Method for controlling continuous granulation and drying process, and apparatus and system therefor

Abstract

The invention relates to a method for controlling a continuous granulation and drying process, in which a model takes into account a combination of granulation by means of a granulator and subsequent drying by means of a dryer, by means of which control parameters or predictive recipe parameters are ascertained as a function of state parameters of the plant; and / or wherein the model has a static part, by means of which a reference value for the respective control parameter or predicted recipe parameter is determined by means of the state parameter, and the model has a dynamic part, by means of which the reference value is optimized by means of prediction.

Description

Technical Field

[0001] The present invention relates to a method for controlling a device, and a device according to the preamble of claim 14 and a system with said device. Background Art

[0002] The background of the present invention is mainly the manufacture of pharmaceutical dosage forms, especially tablets, capsules or granules. In principle, however, the present invention can also be used in other technical fields, especially if the formulation has predetermined or predeterminable properties related to the particles of the formulation, such as a predetermined particle size distribution and / or humidity. This is the case, for example, if the formulation is to be made into tablets, which is particularly preferred in the pharmaceutical field, but can in principle also be used for detergents, foods or the like. Summary of the Invention

[0003] The present invention particularly relates to a method and a device for continuously manufacturing a formulation from a feed, preferably. It is highly particularly preferred that the device is controlled according to the proposed method, or the device is configured to produce a formulation having predetermined or predeterminable characteristics related to the particles of the formulation, such as homogeneity or particle distribution, and preferably a predetermined or predeterminable (relative) humidity from the feed.

[0004] It has been proven that combining a granulator and a dryer is beneficial for manufacturing a formulation from a feed. Thus, an intermediate product (granules) having a predetermined or predeterminable particle distribution can first be obtained from the feed using a granulator, and the intermediate product is then adjusted to a predetermined or predeterminable (relative) humidity by means of a dryer.

[0005] The difficulty here is that the changing characteristics of the intermediate product affect drying and thus the formulation.

[0006] Fluidized bed granulators are known in principle. In the solution disclosed herein, a determined particle distribution and the adjustment of the relevant relative humidity are achieved in the same step. However, the solution based on a fluidized bed granulator has disadvantages in reliably generating an accurate particle size distribution and humidity, and is also only applicable to certain feeds.

[0007] In contrast, it has proven more advantageous to first use a granulator and then a separately implemented dryer, because the components can also be used separately here, and a more accurate particle size distribution and relative humidity can be achieved if necessary, preferably at least substantially independently of each other. However, the proposed method is also advantageous in principle for controlling a fluidized bed granulator.

[0008] The combination of a granulator and a dryer is also known in principle, initially operating in batches. In a batch process, a batch of feed first undergoes a first manufacturing step, and then the entire batch undergoes another second manufacturing step, and finally a result, i.e., a formulation, is produced from the manufacturing process.

[0009] In contrast, in a continuous manufacturing process preferably based on the present invention, after the start phase, feeds are added simultaneously, the previously added feeds are placed in the manufacturing process, and the previously added feeds that have been fully placed in the manufacturing process are removed. Thus, in a continuous process, feeds are added simultaneously and the result in the form of a formulation is removed.

[0010] The present invention preferably relates to a continuous manufacturing or continuous method of a formulation and an apparatus therefor or a system having the apparatus, preferably as opposed to a batch process. The advantages of a continuous process are:

[0011] - Simple scalability (scalable according to time and throughput)

[0012] - Small space requirement of the apparatus

[0013] - Shorter downtime of the apparatus compared to batch apparatus

[0014] - Achieving a higher degree of automation

[0015] - Higher product quality

[0016] In this context, the object of the present invention is to provide a method, an apparatus and a system by means of which the process of manufacturing a formulation from feeds can be improved with respect to the reliable and stable achievement of various properties, such as in particular particle size distribution and humidity.

[0017] This object is solved by the method according to claim 1, the apparatus according to claim 14 or the system according to claim 15. Advantageous developments are the subject matter of the dependent claims.

[0018] On the one hand, the present invention relates to a method for controlling an apparatus for manufacturing a formulation from feeds, wherein the manufacturing comprises processing the feeds using a granulator and drying an intermediate product produced from the feeds using the granulator with the aid of a dryer.

[0019] In a first variant of the present invention, the control parameters of the apparatus are determined based on a model. Here, the model takes into account predetermined target formulation parameters which represent the desired properties of the formulated product that has been produced or is to be produced.

[0020] The desired properties of the formulated product that has been produced or is to be produced are in particular the expected particle size properties and humidity.

[0021] Here, the model can be transmitted with variably predeterminable target formulation parameters, and the model uses the target formulation parameters to determine the control parameters. However, alternatively or additionally, the target formulation parameters can also be or have been taken into account when forming the model.

[0022] A control parameter is or represents a manipulation parameter for actuating an actuator of a device.

[0023] Determine the state parameters of the device and process them through a model. For this purpose, preferably transmit the state parameters to the model, and the model uses the state parameters to determine the control parameters.

[0024] The state parameters respectively represent the states of the device that affect manufacturing, preferably sensor values.

[0025] In addition, feed parameters are considered by the model. In particular, variably pre-given feed parameters are transmitted into the model, and the model uses the feed parameters to determine the control parameters. However, alternatively or additionally, when forming the model, the feed parameters can also be or have been at least partially considered.

[0026] The feed parameters represent the properties of the feed, especially the humidity and / or particle size characteristics.

[0027] In one aspect, the dryer is connected to the granulator in such a way that the intermediate product is automatically and continuously conveyed from the granulator to the dryer. According to the suggestion, the model takes into account the combination of continuous operation of granulation with the granulator and drying with the dryer.

[0028] In a second aspect that can be combined with the first aspect, the model has a static part and a dynamic part. With the help of the static part, a reference value for the corresponding control parameter is obtained, and the reference value is optimized by the dynamic part with the help of prediction.

[0029] In a second variant of the present invention, the actual formulation parameters are based on a (same or different) model prediction.

[0030] The model takes into account the control parameters predetermined or predeterminable in this regard. Preferably, variably predeterminable control parameters are transmitted to the model, and the model uses the control parameters to predict the actual formulation parameters. However, alternatively or additionally, when forming the model, the control parameters are at least partially considered or have been considered.

[0031] The actual formulation parameters represent the actual characteristics of the formulated product that has been manufactured or to be manufactured, especially the actual particle size characteristics and humidity.

[0032] Determine the state parameters as in the first variant, and process them by the model, and the feed parameters are considered by the model.

[0033] In addition, as in the previous variant, in one aspect, the dryer is coupled to the granulator such that the intermediate product is automatically and continuously conveyed from the granulator into the dryer. According to the suggestion, the model takes into account the combination of continuous operation of granulation with the granulator and subsequent drying with the dryer.

[0034] In a second aspect that can be combined with the first aspect, the model also has a static part and a dynamic part in a second variant, and a reference value for a corresponding control parameter is obtained using the static part, and the reference value is optimized using the dynamic part by prediction.

[0035] In the proposed method, first, feed parameters are predetermined or considered for controlling an apparatus for manufacturing a formulation from the feed or for supporting the control, and the feed parameters represent the state of the feed, in particular the humidity and / or particle characteristics, such as particle size distribution.

[0036] The feed referred to in the present invention is preferably a granulable material, i.e., a material that can be processed into pellets by a granulation process. The feed is very particularly preferably a powder or pellets, the particle characteristics of which can be changed by the granulation process.

[0037] Furthermore, the feed is preferably a material mixture, i.e., at least substantially homogeneous mixture of different components (preferably solid components). These components may include active materials (in particular pharmacologically active materials or other active materials), fillers and / or disintegrants. In particular, the feed is at least a substantially homogeneous powder mixture.

[0038] When manufacturing a formulation using the feed, the material parameters of the feed passing through the apparatus, via its intermediate products to the final product (formulation) are preferably not determined by sampling and analysis next to the apparatus. Instead, only parameters measurable during continuous, ongoing operation are used.

[0039] Either the material parameters are not determined at all, or at most the material parameters measurable online are determined, such as the measurement values of non-contact measurement methods, reflection and / or transmission measurements, in particular using infrared radiation, for example as an indicator of the material humidity. The particle size distribution is preferably not measured during the manufacturing process of at least the intermediate products.

[0040] The control parameters of the apparatus are or represent the manipulation parameters of an actuator for controlling the apparatus, and the control parameters can be determined for the proposed control. Thus, the apparatus can be controlled using these control parameters. Preferably, it is implemented such that different actuators of the apparatus are controlled using the control parameters, so that the control parameters have an impact on the feed or the intermediate products.

[0041] Alternatively or additionally, actual formulation parameters representing the characteristics of the manufactured formulation under predetermined boundary conditions can also be predicted to support the control. This can be done on the basis of predetermined or predeterminable control parameters.

[0042] In other words, the actual formulation parameters can be predicted by preferably manually predefining or inputting control parameters and preferably outputted. In this way, the user can compare with the target formulation parameters (preferably manually again) and predefine the changed control parameters in order to adjust the predicted actual formulation parameters to the target formulation parameters. The basis for the control of the device is preferably based on the changed control parameters here.

[0043] Preferably, at least one granulator drive and the feeding of desiccant, in particular (conditioned) air, are controlled with the control parameters. Additionally, the feeding device for the feed, the injection of liquid during granulation, one or more temperature control devices of the granulator, the conveying device for the desiccant for setting the volume flow rate of the desiccant, and / or the temperature control device for the temperature control of the desiccant can also be controlled with the control parameters.

[0044] In order to control the device or for supporting said control, state parameters of the device are determined, in particular one or more sensor values, and the state parameters preferably each represent the following states of the device, and these states affect the process of manufacturing the formulation from the feed. This especially includes temperature and / or pressure or pressure difference and / or torque and / or volume flow rate. However, other parameters or sensor values for describing the state of the device are also conceivable.

[0045] However, the state parameters preferably do not describe the material properties of the feed or the intermediate product (granules) or the final product (formulation) formed from the feed, or at least do not directly describe them.

[0046] In any case, preferably, during the continuously operating granulation and drying process, the particle size distribution, size, shape, density, and / or active ingredient content of the feed or the intermediate product formed therefrom are not determined. In this regard, the present invention adopts a completely different solution from the prior art. On the contrary, the characteristics of the feed can be determined in advance, and the characteristics of the final product (i.e., the formulation) can be determined for verification and / or model formation after completion.

[0047] The target formulation parameters are or are preferably predefined for control, and the target formulation parameters represent the desired characteristics of the formulated or to-be-formulated formulation, in particular one or more physical characteristics, such as the particle size, particle size distribution, particle shape, or density and / or humidity of the formulation.

[0048] Furthermore, for model generation and / or verification, the actual formulation parameters are preferably determined by characterizing the formulation, and the actual formulation parameters represent the actual characteristics of the formulated or to-be-formulated formulation, in particular one or more physical characteristics, such as the particle size, particle size distribution, particle shape, or density and / or humidity of the formulation.

[0049] Finally, a control parameter is determined based on the model. Here, during the process for manufacturing the formulation, the device can be controlled with the control parameter, which is preferably determined by processing the state parameter with the model.

[0050] Preferably, (measured) actual formulation parameters are used to derive or define the model. However, the (measured) actual formulation parameters are preferably not used as the basis for controlling the device during the running process.

[0051] Therefore, the control parameter is preferably not determined by processing the (measured) actual formulation parameters, nor is it derived from the actual formulation parameters. This is because a surprising fact shows that deriving the control parameter from the (measured) actual formulation parameter starts too late. If the (measured) actual formulation parameter deviates from the target formulation parameter during the running process, a large amount of waste products have already been produced. However, the object of the present invention is to avoid such waste products. To this end, on the contrary, it is preferably to control the device (at least substantially) independently of the (measured) actual formulation parameter, or the device is constructed for this purpose.

[0052] Preferably, (measured) actual formulation parameters are used to determine the model or derive a model form scheme, from which the control parameter is derived from the state parameter of the device during the process.

[0053] The model preferably has a machine learning-based structure, especially a neural network.

[0054] As described above, according to the first aspect of the present invention, the device has a granulator and a dryer for processing the feed into an intermediate product, and the dryer is coupled with the granulator to automatically and continuously convey the intermediate product from the granulator into the dryer.

[0055] The model takes into account, especially describes, the combination of granulation with the granulator and subsequent drying with the dryer. With this model, the control parameter, especially the reference value of the control parameter, can be determined according to the state parameter of the device, and preferably the feed parameter.

[0056] Alternatively or additionally, predicted (i.e., unmeasured) actual formulation parameters can also be determined according to the model. Here, the model can be different from the model for determining the control parameter. The predicted actual formulation parameters can be output for (manual) comparison with the target formulation parameter, and the control parameter can be adjusted (manually or automatically) if necessary. In this case, the control parameter is preferably predetermined to the model.

[0057] Optionally, one or more online measurable characteristics of the intermediate product, especially one or more optically measurable parameters on the intermediate product, can be additionally taken into account. However, it is preferably to avoid that determining the measured value requires sampling, separate analysis from the device, or interruption of the manufacturing process.

[0058] Therefore, in addition to the status parameters of the equipment, it is not excluded to obtain certain (physical) online-measurable characteristics (actual recipe parameters) of the feedstock and / or intermediate products and / or recipes, or to use the online-measurable characteristics as input parameters for regulation or for the model, such as the humidity and / or particle size distribution of Recipe 7 or parameters determined therefrom, especially for online measurement in the case of non-tip continuous granulation and drying.

[0059] Preferably, the measurement of the particle size distribution is carried out mostly or only in the recipe and not in the intermediate product. Accordingly, the equipment may have sensors for determining the properties of the particles describing the recipe, such as the particle size distribution, but preferably only at the recipe outlet or downstream thereof for discharging the recipe after the intermediate product has been dried. Sensors provided as part of the equipment or a system with the equipment are in particular online probes with spatial filtering anemometers for particle size measurement. However, other principles may also be used here.

[0060] The humidity is preferably determined from the recipe, but alternatively or additionally may also be determined from the intermediate product. For this purpose, one or more sensors may be provided. In particular, optical sensors, especially sensors based on infrared radiation, are particularly preferred. In this case, sensors based on near-infrared radiation, especially sensors of the NIR-2 spectrum with wavelengths of 860 to 1040 nm, have proven to be particularly advantageous.

[0061] It has been proven that it is particularly advantageous if the model has a static part that is used to obtain reference values for the corresponding control parameters or predicted actual recipe parameters from the feed parameters and status parameters, and the model also has a dynamic part that is used to optimize or can optimize the reference values by means of prediction, so that the model generation of the combination of the granulator and dryer is both independent and mutually promoting.

[0062] The static part of the model can be defined by determining model parameters that are determined correspondingly in a plurality of respective stationary (steady state, at least substantially static) states of the equipment, especially by measurement.

[0063] The static part of the model can be used to determine reference values for control parameters and / or predict actual recipe parameters from status parameters. The static part of the model is preferably predetermined invariantly.

[0064] The dynamic part of the model can continuously adjust the behavior of the model. For example, due to wear, material expansion, aging, or similar reasons, the behavior of the device changes over time. Instead of directly considering such or similar transient effects by changing control parameters or the actual recipe parameters of the prediction (especially by scaling and / or correcting by adding / subtracting correction terms), according to the suggestion, the model is adjusted so that the model generates corrected control parameters or actual recipe parameters from the state parameters.

[0065] One or more (possibly different) parameters (such as state parameters) are processed by the model. Here it can be set that the parameters form the input values of the model, i.e., especially the input values of the static part of the model and / or the dynamic part of the model.

[0066] The model (especially its static part) can have an artificial neural network. The parameters can be passed to the nodes of the input layer of the artificial neural network, and the artificial neural network can utilize the parameters to generate numerical values at the nodes of the output layer of the artificial neural network.

[0067] The values generated at the nodes of the neural network of the static part of the model in the output layer can be optimized by the dynamic part of the model, for example, by processing by considering the same and / or other parameters.

[0068] Another aspect of the invention that can also be implemented independently relates to a computer program product or a computer-readable storage medium that contains instructions which, when executed by a computer, cause the computer to execute the method according to one of the implemented aspects.

[0069] Another aspect of the invention that can also be implemented independently relates to a device for manufacturing a recipe from a feedstock, wherein the manufacturing includes processing the feedstock with a granulator and drying the intermediate product manufactured with the granulator by means of a dryer:

[0070] The device has sensors for detecting the state parameters of the device, and the state parameters respectively represent the state of the device that affects the manufacturing. The device has actuators for directly or indirectly influencing the feedstock. In addition, the device has a control device and a model, and the control device is used to control the actuators with control parameters, wherein target recipe parameters are predetermined or can be predetermined for the control device, and the control parameters can be obtained based on the model.

[0071] The granulator and the dryer are coupled such that the intermediate product is automatically and continuously conveyed from the granulator to the dryer, wherein the model takes into account the combination of granulating with the granulator and the subsequent continuous drying with the dryer, and the device is configured to enable the control device to obtain the control parameters based on the state parameters of the device using the model.

[0072] Alternatively or additionally, the model has a static part, wherein the control device is configured to determine a reference value for a respective control parameter from feed parameters and status parameters using the static part, and the model has a dynamic part, wherein the control device is configured to optimize the reference value using the dynamic part by means of prediction. The device is preferably controlled by means of the generated optimized reference value, or the device can be controlled hereby.

[0073] Another aspect of the invention, which can also be implemented independently, relates to a system, which comprises: the proposed device; and a device for forming a feed from a plurality of components (preferably powders, preferably by sieving); and / or a device for further processing the formulation, preferably tabletting.

[0074] A "formulation" in the sense of the invention is a material that has passed through the device and has been changed in its physical properties hereby. Thus, the formulation is preferably a product that combines granulation and drying. Further processing of the formulation, for example by tabletting, is not excluded, and the formulation preferably exists as (dried or preferably humidity-adjusted in its (relative) humidity) pellets.

[0075] A "feed" in the sense of the invention is preferably a material that is fed into a device or a granulator to change its physical properties. The feed is preferably an active material - filler mixture. However, this is not necessary. The feed can be pre-treated, for example by mixing at least substantially homogeneously one powder with another powder or other substances, where one can be or have an active material. The active material is preferably a pharmacologically active substance.

[0076] A "granulator" in the sense of the invention preferably relates to a device that mechanically processes a feed to change its physical properties. Particularly preferably, the granulator can convert the feed into pellets, especially coarse (granular) powders. For this purpose, the feed can be fed into the granulator as a powder in order to process the powder into a powder with coarser or finer particles. The intermediate product preferably relates to solids.

[0077] The granulator preferably conveys the feed while acting on it. The granulator preferably generates pressure and / or friction in the feed, preferably under temperature regulation / heating and / or provision of humidity. Particularly preferably, the granulator can change the particle size, particle diameter or particle size distribution of the feed. In the present invention, the granulate is an intermediate product that is subsequently further processed.

[0078] The granulator can have or consist of an extruder. In particular, it relates to a screw extruder, preferably a twin-screw extruder, or the granulator has a screw extruder or is similar to a screw extruder. However, other solutions are also conceivable.

[0079] The granulator is preferably a worm granulator, such as a twin-screw granulator. In a worm granulator, the feed is conveyed and processed by a worm shaft rotating around a rotation axis. For this purpose, the worm shaft can have different leads and / or worm channels with different processing structures. In a twin-screw granulator, two worm shafts are provided, which are arranged parallel to each other and / or cooperate with each other and effect conveyance and processing. In principle, other solutions can also be used here. The conveyance is preferably effected in an extrusion manner. The worm is thus or forms one or more extruders, or effects the extrusion of the feed.

[0080] The granulator preferably relates to a granulator for wet granulation. For this purpose, the granulator can have a liquid injection for humidifying the feed, and then the humidified feed is processed into an intermediate product by means of the granulator. The granulator can have openings, in particular nozzles or valves, for adding a liquid, in particular water, ethanol, isopropanol and / or mixtures thereof, to the feed to (temporarily) increase the humidity.

[0081] The "dryer" in the sense of the present invention relates to a device for reducing the (relative) humidity or water content of a substance, and is currently used for reducing the (relative) humidity or water content of the feed processed into an intermediate product by a granulator. The dryer is preferably a device for removing moisture from the substance / intermediate product. The device preferably enables the substance / intermediate product to come into contact with a desiccant that can remove the moisture in the substance.

[0082] The desiccant preferably relates to air or other gases with a relative air humidity that enables the reception of water from the substance. The desiccant is preferably temperature-controlled, in particular to a temperature higher than the ambient temperature. The desiccant thus particularly relates to pre-treated, warm and / or dry air, also called process air. However, in principle, it can also be other substances, in particular inert gases or others, preferably gaseous desiccants.

[0083] The dryer preferably relates to a fluidized bed dryer, also called a vortex layer dryer. The "fluidized bed dryer" in the sense of the present invention is a device for generating an air cushion or gas (process air) for the substance to be dried (here the intermediate product / granules). In this case, the air or gas is preferably a gaseous desiccant.

[0084] The gaseous desiccant is preferably fed into the bed of the dryer from the intermediate product / granules, preferably through a perforated distribution plate. The gaseous desiccant preferably flows through the bed at a speed such that the particles of the intermediate product / granules are kept in a fluidized state despite their large weight. The fluidized particles of the intermediate product / granules form a fluidized bed. It is dried by the preferably gaseous desiccant.

[0085] Bubbles can form and collapse within the fluidized bed to promote the intensive movement of the particles. In this state, the solid behaves like a freely flowing boiling liquid. Due to the close contact between the individual particles and the preferably gaseous desiccant, the heat and mass transfer rates are very high. Although the fluidized bed dryer has proven to be particularly advantageous in the context of the present invention, other dryer concepts can in principle also be used.

[0086] In the context of the present invention, a sensor for detecting a status parameter of a device (where the status parameter represents a state of the device that affects manufacturing) is preferably a sensor that characterizes the device status, i.e., designed and set to measure one or more status parameters.

[0087] The "status parameter" in the context of the present invention can be or represent one or more of the following attributes:

[0088] · The torque and / or speed of a drive device, tool, and / or conveying device (especially a propeller, turbine, worm / extruder), or the corresponding parameters here, such as current consumption and / or rotational speed

[0089] · Mass flow (mass flow of desiccant / process gas, feed, additive / granulating liquid, intermediate product, exhaust gas, final product / formulation) or the corresponding parameters here that characterize, for example, the position state of a valve, flap, rotor speed, pressure difference, etc.

[0090] · The temperature of components of the device that are preferably in direct or indirect contact with the feed being processed, with the additive / granulating liquid, or with the desiccant, the temperature of the tool for processing the feed or intermediate product, the temperature of the device for adding the additive or introducing the desiccant

[0091] The status parameter preferably does not (directly) characterize the (physical or chemical) properties of the feed or the intermediate product or final product or formulation formed therefrom. Therefore, measurement parameters that characterize the chemical composition or particle size of the feed or the intermediate product or final product (formulation) formed therefrom are not considered status parameters of the device.

[0092] In principle, two groups of status parameters can be distinguished.

[0093] The first group includes status parameters that are at least substantially independent of the feed, particularly in such a way that the status parameters are directly predetermined by the actuators of the device. From now on, these status parameters are also referred to as pre - determinable or "feed - independent status parameters". For example, a pre - determinable temperature or rotational speed is such here. They are particularly useful for actuator regulation.

[0094] The second set of status parameters can be distinguished from the "status parameters not affected by the feedstock", and the second set of status parameters is affected by the interaction with the feedstock or with the intermediate products formed from the feedstock. The second set of status parameters is referred to as "status parameters affected by the feedstock". For example, here are the torque set according to the consistency of the feedstock or the (relative) humidity of the exhaust gas moistened by the drying process.

[0095] One or more status parameters affected by the feedstock are preferably used as input parameters of the model. Therefore, the status parameters processed by the model or transmitted to the model for this purpose and used by the model to determine control parameters or predict actual formulation parameters are preferably status parameters affected by the feedstock.

[0096] The status parameters processed by the model or transmitted to the model for this purpose and used by the model to determine control parameters or predict actual formulation parameters are preferably at least one status parameter affected by the feedstock of the granulator, further preferably at least one status parameter affected by the feedstock of the granulator and the dryer, especially at least two status parameters affected by the feedstock of the granulator and at least one (preferably at least two) status parameter affected by the feedstock of the dryer.

[0097] One or more parameters not affected by the feedstock are preferably used to adjust the corresponding actuators. Alternatively or additionally, one or more status parameters affected by the feedstock are used to determine one or more control parameters, preferably determined with the aid of a model. The control parameters can be the specified targets for the adjustment of one or more actuators, based on which the actuators are adjusted.

[0098] The "on-line measurable characteristics" in the sense of the present invention are preferably characteristics that can be occasionally measured during the continuous manufacturing process of uninterrupted operation.

[0099] The "actuators" acting on the feedstock in the sense of the present invention are preferably drive devices of tools and / or conveying devices. Here it can be a motor, which is used, for example, to drive a fan, a turbine, a conveyor belt and / or a worm, and can also be a temperature control device, a heating or cooling device (such as a cooling water conveying device) for the temperature control of the housing part of the granulator for conveying the feedstock. The temperature control device for the temperature control of the process gas, especially a heating or cooling device (such as a heating jacket), can also be an actuator, because the actuator indirectly acts on the feedstock through the temperature of the desiccant / process gas.

[0100] The "control device" for controlling the actuator in the sense of the present invention is preferably an electronic component for influencing the operation of the actuator, the control of the motor speed (worm drive device and / or fan motor) and / or for controlling or adjusting the temperature control device or the heating of the granulator or the desiccant.

[0101] In the context of the present invention, a "control parameter" is or represents a manipulation variable for controlling an actuator, and the control parameter is preferably a value that serves as a specification for the operation of the actuator, in particular one or more predetermined motor speeds (of a worm drive and / or a fan motor) or the corresponding current consumption therefor and / or a specification for controlling or regulating a granulator or a desiccant or a similar temperature-regulating device or heating section.

[0102] A "feed parameter" in the context of the present invention is preferably a parameter that characterizes a preferred physical property of the feed, such as the particle size, grain size, grain size distribution, (relative) humidity and / or temperature of the feed.

[0103] The state of the feed in the context of the present invention is preferably determined at least by its humidity and / or grain size distribution, optionally supplemented by other properties, or replaced by corresponding data that directly or indirectly indicate the humidity and / or grain size distribution or are derived therefrom.

[0104] A "target formulation parameter" in the context of the present invention is preferably a parameter that represents the desired properties of a formulated product that has been or is to be manufactured, in particular the grain size or grain size distribution and humidity, which may also be supplemented by other properties, or replaced by corresponding data that directly or indirectly represent the humidity and / or grain size distribution or are derived therefrom.

[0105] An "actual formulation parameter" in the context of the present invention refers to a parameter measured on a formulation, which represents the properties of a formulated product that has been or is to be manufactured, in particular the particle size, particle size distribution and humidity, optionally supplemented by other properties, or replaced by corresponding data that directly or indirectly represent the humidity and / or particle size distribution or are derived therefrom.

[0106] Instead of measuring, the actual formulation parameter can be predicted, but is hereinafter referred to as the predicted actual formulation parameter.

[0107] A "model" in the context of the present invention is preferably an (abstract) representation, preferably limited to basic properties. Here, the model is preferably a model of a device and mathematically represents the characteristics of the device directly or indirectly through its influence on the feed and / or intermediate products. For this purpose, the model can have a mathematical description of a granulator, a dryer and / or their influence on the feed and / or intermediate products or be formed therefrom. The model preferably enables the determination of control parameters or the prediction of formulation parameters based on the state parameters of the device and the feed parameters (where applicable).

[0108] The "coupling device" of the granulator and dryer in the sense of the present invention is preferably a device for conveying the intermediate product to the dryer, and may have a conveying device for such conveying, such as a chute, a conveyor belt, a worm or the like. The coupling device preferably ensures that the intermediate product is automatically, continuously and uninterruptedly, that is, "on-line", conveyed from the granulator to the dryer. For this purpose, the granulator may also have an outlet directly leading into the dryer. In this case, the coupling device is realized by the granulator or by the conveying effect of the granulator.

[0109] The "reference value" of the control parameter in the sense of the present invention is preferably a control parameter set as a basic setting, and is preferably determined or preset independently of the measured characteristics of the final product (formulation). The reference value of the predicted actual formulation parameter is the starting point of the prediction.

[0110] The "static part" of the model in the sense of the present invention is preferably the part of the model based on empirical values, and the reference value can be determined or predicted by means of the part based on empirical values.

[0111] The "dynamic part" of the model in the sense of the present invention is preferably the part of the model that can be dynamically changed according to the state parameters of the equipment, that is, the part that can be changed during the operation of the equipment when the feed is processed into the intermediate product and the final product (formulation). In order to make the model adapt to any (future) state change of the equipment or process accordingly, it is preferably to make the model represent the behavior of the equipment or process for manufacturing the formulation from the feed with sufficient accuracy. The dynamic part of the model preferably can optimize the reference value by means of prediction. For this purpose, past developments can be considered. Description of the Drawings

[0112] Other aspects, advantages and features of the present invention are derived from the claims and the following description of the preferred embodiments with reference to the drawings. It shows:

[0113] Figure 1 A schematic cross-section of the proposed equipment;

[0114] Figure 2 According to Figure 1 A simplified block diagram view of the control-related components of the shown equipment;

[0115] Figure 3 A simplified schematic view of an artificial neural network;

[0116] Figure 4 A schematic diagram of the past and predicted temporal trends;

[0117] Figure 5 A schematic diagram of the result of the control with constant control parameters over time;

[0118] Figure 6 Schematic diagram of predictions offset in time from each other;

[0119] Figure 7 Schematic diagram of process characteristics changing over time;

[0120] Figure 8 Schematic flow chart;

[0121] Figure 9 Devices embedded in the system. Detailed implementation mode

[0122] In the drawings, the same or similar components are denoted by the same reference numerals, and the same or similar characteristics and advantages can be achieved even if repeated descriptions are omitted for clarity.

[0123] Figure 1 Schematic cross-sectional view showing the device 1 according to the suggestion for manufacturing the formulation 2 from the feed 3. The device 1 is preferably configured to perform a method for manufacturing the formulation 2 from the feed 3.

[0124] The feed parameters 4 are predetermined for the feed 3, and the feed parameters represent the state of the feed 3, in particular humidity, composition and / or particle size characteristics, such as particle size distribution.

[0125] The device 1 is preferably configured to be able to determine the control parameters 5 of the device 1, and the control parameters are or represent the operating parameters for controlling the actuator 6 of the device 1.

[0126] In addition, the state parameters 7 of the device 1 are preferably determined, or are determinable by means of the device 1, for example determined by the sensor 8, in particular the sensor values or the parameters derived therefrom as the state parameters 7, wherein the state parameters 7 respectively represent the state of the device 1 affecting the manufacturing.

[0127] The actuator 6 of the device 1 can be controlled by means of the control device 9 of the device 1 to control or affect the manufacturing process of forming the formulation 3 from the feed 2.

[0128] The target formulation parameters 10 can be predetermined or can be predetermined, and the target formulation parameters represent the desired characteristics of the formulated product 2 that has been manufactured or is to be manufactured, in particular physical characteristics, such as particle size characteristics, in particular particle size distribution and / or humidity.

[0129] The target formulation parameters 10 can be stored and / or retained in the database 10A. The database 10A can be read out by the control device 9, or the control device 9 can retrieve the target formulation parameters 10 from the database 10A and use them for control.

[0130] Actual recipe parameters 11 may be provided or measured, which represent actual properties of the recipe 3 manufactured or to be manufactured, in particular physical properties, such as particle size properties (such as particle size distribution) and / or moisture. The actual recipe parameters 11 are or preferably have data on properties of the recipe 2, which are measured in a separate, in particular separate, analysis process from the device 1, preferably not online or real-time / delayed measurements.

[0131] The device 1 , in particular the control device 9 , is preferably designed to ascertain the control variable 5 based on the model 12 .

[0132] The proposed model 12 can alternatively or additionally be used independently of the control device 9 or without directly influencing or preferably automatically controlling the actuator 6, preferably for predicting and / or outputting predicted (expected under given boundary conditions) actual recipe parameters 11 of the recipe 2. For this purpose, one or more properties of the recipe 2 can be predicted from one or more state parameters 7 and control parameters 5 and preferably feed parameters 4 by means of the model 12.

[0133] In one aspect of the invention, the apparatus 1 has a granulator 13 for processing the feed material 2 into an intermediate product 14 and a dryer 15 for the intermediate product 14 .

[0134] The dryer 15 is coupled to the granulator 13 so that the intermediate product 14 is automatically and / or continuously conveyed from the granulator 13 to the dryer 15. Granulation and drying preferably form a common continuous process.

[0135] Figure 2 A simplified block diagram of control-related components of the device 1 is shown in FIG.

[0136] The control device 9 is preferably designed to control the system 1 or the combination of a granulator 13 and a dryer 15. The control is preferably based on a model 12, which describes the behavior of the system 1 or the combination of a granulator 13 and a dryer 15 and enables the determination of control parameters 5 based on at least one or more state parameters 7 for controlling actuators 6 of the system 1 or the combination of a granulator 13 and a dryer 15. It is particularly preferred that the model 12 takes into account the combination of granulation with the granulator 13 and subsequent drying with the dryer 15.

[0137] The control parameter 5 , in particular an initial reference value for the control parameter 5 , is preferably ascertained using the model 12 , preferably based on the state parameter 7 of the system 1 and further preferably based on the feed parameter 4 .

[0138] To this end, the model 12 can have a static part 12A, with which a reference value for the respective control parameter 5 is determined taking into account the feed parameters 4 and the state parameters 7; and the model 12 can have a dynamic part 12B, with which the (respective) reference value can be adjusted, preferably optimized by means of prediction.

[0139] In other words, the static part 12A of the model 12 is preferably formed by means of machine learning, using which the reference value (basic setting) of the control parameter 5 is determined, and then the control parameter can be adjusted and finally determined by means of the dynamic part 12B of the model 12, so as to ultimately be used to control the actuator 6.

[0140] The dynamic part 12B of the model 12 preferably fine-tunes the reference value or basic setting determined by the static part 12A of the model 12. Here, the reference value of the control parameter 5 determined by the static part 12A is preferably independent of the transient behavior, i.e., the behavior during operation, such as the behavior under the influence of environmental effects, tolerance changes, wear of the device 1, or the manufacturing process of the formulation 2.

[0141] In contrast, the adjusted value or the correspondingly adjusted setting or control parameter 5 of the control parameter 5 obtained by the dynamic component 12B is a value taking into account the instantaneous or runtime effects, for example, considered by means of one or more predicted values. This is preferably done taking into account past developments, especially by means of time series prediction.

[0142] To this end, an adjusted value can be determined, which is used for the reference value of the control parameter 5 or the correspondingly adjusted setting or control parameter 5. The adjusted value preferably takes into account the comparison of the predicted actual formulation parameter 11 with the target formulation parameter 10, and makes the actual formulation parameter 11 approximate the target formulation parameter 10 by dynamically adjusting the reference value of the control parameter 5, especially based on the current state parameter 7 during operation.

[0143] An advantageous feature of this method is that the reference value and the adjusted value can be determined by different methods. Alternatively or additionally provided: The reference value is determined at least by means of artificial intelligence or machine learning methods, preferably differently from the adjusted value. Furthermore, it is particularly preferred that the reference value is also determined according to the state parameter 7, that is, both the reference value and the adjustment term (correction term) for determining the adjusted or corrected reference value are variable.

[0144] Therefore, it is feasible and preferred that the reference value depends on the state parameter 7, but is independent of the history and predicted development of the state parameter 7. In contrast, both the reference value of the control parameter 5 obtained by the static part 12A of the model 12 according to the (current) state parameter 7 and the correction term or adjustment term obtained by the dynamic part 12B of the model 12 can be changed, and the reference value can be adjusted or corrected.

[0145] Here, the static part 12A of the model 12 preferably differs from the dynamic part 12B of the model 12 in that a reference value is obtained or obtainable with the static part 12A without predicting the change of the state parameter 7 with the control parameter 5 and / or the influence of past changes, while the dynamic part 12B precisely takes into account the influence of future changes of the state parameter 7 with the control parameter 5 and / or the prediction resulting from past changes of the control parameter 5 or the state parameter 7, and preferably also takes into account previous changes.

[0146] In summary, the difference between the static part 12A of the model 12 and the dynamic part 12B of the model 12 preferably lies in that they use different methods (using or from the state parameter 7 and preferably considering the feed parameter 4) to determine the reference value of the control parameter 5 (in the case of the static part 12A), as well as the correction term of the reference value of the control parameter 5 or the corrected / adjusted reference value, as the final control parameter 5.

[0147] The state parameter 7 considered by the model 12 as an input parameter preferably relates to the state parameter 7 affected by the feed 3. Here, the model 12 preferably uses the corresponding at least one state parameter 7 affected by the feed 3 of the granulator 13 and the dryer 15.

[0148] As an input parameter, the model 12 optionally also considers (only) the recipe - characteristic parameter 7K that can be measured online, which describes the physical or chemical characteristics of the recipe 2, and the physical or chemical characteristics can be measured during the uninterrupted continuous operation.

[0149] Alternatively or additionally, the model 12 additionally considers (only) the intermediate - product - characteristic parameter 7L that can be measured online as an input parameter, which describes the physical or chemical nausea of the intermediate product 14, and the physical or chemical characteristics can be measured during the uninterrupted continuous operation. The measurement can be carried out by the intermediate - product - characteristic sensor 8L, especially the (NIR) sensor for determining the moisture (water content) index of the intermediate product 14.

[0150] Conversely, the consideration of other parameters is not excluded. In addition, the definition of the model 12 can consider parameters that cannot be measured online, while the model 12 preferably does not require parameters that cannot be measured online as input parameters during operation.

[0151] The static part 12A of the model 12 can implement artificial intelligence or machine - learning methods, while the dynamic part 12B of the model 12 preferably implements different, preferably non - artificial - intelligence - or machine - learning - based methods, or does not implement the neural network 32. Conversely, it is not excluded that both different methods are based on artificial intelligence or machine learning, or neither is.

[0152] Surprisingly, when the above aspects are combined with each other, it is found that the control of device 1 is particularly reliable and accurate, i.e., the actual formulation parameter 11 is particularly close to the target formulation parameter 10. In summary, it is surprisingly entirely particularly advantageous:

[0153] · Controlling the combination of the granulator 13 and the dryer 15 or implementing said control, and / or

[0154] · When there is a direct coupling of the device components, without external or off-line analysis of the intermediate product 14, and / or

[0155] · Excluding the use of off-line measurements (i.e., not during the continuously running process and not on the continuously moving intermediate product 14) or the analysis results of the measurable intermediate product 14, and / or

[0156] · The reference value of the control parameter 5 has been / can be determined by the static part 12A of the model 12 using a first (preferably machine learning-based) method, and / or

[0157] · When determining / can determine the reference value of the control parameter 5, the prediction of the influence of the change of the state parameter 7 with the control parameter 5 during the determination process is not considered,

[0158] · The dynamic component of the control parameter 5 is determined or can be determined by the dynamic part 12B of the model 12; or the reference value has been corrected with said dynamic component, or the device 1 is constructed for this purpose, and / or

[0159] · The dynamic part of the control parameter 5 is determined or can be determined by the dynamic part 12B of the model 12 by means of a method different from (preferably time series prediction or non-machine learning-based) the method used or supported by the static part 12A of the model 12, and / or

[0160] · The dynamic part 12B of the model can predict the influence of the change of the state parameter 7 with the control parameter 5 and take it into account, or the device 1 is constructed for this purpose, and / or

[0161] · The reference value of the control parameter 5 is corrected or adjusted with the dynamic component, or a correction term has been determined / can be determined for this purpose, or the device 1 is constructed for this purpose, and / or

[0162] · The final control parameter 5 is the reference value adjusted with the dynamic component, and the device 1 is controlled by these control parameters 5.

[0163] As Figure 1 shown, the device 1 can have a plurality of different actuators 6 that act directly or indirectly on the feed 3, thereby forming the intermediate product 14 and ultimately the formulation 2.

[0164] The control parameter 5 corresponds to the respective actuator 6. The control parameter can be a control variable for the actuator 6 or can correspond thereto to control the behavior of the actuator 6.

[0165] The actuators 6 of the granulator 13 that can be controlled by the respective control parameter 5 include one or more of the following actuators 6:

[0166] A conveyor drive 6A that can be controlled by means of a conveyor drive control parameter 5A, preferably for metering or conveying the feed 3 to the granulator 13, and / or

[0167] A granulator drive 6B that can be controlled by means of a granulator drive control parameter 5B, preferably for driving the granulation mechanism 17 or one or more worms, and / or

[0168] An injection device 6C that can be controlled by means of an injection device control parameter 5C, preferably for injecting the granulation liquid 13B, and / or

[0169] A granulator temperature control device 6D that can be controlled by means of a granulator temperature control parameter 5D, preferably for temperature control (heating and / or cooling) of the part of the granulator 13 that comes into contact with the feed 3.

[0170] The actuators 6 of the dryer 15 that can be controlled by the respective control parameter 5 include one or more of the following actuators 6:

[0171] An air supply conveyor drive 6E that can be controlled by means of an air supply conveyor drive control parameter 5E, preferably a fan, for supplying air 27 to the dryer 15, and / or

[0172] An air supply temperature control device 6F that can be controlled by means of an air supply temperature control parameter 5F, preferably for temperature control of the air 27, and / or

[0173] An exhaust conveyor drive 6G that can be controlled by means of an exhaust conveyor drive control parameter 5G, preferably for extracting the air 27 after the drying process, and / or

[0174] A fluidized bed drive 6H that can be controlled by means of a fluidized bed drive control parameter 5H, preferably for rotating the turntable of the dryer 15.

[0175] The sensors 8 for measuring the respective corresponding state parameters 7 of the granulator 13 include one or more of the following sensors 8:

[0176] A conveyor sensor 8A for measuring a conveyor parameter 7A, which describes the characteristics of the conveying, in particular the throughput and / or speed of the conveying, preferably a conveying rate or metering that can be expressed in [kg / h]; and / or

[0177] A granulator sensor 8B for measuring granulator parameter 7B, which describes the functional characteristics of granulator 13, especially the torque and / or speed of its drive device 6B, and / or throughput - preferably the conveying rate of the granulator that can be expressed in [kg / h], or the corresponding rotational speed [rpm] of the extruder / worm; and / or

[0178] An injection sensor 8C for measuring injection device parameter 7C, which describes the situation of adding liquid 13B (especially water and / or alcohol) to the feed 3 in the area of granulator 13, especially the throughput and / or quantity ratio compared with the feed, preferably the injection rate that can be expressed in [g / min]; and / or

[0179] One or more granulator temperature sensors 8D for measuring one or more granulator temperatures 7D, especially the temperature of granulator 13, or (indirectly) measuring the temperatures at different positions along the conveying path of feed 3 through granulator 13, preferably expressed in [°C] or corresponding units.

[0180] The sensors 8 for measuring the respective corresponding state parameters 7 of dryer 15 include one or more of the following sensors 8:

[0181] An air supply conveying sensor 8E for measuring air supply conveying parameter 7E, which describes the air supply conveying situation of dryer 15, especially (mass) throughput, (corresponding) pressure difference, and / or the speed of supplied air 27, preferably expressed in [m^3 / h]; and / or

[0182] An air supply sensor 8F for measuring air supply parameter 7F, which describes the characteristics of supplied air 27, especially temperature (preferably expressed in [°C]) and / or the humidity of supplied air 27 (dew point preferably expressed as a percentage, weight percentage, or [°C]); and / or

[0183] An exhaust sensor 8G for measuring exhaust parameter 7G, which describes the exhaust conveying of dryer 15, especially (mass) throughput, (corresponding) pressure difference, the temperature of air 27 discharged from dryer 15, and / or its humidity; and / or

[0184] A fluidized bed sensor 8H for measuring fluidized bed parameter 7H, which preferably characterizes the drying-related change characteristics of the fluidized bed related to dryer 15, especially speed, such as the rotational speed of the rotating part of the dryer, preferably expressed in [rpm].

[0185] Instead of or in addition to using sensor 8, in some cases, in principle, the characteristics or control parameters 5 of actuator 6 can also be considered to determine or derive one or more of the above or corresponding parameters. The parameters can be used as the basis for control here.

[0186] Conversely, when determining the state parameter 7 affected by the feed 3, it must be measured by means of sensor 8. These state parameters 7 are also used as the unit parameters of the model 12.

[0187] Optionally as part of the device 1, alternatively or additionally, a recipe temperature sensor 8I is externally provided for measuring the recipe temperature 7I characterizing the temperature of the recipe 2; and / or

[0188] A recipe outlet quantity sensor 8J for measuring the recipe outlet quantity parameter 7J, which describes the characteristics of the manufacturing and / or the device 1 in terms of the manufacturing of the recipe 2, in particular the outlet quantity of the recipe 2; and / or

[0189] A recipe characteristic sensor 8K, which characterizes the recipe characteristic parameter 7K of the recipe 2 that can be measured online, in particular the particle size, particle size distribution, particle size and / or humidity, particularly preferably the particle size or particle size distribution (XD10, XD50, XD90) and / or the residual humidity and / or drying loss of the intermediate product 14 when processed into the recipe 2.

[0190] The above control parameters 5, actuators 6, state parameters 7 and sensors 8 are all particularly preferred embodiments of a particularly preferred combination of the granulator 13 (preferably a (double) screw granulator 13) and the dryer 15 (preferably a fluidized bed dryer 15).

[0191] It should be understood that other control parameters 5, actuators 6, state parameters 7 and sensors 8 are alternatively or additionally implemented in other granulators 13 and / or dryers 15. Therefore, the present invention is preferably not limited to the above control parameters 5, actuators 6, state parameters 7 and sensors 8. In addition, it is not required to implement or use all of the control parameters 5, actuators 6, state parameters 7 and sensors 8 below. There can be different selections here.

[0192] The following will refer to Figure 1 the embodiments shown in

[0193] Therefore, it is also conceivable that the present invention can also be used in cases where other granulator technologies and / or dryer technologies are used. Therefore, in addition to the twin-screw granulator 13, other types of granulators can also be used. Alternatively or additionally, in addition to the fluidized bed dryer 15, other types of dryers can also be used. In any case, the present invention has proven to be particularly advantageous in this case.

[0194] In an embodiment according to Figure 1 the device 1 has a granulator 13 for processing the feed 3 into an intermediate product 14 and a dryer 15 for forming the formulation 2 from the intermediate product 14.

[0195] The granulator 13 has a conveyor 16 for conveying and / or metering the feed 3. The conveyor 16 can have a storage container 16A, especially in the shape of a funnel, for storing the feed 3. In addition, the conveyor 16 can have a conveying device 16B that feeds the feed 3 into the granulation mechanism 17 of the granulator 13 coupled to the conveyor 16 at a determined conveying rate (quantity per unit time).

[0196] The conveyor 16 can have a drive device 6A for the conveying device 16B, especially a motor for driving the worm 16C. In principle, however, in addition to having a worm 16C, other conveying principles can also be used, such as a conveyor belt. As mentioned above, the drive device 6A can preferably be controlled by means of conveyor parameters 7A. The conveying rate can preferably be set with the conveyor drive control parameter 5A.

[0197] The feeder 16 preferably has one or more sensors 8A for measuring the throughput and / or (preferably corresponding thereto) the speed.

[0198] The granulation mechanism 17 of the granulator 13 is preferably connected to the conveyor 16. The granulation mechanism 17 is configured to change the particle size or grain size of the feed 3, respectively. For this purpose, the granulation mechanism 17 can act physically on the feed 3, especially by kneading and / or rubbing.

[0199] In the example shown, the granulation mechanism 17 has at least one worm 18, preferably a twin-screw. The worms 18 of the twin-screw preferably cooperate with each other and convey the feed 3, while the worms act physically on the feed to change the particle size or particle size distribution of the feed 3.

[0200] The granulation mechanism 17 or the worm 18 can have different processing zones 19. In particular, different worm leads and / or worm surfaces can be provided to achieve ideal processing of the feed 3.

[0201] The granulation mechanism 17 may include a granulator drive device 6B for driving the granulation mechanism 17, in particular the worm 18. As described above, the granulator drive device 6B can be controlled by granulator drive control parameters 5B, preferably related to throughput and / or speed, in particular the rotational speed of the worm 18.

[0202] The granulator 13 preferably has a granulator sensor 8B with which parameters representing the granulation speed and / or processing intensity can be measured, in particular the speed, throughput or (very particularly preferably) torque (of the granulation mechanism 17 / worm 18).

[0203] The granulator 13 may have an injection device 6C for injecting a liquid 13B to mix with or incorporate into the feed 3. This can be a sprayer, alternatively however also a drip device, or generally a device for adding a liquid substance.

[0204] The injection device 6C is preferably arranged in the region of the inlet 13A or in the first half or the first third of the transport path of the feed 3 formed by the granulation mechanism 17 between the inlet 13A and the intermediate product outlet 21 of the feed 3 processed by the granulator 13.

[0205] The granulator 13 preferably has at least one (preferably a plurality of) granulator temperature sensors 8D for measuring the temperature 7D of the feed 3 or the corresponding temperature 7D at different positions of the granulator 15 or its housing 22.

[0206] In the example, more than three and / or fewer than ten granulator temperature sensors 8D are provided for measuring the temperature 7D of the feed 3 or the corresponding temperature 7D of the granulator 15. The temperature sensors 8D are preferably arranged (at least substantially equidistantly) along the transport path of the feed 3 in the granulator 15 or the granulation mechanism 17.

[0207] The coupling device 20 is preferably able to continuously and / or uninterruptedly transfer the intermediate product 14 from the intermediate product outlet 21 (which may be formed by the housing 22 of the granulator 13) to the intermediate product inlet 24 of the dryer 15 (preferably formed by the housing 23 of the dryer 15).

[0208] Preferably not provided: stopping the intermediate product 14 for temporary storage, at least not for more than 1 minute, 2 minutes or 5 minutes. Thus, in particular not provided: taking a sample of the intermediate product 14 and analyzing it separately from the equipment or with a delay in the manufacturing process. On the contrary, in the sense of the present invention, it is preferably to transfer the intermediate product 14 to the dryer 14 at least substantially uninterruptedly and / or continuously.

[0209] One or more of the granulator temperature sensors 8D can be set to measure the temperature 7D of the intermediate product 14 or a corresponding temperature 7D, in particular the temperature of the granulator 15 in the region of the coupling device 20.

[0210] The dryer 15 preferably has a fluidized bed 25. In addition, the dryer 15 preferably has an air inlet 26 for the entry of (process) air 27, an optional diffuser 28 for uniformly delivering the air 27 to the fluidized bed 25, and an air outlet 29 for discharging the air 27 after it has passed through the fluidized bed 25. In the region of or upstream of the air outlet 29, the dryer 15 optionally has a separator 30, such as a cyclone separator or a filter, for capturing the particles of the intermediate product 14 or the formulation 2 from the air 27.

[0211] Finally, the dryer 15 preferably has a formulation outlet 31 for discharging the formulation 2, i.e., the intermediate product 14 dried by the dryer 15. The dryer 15 dries the intermediate product 14 entering through the intermediate product inlet 24 with the air 27 and then discharges the resulting formulation 2 through the formulation outlet 31, preferably after passing through the fluidized bed 25.

[0212] To convey the air 27 to the dryer 15, in particular to the fluidized bed 25, the dryer 15 preferably has an air supply conveyor 15A with an air supply conveyor drive 6E. This can involve a fan or a device generally used for transporting and / or compressing the air 27.

[0213] The dryer 15 can have an air supply conveyor sensor 8E with which the throughput of the air 27 or a corresponding parameter can be measured. In particular, the air supply conveyor sensor 8E is a pressure sensor for measuring the air pressure on the discharge side or on the side of the air supply conveyor drive 6E facing the fluidized bed 25, or a differential pressure sensor for determining the pressure difference across the air supply conveyor 15A. Alternatively or additionally, the air supply conveyor sensor 8E can also be or have a parameter assigned to the air supply conveyor drive 6E, such as the rotational speed (fan speed), current consumption, torque, etc.

[0214] The air supply conveyor 15A or its air supply conveyor drive 6E can be controlled by means of air supply conveyor drive control parameters 5E, in particular with respect to throughput, (fan) rotational speed, current consumption, and / or pressure or pressure difference. The pressure difference can be the pressure across the air supply conveyor 15A, however alternatively or additionally it can also be the pressure on the fluidized bed 25, etc.

[0215] Before the air 27 is fed into the fluidized bed 25 or other drying device of the dryer 15, it is preferably temperature-adjusted, especially heated. For this purpose, the dryer 15 preferably has an air supply temperature-adjusting device 6F, preferably a heating jacket. The air supply temperature-adjusting device 6F can be controlled by means of the air supply temperature-adjusting device control parameter 5F, or is constructed for this purpose.

[0216] Preferably, the temperature and / or (relative) humidity of the conditioned air 27 that has come into contact or is to come into contact with the intermediate product 14 for drying purposes is measured. For this purpose, the dryer 15 can have an air supply sensor 8F, which can measure the temperature of the air 27 and alternatively or additionally the (relative) humidity, as the air supply parameter 7F, or is constructed for this purpose.

[0217] The air supply sensor 8F can be arranged between the fluidized bed 25 and the air supply temperature-adjusting device 6F or the air supply conveyor 15A, or measure the characteristics of the air 27.

[0218] After the air 27 comes into contact with the intermediate product 14 for drying purposes, the air 27 is discharged through the dryer 15. This is preferably done through the air outlet 29, which is different from the formulation outlet 31 for discharging the formulation 3.

[0219] The air 27 can be conveyed from the air outlet 29 by means of an exhaust conveyor 15B, especially a (second) fan. For this purpose, an exhaust conveyor drive device 6G can be provided, which can effect the conveyance or drive of the exhaust conveyor 15B.

[0220] The exhaust conveyor 15B can be controlled or adjusted by means of the exhaust conveyor drive control parameter 5G. In particular, it is provided that the exhaust conveyor 15B is adjusted such that no air 27 leaks through the intermediate product outlet 21 and the formulation inlet 31. For this purpose, the conveying capacity of the exhaust conveyor 15B can be the same as or exceed the conveying capacity of the air supply conveyor 15A.

[0221] The throughput, speed, temperature and / or humidity of the air 27 discharged from the dryer 15 after the drying process, and / or the corresponding parameters such as the speed of the impeller of the exhaust conveyor drive device 6G or the fan, can be measured by means of an exhaust conveyor sensor 8G as the exhaust gas parameter 7G.

[0222] The fluidized bed 25 can have or form a processing zone, especially on the side facing away from the air inlet 26. The fluidized bed 25 or the structure (such as a sieve or perforated plate or rotating part) bounding the fluidized bed in the direction of the air inlet 26 can be driven, especially made to move. For this purpose, a fluidized bed drive device 6H can be provided, which can be controlled with the fluidized bed drive control parameter 5H.

[0223] The fluidized bed parameters 7H can be measured using the fluidized bed sensor 8H, and the fluidized bed parameters can describe the characteristics of the fluidized bed 25, such as the movement of the rotating part.

[0224] After drying the intermediate product 14 with the dryer 15, the processed feed 3 is output as the formulation 2.

[0225] Here or hereafter, one or more actual formulation parameters 11 can be determined from the formulation 2, that is, the parameters describing the physical or chemical characteristics of the resulting formulation 2. In particular, this can be done online, that is, during the uninterrupted operation process. Alternatively or additionally, however, the actual formulation parameters 11 can also be determined subsequently by means of laboratory tests. The actual formulation parameters 11 that cannot be measured online are preferably used as the basis for the model 12, so that the non-online measurable actual formulation parameters 11 are considered in the modeling. In turn, the non-online measurable actual formulation parameters 11 are not directly used to control the device 1 or as input parameters for the control 9 or the model 12.

[0226] The device 1 can have one or more sensors 8 for online describing the characteristics of the formulation 2. For this purpose, it includes one or more formulation temperature sensors 8I for measuring the formulation temperature 7I, a formulation outlet quantity sensor 8J for measuring the formulation parameter 7J (which describes the outlet quantity or throughput (mass flow) of the formulation), and / or a formulation characteristic sensor 8K for measuring one or more characteristics of the formulation 2 and outputting them as formulation characteristic parameters 7K, preferably (relative) humidity or residual humidity and / or drying loss.

[0227] The formulation parameters 10, 11 preferably relate to at least one parameter characterizing the particle characteristics of the formulation 2, such as particle size or particle size distribution (XD10, XD50, and / or XD90) or corresponding parameters.

[0228] Alternatively or additionally, the formulation parameters 10, 11 are preferably (relative) humidity or residual humidity and / or drying loss (LoD – Loss on drying) and / or corresponding parameters.

[0229] The supplementary formulation parameters 10, 11 can be obtained as needed by the formulation temperature sensor 8I, the formulation outlet quantity sensor 8J, and / or the formulation characteristic sensor 8K, or the formulation temperature 7I, the formulation outlet quantity parameter 7J, and / or the formulation characteristic parameter 7K.

[0230] In principle, the measured parameters considered, such as the particle size or particle size distribution (XD10, XD50, and / or XD90) of the formulation 2, can be used to form the model 12. However, it is not necessarily or in all cases set to determine the corresponding parameters during the operation of the device 1 or input them to the control of the device.

[0231] Optionally but preferably, the online-measurable properties of the intermediate product 14, in particular the humidity, can be determined by means of the intermediate product (humidity) sensor 8L as the intermediate product characteristic parameter 8L. If available, this can also be taken into account in the model 12, in particular used as an input parameter, or the control of the device 1 can also be (supplementarily) based thereon.

[0232] The actual recipe parameters 11 can be measured downstream of the device 1. This will be discussed in detail below. However, it has already been pointed out here that preferably the parameters characterizing the particle shape, size or distribution of the recipe 2 are also measured online.

[0233] Conversely, preferably the measurement of the particles characterizing the intermediate product 14 is avoided.

[0234] The various sensors 8 for determining the status parameters 7 of the device 1 have been described in connection with Figure 1 the embodiments. However, it is not mandatory to provide all sensors 8 or to use the status parameters 7.

[0235] Preferably, at least two or at least three status parameters 7 or sensors 8 are used for the granulator 13 and the dryer 15 respectively.

[0236] The control parameters 5 for device control, particularly preferably for the dynamically adjustable control parameters 5 of device control (the selection of the control parameters in other aspects of the present invention can be an independent concept of the present invention) include:

[0237] In terms of the granulator 13:

[0238] The conveyor drive control parameter 5A, preferably characterizing the metering of the feed 3; and / or

[0239] The granulator drive control parameter 5B, preferably characterizing the extrusion speed of the extruder of the granulator 13 (the extruder can constitute the granulating mechanism 17 of the granulator 13), or characterizing the (other) parameter corresponding to the conveying or processing speed of the granulator 13; and / or

[0240] The injection device control parameter 5C, in particular the injection rate, characterizing the amount of the liquid 13B conveyed per unit time.

[0241] In terms of the dryer 15:

[0242] The fluidized bed drive control parameter 5H, preferably characterizing the speed of the rotating part of the dryer 15;

[0243] The air supply conveyor drive control parameter 5E, preferably representing the air flow on the inlet side, and / or

[0244] The air supply temperature control device control parameter 5F, in particular representing the temperature of the air flow on the inlet side of the air 27.

[0245] Particularly preferred parameters 3 and 7 are used as input parameters for device control or models, and their selection can be an independent aspect of the present invention in other aspects of the present invention. The parameters include:

[0246] In terms of the granulator 13:

[0247] The granulator parameter 7B, in particular, characterizes the extrusion torque of the extruder of the granulator 13 or other parameters of the granulator drive device 6B, and this parameter depends on or the consistency and / or conveying rate of the processed feed 3; and / or

[0248] One or more granulator temperatures 7D, in particular, characterize one or more temperatures at preferably different positions along the material flow of the feed 3 being processed in the granulator 13 or the corresponding temperatures; and / or

[0249] The granulator temperature 7D or the intermediate product characteristic parameter 7L, where the temperature or parameter is the temperature of the intermediate product or the corresponding temperature.

[0250] In terms of the dryer 15:

[0251] The exhaust gas parameter 7G, in particular, characterizes the exhaust gas temperature and / or (relative) humidity and / or the throughput of the exhaust gas 27 (e.g., represented by a pressure difference); and / or

[0252] The formulation outlet quantity parameter 7J, in particular, characterizes the outlet quantity and / or pressure difference related to the formulation output, such as the outlet quantity and / or pressure difference through a filter or sieve.

[0253] In terms of the feed 3:

[0254] The feed parameter 4, preferably characterizing the humidity and / or particle size of the feed 3.

[0255] Optionally, for the control of the device 1, aspects of the formulation 2 can be additionally considered:

[0256] The formulation characteristic parameter 7K, preferably characterizing the particle size distribution, in particular D10, D50, and / or D90, and / or the drying loss.

[0257] When using the above parameters 4, 5, and 7, one or more of the parameters 4, 5, and 7 discussed above and below can be additionally used.

[0258] Figure 3 A simplified schematic view of the artificial neural network 32 for obtaining the control parameter 5 is shown, and this control parameter is used to control the device 1 or its actuator 6.

[0259] According to one aspect of the present invention, the model 12, in particular the static part 12A of the model 12, is constituted by an artificial neural network 32 or has an artificial neural network 32. An example of the structure of the artificial neural network 32 is as shown in Figure 3 as shown in

[0260] In the input layer 33, the artificial neural network 32 has nodes 36 in the form of input nodes, and the nodes respectively correspond to one or more of the feed parameters 4, one or more of the state parameters 7 of the granulator 13, one or more of the state parameters 7 of the dryer 15, and / or one or more of the target formulation parameters 10.

[0261] Preferably, the artificial neural network 32 has nodes 36 in one or more hidden layers 34, through which the input layer 23 can be linked to the output layer 35.

[0262] The artificial neural network 32 may have nodes 36 in the output layer 35 in the form of output nodes, and the nodes correspond to one or more control parameters 5.

[0263] The nodes 36 of the different layers 33, 34, 35 can be interconnected by means of edges 37. The nodes 36 can form a graph by means of the edges 37. The nodes 36 and / or the edges 37 preferably have weights 38, and the weights can specify the links of the nodes 36, and the links can be represented by means of the edges 37.

[0264] The artificial neural network 32 is preferably trained by means of a data set composed of different combinations of the feed parameters 4, the state parameters 7, the control parameters 5, and the actual formulation parameters 11, and the actual formulation parameters can be set under the condition of presetting the feed parameters, the state parameters, and the control parameters.

[0265] The training data sets respectively represent the stationary states of the device 1, wherein the actual formulation parameters 11 and the state parameters 7 have adopted at least substantially stationary values based on the constant control parameters 5 and the feed parameters 4.

[0266] The artificial neural network 32 is preferably trained by applying at least one (preferably multiple) state parameter 7 of the granulator 13 that is preferably affected by the feed 3 and at least one (preferably multiple) corresponding state parameter 7 of the dryer 15 that is preferably affected by the feed 3 in the training data set to the input nodes 36.

[0267] The input nodes (the nodes 36 in the input layer 33) preferably also respectively apply one or more corresponding actual formulation parameters 11 and one or more corresponding feed parameters 4 of the corresponding training data set.

[0268] By presetting the parameters 4, 5, 7, 11 of the training data set in the input layer 33, a control parameter 5 (the value of the neural network 32) is given at the output node (node 36 of the output layer 35). Preferably, by comparing with the control parameter 5 of the corresponding training data set, an error can be determined from the control parameter, and the error can be reduced or compensated by adjusting the weights 38 of the artificial neural network 32 (preferably by means of backpropagation and / or in a continuous manner).

[0269] In order to finally obtain the control parameter 5 with the help of the neural network 32 here, the target recipe parameters 10 and other current parameters 4, 7 are preset to the artificial neural network 32 (input layer 33), rather than the actual recipe parameter 11, so as to generate (on the output layer 35) the control parameter 5 for controlling the device 1. On this basis, the device 1 can be controlled or (automatically) controlled.

[0270] The parameters 4, 5, 7, 10 provided for its node 36 are preferably at least:[[]]END]

[0271] Preferably in the input layer 33:[[]]END]

[0272] At least one node 36 for at least one corresponding feed parameter 4, preferably the humidity of the feed 3 and / or the characteristics of the particles of the feed 3, especially its particle size distribution; and / or[[]]END]

[0273] · A node 36 for the granulator parameter 7B, especially the worm or extruder torque of the granulator 13; and / or[[]]END]

[0274] · A node 36 for the granulator temperature 7D, especially multiple nodes 36 for multiple granulator temperatures 7D; and / or[[]]END]

[0275] · A node 36 for the exhaust parameter 7G, especially the temperature and humidity of the air 27 and / or[[]]END]

[0276] Or the pressure difference of the air 27 on the separator 30;[[]]END]

[0277] Preferably in the output layer 35:[[]]END]

[0278] · A node 36 for the granulator drive control parameter 5B; and / or[[]]END]

[0279] · A node 36 for the injection device control parameter 5C; and / or[[]]END]

[0280] · A node 36 for the fluidized bed drive control parameter 5H; and / or[[]]END]

[0281] · A node 36 for the gas supply conveyor drive control parameter 5E; and / or a node 36 for the gas supply temperature control parameter 5F.[[]]END]

[0282] Optionally, nodes for one or more of the following state parameters 7 are provided in the input layer 34:

[0283] · A node 36 for the injection device parameter 7C;

[0284] · A node 36 for the gas supply parameter 7F;

[0285] · A node 36 for the gas supply conveyor parameter 7E;

[0286] · A node 36 for the fluidized bed parameter 7H;

[0287] · A node 36 for the formulation temperature 7I;

[0288] · A node 36 for the formulation outlet quantity parameter 7J; and / or

[0289] · A node 36 for the formulation characteristic parameter 7K,

[0290] And / or nodes for one or more control parameters 5 are provided in the output layer 35:

[0291] · A node 36 for the conveyor drive control parameter 5A;

[0292] · A node 36 for the granulator temperature control device control parameter 5D; and / or

[0293] · A node 36 for the exhaust conveyor drive control parameter 5G.

[0294] Thus, preferably, the state parameters 7 corresponding to the respective nodes 36 include the granulator parameter 7B, in particular the torque of the granulating mechanism of the granulator 13, one or more granulator temperatures 7D at different positions along the conveying path of the granulator 13 for the feed 3, the formulation temperature 7I, in particular the temperature of the formulation 2 at the formulation outlet 31 of the dryer 15, the formulation characteristic parameter 7K, in particular the humidity, in particular the relative humidity, of the formulation 2 at the formulation outlet 31 of the dryer 15, and / or the waste gas parameter 7G, in particular characterizing the pressure loss on the filter (here exemplified by the (cyclone) separator 30) of the dryer 15.

[0295] The artificial neural network 32 is preferably constructed by training to generate the control parameter 5 from the parameters 4, 7, 10 fed into the nodes 36 of the input layer, and the device 1 or the combination of the granulator 13 and the dryer 15 can be controlled with the control parameter. Before being used as the control basis for the device 1, the control parameter 5 generated by the artificial neural network 32 is preferably (but not necessarily) optimized with the aid of the dynamic model 12B.

[0296] With the aid of the model 12, the control parameter 5 is preferably determined only according to or mainly according to the feed parameter 4, the state parameter 7 and the target formulation parameter 10.

[0297] Here, even in this case, the characteristics of the intermediate product 14 that are not measured online are preferably disregarded. Preferably, at most, the temperature and / or humidity of the intermediate product 14 are considered.

[0298] The model 12 preferably takes into account the future influence of changes in the state parameter 7 on the actual formulation parameter 11, preferably by means of prediction. For this purpose, the dynamic part of the model 12 is preferably constructed to take into account the changes in the actual formulation parameter 11 caused by long-term influences.

[0299] The controller 9 preferably takes into account the future influence of changes in the control parameter 5 on the actual formulation parameter 11 and / or the state parameter 7, preferably by means of prediction. In particular, the dynamic part of the model 12 is constructed to pre-compensate for the changes in the actual formulation parameter 11 caused by long-term influences.

[0300] For this purpose, the model 12 can use the prediction of the future development of the actual formulation parameter 11 as a basis for determining or adjusting the control parameter 5. For this purpose, in addition to the static part 12A (using the static part to obtain a reference value for the corresponding control parameter 5 from the state parameter 7 and preferably the feed parameter 4 and the target formulation parameter 10), the model 12 can also have a dynamic part 12B, using which the reference value is optimized by means of prediction.

[0301] Therefore, when the current state parameter 7 changes, with the dynamic part 12B of the model 12, the change in the actual formulation parameter 11 can be predicted based on the feed parameter 4 and the current state parameter 7, and based on the prediction, the reference value of the control parameter 5 can be adjusted, and the device 1 can be controlled with the adjusted control parameter 5.

[0302] Therefore, the model 12, in particular the dynamic part 12B of the model 12, is constructed to predict the long-term influence of changes in the control parameter 5 on the actual formulation parameter 11 and / or the state parameter 7. Therefore, the control unit 9 is set up by the model 12 to control the device 1 or the combination of the granulator 13 and the dryer 15 while compensating for long-term influences. Therefore, even though the granulator 13 and the dryer 15 are directly coupled, manufacturing can still be achieved in a surprising manner while maintaining a small / allowed deviation between the actual formulation parameter 11 and the target formulation parameter 10.

[0303] If only the prediction of the characteristics of the formulation 2 is desired or achieved with the model 12, the model 12 or the artificial neural network 32 can be constructed differently from the case where preferably fully automatic control is achieved with the model 12, that is, preferably, the characteristics characterizing the formulation 2, that is, in particular one or more predicted actual formulation parameters 11, can be determined and / or output.

[0304] In this case, the model 12 preferably determines or predicts one or more actual recipe parameters 11 based solely or mainly on the feed parameters 4, the status parameters 7, and the predefined control parameters 5.

[0305] Thus, the dynamic part 12B of the model 12 can preferably predict the changes in the actual recipe parameters 11 based on the feed parameters 4, the current status parameters 7, and / or the control parameters 5 or their reference values, while taking into account the changes in the status parameters 7 associated with the changes in the control parameters 5, and based on said prediction, can predict and preferably output the generated actual recipe parameters 11.

[0306] Therefore, the model 12, especially the dynamic part 12B of the model 12, is preferably configured to predict the long-term impact of the changes in the status parameters 7 on the actual recipe parameters 11. Thus, in a surprising manner and although the granulator 13 and the dryer 15 are directly coupled with the predicted actual recipe parameters 11, an indicator can be provided for the user to select the appropriate control parameters 5.

[0307] Regardless of whether the model 12 uses the control parameters 5 or the predicted actual recipe parameters 11, the characteristics of the intermediate product 14 that are not measured online are preferably ignored, such as the physical characteristics of the particles that characterize the intermediate product 14. In particular, all characteristics of the intermediate product 14 are not considered except for the temperature and humidity of the intermediate product 14.

[0308] Therefore, the model 12 preferably only considers the parameters of the feed 2 or the intermediate product 14 that are in the process and can be measured online, that is, the parameters that do not require sampling and analysis separately from the equipment.

[0309] As described above, the model 12 can alternatively be used to predict the actual recipe parameters 11. In this case, the model 12 or the artificial neural network 32 is configured differently from the artificial neural network 32 used to determine the control parameters 5.

[0310] Nodes 36 are provided for the parameters 4, 5, 7, 10, which are preferably at least when the model 12 determines the predicted actual recipe parameters 11:

[0311] Preferably in the input layer 33:

[0312] · Nodes 36 for at least one corresponding feed parameter 4, preferably the humidity of the feed 3 and / or the characteristics of the particles of the feed 3, especially its particle size distribution; and / or

[0313] · Nodes 36 for the granulator parameters 7B, especially the worm or extruder torque of the granulator 13, and / or

[0314] · Node 36 for the granulator temperature 7D, especially for multiple granulator temperatures 7D; and / or

[0315] · Node 36 for the exhaust parameters 7G, especially for the temperature and humidity of the air 27 and / or the pressure difference of the air 27 on the separator 30; and / or

[0316] · Node 36 for the granulator drive control parameter 5B; and / or

[0317] · Node 36 for the injection device control parameter 5C; and / or

[0318] · Node 36 for the fluidized bed drive control parameter 5H; and / or

[0319] · Node 36 for the gas supply conveyor drive control parameter 5E; and / or

[0320] · Node 36 for the gas supply temperature control device control parameter 5F.

[0321] Preferably in the output layer 35:

[0322] · One or more nodes 36 for a (corresponding) one (predicted) actual recipe parameter 11.

[0323] Optionally, nodes 36 for one or more of the following state parameters 7 are provided in the input layer 34:

[0324] · Node 36 for the injection device parameter 7C and / or

[0325] · Node 36 for the gas supply parameter 7F and / or

[0326] · Node 36 for the gas supply conveyor parameter 7E and / or

[0327] · Node 36 for the fluidized bed parameter 7H and / or

[0328] · Node 36 for the recipe temperature 7I and / or

[0329] · Node 36 for the recipe parameter 7J and / or

[0330] · Node 36 for the recipe characteristic parameter 7K and / or

[0331] · Node 36 for the conveyor drive control parameter 5A and / or

[0332] · Node 36 for the granulator temperature control device control parameter 5D and / or

[0333] · Node 36 for the exhaust conveyor drive control parameter 5G.

[0334] In this case, the artificial neural network 32 can be trained using a data set corresponding to or identical to the artificial neural network 32 used for determining the control parameter 5.

[0335] Figure 4 A schematic diagram showing the past and predicted trends of one or more of the parameters 5, 7, 10, and 11.

[0336] To the left of the Y-axis representing the values of the parameters 5, 7, 11, as indicated by the arrow for the past P, is the past development of one or more of the parameters 5, 7, 11, as well as the reference trajectory 39 and the past trends of the measured parameters 7, 11. Additionally, as indicated by the arrow for the future F, its future development is shown in the prediction range 40.

[0337] It is proposed that: at discrete time points t k+p one or more of the parameters 5, 7, 10 are changed or the change is predicted. In the example, the time points t k+p are spaced apart from each other by the sampling time Δt. However, in principle, the sampling time Δt does not necessarily have to be constant (even if this is possible), and the sampling time Δt can also be chosen to be very short such that the trends of one or more of the parameters 5, 7, 11 can be at least substantially continuous.

[0338] As can be seen in Figure 4 it is feasible that the predictions of one or more of the parameters 5, 7, 11 can have different trends, and can also have rising and falling trends, with the aim of approximating the measured values of the state parameter 7 and / or the actual recipe parameter 11 to the reference trajectory 39.

[0339] Ultimately, the trend according to Figure 4 represents a possible system behavior that takes into account the advantageous optimization of the control of the device 1 or the manufacturing process by means of the model 12, and takes into account the past development as well as the predicted development of different parameters in the future.

[0340] Figure 5 A schematic diagram showing the result of the control with a constant control parameter 5 over time. In a predetermined, preferably fixed time window 41, one or more state parameters 7 are derived from the one or more at least substantially constant control parameters 5. Taking into account the quality dynamics 42, one or more state parameters 7 preferably approach a fixed value asymptotically, and the quality dynamics can be measured at discrete time points.

[0341] The static part of model 12A can be determined based on the combination of constant control parameter 5 and one or more generated state parameters 7. In particular, a machine learning-based model 12A of static device behavior can be generated based on this. As described above, this can be an artificial neural network 32, but in principle it can also be other machine learning-based methods.

[0342] Figure 6 A schematic diagram showing predictions that are temporally offset from each other at time t is presented. At each time point t k+p the future measurement value κ depends on the previous development and the current control parameter 5 or state parameter 7. As Figure 6 shown, this regression can be solved by means of time series prediction. The trend of measurement value κ at different relative times ζ is shown, where the relative times are temporally offset from each other by a time difference or sampling time Δt, and this is represented by means of the time difference or sampling time as a prediction of a time series.

[0343] Specifically and preferably, the prediction by means of model 12 is updated at regular intervals Δt. Once the state of device 1 changes and thus one or more state parameters 7 deviate from the previous values, a changed or adjusted prediction can be generated (preferably by means of model 12), in particular one or more control parameters 5, state parameters 7 and / or actual recipe parameters 11.

[0344] Figure 7 A schematic diagram showing the change of process characteristics over time is presented. The basic idea is to combine the static behavior (e.g., as explained with reference to Figure 5 ) with the dynamic behavior and the development of the prediction considering the past trend, so that, as Figure 7 exemplarily shown, the process starts with a static reference value (preferably determined by the static part 12A of model 12) and thus during the production process, through repeated iterative optimization, in particular by means of a time prediction scheme, the desired properties, in particular the actual recipe parameters 11, can be achieved in a short time and then at least remain substantially unchanged.

[0345] Therefore, in principle, when starting the continuous processing of feed 3 to recipe 2, the control parameter 5 can be determined according to the static part 12A of model 12, and during the operation of device 1, the control parameter 5 can be conveniently tracked by means of the dynamic part 12B of model 12. This can be seen in the time period in the region where the sampling time Δt is input in terms of time in Figure 7 .

[0346] Therefore, before activating the compensation regulation by means of the dynamic part 12B of model 12, device 1 first reaches a quasi-steady state and then preferably uses the compensation regulation by means of time series prediction at intervals of the corresponding sampling time Δt. However, in principle, other control strategies can also be used.

[0347] The proposed device 1 can particularly advantageously be used in a system 45 for manufacturing tablets. For this purpose, Figure 8 An extended method is represented by a schematic flow chart. In the system 45, components for processing the feed 3 and / or for post-processing the formulation 2 can be provided, such as preferably a pneumatic conveying device 46, which add additional functions to the device 1. Now, with reference to Figure 9 a more detailed description will be given.

[0348] Figure 9 The device 1 embedded in the system 45 is shown. The feed 3 can be prepared prior to the device 1. In the example according to Figure 9 with reference to the method according to Figure 8 the feed 3 is produced from the components 47 of the feed 3 by sieving and / or mixing in a preparation step 48, and currently exemplarily and also generally preferably a powder mixture of the components 47.

[0349] In a granulation step 49 (preferably continuously according to the proposal and followed by a subsequent drying step 50), the feed 3 is then processed into an intermediate product 14 by means of a granulator 13, and preferably a liquid 13B (also called granulation liquid) is added. In a continuous process, the intermediate product 14 is then dried by means of a dryer 15, thereby finally producing the formulation 2. Regarding the continuous processing from the feed 3 to the formulation 2, see also the previous paragraph for supplementation.

[0350] The formulation 2 can then be further processed. In one or more post-processing steps 51, the formulation 2 can be sieved, for example, in a post-processing device 52, especially for particle picking, to form a post-processed (especially sieved) formulation 53. Alternatively or additionally, in a mixing process 55, the formulation 2 or the post-processed formulation 53 is mixed with additives such as disintegrants and / or binders 56 to form a final mixture 54.

[0351] Finally, by means of a tableting process 57, especially compression, the formulation 2 or the post-processed formulation 53 or the final mixture 54 can be made into a dosage form 58, especially one or more tablets.

[0352] As shown in the schematic flow chart according to Figure 8 the proposed preferred pharmaceutical process focuses on the continuous process steps for manufacturing solid oral dosage forms, and will be supplemented and explained below according to the system 45 of Figure 9 .

[0353] After the optional (preferably batch-wise) preparation of a homogeneous powder premix of component 47 as feed 3 (preferably forming a powder as component 47 of feed 3 by sieving and / or mixing), an intermediate product 14, preferably a wet granulate (granulate with wetness or humidity), is formed in a continuous granulation step using a granulator 13, preferably a so-called twin-screw granulator (TSG). For this purpose, a liquid 13B can be added to feed 3 in the granulator 13 or during the granulation process. This can be done using an injection device 6C.

[0354] The intermediate product 14 is preferably directly conveyed in a continuous product stream to a continuously operating dryer 15, here a fluidized bed dryer. After these two consecutive process steps of granulation and drying, a formulation 2, preferably a dry granulate (dried wet granulate), is produced.

[0355] The formulation 2 is optionally and preferably then sieved (to a post-treatment formulation 53) and / or mixed with an extra-granular phase (additive / splitting and / or binder 56) to obtain a final mixture 54. This final mixture 54 (final blend) preferably forms the starting material for a tableting process 57, for tableting or the system 45 is constructed for this purpose.

[0356] Therefore, preferably, the two consecutive process steps of granulation and drying are provided as equipment 1. In turn, the system preferably combines the equipment 1 into a total of at least three, especially four different process units or production steps.

[0357] The process unit of feeding and twin-screw wet granulation is responsible for the continuous granulation process. The next two process units - the continuous fluidized bed drying and the pneumatic conveying system - are responsible for the continuous drying process.

[0358] Overall, (preferably six) control parameters 5 (main input variables) are defined for the control / control device of the continuous granulation and drying equipment 1, or for control.

[0359] At least two, preferably at least three control parameters 5 or main input variables of the granulator 13 preferably are or include the metering [kg / h] of (feed 3), the extruder speed [U / min] of (the granulator 13), and / or the injection rate [g / min] of (the injection device 6C).

[0360] For the dryer 15, at least two, preferably at least three control parameters 5 or main input variables are or include the supply air flow [m 3 / h] into the dryer 15, the supply air temperature [°C] of the supply air into the dryer 15, and / or the rotational speed [U / min] of the rotating part of (the dryer 15 or its rotating part).

[0361] Since it is generally preferred to define or use at least six main input parameters or control parameters 5 for the device 1 for control, it is not difficult to understand that the understanding, control, and monitoring of the interconnections of such a multi-factor device 1 can be a challenge.

[0362] Figure 8 and Figure 9 shows a schematic view of the material and data flow of the continuous granulation and drying device 1 or the system 45 formed therefrom.

[0363] In terms of the material flow, it is preferred that the system 45 operates or is constructed in a bin-to-bin manner. This means that the first container feeds a preferably homogeneous powder premix as feed 3 into the continuous production line (consisting of a granulator 13 and a dryer 15) so that after two successive process steps (preferably twin-screw wet granulation and fluidized bed drying), dried granules as formulation 2 are obtained in the second container.

[0364] The powder premix as feed 3 is preferably prepared batchwise by the system 45. The further processing of the dry granules (formulation 2) is preferably also carried out batchwise, as Figure 9 shown.

[0365] Contrary to fully continuous manufacturing (from raw materials to finished tablets), this bin-to-bin method offers greater flexibility because the system 45 itself can be used modularly. Thus, the advantage is that the continuous process consisting of granulation and drying is embedded in the bin-to-bin method, thereby achieving the above advantages with a high degree of flexibility at the same time.

[0366] When controlling the device 1 or the system 45, three types of data are preferably considered.

[0367] Firstly, the control parameters 5, preferably including the six relevant control parameters 5 described above.

[0368] The second type of data is the state parameters 7 related to the process state (such as temperature, pressure loss, torque), and the process state can be measured by a plurality of sensors 8 throughout the device 1, preferably in an online and / or real-time mode.

[0369] The third type of data preferably includes one or more feed parameters 4 and / or actual formulation parameters 11, and these key material properties (material properties of the feed 3, intermediate product 14, and / or formulation 2) are preferably measured separately from the device 1, discontinuously, and / or based on random samples as in-process control, and thus form the third type of data with a time delay.

[0370] The complete data set preferably consisting of all three types of data forms the basis for creating the model 12 used in the present invention.

[0371] As a particularly surprising and important advantage of the continuous granulation and drying device 1 according to the present invention, two related aspects can be emphasized in particular, even if these two aspects are not mandatory:

[0372] It should be understood the multi-factor interaction of, for example, six main input parameters (control parameter 5 and state parameters 7 that can be directly affected by the control parameter 5 or are not affected by the feed 3) and the resulting output parameters. The state parameters 7 that are generated or affected by the feed (for example, 10 states; measured online and in real time) and the material properties (preferably the properties characterizing the feed 3 and the formulation 2, described by, for example, a total of four material parameters or actual formulation parameters 11; measured offline and with a time delay) can be defined as output parameters.

[0373] Furthermore, the continuous granulation and drying device 1 preferably combines different (continuous) process steps. Therefore, the process parameters or control parameter 5 of one process unit also affect the process state or state parameter 7 of subsequent units and ultimately affect the material properties or actual formulation parameters 11.

[0374] If these two aspects are taken into account, it will be found that optimal manual control of the process is a challenge and almost no good results will be achieved. For this reason, the great advantages of machine learning-based methods are clearly visible, which can predict and control the adjustment of the continuous process.

[0375] In summary, the following aspects can be implemented individually or in different feasible combinations in the device 1:

[0376] - Integrated continuous process: Instead of a completely continuous process (from active materials and excipients to the final product), at least two traditional batch processes are replaced by one continuous process step; the combination of batch processes and continuous steps = integrated continuous process

[0377] - Use of a flat-bottom metering device as the conveyor 16, which enables a very precise and well-controlled mass flow;

[0378] - Use of a double-screw wet granulation device as the granulator 13;

[0379] - Use of a continuous fluidized bed dryer as the dryer 15 enables dual functions; the fluidized bed dryer can be converted into a fluidized bed granulator. This more modular and variable system enables the dual use of the main equipment, thereby increasing the productivity of the equipment, effectively utilizing the floor area of the equipment, and reducing downtime;

[0380] - Use a special continuous dryer 15 with a slowly rotating rotating part that can divide a large fluidized bed chamber into multiple small chambers, thus avoiding the formation of partial batches;

[0381] - Reduce the possible significant congestion of wet granules by shortening the transportation route and avoiding valves for wet granules (intermediate product 14);

[0382] - Generated from on-line data of NIR probes and / or particle size measurement systems (intermediate product sensors 8L / formulation property sensors 8K) for measuring the material properties of wet granules (intermediate product 14)

[0383] and / or dry granules (formulation 2).

[0384] The present invention also relates to a computer program product or a computer-readable storage medium that contains instructions which, when executed by a computer, enable the computer to execute the proposed method / the steps of the method or parts thereof.

[0385] The various aspects of the present invention can be implemented separately from each other, but can also be implemented in different combinations. In particular, the aspects related to the system 45 can be combined or can advantageously be combined with the aspects previously described in connection with the combination of the granulator 13 and the dryer 15.

[0386] List of Reference Numerals

[0387] 1 Equipment

[0388] 2 Formulation

[0389] 3 Feed

[0390] 4 Feed parameters

[0391] 5 Control parameters

[0392] 5A Conveyor drive control parameters

[0393] 5B Granulator drive control parameters

[0394] 5C Injection device control parameters

[0395] 5D Granulator temperature control device control parameters

[0396] 5E Gas supply conveyor drive control parameters

[0397] 5F Gas supply temperature control device control parameters

[0398] 5G Exhaust conveyor drive control parameters

[0399] 5H Fluidized bed drive control parameters

[0400] 6 Actuators

[0401] 6A Conveyor Drive

[0402] 6B Granulator Drive

[0403] 6C Injection Device

[0404] 6D Granulator Temperature Control Device

[0405] 6E Gas Supply Conveyor Drive

[0406] 6F Gas Supply Temperature Control Device

[0407] 6G Exhaust Conveyor Drive

[0408] 6H Fluidized Bed Drive

[0409] 7 Status Parameters

[0410] [7] Status Parameter Values

[0411] 7A Conveyor Parameters

[0412] 7B Granulator Parameters

[0413] 7C Injection Device Parameters

[0414] 7D Granulator Temperature

[0415] 7E Gas Supply Conveyor Parameters

[0416] 7F Gas Supply Parameters

[0417] 7G Exhaust Gas Parameters

[0418] 7H Fluidized Bed Parameters

[0419] 7I Formulation Temperature

[0420] 7J Formulation Outlet Quantity Parameters

[0421] 7K Formulation Characteristic Parameters

[0422] 7L Intermediate Product Characteristic Parameters

[0423] 8 Sensors

[0424] 8A Conveyor Sensors

[0425] 8B Granulator Sensors

[0426] 8C Injection Sensors

[0427] 8D Granulator Temperature Sensors

[0428] 8E Gas Supply Conveyor Sensors

[0429] 8F Air Supply Sensor

[0430] 8G Exhaust Conveyor Sensor

[0431] 8H Fluidized Bed Sensor

[0432] 8I Formulation Temperature Sensor

[0433] 8J Formulation Outlet Quantity Sensor

[0434] 8K Formulation Characteristic Sensor

[0435] 8L Intermediate Product Sensor

[0436] 9 Control Unit

[0437] 10 Target Formulation Parameter

[0438] 10A Database

[0439] 11 Actual Formulation Parameter

[0440] 12 Model

[0441] 12A Static Component

[0442] 12B Dynamic Part

[0443] 13 Granulator

[0444] 13A Granulator Inlet

[0445] 13B Granulation Liquid

[0446] 14 Intermediate Product

[0447] 15 Dryer

[0448] 15A Air Supply Conveyor

[0449] 15B Exhaust Conveyor

[0450] 16 Conveyor

[0451] 16A Storage Container

[0452] 16B Input Device

[0453] 16C Worm

[0454] 17 Granulation Mechanism

[0455] 18 Worm

[0456] 19 Processing Area

[0457] 20 Coupling Device

[0458] 21 Intermediate Product Outlet

[0459] 22 housing

[0460] 23 housing

[0461] 24 intermediate product inlet

[0462] 25 fluidized bed

[0463] 26 air inlet

[0464] 27 air

[0465] 28 diffuser

[0466] 29 outlet

[0467] 30 separator

[0468] 31 formulation outlet

[0469] 32 artificial neural network

[0470] 33 input layer

[0471] 34 hidden layer

[0472] 35 output layer

[0473] 36 node

[0474] 37 edge

[0475] 38 weight

[0476] 39 reference trajectory

[0477] 40 prediction range

[0478] 41 fixed time window

[0479] 42 quality dynamics

[0480] 43 planning dimension

[0481] 44 process status

[0482] 45 system

[0483] 46 conveyor equipment

[0484] 47 component

[0485] 48 preparation step

[0486] 49 granulation step

[0487] 50 drying step

[0488] 51 post - treatment step

[0489] 52 post - treatment device

[0490] 53 Post-treatment formulation

[0491] 54 Final mixture

[0492] 55 Mixing process

[0493] 56 Splitting and / or binder

[0494] 57 Tabletting process

[0495] 58 Dosage form

[0496] t Time

[0497] t k+p Time point

[0498] Δt Sampling time

[0499] P Past

[0500] F Future

[0501] ζ Relative time

[0502] κ Measured value

Claims

1. A method for controlling an apparatus (1) for manufacturing a formulation (2) from a feedstock (3), wherein, Said manufacturing includes processing said feedstock (3) with a granulator (13) and drying an intermediate product (14) produced from said feedstock (3) by said granulator (13) by means of a dryer (15). Wherein, based on a model (12): (a) Control parameters (5) of said device (1) are determined, said control parameters being or representing manipulation variables for actuators (6) for controlling said device (1), wherein said model (12) takes into account predetermined target recipe parameters (10), said target recipe parameters representing desired properties of a recipe (2) that has been or is to be produced, in particular expected particle size properties and humidity, or (b) Actual recipe parameters (11) are predicted, said actual recipe parameters representing actual properties of the recipe (2) that has been or is to be produced, in particular actual particle size properties and humidity. Wherein said model (12) takes into account predetermined or predeterminable control parameters (5), said control parameters being or representing manipulation variables for actuators (6) for controlling said device (1). Wherein state parameters (7) of said device (1), in particular sensor values, each represent a state of said device (1) affecting said manufacturing, said state parameters are determined and processed by said model (12), wherein feed parameters (4) represent properties of said feedstock (3), in particular humidity and / or particle size properties, and said model (12) takes into account said feed parameters, and Wherein said dryer (15) is coupled to said granulator (13) such that said intermediate product (14) is automatically and continuously conveyed from said granulator (13) into said dryer (15), and said model (12) takes into account a combination of continuous operation consisting of granulation with said granulator (13) and subsequent drying with said dryer (15), and / or Wherein said model (12) has a static part (12A), and reference values for corresponding control parameters (5) or predicted actual recipe parameters (11) are determined using said static part, wherein said model (12) has a dynamic part (12B), and said reference values are optimized using said dynamic part by means of prediction.

2. The method according to claim 1, characterized in that, Said model (12), in particular said static part (12A) of said model (12), is or has an artificial neural network (32), wherein said artificial neural network (32) includes nodes (36) in an input layer (33), said nodes corresponding to at least one state parameter (7) of said granulator (13), at least one state parameter (7) of said dryer (15), and preferably to said feed parameters (4).

3. The method according to claim 1 or 2, characterized in that, Said artificial neural network (32) has nodes (36) in an output layer (35) corresponding to said control parameters (5), and said artificial neural network (32) has nodes (36) in said input layer (33) corresponding to said target recipe parameters (10); or The artificial neural network (32) has nodes (36) corresponding to the control parameters (5) in the input layer (33), and the artificial neural network (32) has nodes (36) corresponding to the predicted actual recipe parameters (11) in the output layer (35).

4. The method according to claim 2 or 3, characterized in that, The artificial neural network (32) is trained with different training data sets, which respectively consist of combinations of the feed parameters (4), the state parameters (7), the control parameters (5), and the actual recipe parameters (11) set under the condition of predetermining these parameters.

5. The method according to claim 4, characterized in that The training data sets respectively represent the stationary states of the device (1), in which the actual recipe parameters (11) and the state parameters (7) adopt at least substantially static values based on the constant control parameters (5) and feed parameters (4).

6. The method according to claim 5, characterized in that, The artificial neural network (32) is trained in such a way that at least one state parameter (7) of the granulator (13) and at least one corresponding state parameter (7) of the dryer (15) of the corresponding training data set are loaded into the nodes (36), and preferably the corresponding feed parameters (4) of the corresponding training data set are also loaded.

7. The method according to claim 6, characterized in that, As long as nodes (36) are provided in the input layer (33) for this purpose, the nodes (36) are also correspondingly loaded with the corresponding control parameters (5) or actual recipe parameters (11).

8. The method according to claim 6 or 7, characterized in that, Values are respectively generated at the nodes (36) of the output layer (35) for the corresponding control parameters (5) or predicted actual recipe parameters (11), and the error is determined by comparing the values with the corresponding control parameters (5) or actual recipe parameters (11) of the corresponding training data set, wherein the error is reduced by adjusting the weights of the artificial neural network (32), preferably by means of backpropagation and / or in a continuous manner from one training data set to another training data set.

9. The method according to any one of the preceding claims, characterized in that, The control parameters (5) have control parameters (5B, 5C, 5D) for controlling the granulator (13) and at least one control parameter (5E, 5F, 5G, 5H) for controlling the dryer (15).

10. The method according to any one of the preceding claims, characterized in that, The state parameters (7) have at least one parameter (7B, 7C, 7D) describing the operating state of the granulator (13) and parameters (7E, 7F, 7G, 7H) describing the operating state of the dryer (15).

11. The method according to any one of the preceding claims, characterized in that, With the aid of the model (12) (a) the control parameters (5) are determined only based on the feed parameters (4), the state parameters (7), and the actual recipe parameters (11) or (b) the predicted actual recipe parameters (11) are determined only based on the feed parameters (4), the state parameters (7), and the control parameters (5), Preferably, wherein the non-online measurable characteristics of the intermediate product (14), especially the characteristics of the intermediate product (14) related to the particles, are not considered.

12. The method according to any one of the preceding claims, characterized in that, The model (12), in particular the dynamic part (12B) of the model (12), takes into account the long-term influence of the change of the control parameter (5) on the actual formulation parameter (11) and / or the state parameter (7); and / or The change of the state parameter (7) is predicted by means of the dynamic part (12B) of the model (12), and the reference value of the control parameter (5) is adapted based on the prediction, and the device (1) is controlled by means of the control parameter (5) optimized in this way.

13. A computer program product or a computer-readable storage medium, which comprises instructions that, when the program is executed by a computer, cause the computer to execute the method according to any one of the preceding claims.

14. An apparatus (1) for manufacturing a formulation (2) from a feed (3), wherein, The manufacturing includes processing the feedstock (3) with a granulator (13) and drying the intermediate product (14) manufactured with the granulator (13) by means of a dryer (15), and the device has: A sensor for detecting the state parameter (7) of the device (1), and the state parameter respectively represents the state of the device (1) that affects the manufacturing. An actuator (6) of the device (1) for directly or indirectly acting on the feedstock (3). A control device (9) for controlling the actuator (6) with a control parameter (5), and the control parameter is or represents a manipulation variable for controlling the actuator (6). Wherein, a target formulation parameter (10) is predetermined or predeterminable for the control device (9), and the target formulation parameter represents the desired characteristics of the formulation (2) that has been manufactured or is to be manufactured, in particular characterizing the physical characteristics and humidity of the granules, and A model (12) based on which the control parameter (5) can be determined and / or the actual formulation parameter can be predicted. It is characterized in that The granulator (13) is coupled to the dryer (15) such that the intermediate product (14) is automatically and continuously conveyed from the granulator (13) into the dryer (15), wherein the model (12) takes into account the combination of granulation with the granulator (13) and subsequent continuous drying with the dryer (15), and the device (1) is constructed for this purpose such that the control device (9) uses the model (12) to determine the control parameter (5) and / or predict the actual formulation parameter (11) based on the state parameter (7) of the device (1); and / or Wherein, the model (12) has a static part (12A), wherein the control device (9) is constructed to use the static part (12A) to determine a reference value for the corresponding control parameter (5) or the actual formulation parameter (11) by means of the state parameter (7), and the model (12) has a dynamic part (12B), wherein the control device (9) is constructed to use the dynamic part (12B) to optimize the reference value by means of prediction.

15. A system (45) comprising the device (1) according to claim 14 and means for forming a feed (3) from a plurality of components (47), preferably powders, preferably by sieving, and / or means for further processing (51, 55, 57), preferably tabletting, of the formulation (2).

Images