RU2782760C1 - Method for monitoring of state of cooling system - Google Patents

Abstract

FIELD: automated control.

SUBSTANCE: invention relates to the field of automated control of technological processes. A method for monitoring of a state of a cooling system includes collection with a given periodicity of current data on operation of a central refrigerating machine (hereinafter – CRM) and consumers, its recording to a database, and provision of availability of preliminary collected archive data. A value of probability of the presence of refrigerant leakage is determined based on current and archive data. The presence of refrigerant leakage in the system is defined as a consequence of that all consumers of cooling energy except for the most remote ones, have stable indicators of a temperature of a cooled medium and performance, while remote consumers show stable growth of the mentioned indicators, or all consumers show growth of the total time of being in a cooling mode, while CRM shows growth of performance. The value of probability of the presence of refrigerant leakage is updated in the database and transmitted to an addressee a signal on change in the mentioned value.

EFFECT: increase in immediacy of reaction to leakage of a refrigerant in a cooling system contour.

9 cl, 3 dwg

Description

Technical fields

The present invention relates to automated process control technologies, namely, to the means of controlling the cooling system. The invention can be used in engineering equipment dispatching systems, building management systems, software and hardware systems for controlling cooling systems.

State of the art

A known method of controlling the cooling system, disclosed in the RF patent for invention No. RU2501715 (published on 12/20/2013, priority dated 09/29/2009, 11/13/2009, PCT application No. WO 2011/038888). The method includes processing signals about the mode of operation of the cooling system and consumers of refrigeration energy. Data on the priority of consumers, depending on the mode of operation, is stored in the database. Depending on the operation mode signals and the priority data, signals are generated that control the supply of refrigeration energy to the consumers.

However, the known method is characterized by low responsiveness to the occurrence of malfunctions in the cooling system, which can lead to the failure of expensive equipment. This is due to the fact that the method does not provide the possibility of timely elimination and localization of the malfunction and does not allow processing data on the functioning of the system, which could indicate the presence of a refrigerant leak in the cooling circuit.

There is a known method for remote monitoring and forecasting the state of individual units and complex technological complexes, disclosed in the RF patent for invention No. RU2677429 (publ. 01/16/2019, priority dated 01/30/2017). The method includes accumulating data on the functioning of the aggregates, obtaining them in the form of sequences of indicators correlated with time points, choosing a model for the functioning of the aggregates, a method for estimating the difference in parameters, a deviation value, and a method for estimating their deviation values. During the operation of the complex, current performance indicators are obtained, deviations from the previously selected functioning model are detected, and a notification signal is generated about the detected deviation and its status (normal/abnormal).

However, the known method is characterized by a narrow scope, limited work with turbine units and complexes of such units. The method does not take into account operating parameters from special areas, and also does not offer integration and methods of communication with refrigeration machines, consumers of refrigeration energy and devices designed to collect data on the functioning of systems consisting of these devices. In addition, the known method uses a rather complex computing apparatus, for the adaptation of which for other areas of application, significant improvements are required.

A control system for controlling a cooling system is known, disclosed in the RF patent for invention No. RU2720585 (published on May 12, 2020, priority dated April 21, 2017, PCT application No. WO 2018/192827). The control system includes a central control unit and object controllers configured to control one of the cooling objects, in particular in relation to the supply of refrigerant to the object's evaporator. The central control unit and object controllers are integrated into a secure communication network with the ability to exchange communication signals.

However, the known method is characterized by low responsiveness to the occurrence of malfunctions in the cooling system, which can lead to the failure of expensive equipment.

In addition, the existing methods of monitoring cooling systems are limited to visual control of the state of the equipment, visual analysis of changes in the parameters of the refrigerant and cooled air, control of the parameters of the refrigerant and cooled air in order to go beyond static boundaries, and installation of additional control systems.

These solutions do not allow to determine the refrigerant leakage in advance and with high accuracy, especially in cases where the cooling system has dozens of consumers located at some distance from each other, for example, across the territory of the enterprise.

Visual monitoring of the state of systems reduces the likelihood of detecting faults, which is due to the human factor. At the same time, there are often difficulties with surveying the routes along which the refrigerant circulates, due to the inaccessibility of the places where the routes pass. In the case of geographically distributed systems, the volume of required labor costs increases many times over.

Static limits for monitoring consumer parameters are selected in such a way as to reduce the number of false alarms, therefore, the output of adjustable consumer parameters (for example, the air temperature inside the consumer) beyond these limits most often occurs either a few hours before a malfunction, or already immediately at the time of a malfunction. .

The installation of additional equipment requires intervention in the operation of the refrigeration unit, additional costs for the installation and maintenance of control systems. For example, when installing a refrigerant leak detector, leak detection is carried out only in the room where it is installed, this method is not suitable if consumers are located, for example, in sales areas with large areas, where the required gas concentration in the event of a leak will not be achieved due to too small volume of gas relative to the volume of the room, in addition, this method is not suitable in case of leakage on routes or connections located on the street.

In view of the foregoing, there is a need for a method for monitoring the condition of cooling systems, which allows to increase the responsiveness to the occurrence of malfunctions in the cooling system.

Disclosure of the essence of the invention

The technical problem underlying the present invention is to provide the possibility of monitoring the state of the cooling system.

The technical result achieved by the implementation of the present invention is to increase the responsiveness to the occurrence of a refrigerant leak in the cooling system circuit.

The invention discloses a computer-implemented method for monitoring the state of a cooling system consisting of a central refrigeration machine (CCM) and associated refrigeration energy consumers that provide air cooling using the cold generated by the CCM. The method includes the steps in which:

- with a given frequency, they collect current data on the work of the CCM and consumers and record them in the database;

- ensure the availability of previously collected archival data on the work of the CCM and consumers;

- on the basis of the mentioned current and archive data, time series are formed, each of which reflects the time distribution of a separate parameter of the operation of the cooling system, while at least one parameter is used as the mentioned operation parameters, which is selected from the group including: air temperature in each consumer, the number of consumers in the cooling mode, the time spent by each consumer in the cooling mode, the value of the performance of the CCM compressor, the value of the pressure at the inlet to the CCM compressor;

- carry out pre-processing of time series, including procedures for smoothing, filling gaps in the data and normalization;

- determine the values of the regression coefficients for each processed time series;

- transmit the values of the regression coefficients to the input of the computational model, which reflects the relationship between the probability of a refrigerant leak in the cooling system, the parameters of its operation and the remoteness of each consumer from the CCM;

- determining the value of the refrigerant leakage probability, updating said value in the database and transmitting a signal to the addressee about the change in said value.

Additional features and advantages of the present invention appear in the following particular embodiments.

In particular, the controllers of the CCM and consumers are the source of current data on the operation of the cooling system.

In particular, the source of current data on the operation of the cooling system is the monitoring unit integrated into the said system and connected to the controllers of the CCM and consumers.

In particular, the collected current and pre-collected archive data are provided with a time stamp of their collection.

In particular, data smoothing is carried out by at least one method selected from the group consisting of: exponential smoothing, moving average method, Holt-Winters method.

In particular, data normalization is carried out by one of the methods selected from the group consisting of division by maximum value, minimax normalization, or Z-normalization.

Specifically, the destination to which the refrigerant leakage probability value change signal is signaled is the refrigeration system monitoring system operator notification module.

Specifically, the destination to which the refrigerant leakage probability value change signal is signaled is the data visualization unit.

The claimed method makes it possible to detect the presence of a refrigerant leak in the circuit of the cooling system, which consists of a large number of consumers connected to the central refrigeration machine, without the need for visual control of the operation of the equipment. The computational model used for this reflects the relationship between the operating parameters of the cooling system, the distance of consumers from the central refrigeration machine, and the probability of a leak.

The presence of a refrigerant leak is defined as a consequence of a situation in which all but the most remote refrigeration consumers have stable temperatures of the refrigerated medium and capacity, while remote consumers show a small but stable increase. As additional factors for the presence of a leak in the system, there is an increase in the total time spent by all consumers in the cooling mode with a simultaneous increase in the performance of the central refrigeration machine.

Brief description of the drawings

FIG. 1 illustrates an example of a cooling system;

FIG.2 illustrates a block diagram of the software implementation of monitoring the state of the cooling system;

FIG. 3 illustrates a flowchart of a method for monitoring the state of a cooling system.

Description of embodiments of the invention

In accordance with FIG.1, the cooling system includes a central chiller (CCM) 101 and associated consumers of refrigeration energy 102. The central chiller 101 provides the generation of refrigeration energy. Consumers of refrigeration energy 102 provide cooling of air volumes using the cold generated by the CCM 101. The refrigeration process is a constantly repeating cycle in which the refrigerant changes from a liquid state to a gaseous state. The refrigerant is the working substance of the refrigeration cycle, the main characteristic of which is a low boiling point. As refrigerants, hydrocarbon compounds are used, which, in particular, contain atoms of chlorine, fluorine, bromine. In addition, ammonia, carbon dioxide, propane, air are used as a refrigerant. In case of damage in the circuit of the cooling system, the tightness of the routes through which the refrigerant is circulated, the amount of refrigerant in the system gradually decreases, reducing the cooling capacity of the system.

The refrigeration energy generated by the CCM 101 is delivered to consumers, which provide cooling of the object to be cooled. The refrigerated object can be a room, a freezer, an equipment, but is not limited to these examples of refrigerated objects. The consumer 102 may be an arbitrary type of consumer, including a heat exchanger (evaporator). In the consumer 102, the refrigerant prepared by the refrigeration machine 101 evaporates and takes away the excess heat of the cooled medium.

The central refrigeration machine 101 is connected to the CKM controllers 103, which form the CKM automation system. Examples of controllers CKM 103 are specialized controllers from Danfoss, Carel, Dixell, but are not limited to these examples of controllers. For the purpose of controlling the CCM 101, controllers of the CCM 103 from different manufacturers can be used, since the control algorithms and the composition of the controlled equipment in the consumers and the refrigeration machine may be different. Refrigeration consumers 102 are associated with consumer controllers 104. Consumer controllers 104 can be configurable controllers of any manufacturer or destination. This is due to the fact that the control algorithms in consumers 102 are relatively simpler than in the refrigeration machine 101, and the number of consumers 102 may vary depending on the cooling capacity of the cooling system.

Controllers CKM 103 and consumer controllers 104 may be connected by information interfaces. Alternatively, the controllers CKM 103 and controllers consumers 104 may not be connected by information interfaces. In this case, said controllers 103, 104 operate autonomously, solving the tasks of regulating the ambient temperature and regulating the parameters of the refrigerant, respectively.

The CKM controllers 103 and consumer controllers 104 may be capable of supporting open communication communication protocols used in automation systems. The mentioned communication protocols can be represented by Modbus, BACnet, Lonworks protocols, but are not limited to these examples of protocols. The CM controllers 103 and consumer controllers 104 are not necessarily capable of supporting the above communication protocols. In particular, if intermediate equipment that is otherwise compatible with said controllers is used to collect controllers 103, 104, support for open communication protocols then becomes optional.

Refrigerant leakage occurs due to depressurization of the cooling circuit. The causes of depressurization may be external damage to the route or circuit elements, poor-quality connections, loosely closed service ports. A leak leads to a decrease in the volume of refrigerant and, as a result, to a decrease in the amount of refrigerant in the liquid aggregate state supplied to consumers, and the further the consumer is from the central refrigeration machine, the higher the likelihood of a shortage of refrigerant for him in the event of a leak. The lack of refrigerant leads to a decrease in the cooling capacity of an individual consumer and an increase in the temperature of the cooled medium.

Most often, with small leaks, most of the consumers continue to operate normally, while consumers located at a distance from the central refrigeration unit begin to experience a lack of refrigerant. This leads to a gradual increase in the operating time of these consumers in cooling mode, and as a result, to an increase in the temperature of the cooled medium.

In accordance with FIG.2, the software implementation of monitoring the state of the cooling system has a modular structure.

Said modules may be implemented by various combinations of hardware and software. Each module may be implemented using at least one processor, RAM and/or non-volatile memory capable of operating program modules, and at least one physical communication interface capable of exchanging data and commands between controllers and at least one server device through communication channels. Said modules may be any number of computer program modules desired and may use any desired number of libraries, shells, applications, software packages, depending on the particular function or method step implemented by the particular module. Computer program modules may be implemented as machine code or as program text in a programming language such as C, C++, C#, Java, Python, Perl, Ruby, but not limited to these examples of programming languages. As database management systems (DBMS), which are necessary for information support of the method, PostgreSQL, InfluxDB, MySQL, Prometheus DBMS can be, but are not limited to these examples of DBMS.

The software implementation of monitoring the state of the cooling system includes a module 'Backend' 201, designed to organize the interaction of the components of the software system. An example implementation of module 201 would be Django software. The module 201 is linked to the 'Frontend' module 202 for data visualization and user interaction. An example implementation of module 202 may be the Grafana software. Modules 201, 202 may be associated with analysis modules 203 that process data from monitoring the condition of the refrigeration system. Modules 201, 202, 203 are associated with databases 204 necessary for the operation of said modules.

The module 201 can be connected to the 'ETL' 205 (Eng. Extract, Transform, Load) module, which provides extraction and transformation of data received from the monitoring units 206. The monitoring units 206 are designed to receive data from the controllers 103, 104. Data collection can be carried out by transferring files of known structure, such as those created using the XML markup language. In addition, data collection can be carried out by sending requests to the application programming interface of the monitoring unit or controller, for example, using the REST API system interaction style.

Module 201 may be connected directly to controllers 103, 104 to provide data collection. Data collection is possible using open communication protocols. For example, when using the Modbus protocol, Zabbix software can be used to provide data collection.

In accordance with FIG. 3, the method for monitoring the state of the cooling system includes the following steps.

At step 301, current data on the operation of the CHM and consumers are collected at a specified frequency and recorded in the database 204. The collected data with time stamps are stored in the database 204, where the table fields are the equipment parameters. Additionally, for each consumer, the parameters associated with the mode of its operation are calculated: the total time ( T ) of the consumer being in the cooling mode for the previous period, for example, for the last hour; as well as the current number ( N ) of consumers in cooling mode. This data is also stored in the database 204 with timestamps of the data received.

At step 302, the pre-collected historical data on the operation of the CCM and consumers are made available.

At step 303, based on the mentioned current and archived data, time series are formed, each of which reflects the time distribution of a separate parameter of the cooling system operation, while at least one parameter is used as the mentioned operation parameters, which is selected from the group including: air temperature in each consumer, the number of consumers in the cooling mode, the time spent by each consumer in the cooling mode, the value of the productivity of the CCM compressor, the value of the pressure at the inlet to the CCM compressor.

The likelihood of a refrigerant leak is assessed based on current and historical data. The change in air temperature in the cooled volume with a small leakage is a rather inertial process, therefore it is sufficient to calculate the probability estimate with a frequency equal to, for example, 6 hours (Freq = 6 hours), evaluating the data for the previous 3 days (Hist = 3 days). The parameters Freq and Hist must be determined empirically on several samples and can be optimized if the method is applied to cooling systems with characteristics different from the standard ones. The algorithm runs at the specified frequency Freq and unloads the last parameters from the database for the selected period Hist.

To calculate the score, the following parameters are unloaded from the database:

- total time (T) of the consumer being in the cooling mode for the last hour;

- number (N) of consumers in cooling mode;

- temperature (t) of air in the cooled volume of the consumer, for each consumer;

- productivity (P) of the compressor of the central refrigerating machine (CHM);

- pressure (p) of the refrigerant at the compressor inlet.

At step 304, data is pre-processed. Pre-processing includes smoothing time series, filling gaps in them, and normalizing data.

Current data may have single outliers and various fluctuations. Estimating the likelihood of a leak, in contrast, works with longer-term changes. To eliminate the influence of fluctuations in the original sample, smoothing is applied to it by the simple moving average (SMA) method according to the formula:

whereSMA _t - the value of the simple moving average at the pointt;n - the number of samples in the selected window;p _ti - parameter value at pointti.

The method showed the most stable results for assessing the probability of a leak with a smoothing window equal to the period Hist . The size of the smoothing window depends on the period of saving data in the database. In our case, the sampling period was 2 minutes, respectively, the window size was 2160 samples.

The smoothing procedure can be carried out using any software package suitable for this purpose, such as Pandas, but not limited to this example. You can also use exponential smoothing or the Holt-Winters method for smoothing. The latter option may give the best result, as it takes into account the seasonal component and the trend of the time series.

The initial data contains at least partially null values, which can be caused, among other things, by data loss due to an unstable network connection. In particular, said null values may be replaced by nearest adjacent values that occur in the sample. In addition, these empty values can be replaced by the value of the arithmetic mean of the data sample.

After smoothing procedures and processing of missing values, performed for all estimated parameters, the data is normalized according to the formula:

where z _i is the standardized estimate of x _i , x _i is the value of the standardized parameter, μ is the mean value and σ is the standard deviation of the parameter in the normalized sample.

At step 305, the values of the regression coefficients for each time series are determined. The determination of said value can be carried out using an arbitrary software package suitable for this purpose, such as Sklearn, but not limited to this example.

At step 306, the obtained values of the regression coefficients are transmitted to the input of a computational model that reflects the relationship between the probability of a refrigerant leak in the cooling system, its operation parameters, and the distance of each consumer from the CCM.

The function used in the computational model can be written analytically as follows:

where ( T _i , t _i ), are the previously calculated regression coefficients for the air temperature in the consumer and the time the consumer is in cooling mode, N is the regression coefficient for the number of consumers in cooling mode, R is the regression coefficient for the compressor performance, p is the regression coefficient for the pressure at the compressor inlet.

For the analysis of consumers, several consumers with the largest (non-negative) terms are selected, for example, 3 consumers. The assignment of weight coefficients k _i for consumers is carried out in accordance with the degree of their remoteness from the CCM. In particular, the weight coefficient k for three consumers can take the following values: 0.1, 0.15 and 0.25 for the most distant one.

In step 307, a refrigerant leakage probability value is determined, the value is updated in the database, and the value is signaled to the destination.

It is worth noting that the above description is given by way of example and should not be construed as limiting the scope of protection of the present invention to be determined solely by the scope of the appended claims.

Although the particular embodiments described above are given with reference to specific steps performed in a certain order, it should be obvious that these steps can be combined, separated, or their order can be changed without deviating from the essence of the present invention. Accordingly, the order and grouping of steps is not limiting to the spirit of the present invention.

Claims (20)

1. A computer-implemented method for monitoring the state of a cooling system consisting of a central refrigeration machine (CCM) and associated consumers that provide air cooling using the cold generated by the CCM, while in the method: - with a given frequency, they collect current data on the operation of the CCM and consumers and record them in the database; – ensure the availability of previously collected archival data on the work of the CCM and consumers; - mathematical processing of current and archived data is carried out, during which the value of the probability of a refrigerant leak is determined, depending on the operating parameters of the cooling system and the remoteness of each consumer from the CCM, and the mentioned current, archived data reflect the operating parameters of the cooling system, which are selected from the group including : air temperature in each consumer, the number of consumers in the cooling mode, the time spent by each consumer in the cooling mode, the value of the productivity of the CCM compressor, the value of the pressure at the inlet to the CCM compressor; – at the same time, the presence of a refrigerant leak in the system is determined as a consequence of a situation in which all consumers of refrigeration energy, except for the most remote ones, have stable indicators of the temperature of the cooled medium and productivity, while remote consumers show a stable increase in these indicators; - while the presence of a refrigerant leak in the system is determined as a consequence of a situation in which all consumers show an increase in the total time spent in cooling mode, while the CCM shows an increase in productivity; – update in the database the probability value of the presence of a refrigerant leak in case at least one of the mentioned situations is observed in the system; – transmit a signal to the addressee about a change in the mentioned value of the refrigerant leakage probability. 2. The method according to claim 1, in which the source of current data on the operation of the cooling system is the controllers of the CCM and consumers. 3. The method according to claim 1, in which the source of current data on the operation of the cooling system is a monitoring unit integrated into the said system, associated with the controllers of the CHM and consumers. 4. The method according to claim 1, in which the collected current and previously collected archived data is provided with a timestamp of its collection. 5. The method according to claim 1, in which the following actions are performed during mathematical processing of current and archived data: - form time series, each of which reflects the distribution in time of a separate parameter of the cooling system; – carry out preliminary processing of time series, including smoothing, filling gaps in data and normalization procedures; – determine the values of the regression coefficients for each processed time series; – transfer the values of the regression coefficients to the input of the computational model, which reflects the relationship between the probability of a refrigerant leak in the cooling system, its operation parameters and the distance of each consumer from the CCM. 6. The method according to claim 5, wherein the data smoothing is carried out by at least one method selected from the group including: exponential smoothing, moving average method, Holt-Winters method. 7. The method according to claim 5, in which the normalization of the data is carried out by dividing by the maximum value. 8. The method of claim 1, wherein the destination is a cooling system monitoring system operator alert module. 9. The method of claim 1, wherein the target is a data visualization module.

Images