Method of operating an industrial air cleaning system

Abstract

A method, controller and computer program product for operating an industrial air cleaning system (1) arranged within a facility (100). Operating the industrial air cleaning system comprises an impact assessment sequence (a) and a preemptive action sequence (b). The impact assessment sequence (a) comprises detecting observation instances of an industrial activity (X), measuring indoor air quality values in the facility (100), and determining impact data corresponding to the industrial activity (X) based on the first indoor air quality values and data representative of the plurality of observation instances of the industrial activity(X). The preemptive action sequences comprise detecting actionable instances of the industrial activity (X) and / or of an indication of commencement of an industrial activity (X), determining one or more set-point parameters of the air cleaning devices (10, 101-n) based on data representative of the actionable instances of the industrial activity (X), the impact data corresponding to the industrial activity (X) and indoor air quality target values; and controlling the air cleaning devices (10, 101-n) in accordance with the one or more set-point parameters.

Description

[0001] METHOD OF OPERATING AN INDUSTRIAL AIR CLEANING SYSTEM

[0002] TECHNICAL FIELD

[0003] The present disclosure relates to a computer implemented method of operating an industrial air cleaning system comprising a plurality of air cleaning devices arranged within a facility. The present disclosure further relates to a control unit for controlling an industrial air cleaning system. The present disclosure further relates to an industrial air cleaning system, comprising a control unit, one or more activity sensors, and one or more air cleaning devices. The present disclosure even further relates to a computer program product comprising computer-executable instructions for operating an industrial air cleaning system.

[0004] TECHNICAL BACKGROUND

[0005] Indoor air quality is a term referring to the air quality within buildings and structures, affecting the health and comfort of building occupants. Indoor air quality is an important topic since people spend as much as 90% of their time indoors, either at home, work, or school. Therefore, the indoor environment is important to health and welfare. Indoor air quality can be affected by microbial contaminants (mold, bacteria), gases (including carbon monoxide, radon, volatile organic compounds), particulates, or any mass or energy stressor that can induce adverse health conditions. Indoor air is becoming an increasingly more concerning health hazard than outdoor air. Furthermore, Indoor air quality not only affects the health and comfort of occupants but has significant impact on industrial processes, greatly affecting the quality of products produced as well as the production machines themselves. For example, the chemical industry, in particular the pharmaceutical industry has very strict standards defined not only to ensure product quality but also to meet regulatory requirements. As a further example, the semiconductor industry, in particular the productions of semiconductor-based circuitry is highly sensitive to contaminants, wherein even dust particles on a scale of nanometers jeopardize the quality of semiconductor circuitry.

[0006] Providing, respectfully maintaining an indoor air quality target value may be achieved using air cleaning devices by ventilation (using ventilation devices), air filtering (using air filtering devices) or a combination of these measures.

[0007] Using ventilation (natural and / or mechanical) to dilute contaminants, filtration, and source control have long been the primary methods for improving indoor air quality in most buildings. Ventilation is the process of supplying fresh air to an enclosed space (a facility) in order to refresh / remove / replace the existing air. Ventilation is commonly used to remove contaminants such as fumes, dusts or vapors and provide a healthy and safe working environment; in other words, it is an engineering control with the purpose to remove ‘stale’ indoor air from a building and its replacement with ‘fresh’ outside air. It is either assumed that the outside air is of reasonable quality, or the outside air is cleaned before being allowed to enter the indoor space. Ventilation can be accomplished by natural means (e.g., opening a window) or mechanical means (e.g. fans or blowers). Ventilation should not be confused with exhaust. For example, in the case of combustion equipment such as water heaters, boilers, fireplaces, and wood stoves, exhausts are provided to carry the products of combustion which have to be expelled from the building in a way which does not cause harm to the occupants of the building. Movement of air between indoor spaces, and not the outside, is called transfer.

[0008] With the increased costs of bringing the outside air to temperature and / or humidity levels to comfortable / regulation compliant levels (e.g. by heating, respectively cooling), the ventilation, i.e. the exchange inside and outside air is to be reduced as far as possible.

[0009] Air filtering by removing (at least a portion of) contaminants from a facility by recirculation is a particular process of improving maintaining indoor air quality. According to air filtering, as opposed to ventilation, contaminants are not diluted by “fresh” air but removed from the facility. It shall be noted, that most installations for management of indoor air quality employ a combination of ventilation and air filtering. Air filtering is usually performed by the use of one or more air filtering devices arranged within a facility, the air filtering devices being configured to remove at least a portion of contaminants from the facility by recirculation. State of the art air filtering devices are configured to draw in air from a facility through an air inlet; force at least a portion of the drawn-in air through one or more air filtering filters to physically capture a portion of contaminants from the portion of the drawn-in air; and to return at least a portion of the filtered air through an air outlet back to the facility.

[0010] In order to adapt to different environments affected by different contaminants, different types and sizes of air filtering filters, respectively different types and sizes of air filtering devices are available to capture the respective contaminants.

[0011] In order to manage indoor air quality of large and / or highly sensitive indoor environments, several, independent air filtering devices are known to be installed within a facility wherein each air filtering device is operated independently from the other air filtering devices arranged in the air volume.

[0012] Indoor air quality within a facility is not static but strongly influenced by a variety of internal factors, such as activities in the room / hall / building and / or external factors, such as open doors, windows, variations in the levels of ventilation. In addition, in a larger facility the internal factors depend on the location within the facility.

[0013] Indoor air quality is a result of the interaction of a complex set of factors. Each of these factors must be considered when managing indoor air quality. Indoor air quality professionals use the four factors listed below as a basis for an investigative approach. Source: The source of contamination or discomfort indoors, outdoors, or within the mechanical systems of the building.

[0014] Ventilation: The ability of the ventilation system to control existing air contaminants and ensure thermal comfort (temperature and humidity conditions that are comfortable for most occupants).

[0015] Pathways: The one or more contaminant pathways connecting the contaminant source to the occupants, production equipment and / or products produced with the existing driving force moving contaminants along the pathway(s).

[0016] Occupants: Building occupants are present and are affected enough to raise indoor air quality concerns.

[0017] However, known methods, systems for operating an industrial air cleaning system are limited in their ability to ensure / maintain indoor air quality target values efficiently in dynamic environments. A first known approach to handle fluctuations in contamination is to provide air cleaning at levels corresponding to a highest expected level or contamination. This approach is highly resource inefficient. In order to avoid such inefficient use of resources, known systems are configured to react to degrading indoor air quality by intensifying air cleaning and / or ventilation and reducing the intensity of air cleaning and / or ventilation upon improvement of indoor air quality.

[0018] However, such air cleaning systems operated in accordance with measured indoor air quality are not always able to maintain the required indoor air quality target values, especially in highly dynamic environments.

[0019] It has been observed that known industrial air cleaning systems struggle to maintain indoor air quality target values in highly dynamic environments at least because air cleaning and / or ventilation measures are triggered in reaction to already degrading indoor air quality. Hence, such methods respectively systems are often “one step behind” the indoor air quality in the sense that indoor air quality first needs to degrade before action is take, resulting in a delay during which indoor air quality does not meet the indoor air quality target values.

[0020] While certain methods for operating an industrial air cleaning system have been developed which do monitor contamination-causing industrial activities within a facility and control air cleaning devices in dependence of detected industrial activities, air cleaning measures within a facility are still performed in reaction to an already ongoing industrial activity within that facility, leading to a delay in initiating air cleaning and / or ventilation.

[0021] SUMMARY

[0022] It is an object of the present disclosure to provide a computer implemented method for operating an industrial air cleaning system that overcomes at least some of the disadvantages of the prior art.

[0023] In particular, it is an object of the present disclosure to provide a computer implemented method for operating an industrial air cleaning system comprising one or more air cleaning devices arranged within a facility able to ensure indoor air quality target values even in a highly dynamic environment with strongly fluctuating levels, sources as well as timing of contamination.

[0024] The industrial air cleaning system comprises at least one control unit, one or more activity sensors, and one or more air cleaning devices arranged within a facility. According to embodiments, the one or more activity sensors are provided as part of the plurality of air cleaning devices. Alternatively, or additionally, the one or more activity sensors are provided as separate units communicatively connected with the control unit with a wired or wireless connection. In a preparatory step, the one or more air cleaning devices are communicatively connected to each other and / or with the control unit - in a star, mesh or daisy-chain network topology. According to embodiments, the control unit is provided as a stand alone unit. Alternatively, or additionally, the control unit is part of one of the plurality of air cleaning devices. Alternatively, or additionally, the control unit is distributed across the one or more air cleaning devices to provide scaling and / or redundancy.

[0025] The one or more air cleaning devices according to the present disclosure are arranged in a facility, in particular an industrial facility such as a production site. The one or more air cleaning devices are configured to influence the indoor air quality within - at least a part of - the facility. According to embodiments, the one or more air cleaning devices are configured to influence the indoor air quality by removing at least a portion of contaminants using one of more filters. Alternatively, or additionally, the air one or more cleaning devices are configured to influence the indoor air quality by ventilation, i.e. extraction of air from the facility and / or introduction of air into the facility from outside the facility.

[0026] The above-identified objects are addressed according to the present disclosure by a computer implemented method of operating an industrial air cleaning system, the method comprising an impact assessment and one or more preemptive action sequences.

[0027] In a first step of the impact assessment sequence one or more observation instances of an industrial activity are detected by the one or more activity sensors. The term ‘industrial activity’ as used herein encompasses any kind of activity which has the degrading, contamination potential to indoor air quality within the facility. Examples of an industrial activity include (non-exhaustive list): welding, soldering, drilling, sawing, milling of various materials, loading and unloading of goods, in particular in a dusty environment, various chemical-related processes but also industrial activities whereby the mere presence of humans is the cause of contamination, which is for instance the case in clean rooms for pharmaceutical production facilities. As used herein, the term ‘observation instances’ are occurrences of events (industrial activities) which are not necessarily acted upon but observed (for gathering training data).

[0028] Thereafter, in a further step of the impact assessment sequence, one or more indoor air quality values are measured by the contaminant sensor during and / or following detecting the plurality of observation instances.

[0029] Having detected observation instances of an industrial activity X and related indoor air quality values, in a further step a3) of the impact assessment sequence a), impact data corresponding to the industrial activity is determined by the control unit based on the one or more first indoor air quality values and data representative of the plurality of observation instances of the industrial activity(X).

[0030] Following the impact assessment sequence, in one or more preemptive action sequence(s) the effects of ongoing or commencing industrial activities are mitigated I preemptively compensated for.

[0031] In a first step of the preemptive action sequence, one or more actionable instances of an industrial activity X are detected by the one or more activity sensors. In contrast to observation instances, actionable instances of an industrial activity X are also acted upon to mitigate its impact. Alternatively, or additionally, in the first step of the one or more preemptive action sequences, indications of commencement of an industrial activity within the facility are detected by the one or more activity sensors. As used herein, the detection of ‘an indication of commencement’ refers to an indirect detection (with a certain degree of probability) that an industrial activity will commence. For example, turning on of lights in a production facility in the morning is an indication that production (an industrial activity will commence soon. Detecting an indication of commencement of an industrial activity even before the industrial activity starts provides the advantage that the air cleaning system can be controlled to preempt degradation of the indoor air quality. In other words, by detecting an indication of commencement of an industrial activity, the air cleaning system can prepare for the contamination-producing industrial activity.

[0032] Thereafter, in a further step of the preemptive action sequence, one or more set-point parameters of the one or more air cleaning devices are determined by the control unit based on data representative of the one or more actionable instances of the industrial activity, the impact data corresponding to the industrial activity and one or more indoor air quality target values.

[0033] Finally, having determined the setpoint parameters, in a further step of the preemptive action sequence, the one or more cleaning devices are controlled by the control unit in accordance with the one or more setpoint parameters. In this way, the air cleaning system can preemptively act to avoid degradation of indoor air quality before the contamination caused by the industrial activity even occurs.

[0034] According to embodiments, in addition to determining the setpoint parameters, the control unit determines whether currently running measures to counteract the impact of the actionable instance of the industrial activity are not suitable. For example, the control unit determines that air is being filtered within with an air filter inappropriate air filter for the contamination-produced by the detected industrial activity, leading to inefficient air cleaning and / or premature clogging of the air filter.

[0035] In order to even further improve the estimation of the propagation of and hence improve the identification of impacted perimeters, according to embodiments, the method employs a first machine learning algorithm trained using a plurality of training datasets. In a first step of a training sequence, data representative of the plurality of observation instances of the industrial activity are provided as input of the datasets. In a further step of training sequence, data representative of the one or more indoor air quality values are provided as expected output of the training datasets. Hence, the training dataset comprises data of observation instances “labelled” with the corresponding levels of contamination caused by the observation instances of industrial activities. Such datasets can then be used for so-called supervised learning of the first machine-learning algorithm, whereby labeled training data is required. Finally, having created the training datasets, in a further step of the training sequence, the first machine-learning algorithm is trained using the plurality of training datasets.

[0036] According to embodiments whereby the method employs a first machine learning algorithm trained using a plurality of training datasets, in a further step data representative of the actionable instance of an industrial activity is fed to an input of the first machine learning algorithm. In a further step, an estimated impact of the one or more actionable instances of the industrial activity at an output of the first machine learning algorithm. Thereafter, the one or more setpoint parameters are determined further using the estimated impact.

[0037] In a particular embodiment, in the first step of the training sequence, data representative of the plurality of observation instances of the industrial activity is fed (input) to one or more nodes of an input layer of a neural network of the first machine learning algorithm. In the further step, data representative of the one or more indoor air quality values is fed (into) to one or more nodes of an output layer of the neural network. Correspondingly, data representative of the actionable instance of an industrial activity is fed to the input layer of the neural network. In the further step, the estimated impact is determined at the output layer of the neural network. Alternatively, or additionally, the one or more setpoint parameters are determined further using the output at the output layer of the neural network.

[0038] In order to employ the most suitable air filter for each type of contamination, according to embodiments, the observation instances and / or actionable instances of industrial activities are classified into one or more categories of industrial activities. Correspondingly, detecting one or more actionable instances of an industrial activity comprises identifying a category of industrial activities corresponding to the detected one or more actionable instances of an industrial activity. The step of determining one or more setpoint parameters of the one or more air cleaning devices further comprises determining a type of air filter of the in dependence of the category of the one or more actionable instances of industrial activity.

[0039] According to further embodiments disclosed herein, the method further comprises determining data representative of the plurality of observation instances and / or the one or more actionable instances of the industrial activity comprising one or more of: an intensity; a propagation speed; a direction; an acceleration; and / or an attenuation of the plurality of observation instances and / or of the one or more actionable instances of the industrial activity.

[0040] According to further embodiments, the step of determining one or more setpoint parameters of the one or more air cleaning devices comprises determining one or more of: a timing of the air cleaning, such as a timing following detection of the one or more actionable instances of the industrial activity; an air cleaning direction; an air cleaning intensity; and / or an air filtering type, such as an air filtering type dependent on a particlesize of contaminants.

[0041] According to embodiments disclosed herein, the instances (actionable and / or observation) of industrial activities are detected directly and / or indirectly based on various indicia, comprising but not limited to:

[0042] Detecting electromagnetic radiation associated with the industrial activity using an electromagnetic sensor comprised by and / or communicatively connected to the one or more activity sensors; Detecting acoustic signals (noise) associated with the industrial activity using an acoustic sensor, such as a microphone, comprised by and / or communicatively connected to the one or more activity sensors.

[0043] Detecting physical motion associated with the industrial activity using motion detectors (such as a GPS sensor, a radar sensor, a LIDAR sensor and / or imaging device) comprised by and / or communicatively connected to the one or more activity sensors;

[0044] Detecting electromagnetic radiation transmitted by an electromagnetic transmitter attached to an industrial machine using a detector comprised by and / or communicatively connected to the one or more activity sensors;

[0045] Detecting an acoustic signal transmitted by an acoustic transmitter attached to an industrial machine using a detector comprised by and / or communicatively connected to the one or more activity sensors;

[0046] Detecting deviations of energy consumption (from a baseline or average consumption) associated with the industrial activity, such as electrical consumption and / or consumption of a combustible energy source.

[0047] According to further embodiments, one or more indoor air quality values in the facility are measured by one or more contaminant sensors during and / or following detection of the one or more actionable instances of the industrial activity. Based thereon, the subset of air cleaning devices are further controlled such that one or more indoor air quality values measured by the one or more contaminant sensors correspond to the one or more indoor air quality target values. The term ‘correspond’ - with respect to air quality values corresponding to indoor air quality target values - encompasses the values being equal with a certain amount of tolerance but also the values being in a functional relationship suitable for the particular application.

[0048] In order to enable a better understanding of the environment, according to further embodiments, the method further comprising an environmental assessment sequence comprising detecting one or more environmental parameters associated with the facility. Correspondingly, the one or more setpoint parameters of the subset of air cleaning devices are determined (in step B2) further based on the one or more environmental parameters. Alternatively, or additionally, the one or more impacted perimeters of the plurality of perimeters are identified further based on the one or more environmental parameters.

[0049] According to embodiments, detecting one or more environmental parameters associated with the facility comprises: detecting an atmospheric pressure within one or more of the plurality of perimeters and / or outside the facility and determining an impact of the atmospheric pressure on a flow of air into or out of the facility; and detecting one or more apertures of the facility, such as open windows or doors and determining an impact of the one or more apertures on a flow of air into or out of the facility;

[0050] Correspondingly, the step of determining the one or more setpoint parameters comprises determining the setpoint parameters such as to ensure an overpressure within one or more of the impacted perimeters as compared to one or more of the plurality of perimeters and / or as compared to atmospheric pressure outside the facility. An overpressure within one or more of the impacted perimeters prevents (at least to a certain degree) air from other perimeters from entering the impacted perimeters.

[0051] Alternatively, the step of determining the one or more setpoint parameters comprises determining the setpoint parameters such as to ensure an underpressure within one or more of the impacted perimeters as compared to one or more of the plurality of perimeters and / or as compared to atmospheric pressure outside the facility. An underpressure within one or more of the impacted perimeters facilitates air from other perimeters to enter the impacted perimeters, providing “fresh” air into the impacted perimeters. In addition to controlling the air cleaning device(s), according to further embodiments, one or more connected systems are also controlled by the by the control unit in accordance with data representative of the actionable instance in order to mitigate an impact of the actionable instance. For example, an industrial machinery (e.g. a saw) which generates dust levels beyond the current air cleaning capacity within the impacted perimeter(s). Once the air cleaning device(s) is running at required capacity (exceeding the levels of dust generated by the saw), the industrial machinery is turned on again. As a further example, in accordance with data representative of the actionable instance, the control unit may be configured to lock doors of the facility for the duration of a certain type of industrial activity, such as locking the doors to a clean room for pharmaceutical production. As a further example, in accordance with data representative of the actionable instance, the control unit may be configured to control a warning system to warn occupants of the facility, e.g. if an actionable instance of an industrial activity has been detected which is estimated to lead to unsafe indoor air quality.

[0052] In order to monitor the efficiency of mitigation of the impact of the industrial activity and to improve mitigation of further actionable instances of the industrial activity, according to further embodiments, the method further comprises an efficiency assessment and improvement sequence. In a first step) of the efficiency assessment and improvement sequence, one or more further indoor air quality values in the facility during and / or after the one or more preemptive actions sequences are measured by the contaminant sensor. Thereafter, in a further step) of the efficiency assessment and improvement sequence, efficiency data corresponding to the industrial activity and the one or more set-point parameters of the one or more air cleaning devices is determined by the control unit using a comparison of the first indoor air quality value and the one or more further indoor air quality values. Based on the efficiency data, the determination of one or more set-point parameters corresponding to the industrial activity is refined by the control unit. In order to optimize the set-point parameters and hence improve the mitigation of the impact of industrial activities, according to embodiments, the method further comprises a set-point optimization sequence. In a first step of the set-point optimization sequence, data representative of the one or more actionable instances of the industrial activity and the one or more set-point parameters corresponding to the industrial activity are fed to a second machine learning algorithm. In addition, in a further step of the set-point optimization sequence, the one or more further indoor air quality values (the ones measured during and / or after the one or more preemptive actions sequences) are also fed to the second machine learning algorithm. Thereafter, in a third step of the set-point optimization sequence, parameters of the second machine learning algorithm are optimized in order to minimize deviations of the indoor air quality values from the indoor air quality target values while also optimizing one or more set-point parameters, such as minimizing a set-point parameter indicative of an intensity of operation of the one or more air cleaning devices. According to embodiments, the optimization sequence comprises optimizing weights of a weighted function such as to minimize deviations of the indoor air quality values from the indoor air quality target values. According to embodiments, the one or more set-point parameters are optimized in view of environmental parameters (such as a seasonal parameter, e.g. winter respectively summer). Alternatively, or additionally the one or more set-point parameters are optimized in view of a particular use case, such as an industrial activity type / sub-type (e.g. the expected contamination is different in a bakery when baking bread as compared to baking buns due to the different baking temperatures and different flour used). Alternatively, or additionally the one or more set-point parameters are optimized in accordance with human parameters, such as in dependence of an identified user / operator, whereby a particular identified user / or operator is assigned a specific activity profile associated with a specific intensity and / or type of contaminating behavior (e.g. the level of contamination caused by welding is highly dependent on the skills and / or work habits of the particular welder). Alternatively, or additionally the one or more set-point parameters are optimized in accordance with identified tools, work products. For example, in an industrial facility whereby machinery is commissioned / tested, the type of commissioned / tested machinery has a great influence on the expected contamination (e.g. upon a first startup of a tractor with an internal combustion engine, the level of expected contamination is highly dependent on the size of its engine and also on the type of fuel used).

[0053] Correspondingly, the step of determining one or more set-point parameters based on the impact data comprises feeding the data representative of the one or more actionable instances of the industrial activity to an input of the second machine learning algorithm; and determining the one or more set-point parameters of the one or more air cleaning devices at an output of the second machine learning algorithm.

[0054] It is a further object of the present disclosure to provide a control unit for operating an industrial air cleaning system comprising one or more air cleaning devices arranged within a volume of that overcomes at least some of the disadvantages of the prior art. In particular, it is an object of the present disclosure to provide a control unit for operating an industrial air cleaning system to ensure indoor air quality target values even in a highly dynamic environment with strongly fluctuating levels, sources as well as timing of contamination. This object is addressed by a control unit configured to control an industrial air cleaning system according to the method of operating an industrial air cleaning system according to any one of the embodiments disclosed herein. In particular, the object of the present disclosure is further addressed by a control unit comprising a processing unit, a data storage for storing computer-readable instructions, which when executed by the processing unit causes the control of an industrial air cleaning system according to the method of operating an industrial air cleaning system according to any one of the embodiments disclosed herein.

[0055] It is a further object of the present disclosure to provide an industrial air cleaning system comprising that overcomes at least some of the disadvantages of the prior art. In particular, it is an object of the present disclosure to provide an industrial air cleaning system able to ensure indoor air quality target values even in a highly dynamic environment with strongly fluctuating levels, sources as well as timing of contamination. This further object is addressed by an industrial air cleaning system comprising a control unit, one or more activity sensors, and one or more air cleaning devices configured to influence the indoor air quality within the facility, the industrial air cleaning system being configured to carry out the method of operating an industrial air cleaning system according to any one of the embodiments disclosed herein. The one or more activity sensors and the one or more air cleaning devices are communicatively connected (directly or indirectly) with the control unit, for example using wired and / or wireless connections.

[0056] According to embodiments disclosed herein, the air cleaning devices comprise a housing with at least one inlet and at least one outlet. Air from the air volume is drawn into the housing through the at least one inlet, in that the at least a fan produces an underpressure at the outlet. The air then exits the housing through the at least one outlet. The at least one outlet may be equipped with at least one air guide flap to influence the direction of the exiting air.

[0057] According to embodiments disclosed herein, one or more of the air cleaning devices are configured to remove at least a portion of contaminants from the facility by ventilation, the air cleaning devices being configured to: draw in air from the facility through an air inlet and force, by air propelling means, at least a portion of the drawn-in air out of the facility; or draw in air from outside the facility through an air inlet and force, by air propelling means, at least a portion of the drawn-in air into the facility. According to further embodiments disclosed herein, one or more of the air cleaning devices are configured to remove at least a portion of contaminants from the facility using an air filter. According to embodiments disclosed herein, the air cleaning devices remove contaminants from the facility by recirculation, that is by drawing in air from the facility through an air inlet; forcing at least a portion of the drawn-in air through one or more air cleaning filters to physically capture a portion of contaminants from the portion of the drawn-in air; and returning at least a portion of the filtered air through the air outlet back to the facility. In particular, the air cleaning devices draw the air in; force the air through air cleaning filters and returning the filtered air aided by air propelling means such as a fan. Good results are achieved by air cleaning devices which comprise a housing and therein arranged one or several filters of the same or different kind (e.g. arranged in a serial manner). According to embodiments of the present disclosure the filters of the air cleaning device(s) further comprise molecular filtration means; UV-light based decontamination means and / or photocatalystic decontamination means. Correspondingly, the step of controlling, by the control unit, the subset of air cleaning devices in accordance with the one or more setpoint parameters comprises controlling one or more air cleaning devices of the subset to: draw in air from the facility through an air inlet; to force, by air propelling means, at least a portion of the drawn-in air through an air filter to physically capture a portion of contaminants from the portion of the drawnin air; and to return at least a portion of the filtered air through an air outlet back to the facility.

[0058] It is a further object of the present disclosure to provide a computer program product for operating an industrial air cleaning system that overcomes at least some of the disadvantages of the prior art. In particular, it is an object of the present disclosure to provide a computer program product for operating an industrial air cleaning system able to ensure indoor air quality target values even in a highly dynamic environment with strongly fluctuating levels, sources as well as timing of contamination. This object is addressed by a computer program product comprising computer-executable instructions, which when executed by a control unit of an industrial air cleaning system causes the air cleaning system to carry out the method of operating an industrial air cleaning system according to any one of the embodiments disclosed herein.

[0059] In particular, the step of controlling the air cleaning system comprises controlling one or more of plurality of air cleaning devices of the air cleaning system such that the data indicative of air quality from the one or more air quality data sources corresponds to an indoor air quality target value. According to embodiments of the present disclosure, the indoor air quality target value is a constant value or a value changing according to a schedule, such as an indoor air quality target value schedule determined corresponding to scheduled activity within the facility.

[0060] The control unit(s) controls the plurality cleaning device(s) using both the data indicative of the indoor air quality and the data indicative of an operational state of the plurality of air cleaning devices. Thereby, the present disclosure provides a method of operating an air cleaning system based not on assumptions but based on data indicative of air quality within the facility as well as data indicative of operational state(s) of the plurality of air cleaning devices.

[0061] According to embodiments of the present disclosure, the operational state(s) comprise data indicative of load level(s) of the plurality of air cleaning devices (e.g. 10% load, 10W of 100Wmax load, 10m3 / h of a max of 100m3 / h load, etc.), the method further comprising the control unit controlling the air cleaning system such as to:

[0062] Achieve a load balance between the plurality of air cleaning devices. The load balance may be an equal load between the plurality of air cleaning devices or a balanced load based on the local type and local concentration of contaminants as measured by the contaminant sensor(s). Set one or more of the plurality of air cleaning devices to a load level at or below a threshold efficiency level. Even though having a higher maximum capacity (e.g. higher amount of clean air delivery rate), certain air cleaning devices are most energy efficient up to a certain load level. For example, the increase of air

[0063] 5 resistance of certain air cleaning filters (the pressure drop between the inlet and outlet side of the air cleaning filters) with the increase of air volume is non-linear. Therefore, in order to improve energy efficiency, according to embodiments of the present disclosure, the load is distributed between several air cleaning devices such as to ensure that each air cleaning device is operating efficiently vs. the entire air cleaning load being carried by a single air cleaning device operating at a high but inefficient load while other air cleaning devices being idle. Energy efficiency is expressed for example as the amount of energy required to deliver a certain flow rate of clean air W / (m3 / h). 5 Set one or more of the plurality of air cleaning devices below an increased wear level. Beyond becoming inefficient, even though having a higher maximum capacity (e.g. higher amount of clean air delivery rate), certain air cleaning devices are prone to increased wear beyond a certain load level. Therefore, in order to prolong the lifetime of air cleaning devices, according to embodiments of the0 present disclosure, the load is distributed between several air cleaning devices such as to ensure that none of the air cleaning devices is operating beyond their increased wear level.

[0064] According to embodiments of the present disclosure, the operational state(s) comprises5 data indicative of a degradation level of a first air cleaning device of the plurality of air cleaning devices, the method further comprising the control unit controlling one or more of the plurality of air cleaning devices other than the first air cleaning device such as to compensate for the degradation level of the first air cleaning device. The feature “compensate for the degradation level of the first air cleaning device” comprises the process of increasing the load level (power setting) of one or more air cleaning devices (other than the first air cleaning device) at least temporarily until servicing / replacement of the degraded first air cleaning device. The temporal increase of the load level of one or more air cleaning devices (other than the first air cleaning device) may go even beyond the above-mentioned balance; threshold efficiency and / or increased wear levels.

[0065] The degradation level comprises data indicative of percentage of remaining contaminant removal efficiency, contaminant holding capacity and resistance to airflow of the air cleaning filters. Additionally, the degradation level comprises data indicative of a degradation of further components of the air cleaning devices, such as the air propelling means. According to embodiments of the present disclosure, the control unit switches the first air cleaning device into a service state if the data indicative of a degradation level is above a service threshold and generates an alert signal identifying the first air cleaning device. The alert signal may be an audible, a visual signal and / or an alert message sent by data communication means, for example to a system owner, a service technician, a back office, or even R&D for statistical purposes.

[0066] According to embodiments of the present disclosure, the data indicative of air quality originates from contaminant sensor(s) and / or air quality data sources. The data indicative of air quality comprises data indicative of a contamination type within the facility, in particular particle size of the contamination. Accordingly, the method of operating an air cleaning system further comprises:

[0067] Receiving, by the control unit, data indicative of a contamination type each of the plurality of air cleaning devices is configured to remove (or is most suited to remove) from the facility. The data indicative of a contamination type each of the plurality of air cleaning devices is configured to remove may originate from the air cleaning devices themselves. Alternatively, or additionally the data indicative of a contamination type each of the plurality of air cleaning devices is con-figured to remove may be retrieved by the control unit from a configuration file corresponding to the air cleaning system.

[0068] The control unit controlling the air cleaning system using the data indicative of a contamination type each of the plurality of air cleaning devices is configured to remove and the data indicative of a contamination type of the facility. In particular, the control unit increases the load level of the air cleaning devices best suited to remove the contamination type detected by the contaminant sensor(s) while maintaining or reducing the load level of air cleaning devices not suited therefor.

[0069] Such embodiments are particularly advantageous as they enable the air cleaning system to dynamically adapt to a changing environment and allow efficient use of the available resources, namely the air cleaning devices.

[0070] Embodiments according to the present disclosure comprise a forward-looking (predictive) aspect, wherein, the control unit receives data indicative of an expected indoor air quality impact (such as an indication of increased activity within the facility. Having such data available, the control unit controls the air cleaning system further using the data indicative of an expected air quality impact. Hence, the air cleaning system is able to pre-emptively adapt to a change of indoor air quality, even before such is detected by the contaminant sensors / data sources. For example, the expected indoor air quality impact may comprise data indicative that a certain industrial process will be started at a scheduled point in time, activity which is associated with a particular contamination of the facility. As a pre-emptive measure, the control unit instructs the air cleaning device best suited to remove the particular contaminant resulting from said activity exactly when the activity is scheduled to start, even before the indoor air quality is affected. In such a way, the control unit is able to take preventive action to avoid degradation of indoor air quality. On the other hand, the expected indoor air quality impact may comprise data indicative that a certain industrial process will be completed at a scheduled point in time hence user presence will be reduced. In order to conserve energy, the control unit instructs the air cleaning devices configured to remove the particular contaminant resulting from said activity even before the activity is scheduled to complete, anticipating a drop in the need to maintain the indoor air quality beyond user presence.

[0071] Furthermore, data indicative of an expected indoor air quality impact may comprise external data such as Indoor Air Quality index forecast data comprising for example actual or expected pollen or NOx concentrations.

[0072] With respect to the layout of the air cleaning system as deployed in a facility, the following topologies are envisaged according to embodiments of the present disclosure:

[0073] Mesh topology: The air cleaning system comprising a plurality of control units distributed in the facility and arranged in a mesh network configuration. In such a system layout, the plurality of control units collaboratively control the air cleaning system. In order to achieve the collaborative control of the air cleaning system, the plurality (i.e. 2 or more) of control units are communicatively connected with each other and exchange data indicative of air quality, data indicative of operational state(s) (of the plurality of air cleaning devices) and / or data indicative of the respective control unit controlling one or more of the plurality of air cleaning devices of the air cleaning system. In such a distributed topology, one or more of the plurality of control units may be integrated into respective air cleaning devices.

[0074] Mesh topology is advantageous as the air cleaning system has no single point of failure. Furthermore, air cleaning devices with an integrated / dedicated control unit can easily be deployed into an existing air cleaning system, thereby extending the mesh network. In a mesh topology, good results can be achieved when each air cleaning device is “smart” in the sense that it has an integrated / dedicated control unit configured to control the respective air cleaning device as part of the air cleaning system.

[0075] Star topology: The air cleaning system comprises at least one “master” control unit communicatively connected to the plurality of air cleaning devices of the air cleaning system and to a plurality of contaminant sensors / data sources and controlling the plurality of air cleaning devices such as to influence the indoor air quality within the facility.

[0076] In a star topology, not all air cleaning devices must be “smart” in the sense that they are self-controlling. Instead, the so called “master” control unit is provided to

[0077] 5 control “non-smart” air cleaning devices (air cleaning devices without an integrated / dedicated control unit). This reduces the complexity of the individual air cleaning devices and is advantageous in particular for upgrading existing air cleaning systems comprising a plurality of existing air cleaning devices, in particular upgrading air cleaning devices of different types or even manufacturers. Having at least one “master” control unit connected to a plurality of air cleaning devices of the air cleaning system simplifies maintenance of the air cleaning system.

[0078] Mixed topology: According to particular embodiments of the present disclosure, the air cleaning system is deployed in a facility in a mixed topology, wherein a first 5 subset of air cleaning devices comprise an integrated / dedicated control unit which form a mesh topology network to exchange data between themselves, while a second subset of air cleaning devices are controlled by a single control unit in a star topology network.

[0079] A mixed topology combines the advantages of a mesh and a star topologies,0 avoiding single points of failure, while allowing easily upgrading existing “nonsmart” air cleaning devices.

[0080] According to embodiments of the method of present disclosure for operating an air cleaning system deployed in a mesh or mixed topology, the method further comprises:5 establishing a data communication link between the plurality of control units; establishing a data communication link between each of the plurality of control units and one or more of the plurality of air cleaning devices; and the plurality of control units exchanging data indicative of the respective control unit controlling one or more of the plurality of air cleaning devices of the air cleaning system.

[0081] The data communication links established between the plurality of control units and / or the data communication links established between each of the plurality of control units are wired (such as Ethernet) and / or radio communication links (such as WiFi, Bluetooth, or mobile telecommunication links).

[0082] According to embodiments of the present disclosure, a remote server is provided to collect and process data from a plurality of air cleaning systems in order to take advantage of the compounded dataset throughout a plurality of environments. Correspondingly, the method further comprises:

[0083] The remote server collecting data indicative of indoor air quality and data indicative of operational state(s) from a plurality of air cleaning systems arranged within a plurality of volumes of air. According to the particular use case, the plurality of volumes of air are located in the same and / or in different buildings.

[0084] The remote server generating control parameters using the data collected from the plurality of air cleaning systems.

[0085] The remote server transmitting the control parameters to the control unit(s).

[0086] The control unit(s) controlling the air cleaning system further using the control parameters transmitted by the remote server. In particular, the control parameters are intended to further refine the control algorithms of the control units based on experience of different environments as reflected by the data collected by the remote server.

[0087] According to the present disclosure, the above-mentioned object(s) are further addressed by a control unit comprising processing means and storage means, the storage means comprising computer-executable instructions, which when executed by the processing means cause the control unit to carry out the method according to one of the embodiments disclosed herein. The control unit may be a stand-alone device intended to be used in a star topology (see above), a device configured to be integrated into an air cleaning device in a mesh topology, and / or a generic device suitable to be used both as a stand-alone device or to be integrated into an air cleaning device.

[0088] According to the present disclosure, the above-mentioned object(s) are further addressed by an air cleaning device for removing at least a portion of contaminants from a facility, the air cleaning device comprising: a control unit according to one of the embodiments disclosed herein; an air inlet; one or more air cleaning filters; air propelling means; and an air outlet. The air cleaning device is configured to: draw in air from the facility through the air inlet; force, by the air propelling means, at least a portion of the drawn-in air through the one or more air cleaning filters to physically capture a portion of contaminants from the portion of the drawn-in air; and return at least a portion of the filtered air through the air outlet back to the facility.

[0089] According to the present disclosure, the above-mentioned object(s) are further addressed by an air cleaning system comprising: an air cleaning system comprising a plurality of air cleaning devices configured to remove at least a portion of contaminants from the facility; one or more air quality data sensors configured to measure an air quality within the facility and to make available data indicative of air quality; and one or more control unit(s) according to one of the embodiments disclosed herein. According to embodiments of the present disclosure, the control unit(s) is / are located physically remote from the air cleaning devices and / or comprised by the plurality of air cleaning devices.

[0090] According to the present disclosure, the above-mentioned object(s) are further addressed by a computer program product comprising computer-executable instructions, which when executed by a processing unit (such as a CPU) of one or more control unit(s) cause the control unit(s) to carry out the method according to one of the embodiments disclosed herein.

[0091] It is to be understood that both the foregoing general description and the following detailed description present embodiments, and are intended to provide an overview or framework for understanding the nature and character of the disclosure. The accompanying drawings are included to provide a further understanding, and are incorporated into and constitute a part of this specification. The drawings illustrate various embodiments, and together with the description serve to explain the principles and operation of the concepts disclosed.

[0092] BRIEF DESCRIPTION OF THE DRAWINGS

[0093] The herein described disclosure will be more fully understood from the detailed description given herein below and the accompanying drawings which should not be considered limiting to the disclosure described in the appended claims. The drawings in which:

[0094] Figure 1A shows a highly schematic perspective view of first embodiment of an air cleaning system according to the present disclosure;

[0095] Figure 1B shows a highly schematic perspective view of further embodiment of an air cleaning system according to the present disclosure, comprising a plurality of air cleaning devices;

[0096] Figure 2 shows a flow diagram illustrating a sequence of steps of a first embodiment of the computer implemented method of operating an industrial air cleaning system according to the present disclosure; Figure 3 shows a flow diagram illustrating a sequence of steps of a training sequence of a machine-learning algorithm for determining the propagation pattern;

[0097] Figure 4 shows a flow diagram illustrating steps of employing the trained machinelearning algorithm for identifying the impacted perimeters respectively for determining the setpoint parameters;

[0098] Figure 5 shows a flow diagram illustrating steps of an efficiency assessment and improvement sequence;

[0099] Figure 6 shows a flow diagram illustrating steps of a set-point optimization sequence;

[0100] Figure 7 shows a flow diagram illustrating steps of employing the trained second machine-learning algorithm for determining one or more set-point parameters

[0101] Figures 8 shows a simplified block diagram of a first embodiment of an air cleaning device according to the present disclosure;

[0102] Figures 9 shows a highly schematic perspective view of a first embodiment of an air cleaning device according to the present disclosure; and

[0103] Figure 10 shows a simplified block diagram of a first embodiment of a control unit according to the present disclosure.

[0104] DETAILED DESCRIPTION

[0105] Reference will now be made in detail to certain embodiments, examples of which are illustrated in the accompanying drawings, in which some, but not all features are shown. Indeed, embodiments disclosed herein may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will satisfy applicable legal requirements. Whenever possible, like reference numbers will be used to refer to like components or parts.

[0106] As used herein, the term “contaminant” refers to any kind of particles of interest within a facility 100. In particle theory contaminants / particles are split based on how the particles have been formed, e.g. dust, mist and aerosols:

[0107] Dust is formed usually by decomposition of solid materials, such as crushing of stone, rock drilling and grinding of metal. Dust consists of solid particles in the range 1 micrometer to a few tenths of a millimeter.

[0108] Mist may be formed either by decomposition of a liquid (atomization) such as the use of cutting fluids or by condensation, for example by cooling of moist air.

[0109] Aerosols are suspensions of solid particles or liquid particles in gases. Aerosols may have particle diametric from 0.01 pm to 100 pm, typical concentrations in workplaces can be 1 pg / m3 to 100 mg / m3, in extreme places up to 10 g / m3.

[0110] Nodular particles are spherical in shape, acicular particles are needle-shaped (fibers of various kinds, such as asbestos, mineral fiber, textile fiber), laminar particles are flat shaped (e.g., talc, graphite). Small particles often form aggregates (cluster) of larger size. This applies, for example to the particles in engine exhaust and welding smoke.

[0111] In order to be able to identify the proper means to remove contaminants, particles are commonly categorized. Among the most common categorizations imposed on particles are those with respect to size, referred to as fractions. As particles are often non- spherical (for example, asbestos fibers), there are many definitions of particle size. The most widely used definition is the aerodynamic diameter. A particle with an aerodynamic diameter of 10 micrometers moves in a gas like a sphere of unit density (1 gram per cubic centimeter) with a diameter of 10 micrometers. PM diameters range from less than 10nanometers to more than 10 micrometers. These dimensions represent the continuum from a few molecules up to the size where particles can no longer be carried by a gas.

[0112] It shall be noted that the above are formal definitions. Depending on the context, alternative definitions may be applied. In some specialized settings, each fraction may exclude the fractions of lesser scale, so that PM 10 excludes particles in a smaller size range, e.g. PM2.5, usually reported separately in the same work. Such a case is sometimes emphasized with the difference notation, e.g. PM10-PM2.5. Other exceptions may be similarly specified. This is useful when not only the upper bound of a fraction is relevant to a discussion. The fact that some particle size ranges require greater air cleaning filter strength and the smallest ones can outstrip the body's ability to keep them out of cells both serve to guide understanding of related public policy, environment, and health topics.

[0113] Furthermore, contaminant particles are categorized by their composition. The composition of aerosol particles depends on their source. Wind-blown mineral dust tends to be made of mineral oxides and other material blown from the Earth's crust; this aerosol is light-absorbing. Sea salt is considered the second-largest contributor in the global aerosol budget, and consists mainly of sodium chloride originated from sea spray; other constituents of atmospheric sea salt reflect the composition of sea water, and thus include magnesium, sulfate, calcium, potassium, etc. In addition, sea spray aerosols may contain organic compounds, which influence their chemistry. Sea salt does not absorb.

[0114] Secondary particles derive from the oxidation of primary gases such as sulfur and nitrogen oxides into sulfuric acid (liquid) and nitric acid (gaseous). The precursors for these aerosols, i.e. the gases from which they originate, may have an anthropogenic origin (from fossil fuel combustion) and a natural biogenic origin. In the presence of ammonia, secondary aerosols often take the form of ammonium salts; i.e. ammonium sulfate and ammonium nitrate (both can be dry or in aqueous solution); in the absence of ammonia, secondary compounds take an acidic form as sulfuric acid (liquid aerosol droplets) and nitric acid (atmospheric gas). Secondary sulfate and nitrate aerosols are strong light scatterers. This is mainly because the presence of sulfate and nitrate causes the aerosols to increase to a size that scatters light effectively.

[0115] Organic matter (OM) can be either primary or secondary, the latter part deriving from the oxidation of VOCs; organic material in the atmosphere may either be biogenic or anthropogenic. Organic matter influences the atmospheric radiation field by both scattering and absorption. Another important aerosol type is constituted of elemental carbon (EC, also known as black carbon, BC): this aerosol type includes strongly lightabsorbing material and is thought to yield large positive radiative forcing. Organic matter and elemental carbon together constitute the carbonaceous fraction of aerosols.

[0116] The chemical composition of the aerosol directly affects how it interacts with solar radiation. The chemical constituents within the aerosol change the overall refractive index. The refractive index will determine how much light is scattered and absorbed.

[0117] Turning now to the figures, specific embodiments of the present disclosure shall be described. Figure 1A shows a highly schematic perspective view of first embodiment of an air cleaning system 1 according to the present disclosure, the air cleaning system 1 comprising a control unit 30 and an air cleaning device 10 arranged within a facility 100.

[0118] In the embodiments illustrated on the figures, the control unit 30 is provided as a standalone unit communicatively connected with the air cleaning devices 10 and the activity sensor 5. A detailed description of the control unit 30 is provided in later paragraphs with reference to Figure 10.

[0119] As illustrated, the air cleaning device 10 is installed within an industrial space such as a production hall (facility 100), the air cleaning device 10 commonly hanging from the ceiling and configured to remove at least a portion of contaminants from the facility 100 by recirculating air within the facility 100.

[0120] Further arranged in the facility 100 is an activity sensor 5. According to various embodiments, the activity sensor 5 comprises one or more of:

[0121] An electromagnetic sensor for detecting electromagnetic radiation associated with the industrial activity X and / or for detecting electromagnetic radiation transmitted by an electromagnetic transmitter attached to an industrial machine;

[0122] An acoustic sensor, such as a microphone, for detecting acoustic signals (noise) associated with the industrial activity X and / or for detecting an acoustic signal transmitted by an acoustic transmitter attached to an industrial machine using ;

[0123] A motion detector (such as a GPS sensor, a radar sensor, a LIDAR sensor and / or imaging device) for detecting physical motion associated with the industrial activity X; Means for detecting deviations of energy consumption (from a baseline or average consumption) associated with the industrial activity X, such as electrical consumption and / or consumption of a combustible energy source.

[0124] The air cleaning device 10 is configured to influence the indoor air quality by removing at least a portion of contaminants from the using one of more filters.

[0125] As shown on the figures, further arranged in the facility 100 is a contaminant sensor 40. In order to enable an accurate and true measurement of the indoor air quality within the facility 100, the contaminant sensors 40 is arranged such as to capture representative samples of air. In particular, the contaminant sensor 40 is not to be arranged so as to capture exclusively cleaned air expelled by the air cleaning device 10. Neither is the contaminant sensor 40 to be positioned such as to capture exclusively contaminated air. Fluid dynamics simulations may be used in order to determine the expected flow of air within the facility 100 so as to determine the positioning of the contaminant sensor 40 so as to be able to generate data indicative of air quality within the facility 100.

[0126] Determination of air quality involves the collection of air samples by the contaminant sensor 40. According to embodiments of the present disclosure, the contaminant sensor 40 use light scattering technology to determine mass concentration, preferably in realtime. A sample is drawn from the facility 100 into a sensing chamber of the contaminant sensor 40 in a continuous stream. One section of the aerosol stream is illuminated with a small beam of laser light. Particles in the aerosol stream scatter light in all directions. A lens at an angle (e.g. 90°) to both the aerosol stream and laser beam collects some of the scattered light and focuses it onto a photo detector. A detection circuitry of the contaminant sensor 40 converts the light into a voltage. This voltage is proportional to the amount of light scattered which is, in-turn, proportional to the mass concentration of the aerosol. The voltage is read by a processor and multiplied by an internal calibration constant to yield mass concentration. This value is made available by the contaminant sensor 40 as data indicative of air quality. The internal calibration constant is determined from the ratio of the voltage response to known mass concentration of a test aerosol. Light scattering-type contaminant sensor 40 responds linearly to the aerosol mass concentration. That is, for a monodisperse aerosol, one particle scatters a fixed amount of light; two particles scatter twice as much light; and 10 particles scatter 10 times as much light. The scattered light is dependent upon particle size. This dependence is most dramatic for particles with diameters (D) less than one third the wavelengths of the laser (~ 0.25 pm). For these small particles, the scattered light decreases as a function of the sixth power of the diameter. According to embodiments of the present disclosure, the laser diode used by the contaminant sensors 40i-k has a wavelength of 780 nanometres nm which allows detection of particles as small as about 0.1 pm. The scattered light is also dependent upon the index of refraction and light absorbing characteristics of the particles. Light scattering from particles can be modelled using a complex set of equations using Mie light scattering theory. The effect of particle size dependence on the mass concentration computed is greatest for monodisperse aerosols. For use cases when very accurate mass concentration readings are needed to monitor an environment where a specific aerosol type predominates, the contaminant sensor 40i-k is recalibrated for that aerosol. According to embodiments of the present disclosure, the contaminant sensor 40 is calibrated against a gravimetric reference using the respirable fraction of standard ISO 12103-1 , A1 test dust (Arizona Test Dust). This test dust has a wide size distribution covering the entire size range of the contaminant sensor 40 and is representative of a wide variety of ambient aerosols. The wide range of particle sizes averages the effect of particle size dependence on the measured signal. The sensing volume of the contaminant sensor 40 is constant and is defined by the intersection of the aerosol stream and the laser beam. Mass is determined from the intensity of light scattered by the aerosol within the fixed sensing volume. Since the sensing volume is known, the information can be easily converted by the contaminant sensor 40 to units of mass per unit volume (mg / m3). The optics inside the contaminant sensor 40 is kept clean by surrounding the aerosol stream in a sheath of clean filtered air. This sheath air confines the aerosol to a narrow stream and prevents particles from circulating around the optics chamber and collecting on the optics. Besides keeping the optics clean, this allows the contaminant sensor 40 to respond quickly to sudden changes in concentration.

[0127] According to particular embodiments of the present disclosure, the contaminant sensor 40 is battery-operated, data-logging, light-scattering laser photometers that provide realtime aerosol mass readings. They use a sheath air system that isolates the aerosol in the optics chamber to keep the optics clean for improved reliability and low maintenance. Suitable for clean office settings as well as harsh industrial workplaces, construction and environmental sites and other outdoor applications. The contaminant sensor 40 measures aerosol contaminants such as dust, smoke, fumes and mists.

[0128] Figure 1 B shows a highly schematic perspective view of a further embodiment of an air cleaning system 1 according to the present disclosure, the air cleaning system 1 comprising a control unit 30, a plurality of air cleaning devices 10i.n, a plurality of activity sensors 5i-m, and a plurality of contaminant sensors 40i-k arranged within a facility 100. In embodiments comprising a plurality of air cleaning devices 10i.n, the plurality of air cleaning devices 10i.nis controlled by the control unit 30 as a cluster such as to collaboratively mitigate the impact of the industrial activity. In embodiments comprising a plurality of activity sensors 5i-m, the plurality of activity sensors 5i-mare used in combination to detect instances of the industrial activity X. In embodiments comprising a plurality of contaminant sensors 40i-k, the plurality of contaminant sensors 40i-k is used in combination such as to accurately measure the first indoor air quality values and / or further indoor air quality values in the facility 100.

[0129] Turning now to figures 2 to 4, particular embodiments of the computer implemented method of operating an industrial air cleaning system according to the present disclosure is described in detail. Figure 2 shows a flow diagram illustrating a sequence of steps of a first embodiment of the computer implemented method of operating an industrial air cleaning system according to the present disclosure. As visible in figure 2, the illustrated embodiment of the method of operating an industrial air cleaning system comprises 2 sequences: an impact assessment a) and one or more preemptive action sequences b) .

[0130] In a first step a1) of the impact assessment sequence a), a plurality of observation instances of an industrial activity X within the facility 100 are detected by the activity sensor 5. In contrast to actionable instances, observation instances are not necessarily acted upon but observed (for gathering training data).

[0131] In a further step a2) of the impact assessment sequence a), one or more indoor air quality values are measured by the contaminant sensor 40 in the facility 100 during and / or following detecting the plurality of observation instances.

[0132] Having detected observation instances of an industrial activity X and related indoor air quality values, in a further step a3) of the impact assessment sequence a), impact data corresponding to the industrial activity X is determined by the control unit 30 based on the one or more first indoor air quality values and data representative of the plurality of observation instances of the industrial activity(X).

[0133] According to embodiments, determining impact data comprises training of a machinelearning algorithm - as will be described with reference to figure 3.

[0134] In a first step b1) of the one or more preemptive action sequences b), actionable instances of an industrial activity X are detected by the one or more activity sensors 5. In contrast to observation instances, actionable instances of an industrial activity X are also acted upon to mitigate its impact. Alternatively, or additionally, in the first step b1 of the preemptive action sequence b), an indication of commencement of an industrial activity X within the facility 100 is detected by the one or more activity sensors 5.

[0135] In a further step b2 of the preemptive action sequence B, one or more setpoint parameters of the air cleaning device is determined by the control unit 30 based on based on data representative of the one or more actionable instances of the industrial activity X, the impact data corresponding to the industrial activity X and one or more indoor air quality target values.

[0136] Finally, having determined the one or more setpoint parameters, in a further step b3 of the preemptive action sequence b, the air cleaning device 10 is controlled by the control unit 30 in accordance with the one or more setpoint parameters. In this way, the air cleaning system 1 can preemptively act to avoid degradation of indoor air quality before the contamination caused by the industrial activity X even occurs.

[0137] Figure 3 shows a flow diagram illustrating a sequence of steps of a training sequence c) of a machine-learning algorithm for determining the impact data. In a first step c1 of a training sequence c), data representative of the plurality of observation instances of the industrial activity X are provided as input of the datasets. In a further step c2) of training sequence c), data representative of the one or more indoor air quality values (measured by the contaminant sensor 40) are provided as expected output of the training datasets. Hence, the training dataset comprises data of observation instances “labelled” with the corresponding locations and levels of contamination caused by the observation instances of industrial activities. Such datasets can then be used for so-called supervised learning of the machine-learning algorithm, whereby labeled training data is required. Finally, having created the training datasets, in a further step c3) of the training sequence c), the machine-learning algorithm is trained using the plurality of training datasets. Figure 4 shows a flow diagram illustrating steps of employing the trained first machinelearning algorithm for determining an estimated impact of the one or more actionable instances of the industrial activity. In a step d1 , data representative of the actionable instance of an industrial activity X is fed to an input of the machine learning algorithm. In a further step d2, an estimated impact of the one or more actionable instances of the industrial activity X at an output of the first machine learning algorithm. Thereafter, in a step d3), the one or more setpoint parameters are determined further using the estimated impact.

[0138] Figure 5 shows a flow diagram illustrating steps of an efficiency assessment and improvement sequence. In order to monitor the efficiency of mitigation of the impact of the industrial activity and to improve mitigation of further actionable instances of the industrial activity, according to further embodiments, the method further comprises an efficiency assessment and improvement sequence. In a first step e1) of the efficiency assessment and improvement sequence, one or more further indoor air quality values in the facility 100 during and / or after the one or more preemptive actions sequences are measured by the contaminant sensor 40i-k. Thereafter, in a further step e2) of the efficiency assessment and improvement sequence, efficiency data corresponding to the industrial activity X and the one or more set-point parameters of the one or more air cleaning devices 10, 10i.nis determined by the control unit 30 using a comparison of the first indoor air quality value and the one or more further indoor air quality values. Based on the efficiency data, the determination of one or more set-point parameters corresponding to the industrial activity X is refined by the control unit 30.

[0139] Figure 6 shows a flow diagram illustrating steps of a set-point optimization sequence. In a first step f1) of the set-point optimization sequence, data representative of the one or more actionable instances of the industrial activity X and the one or more set-point parameters corresponding to the industrial activity X are fed to a second machine learning algorithm. In a further step f2) of the set-point optimization sequence, the one or more further indoor air quality values (the ones measured during and / or after the one or more preemptive actions sequences) are also fed to the second machine learning algorithm. Thereafter, in a third step f3) of the set-point optimization sequence, parameters of the second machine learning algorithm are optimized in order to minimize deviations of the indoor air quality values from the indoor air quality target values while also optimizing one or more set-point parameters, such as minimizing a set-point parameter indicative of an intensity of operation of the one or more air cleaning devices 10, 10l-n

[0140] Figure 7 shows a flow diagram illustrating steps of employing the trained second machine-learning algorithm for determining one or more set-point parameters. In a first step g1) - of employing the trained second machine-learning algorithm - the data representative of the one or more actionable instances of the industrial activity X are fed to an input of the second machine learning algorithm. In a second step g2) - of of employing the trained second machine-learning algorithm - the one or more set-point parameters of the one or more air cleaning devices 10, 10i.nare determined at an output of the second machine learning algorithm.

[0141] Figures 8 and 9 show a simplified block diagram, respectively a highly schematic perspective view of a first embodiment of an air cleaning device 10 according to the present disclosure. As illustrated in Figure 8, the air cleaning device 10 comprises a control unit 30; an air inlet 12; one or more air cleaning filters 14; air propelling means 15; and an air outlet 16. The air cleaning device 10 is configured to draw in air from the facility 100 through the air inlet 12; force at least a portion of the drawn-in air through the one or more air cleaning filters 14 to physically capture a portion of contaminants from the portion of the drawn-in air; and return at least a portion of the filtered air through the air outlet 16 back to the facility 100. The removal of airborne particulate is accomplished through mechanical, aerodynamic, and / or electrostatic means. According to embodiments of the present disclosure, the air cleaning device(s) 10 are configured to remove, i.e. filter out gas phase contaminants in the facility 100, in particular by molecular phase filtration. Molecular phase filtration refers to the filtration of gaseous contamination having size at the molecular scale (also called Gas Phase Filtration).

[0142] Mechanical air cleaning filters remove particles from the airstream as particles come in contact with the surface of fibres in the filter media and stick on to the fibres. Mechanical air cleaning filters operate based on sieving / straining, impaction / impingement, interception and / or diffusion of the contaminant particles.

[0143] Electrostatic filtration is a method for removing dust by letting air passing through an ionizer screen where electrons colliding with air molecules generate positive ions which adhere to dust and other small particles present, giving them a positive charge. The charged dust particles then enter a region filled with closely spaced parallel metal plates alternatively charged with positive and negative voltages. Positive plates repel the charged particles which are attracted by and retained on the negative plates by electrostatic forces, further supplemented by intermolecular forces, causing the dust to agglomerate. According to embodiments of the present disclosure, fibers are electrostatically pre charged and attract particles without an ionizer pre step.

[0144] According to embodiments of the present disclosure, the air cleaning filter 14 is a device composed of fibrous materials which removes solid particulates such as dust, pollen, mold and bacteria from the air flowing through it. Whether particulate or gas phases filters, they rely on a complicated set of mechanisms to perform their function. In many cases, more than one of these mechanisms comes into play. Many new technologies have been employed in the effort to improve on the quality and performance of air cleaning filters 14, and in some cases to reduce their Life Cycle Cost LCC. Some notable areas where advancement has been pursued are reduction in pressure drop and the application of various treatments to filter fibers.

[0145] An air cleaning filter 14 deteriorates over the course of its lifetime with respect to efficiency on the target contaminants, contaminant holding capacity and energy input requirement, etc. Filter efficiency, dust holding capacity and differential pressure can be measured in many ways, as the performance of an air cleaning filter 14 changes over time. The challenge imposed on air cleaning filters 14 changes as the environment inside and outside of a building changes. Many air cleaning filter testing methods have been developed by various organizations for predicting the in-use performance of filters and for comparing the performance of air cleaning filters of different designs.

[0146] The air cleaning devices 10 are configured to make available (e.g. though a data communication link) data indicative of their operational state. According to embodiments of the present disclosure, the operational state comprises data indicative of air cleaning filter performance. The performance of an air cleaning filter is generally evaluated based on four parameters. These include:

[0147] Contaminant removal efficiency: Determined by challenging the air cleaning filter 14 with contaminant on the upstream side and measuring the residual contaminant on the downstream side of the air cleaning filter 14 after the air has passed through the media.

[0148] Contaminant holding capacity: Determined by measuring the mass of contaminant removed before the air cleaning filter 14 reaches its maximum differential pressure in the case of particle filters 14 or before the contaminant breaks through the air cleaning filter 14 in the case of a gaseous filter.

[0149] Resistance to airflow: Determined by measuring the pressure of the air upstream of the air cleaning filter 14 and downstream of the filter 14 and comparing those values. The value of resistance to airflow must be accompanied by the value of airflow velocity in order to characterize the performance of the air cleaning filter 14. Safety: Measured by an air cleaning filter’s 14 resistance to fire when no other fuel source is present (the filter is therefore a fuel source), the amount of smoke generated by the filter 14, and the release of sparks by the filter 14 when exposed to the heat of a flame.

[0150] According to embodiments of the present disclosure, the operational state comprises data indicative of air cleaning filter 14 changing interval. The changing interval (lifetime) of the air cleaning filter 14 is highly dependent on how dirty the environment it is installed in. According to embodiments of the present disclosure the changing interval is between 6-12 months in most environments. In hard-contaminated environments the lifetime may be reduced down to 2-3 months, or even down to 1-2 weeks or days.

[0151] As symbolically illustrated on Figure 8, the air cleaning device 10 further comprises air propelling means 15 such as a fan. The air propelling means 15 is an electrically powered device used to produce an airflow for the purpose of drawing in air from the facility 100 through the air inlet 12; forcing at least a portion of the drawn-in air through the one or more filters 14 to physically capture a portion of contaminants from the portion of the drawn-in air; and returning at least a portion of the filtered air through the air outlet 16 back to the facility 100. According to embodiments of the present disclosure, the air propelling means 15 comprises a fan having a revolving vane or vanes used for producing an air current. The type and size of the fan of the air propelling means 15 is determined based on the amount of air that needs to be moved (e.g. based on the required Clean Air Delivery Rate of the air cleaning device 10). Furthermore, the type of fan is determined based on the required pressure differential between the inlet 12 and outlet 16. The air propelling means 15 further comprises an electrical motor for driving the fan. According to embodiments of the present disclosure, the air propelling means 15 comprises an Electronically Commutated EC motor, a brushless DC motor. Basic DC motors rely on carbon brushes and a commutation ring to switch the current direction, and therefore the magnetic field polarity, in a rotating armature. The interaction between this internal rotor and fixed permanent magnets induces its rotation. In an EC motor, the mechanical commutation has been replaced by electronic circuitry which supplies the right amount of armature current in the right direction at precisely the right time for accurate motor control. The electric motor according to embodiments of the present disclosure is further simplified by using a compact external rotor design with stationary windings. The permanent magnets are mounted inside the rotor with the fan impeller attached.

[0152] According to embodiments of the present disclosure, the operational state of the air cleaning devices 10 comprises data indicative of the clean air delivery rate CADR of the respective air cleaning device 10. Clean Air Delivery Rate CADR is a figure of merit that is the facility 100 delivered in a time period (e.g. m3 / h) that has had all the contaminant particles of a given size distribution removed.

[0153] For air cleaning devices 10 that have air flowing through its filters 14, CADR is the fraction of particles (of a particular size distribution) that have been removed from the air, multiplied by the air flow rate (in m3 / h) through the air cleaning device 10.

[0154] Air volume is often described as air exchange (the number of times the total facility 100 in a facility is processed by the air cleaning device 10 within a given period of time). CADR on the other hand not just shows how much air is cleaned nor just what percentages of particles are removed, but the overall performance of the filtration system 14 when both factors are examined. In other words, CADR shows how much volume of clean air the air cleaning device 10 is actually delivering to the facility 100.

[0155] In summary, according to embodiments of the present disclosure, the operational state of the air cleaning devices 10 comprises data indicate of: Revolutions per minute RPM of the fans of the air propelling means 15;

[0156] Air flow (Speed settings);

[0157] Date / Time logging events with respect to the operation of the air cleaning device 10;

[0158] Pressure drop (Pascal) over the filters 14;

[0159] Energy consumption of the air cleaning device 10;

[0160] Temperature(s) in the motor / electronics of the air propelling means 15;

[0161] Motor state / alarm(s); air cleaning filter 14 clogging calculations on an individual air propelling means 15; and

[0162] Sensor connectivity.

[0163] Figure 10 shows a simplified block diagram of a first embodiment of a control unit 30 according to the present disclosure. As schematically illustrated, the control unit 30 comprises processing means 32 and storage means 36, the storage means 36 comprising computer-executable instructions, which when executed by the processing means 32 cause the control unit 30 to carry out the method according to one of the embodiments disclosed herein. In the embodiment shown on Figure 10, the control unit 30 further comprises a communication unit 36 configured to establish data communication links with other control units 30.1 - 30. n; with air cleaning devices 10; and with air quality data sources and / or contaminant sensor(s) 40i-k. According to further embodiments of the present disclosure, the communication unit 36 is further configured to establish a data communication link with a remote computer 50.

[0164] It should be noted that, in the description, the sequence of the steps has been presented in a specific order, one skilled in the art will understand, however, that the computer program code of the computer implemented method may be structured differently and that the order of at least some of the steps could be altered, without deviating from the scope of the disclosure. LIST OF DESIGNATIONS air cleaning system 1 activity sensor 5, 5i-mair cleaning device 10i.ninlet (of air cleaning device) 12 air cleaning filter (of air cleaning device) 14 air propelling means (of air cleaning device) 15 outlet (of air cleaning device) 16 control unit 30 data storage (of control unit) 32 processing unit (of the control unit) 34 communication unit (of the control unit) 36 contaminant sensor 40, 40i-k remote computer 50 facility (100) 100 external ventilation 200 industrial activity

Claims

PATENT CLAIMS1. A computer implemented method of operating an industrial air cleaning system (1), comprising a control unit (30), one or more activity sensors (5, 5i_m), a contaminant sensor (40, 40i-k), and one or more air cleaning devices (10, 10i.n) configured to influence the indoor air quality within a facility (100), the method comprising: an impact assessment sequence comprising the steps of: a1) detecting, by the one or more activity sensors (5, 5i-m), one or more observation instances of an industrial activity (X) within the facility (100); a2) measuring, by the contaminant sensor (40, 40i-k), one or more first indoor air quality values in the facility (100) during and / or following detecting the plurality of observation instances; and a3) determining, by the control unit (30), impact data corresponding to the industrial activity (X) based on the one or more first indoor air quality values and data representative of the plurality of observation instances of the industrial activity (X); one or more preemptive action sequences comprising the steps of: b1) detecting, by the one or more activity sensors (5, 5i-m), one or more actionable instances of the industrial activity (X) and / or of an indication of commencement of an industrial activity (X) within the facility (100); b2) determining, by the control unit (30), one or more set-point parameters of the one or more air cleaning devices (10, 10i.n) based on data representative of the one or more actionable instances of the industrial activity (X), the impact data corresponding to the industrial activity (X) and one or more indoor air quality target values; and b3) controlling, by the control unit (30), the one or more air cleaning devices (10, 10i-n) in accordance with the one or more set-point parameters.

2. The method according to claim 1 , wherein:step a3) of determining impact data corresponding to the industrial activity (X) comprises a training sequence (c) of a first machine learning algorithm comprising: c1) providing data representative of the plurality of observation instances of the industrial activity (X) as input of a plurality of training datasets; and c2) providing data representative of the one or more first indoor air quality values as expected output of the training dataset; c3) training the machine-learning algorithm using the plurality of training datasets, wherein step b2) of determining one or more set-point parameters based on the impact data comprises: d1) feeding the data representative of the one or more actionable instances of the industrial activity (X) to an input of the first machine learning algorithm; d2) determining an estimated impact of the one or more actionable instances of the industrial activity (X) at an output of the first machine learning algorithm; d3) determining the one or more set-point parameters of the one or more air cleaning devices (10, 10i.n) further based on the estimated impact.

3. The method according to one of the claims 1 or 2, further comprising an efficiency assessment and improvement sequence comprising the steps of: e1) measuring, by the contaminant sensor (40, 40i-k), one or more further indoor air quality values in the facility (100) during and / or after the one or more preemptive actions sequences; e2) determining, by the control unit (30), efficiency data corresponding to the industrial activity (X) and the one or more set-point parameters of the one or more air cleaning devices (10, 10i_n), using a comparison of the first indoor air quality value and the one or more further indoor air quality values; and e3) refining, by the control unit (30), the determination of one or more set-point parameters corresponding to the industrial activity (X) in accordance with the efficiency data corresponding to the industrial activity (X).

4. The method according to claim 3, further comprising a set-point optimization sequence comprising: f1) feeding data representative of the one or more actionable instances of the industrial activity (X) and the one or more set-point parameters corresponding to the industrial activity (X) to a second machine learning algorithm; f2) feeding the one or more further indoor air quality values to the second machine learning algorithm; and f3) optimizing parameters of the second machine learning algorithm in order to minimize deviations of the indoor air quality values from the indoor air quality target values while also optimizing one or more set-point parameters, such as minimizing a set-point parameter indicative of an intensity of operation of the one or more air cleaning devices (10, 10i.n) whereby step b2) of determining one or more set-point parameters based on the impact data comprises: g1) feeding the data representative of the one or more actionable instances of the industrial activity (X) to an input of the second machine learning algorithm; g2) determining the one or more set-point parameters of the one or more air cleaning devices (10, 10i.n) at an output of the second machine learning algorithm.

5. The method according to one of the claims 1 to 4, further comprising classification of the observation instances and / or actionable instances of industrial activities into one or more categories of industrial activities, wherein detecting one or more actionable instances of an industrial activity (X) comprises identifying a category of industrial activities corresponding to the detected one or more actionable instancesof an industrial activity (X) and wherein determining one or more set-point parameters of the one or more air cleaning devices (10, 10i.n) comprises determining a type of filtering in dependence of the category of the one or more actionable instances of industrial activity (X).

6. The method according to one of the claims 1 to 5, wherein the step b3) of controlling, by the control unit (30), the one or more air cleaning devices (10, 10i- n) in accordance with the one or more setpoint parameters comprises controlling the one or more air cleaning devices (10, 10i.n) to: draw in air from the facility (100) through an air inlet and force, by air propelling means, at least a portion of the drawn-in air out of the facility (100); or draw in air from outside the facility (100) through an air inlet and force, by air propelling means, at least a portion of the drawn-in air into the facility (100).

7. The method according to one of the claims 1 to 6, wherein the step b3) of controlling, by the control unit (30), the one or more air cleaning devices (10, 10i- n) in accordance with the one or more setpoint parameters comprises controlling the one or more air cleaning devices (10, 10i.n) to: draw in air from the facility (100) through an air inlet; force, by air propelling means, at least a portion of the drawn-in air through an air filter to physically capture a portion of contaminants from the portion of the drawnin air; and return at least a portion of the filtered air through an air outlet back to the facility (100).

8. The method according to one of the claims 1 to 7, further comprising determining data representative of the plurality and / or the one or more actionable instances of the industrial activity (X) comprising one or more of: an intensity; a propagation speed, a direction, an acceleration; and / or an attenuation, of the plurality and / or of the one or more actionable instances of the industrial activity (X).

9. The method according to any of the claims 1 to 8, wherein determining one or more set-point parameters of the one or more air cleaning devices (10, 10i.n) comprises determining one or more of: a timing of air filtering, such as a timing after detection of the one or more actionable instances of the industrial activity (X); an air filtering direction; an air filtering intensity; and / or an air filtering type, such as an air filtering type dependent on a particle-size of contaminants.

10. The method according to any of the claims 1 to 10, wherein detecting, by the one or more activity sensors (5, 5i-m), observation instances and / or actionable instances of the industrial activity (X) comprises one or more of: detecting electromagnetic radiation associated with the industrial activity (X) using an electromagnetic sensor comprised by and / or communicatively connected to the one or more activity sensors (5, 5i.m);detecting acoustic signals associated with the industrial activity (X) using an acoustic sensor, such as a microphone, comprised by and / or communicatively connected to the one or more activity sensors (5, 5i.m); detecting physical motion associated with the industrial activity (X) using5 motion detectors, such as a GPS sensor, a radar sensor, a LIDAR sensor and / or imaging device , comprised by and / or communicatively connected to the one or more activity sensors (5, 5i.m); detecting electromagnetic radiation transmitted by an electromagnetic transmitter attached to an industrial machine using a detector comprised by and / or communicatively connected to the one or more activity sensors (5, 5i- m); detecting an acoustic signal transmitted by an acoustic transmitter attached to an industrial machine using a detector comprised by and / or communicatively connected to the one or more activity sensors (5, 5i.m); 5 detecting deviations of energy consumption associated with the industrial activity (X), such as electrical consumption and / or consumption of a combustible energy source.11 . The method according to one of the claims 1 to 10, further comprising: associating one of the one or more activity sensors (5, 5i-m) with the plurality0 of air cleaning devices (10, 10i_n); communicatively connecting each of the plurality of air cleaning devices (10, 10i.n); providing the control unit (30) as part of one or more of the plurality of air cleaning devices (10, 10i.n). 5 12. The method according to one of the claims 1 to 11 , further comprising providing the one or more contaminant sensors (40, 40i-k) as part of the plurality of air cleaning devices (10, 10i.n).

13. A control unit (30) for operating an industrial air cleaning system (1), the control unit (30) being configured to control an industrial air cleaning system (1) according to the method according to one of the claims 1 to 12.

14. An industrial air cleaning system (1), comprising a control unit (30), one or more activity sensors (5, 5i_m), a contaminant sensor (40, 40i-k), and one or more air cleaning devices (10, 10i.n) configured to remove at least a portion of contaminants from a facility (100) using a filter (14), the industrial air cleaning system (1) being configured to carry out the method according to one of the claims 1 to 12.

15. The industrial air cleaning system (1) according to claim 14, wherein the one or more air cleaning devices (10, 10i_n) are configured to: draw in air from the facility (100) through an air inlet (12); force, by air propelling means (15), at least a portion of the drawn-in air through the air cleaning filter (14) to physically capture a portion of contaminants from the portion of the drawn-in air; and return at least a portion of the filtered air through an air outlet (16) back to the facility (100).

16. The industrial air cleaning system (1) according to claim 14 or 15, wherein the one or more air cleaning devices (10, 10i_n) are each configured to: draw in air from the facility (100) through an air inlet and force, by air propelling means, at least a portion of the drawn-in air out of the facility (100); or draw in air from outside the facility (100) through an air inlet and force, by air propelling means, at least a portion of the drawn-in air into the facility (100).

7. Computer program product comprising computer-executable instructions, which when executed by a control unit (30) of an industrial air cleaning system (1) according to one of the claims 14 to 16 causes the air cleaning system (1) to carry out the method according to one of the claims 1 to 12.

Images