ITUA20163737A1 - Procedure for detecting and diagnosing faults or anomalies in a liquid chiller device or in a heat pump. - Google Patents

Description

"Procedure for detecting and diagnosing faults or anomalies in a liquid chiller device or in a heat pump".

The present invention relates to a method for detecting and diagnosing faults or anomalies in a liquid chiller device or in a heat pump.

More specifically, the method according to the present invention is carried out for the purposes of detecting and diagnosing faults or anomalies with reference to so-called HVAC devices, i.e. relating to heating, ventilation (Ventilating) and air conditioning ( Air Conditioning) within usually large structures such as hospitals, offices, industries or the like, but also within residential buildings and the like.

It has been found that, with reference to said devices, faults in particular relating to the components of the chillers (so-called chillers) essentially affect 42% of the service requests forwarded and 26% of the overall repair costs. Consequently, it is a sensibly perceived need to optimize the operating conditions of the aforementioned chillers and heat pumps, as well as to promptly identify any faults thereto, in order to be able to proceed in a timely and punctual manner to the related repair.

In this way, in addition to ensuring greater longevity of the liquid chiller device or heat pump, it is also possible to optimize its overall efficiency and achieve significant energy savings, as well as costs.

For these purposes, it is not only of interest to carry out the detection and diagnosis of faults or so-called "catastrophic" anomalies (for example the breakdown of the engine or other problems with the mechanical and electrical parts), which are essentially simple to detect as they necessarily involve the interruption of the operational functioning, but also the failures or anomalies connected to the degradation of one or more components which, even if they do not cause the interruption of the operations, however, can be harbingers of significant impairments in performance, as well as of perceptible increases in consumption overall. Faults and anomalies of this type are, for example, the reduction of water in the heat exchangers, the loss or overload of refrigerant gas, or a continuous loss of oil or lubricant.

Obviously, these types of non-catastrophic failures and anomalies are less easy to detect, as the chiller or heat pump device probably continues to operate in conditions of apparent normality.

In the state of the art, different types of procedures are currently known for detecting and diagnosing faults or anomalies (known as Fault Detection and Identification - FDI) in liquid chilling devices or heat pumps.

Said known procedures are primarily aimed at carrying out the timely detection and diagnosis of faults or anomalies in liquid chiller devices or heat pumps in an automated way, for the interception of any presence of faults or, more generally, anomalies, and provide operators with information relating to the possible causes of the same, thus making the necessary maintenance intervention easier and more timely.

With reference to said known processes, it is noted that the detection of faults or anomalies is usually based on the generation of residues, or signals which reflect the presence of the anomalies; in particular, an alarm is usually activated when the residues exceed a predefined threshold.

According to said known procedures (mostly known as model-based), the production of residues is obtained through the use of mathematical models that describe the normal operation of the system (therefore, in a condition of absence of failures); examples of this type are representations in state space or transfer functions to which failure and anomaly detection techniques based on Kalman filters or inserted in a scheme with a PID-type controller can be applied.

From document no. CA 2344 908 A1 a method is known for detecting a fault in the HVAC system, including sensors for measuring the operation of a condenser, a compressor, an evaporator, and a chiller. The known system is equipped with a thermodynamic pre-processor to calculate the characteristic quantities (CQ's) of a plurality of measured inputs; a table of guide cases is also provided for storing a plurality of first configurations of QC values generated by the thermodynamic pre-processor in an initial period in which the HVAC system is supposed to operate in a fault-free condition. An interpolator means is further provided for interpolating a set of guide case QC values, stored in the guide case table, for a predetermined set of measured inputs. A fault detector detects the presence of a fault when the difference between real QC values and guide-case QC values exceeds a predetermined threshold with reference to at least one of said QC values. A fault classifier classifies a certain detected fault according to which QC values are exceeding said threshold value.

In addition, document no. US 2006/0042277 A1 discloses a method for detecting at least one system failure, including the steps of identifying the sensors available within the system, in which the identified sensors generate sensor data relating to the operating characteristics of the system; analyzing sensor data to identify indicative patterns of at least one fault state; determine a failure report based on identified patterns; comparing at least one operating characteristic of the system with a failure report, to identify a system failure; generate at least one fault code, identifying at least one fault mode, when a system fault is detected.

Furthermore, from document no. WO 2012/118550 A1 a method is known for diagnosing a fault condition in an air conditioning system, the method comprising the steps of: receiving a current parameter from an air conditioning system in a fault state, the current parameter comprising a plurality of current data measured and estimated by the air conditioning system; determining a first set of transition probabilities based on current parameters; determining a second set of transition probabilities based on historical parameters from the air conditioning system operating under normal conditions, the historical parameters including a plurality of historical data measured and estimated by the air conditioning system; calculate an anomaly value for the conditioning system, for the first set of transition probabilities and the second set of transition probabilities and, when the anomaly value is above a predetermined threshold, automatically generate the diagnosis of a first problem causing the fault condition.

On the other hand, document no. US 2005/0251364 A1 discloses a system, and related method, for diagnosing sensor failures in an air conditioning system, comprising the steps of generating a model and determining an expected measurement value of a sensor, detecting a real value by means of the sensor, compare said expected value and said real value and determine the difference between the two.

Further procedures for the detection and diagnosis of faults or anomalies in liquid chillers or heat pumps are known from documents no. US 2012/0125592 A1, US 2011/0173496 A1, WO 2011/049890 A1 and US 2006/007496 A1.

However, it is noted that said known processes have certain kinds of drawbacks. In the first place, it is noted that the elaboration of these models is essentially complex and also requires the use of sensitive approximations, which contribute to jeopardizing the reliability and effectiveness of the detection and diagnosis of said faults or anomalies, especially when dealing with particularly complex devices, having an essentially high number of variables.

For the same reasons, by means of said known procedures, it is not always possible to achieve a precise and timely intervention towards the detected fault or anomaly.

In addition, said known processes are strictly referred to single devices or systems; therefore, the relative implementation with respect to devices or systems having different characteristics involves adaptation interventions that are often very complex.

Likewise, with reference to these known procedures, the updating of parameters and models is significantly complex, with what follows in terms of further prejudice to the reliability of the final detection and diagnosis results.

On the other hand, it is noted that the use of these models, for the purpose of detecting and diagnosing faults or anomalies, does not always allow the achievement of an appreciable level of precision and optimization of the result obtained.

Moreover, the hardware component necessary for the implementation of said known procedures is essentially complex.

The present invention, starting from the notion of the aforementioned drawbacks, intends to remedy them.

An object of the present invention is to provide a method for detecting and diagnosing faults or anomalies in a liquid chiller device or in a heat pump, which allows to detect and diagnose said faults or anomalies, greatly limiting the recourse to operations. approximation and therefore with a high level of reliability and precision of the results obtained, even in the case of particularly complex devices, with an essentially high number of variables.

Another object of the present invention is to provide a process as said, which can be easily carried out, without this involving particularly significant adaptation interventions, with reference to devices having characteristics that are also significantly different from each other.

A further object of the present invention is to provide a method as mentioned, in which the different parameters involved in the processing operations are easy to update, so as to optimize the reliability and effectiveness of the final detection and diagnosis results.

On the other hand, the purpose of the present invention is to provide a procedure as specified, which allows the operator to intervene in a precise and timely manner against the detected fault or anomaly.

Moreover, it is also an object of the present invention to provide a method as mentioned which requires, for the purposes of its implementation, an essentially simple hardware component.

In view of these purposes, the present invention provides a method for detecting and diagnosing faults or anomalies in a liquid chiller device or in a heat pump, the essential characteristic of which forms the subject of claim 1.

Further advantageous characteristics of the aforesaid process are described in the subsequent dependent claims.

The aforementioned claims are intended as fully reported here.

The present invention will become clearer from the detailed description which follows, with reference to the attached drawing, illustrating an exemplary embodiment of the method for detecting and diagnosing faults or anomalies in a liquid chiller device or in a heat pump, in which :

- Figure 1 is a flow diagram relating to the implementation of a multivariate analysis system of the PCA (Principal Component Analysis) type for the identification and diagnosis of faults and anomalies in a heat pump, in an exemplary embodiment of the procedure for the detection and diagnosis of faults or anomalies in a liquid chiller device or in a heat pump, according to the present invention;

Figure 2 is a functional diagram relating to the implementation of a preliminary phase of processing signals acquired by means of supervision relative to the current operating state of said liquid chiller device or said heat pump and sampled by dedicated software means, for the exclusion of any signal sections that are not of interest or connected to erroneous measurements.

In what follows, the description of an exemplary embodiment of the method for detecting and diagnosing faults or anomalies in a liquid chiller device or in a heat pump, according to the present invention, is provided.

By way of example, said method according to the invention is implemented for the detection and diagnosis of faults or anomalies in liquid chiller devices or in heat pumps, presenting:

1) different refrigerator lay-out configurations:

to. single circuit with one or more compressors per circuit;

b. multi-circuit with one or more compressors per circuit;

2) different refrigerants:

to. belonging to the halogenated hydrocarbon family;

b. the most common azeotropic mixtures;

c. the most common zeotropic blends;

d. organic (R717, R744, etc.)

3) different types of the following components:

And. evaporator / I.

the. single circuit shell and tube;

ii. multi-circuit shell and tube;

iii. plates

f. compressor / I.

the. scrool;

ii. screw;

iii. Turbocor;

iv. alternative;

g. capacitor

the. with air (battery / and alletta, battery / and allure to microchannels, battery / and allure to mini-channels);

ii. liquid (single-circuit or multi-circuit shell and tube, with plates);

h. lamination valves

the. electronic expansion valves (EEV);

ii. mechanical electronic expansion valves;

the. constant back pressure valves

the. electronic valves;

ii. mechanical valves;

j. flash-gas valves

the. electronic valves;

ii. mechanical valves;

k. pumps

the. at constant flow;

ii. variable flow;

L. condenser fans

the. at constant flow;

ii. variable flow;

m. heat recovery exchangers (partial, complete);

the. single circuit shell and tube;

ii. multi-circuit shell and tube;

iii. plate;

n. liquid subcoolers;

or. regenerative exchangers;

p. finned coils for making the free-cooler;

q. systems for adiabatic air cooling.

Said procedure includes, according to the invention, the following steps consisting of:

- generate, by means of dedicated software means and resident in electronic processor means, a plurality of indicators of correct operation of said liquid chiller device or said heat pump, on the basis of values relating to one or more normal operating conditions in the absence of anomalies, - saving said plurality of indicators of correct functioning in a database, resident in said electronic processor means;

- elaborating by means of said dedicated software means, on the basis of said plurality of indicators, a statistical model of correct operation of said liquid chiller device or said heat pump, in normal operating conditions in the absence of anomalies;

- activating supervision means of the current and real-time operating state of said liquid chiller device or said heat pump, for a supervision period of predetermined duration, said supervision means being operatively connected with respect to said electronic processor means ;

- acquiring, by means of said supervision means, a plurality of measurement signals of the current operating state of said liquid chiller device or said heat pump, and forwarding the same to said electronic processor means;

- processing, by means of said dedicated software means and for each a supervision period of predetermined duration, said measurement signals, to obtain a plurality of corresponding indicators of current operation of said liquid chiller device or said heat pump, in said period of supervision;

- storing said indicators of current operation in said database; - performing, by means of said dedicated software means, a comparison between said indicators of correct operation and said indicators of current operation, for the detection and diagnosis of faults or anomalies in said liquid chiller device or in said heat pump.

Advantageously, according to the present invention, said plurality of indicators of correct functioning is generated, separately or jointly, on the basis of:

- plate values of said liquid chiller device or said heat pump;

- values detected during the testing of said liquid chiller device or said heat pump;

- first start values of said liquid chiller device or said heat pump;

- values detected during the correct operation, in the absence of anomalies, of said liquid chiller device or said heat pump.

For example, for the purpose of generating said plurality of indicators of correct functioning, a multivariate analysis system of the PCA (Principal Component Analysis) type is used, capable of determining a correlation between a multiplicity of variables, assuming the absence of independent factors and contingents, according to the following algorithm:

Where is it:

Number of observations for each signal

Number of acquired signals

Observation matrix

Sample covariance matrix

Number of main components

Matrix of the eigenvectors of ∑ relative to the principal eigenvalues

Matrix of the eigenvectors of ∑ relative to the residual eigenvalues Matrix of the scores relative to the principal components Matrix of the scores relative to the residual components Model of X estimated by the principal components Model residual in the estimate of X

Diagonal matrix containing the principal eigenvalues of ∑

Using the following algorithm:

the sample covariant matrix of X is obtained, where: - the columns of are the eigenvectors of associated with its highest eigenvalues l, while

- the columns of are the remaining eigenvectors of.

In this way, the PCA divides the measurement space into two orthogonal subspaces:

- a Principal Component Subspace - PCS, and - a residual subspace (Residual Subspace - RS),

where the first (PCS) includes the normal variations of the system, while the second (RS) includes the anomaly values.

On the other hand, for the hypothesis in which independent and contingent elements arise in the operation of said liquid chiller device or heat pump, the activation of a command signal aimed at defining the presence of a new normal operating condition is envisaged. of said liquid chiller device or heat pump, as such not symptomatic of the presence of a fault or anomaly situation. On the other hand, in such a hypothesis, the updating of said indicators of correct operation with the data corresponding to said new operating mode is therefore provided so that, upon the occurrence of further operating conditions of said liquid chiller device or pump corresponding to said normal operating mode the risk of activation of false alarm signals is avoided.

The definition of said new operating step is preferably carried out automatically, on the basis of the measurement signals associated with the exceeding of one or more threshold values.

It is obviously understood that said new operating phase can also be defined manually, by modifying, by the operator, the configuration parameters of said chiller device or heat pump (for example, changing the condensing pressure set-points or of the LWT. Etc.)

Said process according to the present invention therefore comprises processing exclusively of the data driven type, as such based solely on experimental data, which can be easily implemented with reference to different types of liquid chiller devices or heat pumps, and such as to allow obtaining results of detection and diagnosis characterized by a high level of reliability, thus overcoming the drawbacks of the procedures known to the state of the art.

In particular, in the present embodiment of the invention, said process carries out elaborations of the Multivariate Statistical Process Monitoring (MSPM) type for the generation of statistical control indices capable of monitoring the behavior of said liquid chiller device or heat pump, with a high level of precision and reliability.

On the other hand, the use of multivariate analyzes such as, for example, the Principal Component Analysis (PCA), allows easy identification of the operating status of the liquid chiller or heat pump device, effectively intercepting the presence of any faults or anomalies.

An example of implementation of a PCA multivariate analysis system for the identification and diagnosis of faults and anomalies in a heat pump is provided in figure 1.

In order to facilitate the discrimination between observations of correct functioning and anomaly, the processing takes place according to a modified version of the PCA. More specifically, system measurements are conveyed into multiple PCA models, where the number of models is defined based on the number and type of components of said liquid chiller device or heat pump to be monitored and the number and type of potential failures. to be detected.

The structure of the individual PCA models is refined in such a way as to be minimally affected by the intrinsic non-linearities of the system.

According to the present embodiment of the procedure, the identification of a fault or anomaly (so-called Fault Detection - FD) takes place according to the following algorithm, where the value x refers to a single monitoring operation of the operating conditions of the refrigeration device. liquid or heat pump

More specifically, the identification of the fault or anomaly can be achieved through Fault Detection indices that monitor the operation of the device such as, for example, Hotelling T2 type indices, for the identification of variability in the PCS:

and indices of the Square Prediction Error (SPC) type, for the representation of the variability in the RS:

It is also possible to process a combined index, where:

The value of x is to be considered representative of a normal operating condition of said liquid chiller device or in a heat pump when:

where they represent statistical thresholds whose expressions are known in the context of multivariate statistical analysis (MSPM).

It should be noted that the aforementioned indices of FD are expressed in quadratic forms.

The related notation can also be simplified by considering a single general index, according to the following algorithm:

In this case, a fault or anomaly situation is identified when one or more indices, or a combination of them, exceed the relative control limits.

When a fault or anomaly condition is identified, it is therefore necessary to diagnose its causes, in particular by verifying which variables reflect the occurrence of faults or anomalies the most.

For this purpose, by way of example, it is possible to use RBC (Reconstruction-Based Contribution) analyzes.

It is obviously understood that further different approaches, as long as they are suitable for the purpose, also fall within the scope of the present invention.

Conveniently, according to the present invention, when said step of carrying out by said dedicated software means, a comparison between said indicators of correct operation and said indicators of current operation, for the detection and diagnosis of faults or anomalies in said liquid chiller device or in said heat pump, it has a negative result, said procedure also comprises the following steps consisting in:

- classifying as non-anomalous, in said database, said current operating indicators, relating to said predefined supervision period;

- updating and storing, in said database, said indicators of correct functioning, corresponding to operating conditions classified as non-anomalous, with the values corresponding to said operating conditions classified as non-anomalous.

In this way, said indicators of correct operation are regularly updated on the basis of current operating values of the liquid chiller device or heat pump.

On the other hand, when said step of carrying out said dedicated software means, a comparison between said indicators of correct operation and said indicators of current operation, for the detection and diagnosis of faults or anomalies in said liquid chiller device or said pump heat, has a positive outcome, said procedure includes the following phases consisting of:

- classifying as anomalous, in said database, said current operating indicators, relating to said predefined supervision period;

- activate a diagnosis phase, on the basis of said current operating indicators classified as anomalous, for the diagnosis of said faults or anomalies.

For the reasons set out above, said diagnosis of said faults or anomalies has a high level of reliability and precision, in order to allow the operator to intervene in a completely timely and effective manner on the fault or anomaly itself.

For the purposes of further optimizing the phase of detection of faults or anomalies (Fault Detection - FD), avoiding any errors or deviations connected to the presence of noises or other disturbing components, said method according to the present invention comprises, for each supervision period, the following stages of:

- filtering said current operating indicators, obtained following the processing of said acquired measurement signals;

- identifying, in said database, new stationary states of said current operating indicators, filtered.

More particularly, according to the invention, said step of processing, by means of said dedicated software means and for each a supervision period of predetermined duration, said measurement signals, comprises a preliminary step of processing the sampled signals by means of said dedicated software means, for the exclusion of any signal sections that are not of interest or connected to erroneous measurements (filtering or Steady State Detection "SSD"), in order to implement their reliability and minimize the overall storage burden (so-called clustering).

An example of implementation of the aforementioned preliminary phase is provided in Figure 2.

Conveniently, according to the invention, said electronic processor means, operatively connected with respect to said liquid chiller device or said heat pump, are located in a remote location with respect to said liquid chiller device or said heat pump.

Alternatively, said electronic processor means, said liquid chiller device or said heat pump, are located on board of said liquid chiller device or said heat pump.

As evident in the light of what has been considered up to now, said method for detecting and diagnosing faults or anomalies in a liquid chiller device or in a heat pump, according to the present invention, allows to detect and diagnose said faults or anomalies. greatly limiting the use of approximation operations and therefore with a high level of reliability and precision of the results obtained, even with reference to particularly complex devices, equipped with an essentially high number of variables.

Moreover, said method as said, can be easily carried out, without this involving particularly significant adaptation interventions, with reference to devices having characteristics that are also significantly different from each other.

In addition, with reference to said process as mentioned, it is noted that the different parameters involved in the processing operations are easy to update, so as to optimize the reliability and effectiveness of the final detection and diagnosis results.

On the other hand, said procedure as specified, allows the operator to intervene in a precise and timely manner against the detected fault or anomaly. Moreover, said process as mentioned requires, for the purposes of its implementation, an essentially simple hardware component.

As it appears from the foregoing, the present invention allows to achieve the purposes set out in the introduction in a simple and advantageous way.

It is to be understood that there may be other embodiments that pertain to the same inventive core, all falling within the scope of the claims set out below.

Claims (10)

CLAIMS 1. Procedure for detecting and diagnosing faults or anomalies in a liquid chiller device or in a heat pump, comprising the following steps consisting of: - generate, by means of dedicated software means and resident in electronic processor means, a plurality of indicators of correct operation of said liquid chiller device or said heat pump, on the basis of values relating to one or more normal operating conditions in the absence of anomalies; - saving said plurality of indicators of correct functioning in a database, resident in said electronic processor means; - elaborating by means of said dedicated software means, on the basis of said plurality of indicators, a statistical model of correct operation of said liquid chiller device or said heat pump, in normal operating conditions in the absence of anomalies; - activating supervision means of the current and real-time operating state of said liquid chiller device or said heat pump, for a supervision period of predetermined duration, said supervision means being operatively connected with respect to said electronic processor means ; - acquiring, by means of said supervision means, a plurality of measurement signals of the current operating state of said liquid chiller device or said heat pump, and forwarding the same to said electronic processor means; - processing, by means of said dedicated software means and for each a supervision period of predetermined duration, said measurement signals, to obtain a plurality of corresponding indicators of current operation of said liquid chiller device or said heat pump, in said period of supervision; - storing said indicators of current operation in said database; - performing, by means of said dedicated software means, a comparison between said indicators of correct operation and said indicators of current operation, for the detection and diagnosis of faults or anomalies in said liquid chiller device or in said heat pump. 2. A method according to claim 1, wherein said step of carrying out, by means of said dedicated software means, a comparison between said indicators of correct operation and said indicators of current operation, for the detection and diagnosis of faults or anomalies in said refrigerating device liquid or in said heat pump, has a negative outcome, characterized by the fact that it includes the following phases: - classifying as non-anomalous, in said database, said current operating indicators, relating to said predefined supervision period; - updating and storing, in said database, said indicators of correct functioning, corresponding to operating conditions classified as non-anomalous, with the values corresponding to said operating conditions classified as non-anomalous. 3. A method according to claim 1, wherein said step of carrying out said dedicated software means, a comparison between said indicators of correct operation and said indicators of current operation, for the detection and diagnosis of faults or anomalies in said liquid chiller device or said heat pump, has a positive outcome, characterized by the fact that it includes the following phases: - classifying as anomalous, in said database, said current operating indicators, relating to said predefined supervision period; - activate a diagnosis phase, on the basis of said current operating indicators classified as anomalous, for the diagnosis of said faults or anomalies. 4. Process according to claim 1, characterized in that it comprises, for each supervision period, the following steps: - filtering said current operating indicators, obtained following the processing of said acquired measurement signals; - identifying, in said database, new stationary states of said current operating indicators, filtered. 5. Process according to claim 1, characterized in that it comprises generating, by means of said dedicated software means and resident in said electronic processor means, said plurality of indicators of correct functioning by implementing a multivariate analysis of the PCA (Principal Component Analysis) type . 6. Process according to claim 5, characterized in that it comprises implementing said multivariate analysis of the PCA (Principal Component Analysis) type according to a plurality of PCA models, the number of said PCA models of which is defined in relation to the number and type of components of said liquid chiller device or heat pump to be monitored and the number and type of potential faults to be detected. 7. Process according to claim 1, characterized in that said step of processing, by means of said dedicated software means and for each a supervision period of predetermined duration, said measurement signals, comprises a preliminary step of processing the signals sampled by said means dedicated software, for the exclusion of any signal sections that are not of interest or connected to erroneous measurements, in order to implement their reliability and minimize the overall storage burden. 8. Process according to claim 1, characterized in that it comprises generating, by means of said dedicated software means and resident in said electronic processor means, said plurality of indicators of correct functioning, jointly or separately, on the basis of: - plate values of said liquid chiller device or said heat pump; - values detected during the testing of said liquid chiller device or said heat pump; - first start values of said liquid chiller device or said heat pump. - values detected during the correct operation, in the absence of anomalies, of said liquid chiller device or said heat pump. 9. Process according to claim 1, characterized in that said step of carrying out, by means of said dedicated software means, a comparison between said indicators of correct operation and said indicators of current operation, for the detection and diagnosis of faults or anomalies in said liquid chiller device or in said heat pump is implemented by RBC (Reconstruction-Based Contribution) type analysis. 10. Process according to claim 1, characterized in that it comprises the following steps: - identification of the presence of independent and contingent elements in the operation of said liquid chiller device or heat pump; - activating a command signal aimed at defining the presence of a new normal operating condition of said liquid chiller device or heat pump, as such not symptomatic of the presence of a fault or anomaly situation; - update said indicators of correct functioning with the data corresponding to said new operating mode, in such a way that, upon the occurrence of further operating conditions of said liquid chiller device or heat pump corresponding to said normal operating mode, the risk of signaling unreal faults is avoided.