Artificial intelligence based provisioning and compliance check of a pump

Abstract

This disclosure relates to autonomously monitoring and configuring a pump, and to performing a compliance check of the pump configuration. The solutions of the disclosure employ artificial intelligence (AI) in the form of a machine learning algorithm (MLA). A pump system for autonomous pump configuration is provided, the pump system comprising a pump for pumping a fluid, and a processor. The processor is configured to execute a MLA, and the MLA is configured to receive input data indicative of one or more performance parameters of the pump, to determine, based on the input data, whether or not a current operating state of the pump is within a target range of operating states, and if the current operating state of the pump is not within the target range of operating states, to adjust at least one operating parameter of the pump.

Description

[0001] ARTIFICIAL INTELLIGENCE BASED PROVISIONING AND COMPLIANCE CHECK OF A PUMP

[0002] TECHNICAL FIELD

[0003] The present disclosure relates to pumps and pump systems, for instance, to smart pumps. The disclosure is concerned with provisioning of a pump, specifically with autonomously monitoring and accordingly configuring the pump. The disclosure is also concerned with performing a compliance check of the pump configuration. The solutions of the disclosure employ artificial intelligence (Al) in the form of a machine learning algorithm (MLA). The Al can autonomously configure a pump in a pump system, and can perform compliance checks of the pump configuration.

[0004] BACKGROUND

[0005] Efficient pump operation is critical for various industrial, buildings, and water utility applications. Conventional methods for configuring pumps often involve manual processes, which may lead to pump inefficiencies, errors, and increased operational costs. Therefore, there is a need for a more intelligent and automated approach for provisioning pump configurations. In particular, an approach is needed, which can adapt to changing conditions in the pump environment, which can understand the context of the pump system of the pump, and which can optimize the pump performance continuously.

[0006] Moreover, conventional methods for ensuring compliance with, for example, local regulations and operational standards often involve manual processes, which may lead to errors, and to increased operational costs. Therefore, there is a need for a more intelligent and automated approach for running compliance checks, before provisioning new pump configurations. There is also a need to enable adaptions to changing regulations and to ensure continuous compliance. SUMMARY

[0007] This disclosure has the objective to provide solutions for to the above-mentioned needs. In particular, an objective of this disclosure is to provide an improved solution for configuring a pump in a pump system, especially for configuring the pump in an autonomous and adaptive manner. Similarly, another objective of this disclosure is to provide an improved solution for performing compliance checks of pump configurations, especially in an autonomous and adaptive manner.

[0008] These and other objectives are achieved by the solutions of this disclosure, which are described in the independent claims. Advantageous implementations are described in the dependent claims.

[0009] The solutions of this disclosure are Al based. The solutions of this disclosure particularly employ at least one MLA, which is configured to continuously monitor a pump's performance parameters, and to dynamically and contextually adjust one or more operating parameters and / or settings of the pump based on the monitoring, in order to achieve a desired configuration state of the pump.

[0010] A first aspect of this disclosure provides a pump system for autonomous pump configuration, the pump system comprising: a pump for pumping a fluid; a processor configured to execute a MLA, wherein the MLA is configured to receive input data indicative of one or more performance parameters of the pump; determine, based on the input data, whether or not a current operating state of the pump is within a target range of operating states; and if the current operating state of the pump is not within the target range of operating states, adjust at least one operating parameter of the pump.

[0011] The pump of the pump system may be a smart pump, e.g., a pump having integrated monitoring and processing capabilities, sensors, software, and / or connectivity features, or the like, which allow the pump system to be able to perform automated monitoring and control of the pumping operation of the pump. The smart pump’s monitoring capabilities may particularly benefit from the advantages provided by the processor and the MLA, respectively, which are configured to monitor the input data being indicative of the performance parameters of the pump. The input data may consist of or comprise the one or more performance parameters, or extracts thereof. The pump may be a rotary pump or a centrifugal pump. The pump may be a fluid pump, or a liquid pump, or an oil pump, or a water pump. That is, the fluid may be a liquid or water. The fluid may also be a gas.

[0012] The processor may be configured to run a computer program, in order to execute the MLA. The computer program may process the input data indicative of the one or more performance parameters of the pump. The performance parameters may be measured by one or more sensors related to the pump and / or the pump environment, or may be derived or calculated from such sensor measurements. The computer program may be the MLA. The MLA may be based on a trained model. The processor may comprise, or may be connected, to a memory, which stores the MLA (its instructions or code) and allows the processor to invoke the MLA from the memory.

[0013] The MLA may generally be referred to as an algorithm or learning procedure. The MLA may comprise a set of rules and / or may employ statistical techniques to identify the current operating state of the pump based on the input data reflecting the one or more performance parameters. The MLA may analyze the input data or the one or more performance parameters, and may learn from determined parameters and the correspondingly made adjustments of the at least one operational parameter. For example, the MLA may learn by processing the input data and using it to train a model, wherein the model may comprise a neural network or similar trainable modes. The model may represent learned patterns and / or correlations in the input data indicating the one or more performance parameters, and the model can be used by the MLA to determine the operating state of the pump and decide whether it is within the target range of operating states or not. In an example, the MLA is trained to determine the current operating state of the pump. Training the MLA may refer to training the model within the MLA. The MLA may accordingly encompass both the definition of the model and the process of training the model. By adjusting the at least one operating parameter of the pump, for example, a most relevant operating parameter of the pump, or adjusting multiple operating parameters that are related to the performance parameters indicated by the input data, the current operating state of the pump can be changed, and can ideally be changed to be within the target range of operating states. In this way, an improved solution for configuring the pump, especially in an autonomous and adaptive manner, is provided. Notably, if the current operating state of the pump is within the target range of operating states, no adjustment of operating parameters of the pump maybe necessary, and the processor may configured to further monitor the pump. That is, the MLA is still executed and will received new input data, for instance, if one or more performance parameters of the pump have changed and / or at regular intervals. As soon as the MLA determines, based on any input data, that the (new) current operating state of the pump is not within the target range of operating states, the adjustment of at least one operating parameter of the pump maybe performed.

[0014] In an implementation of the pump system, the MLA is configured to adjust the at least one operating parameter of the pump until a new operating state of the pump is within the target range of operating states.

[0015] For example, by executing the MLA by the processor, the current operating state of the pump maybe continuously or intermittently determined, based on the latest input data indicating the latest one or more performance parameters of the pump, and may be compared with the target range of operating states, until the target range of operating states is reached. The MLA may continuously monitor the one or more performance parameters by monitoring the input data, and may also react based on the input data, should the current operating state of the pump move out of the target range of operating states according to its determination.

[0016] In an implementation of the pump system, for adjusting the at least one operating parameter of the pump, the MLA is configured to generate a software configuration for the pump; and apply the software configuration to the pump. The software configuration, once applied to the pump, may cause the adjustment of the at least one operating parameter of the pump. For example, processing circuitry of the pump may run the software configuration, which may include corresponding instructions, thereby changing the at least one operating parameter of the pump. In this way, the operating state of the pump can be moved to within the target range of operating states of the pump. Applying the software configuration to the pump may comprise uploading, and / or installing, and / or running the software configuration to / on the pump.

[0017] In an implementation of the pump system, the processor is further configured to execute a second MLA, wherein the second MLA is configured to receive the generated software configuration as an input; and before applying the software configuration to the pump, determine whether the software configuration is in compliance with an operating protocol of the pump and / or with a regulation related to an operator of the pump or an environment of the pump.

[0018] The first MLA and the second MLA may be parts of the same computer program, and may be stored in the same memory. However, the first and the second MLA may also be separate computer programs or codes, which maybe stored in separate memories. By employing the second MLA, an improved solution for performing compliance checks of a pump configuration, especially in an autonomous and adaptive manner, is provided.

[0019] In an implementation of the pump system, the processor is configured to determine the at least one operating parameter of the pump from the input data indicative of the one or more performance parameters of the pump.

[0020] For instance, the MLA may be trained to select the correct one or more operating parameters of the pump, based on the input data that related to the one or more performance parameters, so as to change the operating state of the pump in an optimal way towards the target range of operating states. In an implementation of the pump system, the pump system further comprises one or more sensors for measuring the one or more performance parameters of the pump.

[0021] The sensors may be a part of the pump. For example, the pump may be a smart pump, and may be integrated with one or more of the mentioned sensors. However, one or more of the sensors may also be external to the pump, and / or may be a part of the processor, which may be a pump controller or control unit.

[0022] In an implementation of the pump system, the one or more sensors comprise at least one of: a vibration sensor for measuring a pump vibration; a power sensor for measuring a power of a motor of the pump; a rotation sensor for measuring a rotation of a motor of the pump; a pressure sensor for measuring a pressure of the fluid; a temperature sensor for measuring a temperature of the fluid and / or of a motor of the pump; a flow rate sensor for measuring a flow rate of the fluid.

[0023] By conducting the respective measurements, the one or more performance parameters of the pump can be obtained. Notably, the measured values may directly be the one or more performance parameters of the pump, or at least correlated to the one or more performance parameters of the pump.

[0024] In an implementation of the pump system, the operating state of the pump is determined by one or more operating parameters of the pump.

[0025] That is, in the simplest case, an operating parameter of the pump can define an operating state the pump.

[0026] In an implementation of the pump system, the target range of operating states is determined by a respective target range of one or more operating parameters of the pump; and / or the target range of operating states is a region in a multidimensional parameter space, which is given by a respective range of each of two or more target operating parameters of the pump. Typically, a set of two or more operating parameters defines an operating state of the pump, and accordingly, the respective ranges of these parameters determine the target range of operating states.

[0027] In an implementation of the pump system, the target range of operating states comprise one or more of the following: a range of a first operating parameter of the pump related to an energy consumption of the pump; a range of a second operating parameter of the pump related to a flow rate of the pump; a range of a third operating parameter of the pump related to a pressure of the pump; a range of a fourth operating parameter of the pump related to a speed of the pump.

[0028] In an implementation of the pump system, the processor is configured to determine at least one feature for each of the one or more performance parameters; and provide the at least one feature as the input data to the MLA.

[0029] In this way, the determination of the current operating state of the pump may become more efficient, as it requires less amount of data. For instance, the at least one feature can be extracted from the one or more performance parameters or sensor measurements. For example, this may comprise selecting or transforming raw sensor measurements into relevant, informative variables (features) that enhance the MLA’s performance. The at least one feature may simplify the input data to the MLA, for instance, highlighting patterns and relationships, which may help the MLA to learn more efficiently and make better predictions.

[0030] In an implementation of the pump system, the processor is configured to execute a training of the MLA based on the one or more performance parameters, the determined current operating state of the pump, and the corresponding adjustment of the at least one operating parameter of the pump.

[0031] The training may be executed during normal pump operation, and may lead to a fine-tuning of the MLA, which may have been pre-trained already.

[0032] In an implementation of the pump system, the training of the MLA is based on the generated software configuration applied to the pump, and a new operating state of the pump determined after the software configuration has been applied to the pump.

[0033] In an implementation of the pump system, the training of the MLA executed by the processor is based on a reinforcement learning algorithm.

[0034] In an implementation of the pump system, the MLA is pre-trained based on operating states of a plurality of pumps and corresponding adjustments of operating parameters of the pumps in view of respective target operating states of the plurality of pumps.

[0035] The pre-training maybe carried out, for example, in a command center server.

[0036] In an implementation form of the pump system, the pump system further comprises a communication interface configured to transmit information to a server, wherein the transmitted information indicates at least one of: the one or more performance parameters and the adjustment of the at least one operating parameter of the pump.

[0037] For instance, the pump system may communicate the information to a control or command center, which may comprise the server as a management entity. The server may be cloud-based, that is, it may be in a cloud. For instance, the server may be implemented on a computer. The server may also be the computer, and may comprise a processor. The server may be configured to manage one or more pumps, specifically the pump of the pump system of the first aspect. The server may, to this end, interact with the pump system wirelessly.

[0038] A second aspect of this disclosure provides a method for autonomously configuring a fluid pump of a pump system, the method comprising: executing a MLA, wherein the MLA receives an input based on one or more performance parameters of the pump; determines, based on the input, whether or not a current operating state of the pump is within a target range of operating states; and if the current operating state of the pump is not within the target range of operating states, adjusts at least one operating parameter of the pump. The method of the second aspect may have implementations that correspond to the implementations of the pump system of the first aspect. The method of the second aspect and its implementations achieve the effects and advantages described above with respect to the pump system of the first aspect and its respective implementations.

[0039] A third aspect of this disclosure provides a computer program comprising instructions which, when the program is executed by a pump system, instructs the pump system to perform the method according to the second aspect or any of its implementations.

[0040] In summary of the above aspects and implementation forms, the present disclosure provides a MLA for continuously processing data derived from sensor data to monitor one or more pump performance parameter of the pump, and to adjust at least one operating parameter of the pump dynamically and contextually to ensure the operation of the pump in a desired state. Notably, in the context of the present disclosure, the term “desired state” may refer to a target operating state defined by one or more operating parameters (e.g., flow rate, energy consumption, etc.) of the pump. How the desired state is determined is not limited in this disclosure, and may for example, be set by the manufacturer of the pump. Alternatively, the desired state maybe derived from the contractual terms between the end customer and the manufacturer.

[0041] In addition, a compliance check can be performed by a MLA, to make sure that any operating state changes of the pump adhere to legal and internal standards without manual intervention.

[0042] A feedback mechanisms allows for continuous learning and improvement based on newly provisioned input and real-time data. The two-phase training approach of the MLA allows the MLA to start with a broad understanding based on multiple similar pumps, and then adapt to specific conditions unique to each pump. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The above described aspects and implementations are explained in the following description of embodiments with respect to the enclosed drawings:

[0044] FIG. 1 shows a pump system according to this disclosure.

[0045] FIG. 2 shows an exemplary pump system according to this disclosure including one or more sensors.

[0046] FIG. 3 shows an exemplary method for training a MLA employed by a pump system according to this disclosure.

[0047] FIG. 4 shows an exemplary decision-tree based method for determining the current operating state of the pump.

[0048] FIG. 5 illustrates a training of the MLA of the pump system.

[0049] FIG. 6 shows an exemplary pump system according to this disclosure, configured for performing pump configuration compliance checks.

[0050] FIG. 7 shows a method for operating a pump system according to this disclosure.

[0051] DETAILED DESCRIPTION OF EMBODIMENTS

[0052] FIG. 1 shows a pump system io according to this disclosure. The pump system io comprises a pump n, which is configured to pump a fluid, e.g. a liquid like water or oil, or to pump a gas. The pump n maybe a centrifugal pump. The pump n may be a smart pump having various integrated sensors and / or processing functionalities. The pump system io may be installed in a building, e.g. a house or a factory, and this building and / or the pump surroundings may be referred to as environment of the pump n. The pump system io further comprises a processor 12, which is configured to perform various processing steps and functionalities. The processor 12 may be a controller or of a control unit for the pump n. The processor 12, or processing circuitry thereof, is configured to perform, conduct, or initiate various operations of the pump system io described in this disclosure. The processing circuitry may comprise hardware and / or the processing circuitry maybe controlled by software. The hardware may comprise analog circuitry or digital circuitry, or both analog and digital circuitry. The digital circuitry may comprise components such as application-specific integrated circuits (ASICs), field-programmable arrays (FPGAs), digital signal processors (DSPs), or multi-purpose processors. The processor 12 may further comprise memory circuitry, which can store one or more instruction(s) that can be executed by the processor 12 or its processing circuitry, in particular, under control of the software. For instance, the memory circuitry may comprise a non-transitory storage medium storing executable software code which, when executed by the processor 12 or its processing circuitry, causes the various described operations of the pump system 10 to be performed.

[0053] In particular, the processor 12 is configured to execute a MLA 13. The MLA 13 may be stored in the processor 12, or in a memory accessible to the processor 12. The MLA 13 may be part of a controller or control unit of the pump 11, which also comprises the processor 13. The MLA 13 is configured to receive input data 15, which is indicative of on one or more performance parameters 14 of the pump 11. That is, the input data 15 is at least based on the one or more performance parameters 14, or may even include the one or more performance parameters 14. The input data 15 could consist of one or more performance parameters 14. However, the one or more performance parameters 14 can also be preprocessed by the processor 12 to generate the input data 15. For example, the processor 12 may extract one or more features from the one or more performance parameters 14. Typically, more than one performance parameter 14 is monitored.

[0054] The MLA 13 is further configured to determine, based on the input data 15, whether or not a current operating state of the pump 11 is within a target range of operating states. If the current operating state of the pump 11 is not within the target range of operating states, the MLA 13 may adjust at least one operating parameter of the pump 11. To this end, the processor 12 may send instructive data 16, e.g. a software configuration 16, to the pump 11. In this disclosure, adjusting the operating parameter by the MLA 13 and sending an instruction by the MLA 13 to cause the adjusting of the operating parameter are treated equivalently. The MLA 13 can perform both steps in one. For instance, a result of the determination of the current operating state of the pump 11 does not have to be output or manifested by the MLA 13, but the MLA 13 can directly initiated the adjustment of the at least one operating parameter of the pump 11 if needed, i.e., if the current operating state of the pump 11 is not within the target range of operating states. In other words, the determining, while being a prerequisite to the adjusting, may be an internal, not visible operation. It may be assumed that if the MLA 13 adjusts at least one operating parameter of the pump 11, it has at least implicitly determined whether the current operating state of the pump 11 is within the target range of operating states, or not. However, the MLA 13 could also be considered as a black box, which is configured to adjust one or more operating parameters of the pump 11 based on the input data 15.

[0055] FIG. 2 shows an exemplary pump system 10 according to this disclosure, which builds on the pump system 10 shown in FIG. 1. Same elements in FIG. 1 and FIG.

[0056] 2 are labelled with the same reference signs, and may be implemented likewise.

[0057] The pump system 10 of FIG. 2 employs the MLA 13 - denoted as first MLA 13 -and also employs a second MLA 21. The first MLA 13 is trained to process the input data 15 indicative of the sensor data of one or more sensors 22, which provide measurements of the one or more performance parameters 14 of the pump 11. The first MLA 13 may be trained to extract meaningful features from the one or more performance parameters 13 as the input data 15, wherein the features can be used to (i) efficiently determine if the current pump operating state corresponds to an operating state in a target range of operating state or even an optimal desired operating state, and (ii) in the negative, generate the necessary adjustment(s) of the at least one operating parameter of the pump 11, for example, by generating the a software configuration 16 for the pump 11, and applying the software configuration 16 to the pump n to achieve desired operating state (discussed in more detail below).

[0058] The second MLA 21 is configured to receive the generated software configuration 16 as an input, and before applying the software configuration 16 to the pump n, to determine whether the software configuration is in compliance with an operating protocol of the pump n and / or with a regulation related to an operator of the pump n or an environment of the pump n. The second MLA 21 may be trained to determine, whether the software configuration 16 generated by the first MLA 13 is in compliance with the operating protocol (discussed in more detail below).

[0059] The processor 12 may be a dedicated hardware, particularly, hardware capable of executing software such as the first MLA 13 and the second MLA 21. The first MLA 13 and the second MLA 21 may respectively be stored in a memory 20, for instance, of a controller or control unit of the pump 11. The controller or control unit may also comprise the processor 12, or may be connected to the processor 12. The memory 20 may comprise one or more storage media and generally provides a place to store computer-executable program instructions, which are executable by the processor 12.

[0060] The pump system 10 maybe further configured to transmit information to a server in the command center - the command center server (CCS) 24 - wherein the transmitted information may indicate at least one of the one or more performance parameters 14, the determined current operating state of the pump 11, and the corresponding adjustment of the at least one operating parameter of the pump 11. The command center server 24 can be implemented as a conventional computer server. It maybe operated by the operator of the pump system 10, or by a building manager of the building in which the pump system 10 is located.

[0061] The transmission maybe done via cloud and / or a communication network 23. For instance, the pump 11, may comprises a communication interface (not shown), which is communicatively coupled to the communication network 23 to transmit the information to the command center server 24. In non-limiting implementations, the communication network 23 can be the internet, or a wide-area communication network, or a local area communication network, or a private communication network, or the like. The communication link to the communication network 23 can be wired or wireless.

[0062] As shown further in FIG. 2, the pump system 10 may include at least one sensor 22, potentially more than one sensor 22. The one or more sensors 22 of the pump system 10 may comprise at least one of: a vibration sensor for measuring a pump 11 vibration, an acoustic sensor 22a (exemplary shown as microphone in FIG. 2) for measuring a sound of the pump 11 or of the pump environment, a power sensor for measuring a power of a motor of the pump 11, a rotation sensor 22b (exemplarily shown in FIG. 2) for measuring a rotation of a motor of the pump 11, a pressure sensor for measuring a pressure of the fluid, a temperature sensor 22c (exemplarily shown in FIG. 2) for measuring a temperature of the fluid and / or of a motor of the pump 11, and a flow rate sensor for measuring a flow rate of the fluid.

[0063] The processor 12 is configured to determine the current operating state based on the sensor data 22 provided by the at least one sensor 22, i.e., by using the one or more pump performance parameters 14 indicated by the sensor measurements and employing the MLA 13. The one or more sensors 22 may provide real-time data related to pump operations to the processor 12 and / or the MLA 13.

[0064] The pump system 10 may include a plurality of data inputs and / or interfaces, such as from the sensors 22, or from actuators, controllers, programmable logic controllers (PLCs), human machine interfaces (HMIs), and the like. The plurality of data inputs and / or interfaces may be configured to collect and transmit data associated with the operation of the pump 11, and as such maybe communicatively coupled to the pump 11 via a communication link that may be wired or wireless. It can be understood that more than the data inputs and / or interfaces (as shown) can be connected to the pump 11.

[0065] FIG. 3 shows an exemplary method 30 for training the MLA 13 of the pump system 10, in particular, training and implementing the first MLA 13 will be described. The life cycle of the first MLA 13 may be divided into two phases, a “near-context training (NCT)” phase and a “specific-context training (SCT)” phase. The NCT may be performed at the server 24, and may be handled using comparable pumps (described in more detail below). The NCT occurs prior to the installation of the first MLA 13 in the pump system 10. The SCT may be performed at the site of the pump 11, i.e., in the pump system 10, after installation of the first MLA 13. That is, in the SCT phase, the first MLA 13 is already deployed for use at the pump system 10, but “fine-tunes” itself via self-learning capabilities. The flowchart shown in FIG. 3 briefly explains the NCT and SCT phases of the first MLA 13.

[0066] The NCT phase may generally comprise comparable pump and desired state determination. In the NCT phase, the server 24 is at first configured to perform a step 31 of identifying at least one comparable pump, i.e. comparable to the pump 11. The at least one comparable pump may be determined from a fleet of pumps. How the comparable pump is determined is not limited in this disclosure, and may be done using any one of the following exemplary methods.

[0067] A first method for the step 31 maybe based on clustering of an existing pump fleet. In this embodiment, the server 24 is configured to identify a set of already deployed pumps having a similar operating environment than the pump 11. In the present context, the operating environment may include building envelope data and / or system data.

[0068] The building envelope data, which is related to a building in which the pump 11 is installed or is to be installed, may include at least one of: a building size (e.g., total area), a number of floors, insulation types, construction materials, occupancy patterns and density, type of building (industrial, office, complex, etc.), pump application type (heating, cooling etc.), geographical location, orientation (North, West, East, South) of the building or room in which the pump is installed, and weather data.

[0069] The system data, which is related to the pump system 10, may include at least one of: a pump application (e.g. water supply, irrigation, sewage, etc.), a rated power and model of pump, a type of drive system, a primary source of energy, a number of cooling and heating devices, a water treatment system, a number of vales and types, the year of pump installation and / or manufacture.

[0070] A second method for the step 31 may be based on a desired state configuration (clustering). In some embodiments, the operator of the server 24 may manually assign desired state parameters for a set / fleet of pumps (such as, for example, “flow rate xxx”, and / or “energy efficiency to be xxxx”, and the like).

[0071] Alternatively, in another embodiment, it may be contemplated for step 31 that clustered pumps should share the same desired operating parameters or states, as they operate in the same environment. As such, the desired state may be automatically derived, such as by calculating an average of the one or more operating parameters for the clustered pumps of the pump fleet, wherein such an average is considered to be the desired state.

[0072] As illustrated in FIG. 4, a tree-based model based on, inter alia, the operating environment of the pump 11 may also be used, for instance, by an operator of the server 24. In the tree-based model, the lowest level child nodes 40 are each associated with a desired state (triangle) assigned by the operator of the server 24. The pump 11 maybe inputted into the decision tree comprising multiple nodes 41, and the desired state of the child node 40 corresponds to the child node of the pump 11. The same may be done with the fleet of pumps to obtain the comparable set of pumps, e.g., pumps sharing the same child node 40 as the pump 11.

[0073] Having identified in step 31 the set of comparable pumps, the first MLA 13 maybe configured to receive, as input data, the operational data (e.g., the desired operating state) associated with the set of comparable pumps, and to perform a step 32 of determining, if the one or more comparable pumps are in the desired operating state based on the collected data. Upon determination that a given pump from the fleet of pumps is not operating in the desired operating state, the operator of the server 24 can iteratively change the operating parameters of said pump to achieve the desired operating state, for instance, by performing the steps 33 and 34 of the method 30. In step 33, a software configuration to achieve the desired state is generated, and in step 34 the software configuration is applied to the fleet of pumps. The actions and outcomes are used as training data for the first MLA 13, which is illustrated in FIG. 5. It should be apparent that, after the training of the first MLA 21, the first MLA 13 is configured to determine (1) if a pump is operating in the desired operating state, and (2) the changes in operating parameter (e.g., software configuration) to achieve the desired operating state.

[0074] Once the NCT phase is over, the method 30 comprises a step 35 of uploading the first MLA 13, for example, into the memory 20 of the pump 11 or in the pump system 10, to begin the SCT phase. In other words, after the initial training on a comparable set of pumps, the first MLA 13 is deployed to the pump system 10 to let it adapt and fine-tune to its specific pump environment. More specifically, the first MLA 13, having been previously trained to do so, is configured to perform the step 36 of determining, if the current pump operating state corresponds to a desired state (i.e., is within the target range of operating states, as determined in the NCT phase), and the step 37 of generating the necessary software configuration 16, i.e., causing the adjustment of at least one operating parameter of the pump 11, to achieve the desired state through provisioning in step 38, i.e., applying the software configuration 16 to the pump 11. In some embodiments, the first MLA 13 implements a reinforcement learning mechanism, where the feedback following the provisioning is used to further train the first MLA 13.

[0075] FIG. 6 shows that in some embodiments, prior to provisioning the software configuration 16, the second MLA 21 can be used to determine if the software configuration 16 is compliant with the operating protocol of the pump 11. Although the second MLA 21 is described as being different from the first MLA 13, it maybe contemplated that the first MLA 13 is configured to do these functions of the second MLA 21, or that these functions below are performed by more than one MLA 13, 21.

[0076] In the context of the present disclosure, the term “operating protocol” may refer to laws or regulations, as well as internal regulations of the client or pump manufacturer or the pump operator. In some embodiments, the operating protocol is uploaded into the memory 20 via a data packet 61 received from the server 24. In another embodiment, the operating protocol is uploaded by the client (or operator of the pump n) via a data packet 62. In another embodiment, the pump 11 is connected to the network 23 of the client site, and can access the operating protocol stored on the client’s device.

[0077] How the second MLA 21 determines if the software configuration 16 is compliant with the operating protocol of the pump 11 is not limited in this disclosure. In a first embodiment, the operating protocol may be formatted into a predefined template, where upper and lower limits of operating parameters of the pump 11 are inputted by the server 24 or the client before uploading the operating protocol to the memory 20. The second MLA 21 is configured to determine if the purported software configuration 16 - causing the adjustment of the at least one operating parameter of the pump 11 - does fall within the constraints, i.e., within the limits of said operating parameter.

[0078] In a second embodiment, the second MLA 21 implements a large-language-model (LLM) configured to analyze the “plain” operating protocol text. This is particularly useful and handy for analyzing an internal operating protocol, as it removes the need to convert the operating protocol to a template format. How the second MLA 21 is implemented is not limited in this disclosure. It maybe a pre-trained language model like GPT-3, GPT-4, ALBERT, or DistilBERT, or the like. In some embodiments, it is possible to combine the LLM with a retrieval model to achieve retrieval augmented generation (RAG). Alternatively, the second MLA 21 may be trained (or refined) using a plurality of known operating protocol, whereas the training data includes operating parameters and the label is indicative of the operating parameter(s) being compliant or not (supervised learning). Alternatively, the second MLA 21 may be trained (or refined) by clustering similar operational changes together (unsupervised), whereas each cluster is then determined by an operator to be compliant or not with the operating protocol.

[0079] Once the second MLA 21 determines that the software configuration 16 is compliant, the processor 12 is configured to apply the software configuration 16 to the pump 11, for instance, to execute the software configuration 16 on the pump 11, or execute the software configuration 16 to control the pump 11, or to upload the software configuration 16 to the pump n for execution. It should be understood that enabling the customer to store on an own device, to which the pump n is connected, is particularly advantageous for situations where customers do not want to share internal operating protocols to third parties.

[0080] FIG. 7 shows a method 70 according to this disclosure, which is to be performed by the pump system 10. The method 70 is for autonomously configuring a fluid pump 11 of the pump system 10. The method 70 comprises a step 71 of executing a MLA 13 wherein the MLA 13 when executed, receives 72 input data 15 indicative of one or more performance parameters 14 of the pump 11, determines 72, based on the input data 15, whether or not a current operating state of the pump 11 is within a target range of operating states, and if not, adjusts 71 at least one operating parameter of the pump 11.

[0081] In summary, the solutions of this disclosure, which are based on executing the MLA 13 and / or MLA 21 by the processor 12, lead to an optimized pump performance, as the desired state of the pump 11 may be continuously achieved, and autonomous monitoring and adjustment of the state may be carried out. The adaptability of the pump system 10 also enhances its scalability. Moreover, compliance of the pump configuration state may be ensured autonomously and adaptively.

[0082] In the claims as well as in the description of this disclosure, the word “comprising” does not exclude other elements or steps and the indefinite article “a” or “an” does not exclude a plurality. A single element may fulfill the functions of several entities or items recited in the claims. The mere fact that certain measures are recited in the mutual different dependent claims does not indicate that a combination of these measures cannot be used in an advantageous implementation.

Claims

Claims1. A pump system (io) for autonomous pump configuration, the pump system (io) comprising:a pump (n) for pumping a fluid;a processor (12) configured to execute a machine learning algorithm, MLA, (13) wherein the MLA (13) is configured toreceive input data (15) indicative of on one or more performance parameters (14) of the pump (11);determine, based on the input data (15), whether or not a current operating state of the pump (11) is within a target range of operating states; and if the current operating state of the pump (11) is not within the target range of operating states, adjust at least one operating parameter of the pump (11).

2. The pump system (10) according to claim 1, wherein the MLA (13) is configured toadjust the at least one operating parameter of the pump (11) until a new operating state of the pump (11) is within the target range of operating states.

3. The pump system (10) according to claim 1 or 2, wherein for adjusting the at least one operating parameter of the pump (11), the MLA (13) is configured to generate a software configuration (16) for the pump (11); andapply the software configuration (16) to the pump (11).

4. The pump system (10) according to claim 3, wherein the processor (12) is further configured to execute a second MLA (21), wherein the second MLA (21) is configured toreceive the generated software configuration (16) as an input; and before applying the software configuration (16) to the pump (11), determine whether the software configuration (16) is in compliance with an operating protocol of the pump (11) and / or with a regulation related to an operator of the pump (11) or an environment of the pump (11).

5. The pump system (10) according to one of the claims 1 to 4, wherein the processor (12) is configured to determine the at least one operating parameter of the pump (11) from the input data (15) indicative of the one or more performance parameters (14) of the pump (11).

6. The pump system (10) according to one of the claims 1 to 5, further comprisingone or more sensors (22) for measuring the one or more performance parameters (14) of the pump (11).

7. The pump system (10) according to claim 6, wherein the one or more sensors (22) comprise at least one of:- a vibration sensor for measuring a pump vibration;- an acoustic sensor (22a) for measuring a sound of the pump (11) or the pump environment;- a power sensor for measuring a power of a motor of the pump (11);- a rotation sensor (22b) for measuring a rotation of a motor of the pump (11);- a pressure sensor for measuring a pressure of the fluid;- a temperature sensor (22c) for measuring a temperature of the fluid and / or of a motor of the pump (11);- a flow rate sensor for measuring a flow rate of the fluid.

8. The pump system (10) according to one of the claims 1 to 7, wherein the operating state of the pump (11) is determined by one or more operating parameters of the pump (11).

9. The pump system (10) according to one of the claims 1 to 8, wherein the target range of operating states is determined by a respective target range of one or more operating parameters of the pump (11); and / orthe target range of operating states is a region in a multi-dimensional parameter space, which is given by a respective range of each of two or more target operating parameters of the pump (11).

10. The pump system (io) according to one of the claims 1 to 9, wherein the target range of operating states comprise one or more of the following:- a range of a first operating parameter of the pump (11) related to an energy consumption of the pump (11);- a range of a second operating parameter of the pump (11) related to a flow rate of the pump (11);- a range of a third operating parameter of the pump (11) related to a pressure of the pump (11);- a range of a fourth operating parameter of the pump (11) related to a speed of the pump (11).

11. The pump system (10) according to one of the claims 1 to 10, wherein the processor (12) is configured todetermine at least one feature for each of the one or more performance parameters (14); andprovide the at least one feature as the input data (15) to the MLA (13).

12. The pump system (10) according to one of the claims 1 to 11, wherein the processor (12) is configured toexecute a training of the MLA (13) based on the one or more performance parameters (14), the determined current operating state of the pump (11), and the corresponding adjustment of the at least one operating parameter of the pump (11).

13. The pump system (10) according to claim 12 and claim 3 or 4, wherein the training of the MLA (13) is based on the generated software configuration (16) applied to the pump (11), and a new operating state of the pump (11) determined after the software configuration (16) has been applied to the pump (11).

14. The pump system (10) according to claim 12 or 13, wherein the training of the MLA (13) executed by the processor (12) is based on a reinforcement learning algorithm.

15. The pump system (10) according to one of the claim 1 to 14, wherein the MLA (13) is pre-trained based on respective operating states of a plurality of pumps (11) and corresponding adjustments of operating parameters of the pumps (11) in view of respective target operating states of the plurality of pumps (11).

16. A method (70) for autonomously configuring a fluid pump (11) of a pump system (10), the method (70) comprising:executing (71) a machine learning algorithm, MLA, (13) wherein the MLA (13)receives input data (15) indicative of one or more performance parameters (14) of the pump (11);determines (72), based on the input data (15), whether or not a current operating state of the pump (11) is within a target range of operating states; and if the current operating state of the pump (11) is not within the target range of operating states, adjusts (71) at least one operating parameter of the pump (11).

17. A computer program comprising instructions which, when the program is executed by a processor (12) of a pump system (10), instructs the pump system (10) to perform the method (70) according to claim 16.

Images