlet temperature in (1).

A data preparation procedure will be described with reference to FIG.
In step S201, the input data management unit 21 of the device diagnostic apparatus 1 receives a diagnosis start instruction. Specifically, the input data management unit 21 displays a menu screen (not shown) on the output device 13. Then, the user accepts that the input device 12 such as a mouse selects the character string “start diagnosis of the cooling / heating machine” displayed on the menu screen.

  In step S202, the input data management unit 21 accepts an actual machine to be diagnosed and a diagnosis object period. Specifically, the input data management unit 21 accepts that the user inputs an actual cooling / heating machine ID and a diagnosis target period of the diagnosis target via the input device 12. The diagnosis target period is, for example, a period in which the current date is the end and the date that is a certain number of days back from the current date is the start. For example, when the current state is October 31, 2013 and the actual state of the past three months is diagnosed, the user inputs “20130801 to 20131031”.

  In step S <b> 203, the input data management unit 21 acquires data from the device operation information 31. Specifically, the input data management unit 21 searches the device operation information 31 (FIG. 3, but ignores the record of the sample machine) using the cooling / heating device ID and the diagnosis target period received in step S202 as search keys. And get the corresponding record.

  In step S204, the input data management unit 21 extracts a record during operation. “During operation” means, for example, a state in which the actual cold water flow rate (instantaneous value) is equal to or greater than a predetermined threshold. This is because when the flow rate of cold water is small, the temperature of cold water or the like is observed to be higher. Specifically, the input data management unit 21 extracts a record whose chilled water flow rate at the time of acquisition is greater than or equal to a threshold value from the records acquired in step S203.

  In step S205, the input data management unit 21 creates a time series graph for each type of data. Specifically, the input data management unit 21 creates a time series graph for each type of data (column 103 to column 107 in FIG. 3) from the record extracted in step S204. FIGS. 11A and 11B show examples of time series graphs created at this time (not completely the same as the examples in columns 103 to 107 in FIG. 3).

  In step S206, the input data management unit 21 extracts stable data. There are various definitions of stable data. Specifically, the input data management unit 21 extracts a portion that is stable data from the time-series graph created in step S205. As an example that is not stable data, the fluctuation range of the value in one day is larger than a predetermined threshold, the time series graph does not become a gentle curve but has a rising or falling point, and the wave period is extremely different depending on the day. There are different things. It should be noted that by analyzing the dynamic characteristics of the actual machine, a portion that is determined to be not stable data at first glance can be handled as stable data. For example, a dynamic model related to the temperature of the actual machine is created in advance using a differential equation or the like, and the coefficient of the dynamic model is identified from the change in time-series temperature data at the start of the actual machine operation. If the identified coefficient value is stable, even if the time-series temperature data itself is not stable, it can be handled as stable data. However, in this case, data handled for diagnosis is not a temperature but a coefficient.

  In step S207, the input data management unit 21 erases noise. Even in time series graphs that are stable data, there are often “outliers” due to temporary changes in the external environment, mere chances (measuring the temperature of bubbles in a liquid), etc. . Specifically, the input data management unit 21 recreates a time series graph using a low-pass filter. If there is no suitable low-pass filter, a time series graph may be re-created by averaging values for several minutes around a certain point in time (moving average method). At this time, for example, if the value at that time and the previous six values (a total of seven values) are averaged, there will be no future value for that time.

In step S208, the input data management unit 21 acquires representative data. Specifically, the input data management unit 21 sets the data at the time when the cooling capacity is maximized in one day from the time series graph in which noise is eliminated in step S207, by the number equal to the number of days in the diagnosis target period. Extract and use as representative data.
When the processing of step S208 is completed, the input data management unit 21 holds representative data for each type of data. Incidentally, one of such representative data (for the high temperature regenerator temperature) is the above-described _THG (A, n).

In step S209, the input data management unit 21 calculates the inclination. Specifically, first, whether or not the representative data acquired in step S208 includes at least one of “high temperature regenerator temperature”, “high temperature regenerator concentration”, and “low temperature regenerator drain temperature”. Determine whether. If it is included, the process proceeds to “second” in step S209. Otherwise, the process proceeds to step S210.
Second, the input data management unit 21 uses the above-described method to determine the slope of ΔT _HG relative to the cooling rated capacity ratio, the slope of ΔX _HG relative to the cooling rated capacity ratio, and / or the slope of ΔT _LGd relative to the cooling rated capacity ratio. get.

  In step S210, the input data management unit 21 creates a probability density function of a normal sample machine. Specifically, first, the input data management unit 21 specifies a normal sample machine that matches the attribute of the real machine received in step S202. Here, the attribute is arbitrary data that characterizes each absorption-type cooling / heating machine 42. Generally, the installation location, the outside air temperature of the installation location, the model, the special specification, the manufacturer name, the manufacturing date Days. Assume that device basic information (not shown) in which these attributes are stored in association with all the absorption-type cooling / heating machines 42 including the actual machine and the sample machine is stored in the auxiliary storage device 15. The input data management unit 21 can specify a normal sample machine suitable as a comparison target of the actual machine by searching the device basic information.

Secondly, the input data management unit 21 performs the same processing as in steps S202 to S209 for the normal sample machine identified in “first” in step S210. However, “the actual cooler / heater ID” in step S202 is read as “the cooler / heater ID of the normal sample machine specified in“ first ”in step S210”.
Third, the input data management unit 21 calculates the mean μ ₀ and the variance σ ₀ ² using the representative data as a population, and creates a probability density function indicating the normal distribution N (μ ₀ , σ ₀ ² ). The probability density function created here is the probability density function p ₀ (x) in FIG. However, the inclination with respect to the cooling rated capacity ratio of [Delta] T _HG, one slope for one normal sample machine is defined. Therefore, the input data management unit 21 specifies a plurality (for example, 10) of normal sample machines whose attributes match the real machine (similar even if not completely matched). Then, mean μ ₀ and variance σ ₀ ² are calculated with a sample of 10 slopes as a population, and a normal distribution curve N (μ ₀ , σ ₀ ² ) is created. The same applies to the slope of ΔX _HG and ΔT _LGd with _{respect to} the cooling rated capacity ratio. This also applies to the creation of the normal distribution curve N (μ ₁ , σ ₁ ² ) of the abnormal sample machine described later.

In step S211, the input data management unit 21 creates a probability density function of the abnormal sample machine. Specifically, first, the input data management unit 21 identifies an abnormal sample machine that matches the attribute of the real machine received in step S202.
Second, the input data management unit 21 performs the same processing as in steps S202 to S209 for the abnormal sample machine identified in “first” in step S211. However, “the actual cooling / heating machine ID” in step S202 is read as “the cooling / heating machine ID of the abnormal sample machine specified in“ first ”in step S211”.
Third, the input data management unit 21 calculates the mean μ ₁ and the variance σ ₁ ² with the representative data as a population, and creates a probability density function indicating the normal distribution N (μ ₁ , σ ₁ ² ). The probability density function created here is the probability density function p ₁ (x) in FIG.
Thereafter, the data preparation processing procedure is terminated.

What kind of data is actually used as diagnosis target data depends on the user's setting. For ease of explanation, it is assumed that the input data management unit 21 has acquired representative data of the following seven types of data when the data preparation processing procedure is completed. Furthermore, it is assumed that the input data management unit 21 has created the probability density function p ₀ (x) and the probability density function p ₁ (x) for the following seven types of data.
1: Cooling capacity 2: Cooling performance coefficient 3: Cooling water outlet temperature 4: High temperature regenerator concentration 5: The gradient of the difference from the reference value of the high temperature regenerator temperature to the cooling rated capacity ratio (hereinafter simply referred to as “high temperature regenerator” "Temperature gradient")
6: The gradient of the difference from the reference value of the high temperature regenerator concentration with respect to the cooling rated capacity ratio (hereinafter simply referred to as the “high temperature regenerator concentration gradient”).
7: The slope of the difference from the reference value of the low temperature regenerator drain temperature with respect to the cooling rated capacity ratio (hereinafter simply referred to as the “low temperature regenerator drain temperature slope”).

(Threshold determination procedure)
A threshold value determination process procedure will be described with reference to FIG.
In step S221, the threshold value determination unit 22 of the device diagnostic apparatus 1 receives the amount of damage. Specifically, the threshold value determination unit 22 accepts that the user inputs the damage amount C ₀ and the damage amount C ₁ via the input device 12. C ₀ and C ₁ are used to calculate the total loss C as described above. As is apparent from Equation 6, C ₀ and C ₁ are weights multiplied by the probability, and the weight level itself does not affect the solution “s”. It is the ratio of C ₀ to C ₁ that affects the solution “s”. Therefore, when receiving an instruction from the user that “there is no appropriate value for the amount of damage C ₀ and the amount of damage C ₁ ”, the threshold value determination unit 22 determines that “the amount of damage C ₀ = the amount of damage C ₁ = 1”. And proceed to the subsequent processing. In this case, the weights for E ₀₀ and E ₁₁ are both “1”.

  In step S222, the threshold determination unit 22 determines a threshold. Specifically, for each of the seven types of data described above, “s” that minimizes the total damage amount C of Equation 6 is calculated by the method described above. Thereafter, the threshold determination processing procedure is terminated.

(Diagnostic procedure)
A diagnostic processing procedure will be described with reference to FIGS.
In step S231, the diagnosis unit 23 of the device diagnosis apparatus 1 determines whether or not the flowing water flow rate has been confirmed. Specifically, the diagnosis unit 23 determines whether or not the user has confirmed the chilled water flow rate at the latest time among the representative data acquisition time points. For example, the diagnosis unit 23 displays a “confirmed” button and an “unconfirmed” button on the output device 13 and accepts that the user presses one of them. If the “not confirmed” button is pressed (step S231 “NO”), the process proceeds to step S232. Otherwise (step S231 “YES”), the process proceeds to step S232b (FIG. 9).

  In step S232, the diagnosis unit 23 sets up a cold water flow rate unconfirmed checker. The cold water flow rate unconfirmed checker is, for example, “There is a possibility that the flow rate of cold water is different from the rated value, so that the device may be diagnosed as being abnormal.”

In step S233, the diagnosis unit 23 determines whether or not the maximum cooling capacity is less than a threshold value. Specifically, first, the diagnosis unit 23 acquires the maximum value (maximum cooling capacity) of the cooling capacity in the diagnosis target period of the actual machine to be diagnosed from the representative data. The maximum cooling capacity is guaranteed to be stable and not noise data during operation (hereinafter, the same applies to other types of data).
Secondly, the diagnosis unit 23 determines whether or not the maximum cooling capacity is less than a predetermined threshold (for example, 50% of the rated value), and if it is less than the threshold (step S233 “YES”), step The process proceeds to S234, and otherwise (step S233 “NO”), the process proceeds to step S245.

In step S234, the diagnosis unit 23 determines whether or not the cooling performance coefficient is less than a threshold value. Specifically, first, the diagnosis unit 23 acquires the cooling performance coefficient at the latest time point of the actual machine to be diagnosed from the representative data.
Second, the diagnosis unit 23 determines whether or not the cooling performance coefficient is less than a predetermined threshold (for example, 60% of the rated value), and if it is less than the threshold (step S234 “YES”), step The process proceeds to S235, and otherwise (step S234 “NO”), the process proceeds to step S239.

  In step S235, the diagnosis unit 23 sets a “cold water flow rate excessive” flag. At this time, the diagnosis unit 23 diagnoses that the actual cooling water flow rate itself is excessive, and that the cooling capacity and the cooling performance coefficient are likely to be apparently small.

In step S236, the diagnosis unit 23 determines whether or not the high temperature regenerator concentration is less than a threshold value. Specifically, first, the diagnosis unit 23 obtains the high-temperature regenerator concentration at the latest time of the actual machine to be diagnosed from the representative data.
Second, the diagnosis unit 23 determines whether or not the high-temperature regenerator concentration is less than a predetermined threshold (for example, 61%). If the concentration is less than the threshold (step S236 “YES”), the process proceeds to step S237. In other cases (step S236 “NO”), the process proceeds to step S239.

In step S237, the diagnosis unit 23 determines whether or not the coolant outlet temperature is equal to or higher than a threshold value. Specifically, first, the diagnosis unit 23 obtains the coolant outlet temperature at the latest time point of the actual machine to be diagnosed from the representative data.
Second, the diagnosis unit 23 determines whether or not the cooling water outlet temperature is equal to or higher than a predetermined threshold (for example, 35 ° C.). If the temperature is equal to or higher than the threshold (step S237 “YES”), the process proceeds to step S238. In other cases (step S237 “NO”), the process proceeds to step S239.

In step S238, the diagnosis unit 23 sets a “refrigerant overflow” flag.
In step S239, the diagnosis unit 23 determines whether the gradient of the high temperature regenerator temperature is equal to or greater than a threshold value. More specifically, the first, the diagnostic unit 23, from the representative data, to acquire the inclination of the high-temperature regenerator temperature in the diagnostic period of actual diagnostic object (inclination with respect to the cooling rated capacity ratio of [Delta] H _HG).
Second, the diagnosis unit 23 determines whether or not the gradient of the high temperature regenerator temperature is equal to or higher than a predetermined threshold (for example, 15). If the temperature is equal to or higher than the threshold (step S239 “YES”), the process proceeds to step S240. In other cases (step S239 “NO”), the process proceeds to step S253.

In step S240, the diagnosis unit 23 sets a “cooling water tube contamination” flag.
In step S241, the diagnosis unit 23 determines whether or not the gradient of the high temperature regenerator concentration is equal to or greater than a threshold value. Specifically, first, the diagnosis unit 23 acquires, from the representative data, the gradient of the high-temperature regenerator concentration (the gradient of ΔX _HG with respect to the cooling rated capacity ratio) in the diagnosis target period of the actual machine to be diagnosed.
Secondly, the diagnosis unit 23 determines whether or not the gradient of the high temperature regenerator concentration is equal to or greater than a predetermined threshold (for example, 10), and if it is equal to or greater than the threshold (step S241 “YES”), the process proceeds to step S242. In other cases (step S241 “NO”), the process proceeds to step S244.

In step S242, the diagnosis unit 23 sets an “evaporator deterioration” flag.
In step S243, the diagnosis unit 23 determines whether the slope of the low temperature regenerator drain temperature is equal to or greater than a threshold value. Specifically, first, the diagnosis unit 23 _acquires , from the representative data, an inclination of the low-temperature regenerator drain temperature (an inclination of ΔT _LGd with _{respect to} the cooling rated capacity ratio) in the diagnosis target period of the actual machine to be diagnosed.
Second, the diagnosis unit 23 determines whether or not the slope of the low temperature regenerator drain temperature is equal to or higher than a predetermined threshold (for example, 15), and if it is equal to or higher than the threshold (step S243 “YES”), step The process proceeds to S254, and otherwise (step S243 “NO”), the process proceeds to step S253.

In step S244, the diagnosis unit 23 determines whether the slope of the low temperature regenerator drain temperature is equal to or greater than a threshold value. The processing in this step is the same as that in step S243. However, if “second” is equal to or greater than the threshold (step S244 “YES”), the process proceeds to step S255, and otherwise (step S244 “NO”), the process proceeds to step S253.
In step S245, the diagnosis unit 23 determines whether or not the cooling performance coefficient is less than a threshold value. The processing in this step is the same as that in step S234. However, if “second” is less than the threshold value (step S245 “YES”), the process proceeds to step S246, and otherwise (step S245 “NO”), the process proceeds to step S247.

In step S246, the diagnosis unit 23 sets a “boiler efficiency decrease” flag.
In step S247, the diagnosis unit 23 determines whether the gradient of the high temperature regenerator temperature is equal to or greater than a threshold value. The processing in this step is the same as that in step S239. However, if “second” is equal to or greater than the threshold value (step S247 “YES”), the process proceeds to step S248; otherwise (step S247 “NO”), the process proceeds to step S253.

In step S248, the diagnosis unit 23 sets a “cooling water tube contamination” flag.
In step S249, the diagnosis unit 23 determines whether or not the gradient of the high temperature regenerator concentration is equal to or greater than a threshold value. The processing in this step is the same as that in step S241. However, if “second” is equal to or greater than the threshold value (step S249 “YES”), the process proceeds to step S250, and otherwise (step S249 “NO”), the process proceeds to step S252.

In step S250, the diagnosis unit 23 sets an “evaporator deterioration” flag.
In step S251, the diagnosis unit 23 determines whether the slope of the low temperature regenerator drain temperature is equal to or greater than a threshold value. The processing in this step is the same as that in step S243. However, if “second” is equal to or greater than the threshold value (step S251 “YES”), the process proceeds to step S254, and otherwise (step S251 “NO”), the process proceeds to step S253.

In step S252, the diagnosis unit 23 determines whether the slope of the low temperature regenerator drain temperature is equal to or greater than a threshold value. The processing in this step is the same as that in step S243. However, if “second” is equal to or greater than the threshold value (step S252 “YES”), the process proceeds to step S255; otherwise (step S252 “NO”), the process proceeds to step S253.
In step S253, the diagnosis unit 23 determines not to set any flags other than the already set flags.

In step S254, the diagnosis unit 23 sets an “absorber deterioration” flag.
In step S255, the diagnosis unit 23 sets a “low temperature regenerator deterioration” flag.
In step S256, the diagnosis unit 23 displays and stores a flag. Specifically, first, the diagnosis unit 23 displays all the flags that are already set on the output device 13. When the flag is not set, the diagnosis unit 23 displays a message “the device is normal” on the output device 13.
2ndly, the diagnostic part 23 memorize | stores the present time and the flag set in the determination column 109 of the record of the apparatus operation information 31 (FIG. 3) acquired in step S203. When the flag is not set, the diagnosis unit 23 stores the current time point and “normal”.
Thereafter, the diagnostic processing procedure is terminated.

The following is a description of the portion of FIG. 9 in the diagnostic processing procedure.
In step S232b, the diagnosis unit 23 sets up a chilled water flow rate confirmed checker. The chilled water flow rate confirmed checker is, for example, “use a threshold when the flow rate of chilled water has been confirmed”.

The subsequent processing of FIG. 9 is almost the same as the processing after step S233 of FIG. The same process is given the same number. The differences are as follows.
In step S233b, the diagnosis unit 23 uses a strict value (for example, 80% of the rated value) as a threshold value compared to step S233.
In step S234b, the diagnosis unit 23 uses a strict value (for example, 80% of the rated value) as a threshold value compared to step S234. If step S234b is “YES”, the process proceeds to step S236.
-In step S245b, the diagnostic part 23 uses a severe value (for example, 80% of a rated value) compared with step S245 as a threshold value.
In FIG. 9, step S235 does not exist.
In step S249b, if the diagnosis unit 23 is “NO” in step S249b, the diagnosis unit 23 proceeds to step S249c.
In FIG. 9, there is step S249c not shown in FIG.
In step S249c, the diagnosis unit 23 sets a “heat exchanger perforated” flag.

(Cooling water inlet temperature / cooling performance coefficient estimation processing procedure)
The cooling water inlet temperature / cooling performance coefficient estimation processing procedure will be described with reference to FIG. The processing procedure is an example of step S208 when the cooling water inlet temperature is unknown.
In step S271, the input data management unit 21 acquires the time when the cooling capacity is maximum in one day. Specifically, the input data management unit 21 obtains the time (reference time) at which the cooling capacity is maximized within a certain day from the time-series graph in which noise is eliminated in step S207.

In step S272, the input data management unit 21 acquires temperature data and the like. Specifically, the input data management unit 21 acquires the coolant outlet temperature at the reference time.
In step S273, the input data management unit 21 acquires weather data. Specifically, the input data management unit 21 acquires the humidity and the outside air temperature at the reference time from the external database via the network 2 and proceeds to step S275. Alternatively, if the humidity and the outside air temperature corresponding to the reference time cannot be acquired, the humidity and the outside air temperature at the observation time immediately before the reference time and the humidity and the outside air temperature at the observation time immediately after the reference time are acquired, and the process proceeds to step S274. move on.

In step S274, the input data management unit 21 interpolates weather data. Specifically, the input data management unit 21 acquires the humidity at the reference time by linearly interpolating the humidity at the immediately preceding observation time and the humidity at the immediately following observation time. Further, similarly, the outside air temperature at the reference time is calculated.
In step S275, the input data management unit 21 calculates the wet bulb temperature. Specifically, the input data management unit 21 obtains the wet bulb temperature by applying the humidity and the outside air temperature at the reference time obtained in step S274 (S273) to the “humidity table”. The humidity table is a table in which the vertical axis represents dry bulb temperature (outside air temperature), the horizontal axis represents “dry bulb temperature-wet bulb temperature”, and humidity is described in the cell at the intersection.

In step S276, the input data management unit 21 estimates the cooling water inlet temperature. Specifically, the input data management unit 21 estimates the cooling water inlet temperature using the wet bulb temperature, the cooling water outlet temperature, and the constant according to the following Expression 13. Note that “C ₁ ” in Equation 13 is different from “Damage C ₁ ” in Equation 5, Equation 6, and Equation 8.

（式１３）

(Formula 13)

Here, T _CDin is a cooling water inlet temperature, T _CDout is a cooling water outlet temperature, C _T is a wet bulb temperature influence coefficient, C ₄ is a water amount ratio correction value, and T _aow Is the wet bulb temperature, and R _GD is the cooling water amount ratio (1 at the time of rating).

  In step S277, the input data management unit 21 estimates the heat radiation amount. Specifically, the input data management unit 21 radiates heat in the cooling tower based on the cooling water inlet temperature estimated in step S276, the cooling water outlet temperature acquired in step S272, the cooling water flow rate, and the cooling water specific heat. Is estimated.

In step S278, the input data management unit 21 estimates the heat input. Specifically, the input data management unit 21 sets a value obtained by subtracting the cooling capacity from the heat release amount as the heat input amount from the boiler (strictly, the heat input amount actually input into the refrigeration cycle).
In step S279, the input data management unit 21 estimates the cooling performance coefficient. Specifically, the input data management unit 21 estimates the cooling performance coefficient at the reference time by applying the heat input from the boiler estimated in step S278 and the cooling capacity to Equation 1. Thereafter, the cooling water inlet temperature / cooling performance coefficient estimation processing procedure is terminated.

  The input data management unit 21 creates representative data on the cooling performance coefficient of the diagnosis target period by repeating the processes of steps S271 to S279 for all days belonging to the diagnosis target period.

(Modification 1: Update of probability density function)
In the above example, the input data managing section 21, the normal distribution _{N (μ 0, σ 0 2} ) all data belonging to a population statistically processed to a probability indicating the density function p ₀ and a normal distribution N (mu ₁ , σ ₁ ² ) is created as a probability density function p ₁ . That is, the average and variance for many values included in the population are actually calculated.
The normal sample machine and the abnormal sample machine generate new sample values every moment. Therefore, it is originally desirable that the input data management unit 21 repeatedly executes such processing (data preparation processing procedure) every time an actual machine is diagnosed. However, in order to reduce the amount of information processing and the processing time, the input data management unit 21 may execute the following processing.

For example, when starting the next processing procedure after the elapse of a predetermined period (full review period) such as one month starting from the end point of the data preparation processing procedure, the input data management unit 21 performs all processing of the processing procedure. Execute. This is because the normal sample machine and the abnormal sample machine can accumulate a statistically significant number of new sample values during the entire review period.
When the input data management unit 21 starts the data preparation processing procedure before the entire review period elapses, the processes in steps S210 and S211 are simplified as follows.
In step S210, first, the input data management unit 21 displays the current value of μ ₀ and the current value of σ ₀ ^{2 on} the output device 13.
Secondly, the input data management unit 21 accepts that the user inputs an update value of μ _{0 and} an update value of σ ₀ ² . The update value input by the user may be an actual value or a value such as “a value increased by 5% based on the current value”.
Thirdly, the input data management unit 21 creates a probability density function p ₀ based on the updated value.

In step S211, first, the input data management unit 21 displays the current value of μ ₁ and the current value of σ ₁ ^{2 on} the output device 13.
Second, input data management unit 21 accepts the user to enter an update value and sigma ₁ ² updated values of mu _1. The update value input by the user may be an actual value or a value such as “a value increased by 5% based on the current value”.
Thirdly, the input data management unit 21 creates a probability density function p ₁ based on the updated value.

In the foregoing, there has been described an example in which there are normal sample machines and abnormal sample machines, and the input data management unit 21 acquires data from them. However, there are cases where there is no normal sample machine or abnormal sample machine to be compared, as in the case of diagnosing equipment in a new field. In this case, the input data managing section 21, from the ^_beginning, μ _0, σ _{0 2,} may execute a process that is simplified by accepting the appropriate value of mu ₁ and sigma ₁ ^2.

(Variation 2: Determination of expected damage amount, etc.)
In the above description, the diagnosis unit 23 determines whether the actual machine is “normal” or “abnormal”. However, it is convenient if not only the binary conclusion but also the expected damage amount, the probability of being normal, and the probability of being abnormal when the actual machine presents the diagnosis target data. Therefore, in step S <b> 256, the diagnosis unit 23 may calculate the expected damage amount, the probability of being normal, and the probability of being abnormal and display it on the output device 13. The diagnosis unit 23 calculates the expected damage amount C ^* based on Equation 14.

“T” is an arbitrary value acquired from the representative data. The representative data may be for any data type. Then, the diagnosis unit 23 calculates a probability P ₀ (t) that is normal and a probability P ₁ (t) that is abnormal.
Here, P ₀ (t) + P ₁ (t) = 1 is not satisfied. Because, P ₀ is an input to calculate the value of actual equipment t against the probability density function p ₀ of the normal sample machine, P ₁ is (is another absorption type cold heat machine) abnormal sample machine This is because it is a value calculated by inputting the actual machine t to the probability density function p ₁ . When μ ₀ > μ ₁ , p ₀ (x) dx and p ₁ (x) dx are interchanged with each other in Equation 14 and C ₀ and C ₁ are interchanged with each other.

(Modification 3: Estimation of high temperature regenerator concentration _XGH )
In some cases, the observed value of the high-temperature regenerator concentration _XGH cannot be obtained. At this time, the input data management unit 21 may estimate the high temperature regenerator concentration _XGH based on the high temperature regenerator temperature and the low temperature regenerator drain temperature using a known method. The input data management unit 21 may use, for example, a method described in JP-A-5-231743 (paragraph 0015).

(Effect of embodiment)
This embodiment has the following effects.
(1) The threshold value determination unit 22 sets the diagnosis target data so as to minimize the sum of the probability that the absorption chiller / heater 42 is normal even though it is normal and the opposite probability. Determine the threshold at which is applied. Therefore, it is possible to make a balanced determination as a whole.
(2) The input data management unit 21 acquires sample data as an observation value. Therefore, the accuracy of diagnosis becomes higher.
(3) The probability (and the opposite probability) is weighted by the amount of damage. Therefore, a balanced determination can be made with a more realistic scale.
(4) The probability (and the inverse probability) follows a normal distribution. Thus, standard statistical processing is possible.
(5) The input data management unit 21 reviews the probability distribution of the probability (and vice versa) by reviewing the mean and variance of the population. Therefore, the processing burden can be reduced.
(6) The input data management unit 21 narrows down the diagnosis target data and the sample data to data about the operation time point, stable data, and data without noise. Therefore, more accurate determination is possible.
(7) The device of the present embodiment is an absorption chiller / heater. A wide variety of diagnostic object data can be acquired from the absorption chiller / heater. Therefore, specific diagnosis results can be prepared according to the type of data.
(8) The diagnosis unit 23 calculates and displays the expected damage amount, the probability that the actual machine is normal, and the probability that the actual machine is abnormal. Therefore, it is possible to know the degree of normality (or abnormality) as well as determining the state of the actual machine with a binary value of “normal” or “abnormal”.
(9) The input data management unit 21 acquires actual machine data via the network. Therefore, for example, it becomes easy for a service company to monitor and diagnose many actual machines operated by many customers.

In addition, this invention is not limited to an above-described Example, Various modifications are included. For example, the above-described embodiments have been described in detail for easy understanding of the present invention, and are not necessarily limited to those having all the configurations described. Further, a part of the configuration of one embodiment can be replaced with the configuration of another embodiment, and the configuration of another embodiment can be added to the configuration of one embodiment. Further, it is possible to add, delete, and replace other configurations for a part of the configuration of each embodiment.
Each of the above-described configurations, functions, processing units, processing means, and the like may be realized by hardware by designing a part or all of them with, for example, an integrated circuit. Each of the above-described configurations, functions, and the like may be realized by software by interpreting and executing a program that realizes each function by the processor. Information such as programs, tables, and files that realize each function can be stored in a recording device such as a memory, a hard disk, or an SSD (Solid State Drive), or a recording medium such as an IC card, an SD card, or a DVD.
In addition, the control lines and information lines are those that are considered necessary for the explanation, and not all the control lines and information lines on the product are necessarily shown. In practice, it may be considered that almost all the components are connected to each other.

DESCRIPTION OF SYMBOLS 1 Apparatus diagnostic apparatus 2 Network 11 Central control apparatus 12 Input apparatus 13 Output apparatus 14 Main storage apparatus 15 Auxiliary storage apparatus 21 Input data management part 22 Threshold value determination part 23 Diagnosis part 31 Equipment operation information 41 Repeater 42 Absorption-type cooling / heating machine 43 Number controller 44 Monitoring terminal device 45 Control device 46 Sensor 51 Evaporator 52 Absorber 53 Condenser 54 Low temperature regenerator 55 High temperature regenerator 56 Low temperature heat exchanger 57 High temperature heat exchanger

Claims (15)

Using a normal sample data of a normal device, creating a first probability distribution for determining that the device is abnormal;
An input data management unit that creates a second probability distribution for determining that the device is normal using the abnormal sample data of the device that is abnormal;
A first probability that is a probability that an actual device that is a diagnosis target device is abnormal based on the first probability distribution, and an actual device that is a diagnosis target device based on the second probability distribution A threshold value determination unit that determines a common threshold value for the first probability distribution and the second probability distribution, which minimizes a sum of second probabilities that are determined to be normal.
A diagnosis unit that determines whether or not the actual machine is normal based on a magnitude relationship between the common threshold and the diagnosis target data acquired from the actual machine;
A device diagnostic apparatus characterized by comprising: The input data management unit
Obtain the normal sample data as an observation value of normal equipment,
Obtaining the abnormal sample data as an observed value of an abnormal device;
The apparatus diagnostic apparatus according to claim 1. The first probability is
Weighted by the amount of damage when the actual machine is normal but determined to be abnormal,
The second probability is
Weighted by the amount of damage in the case where the actual machine is determined to be normal despite being abnormal,
The apparatus diagnostic apparatus according to claim 2. The first probability distribution and the second probability distribution are:
Exhibit a normal distribution,
The apparatus diagnostic apparatus according to claim 1. The input data management unit
Updating the first probability distribution by reviewing the mean and variance of the first probability distribution;
Updating the second probability distribution by reviewing the mean and variance of the second probability distribution;
The apparatus diagnostic apparatus according to claim 1. The input data management unit
The diagnostic object data, the normal sample data, and the abnormal sample data,
Limiting the data to the actual machine, the normal device and the abnormal device when the device is actually operating;
The apparatus diagnostic apparatus according to claim 3. The input data management unit
The diagnostic object data, the normal sample data, and the abnormal sample data,
Limited to time series information that is stable enough to meet certain criteria,
The apparatus diagnosis apparatus according to claim 6. The input data management unit
Removing noise of the diagnosis object data, the normal sample data, and the abnormal sample data;
The apparatus diagnostic apparatus according to claim 7. The equipment is
An absorption-type cooling / heating machine,
The diagnosis target data, the normal sample data, and the abnormal sample data are:
Including at least one of cooling capacity, cooling performance coefficient, high temperature regenerator temperature, high temperature regenerator concentration, and low temperature regenerator drain temperature for the device,
The apparatus diagnostic apparatus according to claim 8. The diagnosis target data, the normal sample data, and the abnormal sample data are:
The slope of the difference from the reference value of the high temperature regenerator temperature with respect to the ratio of the cooling capacity to the rated cooling capacity,
The slope of the difference of the high temperature regenerator concentration from its reference value with respect to the ratio of the cooling capacity to the rated cooling capacity; and
The slope of the difference from the reference value of the low temperature regenerator drain temperature with respect to the ratio of the cooling capacity to the rated cooling capacity,
Including at least one of
The apparatus diagnosis apparatus according to claim 9. The diagnostic unit
A maximum value of the cooling capacity in a predetermined period is less than a predetermined threshold, the cooling performance coefficient is not less than a predetermined threshold, and the cooling capacity of the difference from the reference value of the high-temperature regenerator temperature is If the slope relative to the percentage of the rated cooling capacity is not equal to or greater than the predetermined threshold,
Determining that the actual machine is normal;
The device diagnostic apparatus according to claim 10. The diagnostic unit
Based on the diagnosis target data, the first probability distribution, and the second probability distribution, calculating an expected damage amount of the actual machine, a probability that the actual machine is normal, and a probability that the actual machine is abnormal,
The apparatus diagnostic apparatus according to claim 11. The input data management unit
Obtaining an observation value of the actual machine including the diagnosis target data via a network;
The apparatus diagnosis apparatus according to claim 12. The input data management unit of the device diagnostic device
Using a normal sample data of a normal device, creating a first probability distribution for determining that the device is abnormal;
Using the abnormal sample data of the device that is abnormal, create a second probability distribution to determine that the device is normal,
The threshold value determination unit of the device diagnostic apparatus is
A first probability that is a probability that an actual device that is a diagnosis target device is abnormal based on the first probability distribution, and an actual device that is a diagnosis target device based on the second probability distribution Determining a common threshold for the first probability distribution and the second probability distribution that minimizes a sum of second probabilities that are probabilities that are determined to be normal;
The diagnostic unit of the device diagnostic apparatus is
Determining whether or not the actual machine is normal based on the magnitude relationship between the common threshold and the diagnosis target data acquired from the actual machine;
A device diagnosis method for the device diagnosis apparatus. For the input data management unit of the device diagnostic device,
Using a normal sample data of a normal device, creating a first probability distribution for determining that the device is abnormal;
Using the abnormal sample data of the device that is abnormal, execute a process of creating a second probability distribution for determining that the device is normal,
For the threshold value determination unit of the device diagnostic apparatus,
A first probability that is a probability that an actual device that is a diagnosis target device is abnormal based on the first probability distribution, and an actual device that is a diagnosis target device based on the second probability distribution Performing a process of determining a common threshold for the first probability distribution and the second probability distribution, which minimizes a sum of second probabilities that are determined to be normal.
Causing the diagnostic unit of the device diagnostic apparatus to execute a process of determining whether or not the actual machine is normal based on a magnitude relationship between the common threshold and the diagnosis target data acquired from the actual machine;
A device diagnosis program for causing the device diagnosis apparatus to function.

Images