EP3448798A1

EP3448798A1 - Method for optimizing the filling of a container

Info

Publication number: EP3448798A1
Application number: EP17719258.0A
Authority: EP
Inventors: Jochen Ohrem; Ludwig Clüsserath
Original assignee: KHS GmbH
Current assignee: KHS GmbH
Priority date: 2016-04-29
Filing date: 2017-04-25
Publication date: 2019-03-06
Also published as: DE102016108053A1; US20190127198A1; US10618790B2; WO2017186708A1

Abstract

The invention relates to a method for optimizing the filling of a container with a water-containing product, in which method the product is fed to the container via an adjustable control valve, the passage cross-section of which can be controlled in order to control the filling rate during the filling, in which method the following is performed in a learning phase: a) at least one filling parameter, in particular the filling rate, is set in accordance with a similar product, b) a filling operation is performed with the set filling parameter, wherein the foaming associated with the set filling parameter and/or the total filling duration of the filling operation is measured as a target value, c) relationships between the filling parameter and the target value are captured and saved in a data model that simulates the behavior of the filling machine for different products, d) the saved target value is evaluated by means of an evaluation function in order to determine a best target value, e) if a termination criterion is achieved, a jump is made to step g), f) the set filling parameter is varied on the basis of an algorithm, in particular by using the data model, and a jump is made back to step b), g) the filling parameter associated with the best target value is saved as an optimal filling parameter for the product in the data model, and h) the optimal filling parameter is used during the filling with the product in a filling machine.

Description

Method for optimizing the filling of a container

The present invention relates to a method for optimizing the

Bottling a container, especially in the context of beverage containers, such as bottles, cans or kegs by filling machines in the beverage industry. When filling a container, the product to be filled flows through the filling valve of a filling machine into the beverage container, which can be, for example but not limited to, a PET bottle. The maximum filling speed that can be achieved when filling the product depends on various factors, such as the density and viscosity of the product and its CO ₂ content.

In general, the product is not accompanied by an additional

Pressurized filling in the container, but it runs due to its own weight and / or the present geodetic slope between level and discharge opening of the filling valve through the filling valve in the container.

The flow rate during filling, in short, the filling rate, can thus be adjusted over the opening cross-section of the open filling valve, wherein the opening cross section optionally by a

Control valve is variable.

In principle, the aim is to keep the entire process time for filling the container as short as possible.

In this context, the filling process time is understood to mean the period of time necessary for opening the filling valve, filling the container, closing the filling valve, calming the contents in the container and closing the filling itself.

A shortening of the process time could theoretically be achieved by making the opening cross-section of the filling valve as large as possible

1 1314 WO makes what is technically possible apparatus and due to the real geometric shape of the container only limited.

Limits in minimizing the actual filling time are in particular set by a CO2 content of the products, which leads to a foaming of the product during filling, regardless of whether the filling under a pressure or without.

Usually, the filling of a CO2-containing beverage under a counter pressure, the so-called filling pressure, at least the

Saturation pressure of the CO ₂ dissolved in the product corresponds to the

Leakage of CO ₂ from the drink is limited. This is necessary because, in particular, when the filling material flows into the container, the CO2 that is physically dissolved in the product is released from the product.

Therefore, the filling speed over the time of the filling process in the manner of a Füllgeschwindigkeitsprofils is varied, such that on the one hand a low foaming occurs and on the other hand, the product is transferred as quickly as possible in the container.

The possible filling speed is further from others

Depending on parameters, such as the viscosity of the product, which is determined for example by the syrup or beverage concentrate content in the water, by the solubility of the CO ₂ in the product, by the

Ambient temperature, the filling pressure and of course apparatus technical parameters of the filling machine.

Instead of the filling speed, the opening position of a Füllbzw. Control valve over the time of the filling process can be used as a control variable. A method of this kind is known, for example, from EP 2 975 486 A2. It is an object of the present invention to provide a method which enables more efficient filling of different products.

This object is achieved by a method according to claim 1. Advantageous developments of the invention are the subject of the dependent claims. Advantageous developments of the invention are also disclosed in the description and in the drawings.

According to the invention, an optimization method is carried out with a learning phase in which the following method steps are used: a) at least one filling parameter, for example the

Filling speed, is set to a starting value, in particular the value of the filling parameter of a similar product, b) with the set filling parameter, the product is filled, with the filling parameter assigned

Foaming and / or total filling time of the filling process, that is preferably to calm down the beverage in the container, as

Target value is measured,

c) between the set filling parameter and the target value, correlations are recorded and stored in a data model simulating the behavior of the filling machine for the product, d) the stored target value is evaluated by means of a valuation function for determining a best target value,

wherein according to step e) upon reaching a termination criterion to step g) in which the fill parameter associated with the best target value is stored as the optimal fill parameter for the product in the data model and h) the optimal fill parameter when filling the product into one

Filling machine is used.

The abort criterion is defined, for example, by the lapse of a certain period of time or when a certain quality of the target value has been reached or at a certain convergence of the target value after several

Optimization rounds b) to f).

As long as the termination criterion is not reached, in accordance with step f), the set fill parameter is based on a particularly self-learning algorithm, in particular using the

Data model varies and it is returned to step b), where a re-filling process with the varied filling parameters

is carried out and in turn the associated target value is determined and evaluated.

By this method can be optimized for a product of the filling process, that is, in particular the time course of the

Filling speed or the time course of the position or lifting height of the filling valve and thus allows the optimization of the Füllgeschwindigkeitsprofile and thus an optimized control of the control valves / filling valves of the filling elements of a filling machine during the filling process, which ultimately reduces the process time is achieved.

The product-specific and optionally machine-specific optimized filling parameter can then be stored in the data model. For example, the data model may be in a memory of a controller Laboratory filling plant in which optimization tests are carried out for the filling process of a new product. In addition, target values and corresponding optimal filling parameters or generally the data of the data model or the data model itself can be stored in a data cloud so that, for example, a bottler or beverage manufacturer has global access to these values. Both can thus be optimized already

Using fill parameters when filling an already known product, or when a new product is present, the fill parameters of a similar product can be used as a starting point for an optimization process. The invention thus does not only allow for known

Products, but also for new products a very fast and effective optimization of the filling process.

In particular, the filling speed or position of the filling valve / regulating valve is optimized as a filling parameter, in particular its time course over the time of the filling process (opening of the filling valve until the filling valve closes).

In an advantageous embodiment of the invention, the / the optimal (n) filling parameters of the product together with his

Product parameters stored in the data model and correlated to fill parameters for products with similar

Derive product parameters. This allows a fast calculation and optimization of filling parameters of new products.

Preferably, a self-learning algorithm is used as the algorithm, for example an evolutionary algorithm, in which self-learning structures, in particular in the manner of a neural

Network are provided. Such a self-learning algorithm develops some cognitive ability and thus can greatly simplify or speed up the optimization of at least one fill parameter of the product. XXX

The mode of operation of evolutionary algorithms or methods are the

Professional since 1968 known. This year, a case study was presented in which it is important to optimize a generator for

Satellite, which operated on the magnetohydrodynamic principle, was necessary to the performance of a two-phase supersonic nozzle

optimize.

Starting from a conventionally shaped Venturi nozzle, a larger number of sections were created, these sections each had holes of different diameters. Due to the random combination of this section and the determination of the respective power of each nozzle formed, a shape could be found after a manageable number of recombinations of the sections whose

Performance, or efficiency was significantly higher than that of a conventional Venturi nozzle. The shape of this powerful new nozzle was unpredictable and unpredictable at the time. In the context of these evolutionary algorithms, random arrangements are tested for their performance and evaluated. Assign these new orders

improved properties, these arrangements expand

Starting point used for further changes of the arrangement, the new arrangement has deteriorated properties, so are the

Changes that led to this worse arrangement have been discarded.

As product parameters of a product are in particular the

Beverage concentrate or syrup content, the temperature of the product, the beverage kekonzentrattyp, to name the viscosity, the density, the CO ₂ content and / or the solubility of the CO2 in the product. Of course, the component of certain soluble or insoluble beverage components can be listed as a further product parameter. In the event that the individual product parameters have different influences on the filling properties of the product, it is preferably provided that these different weightings are taken into account in the algorithm by means of corresponding weighting factors.

In this way, the determination of the filling characteristics of the product is simplified for the data model. An important product characteristic is, for example, the viscosity of the product because it significantly affects the filling rate when passing through the filling valve.

For the determination of the optimum filling parameter or for determining the optimum time profile of a filling parameter, the associated optimum filling parameter of at least one similar product is preferably interrogated and used as the starting point in step a) of the method.

In this way, with the starting point for the optimization process, you are already in a favorable range, so that you can get through the

Optimization process reaches an optimal filling parameter faster.

Of course, as a starting point, first defaults or values from the experience of a developer can be used. Of course, can be simultaneously optimized by the inventive method multiple filling parameters. In this case, however, preferably the filling parameters are optimized one after the other and not several filling parameters at the same time, because this would be too complex. However, it is possible to re-optimize an already optimized fill parameter after optimizing other fill parameters because after changing other fill parameters, the optimum value of the fill parameter can now be easily shifted. Thus, to determine a further filling parameter, the method according to the invention is returned to step a) after step g), the further or new filling parameter then being optimized in the further steps a) to g). That way you can

successively optimize any number of fill parameters.

Other parameters, such as the temperature, can be computationally captured and corrected by the data model, for example. Thus, for example, an optimized filling curve at 25 ° C can be used to calculate an optimized filling curve at 20 ° C, without the need for a separate optimization process. However, a separate temperature optimization process may be run through, where the data model may use known temperature relationships from other products. Here, too, the data model, in particular taking into account self-learning algorithms and structures, leads to a fast achievement of an optimized filling parameter.

Although as filling parameters to be optimized in general the

Filling speed or the time course of the filling speed or filling valve position is used, other filling parameters can be optimized, such as filling time, temperature, filling pressure, level of the product in the container filled, CO ₂ release during filling. For example, in the course of the optimization, the variation of a filling parameter may be either the size or the time profile of the

Include fill parameters or both at the same time.

If, for example, the filling speed is varied as a filling parameter, the time profile of the filling speed over the filling process, that is to say a filling speed profile, is preferably optimized. On the basis of this filling speed profile, the filling behavior of a product can be recognized relatively quickly by the person skilled in the art. Even for the data model, such a filling speed profile is easy to process and evaluate. In the fill rate profile, for example, different features such as inflection points, maxima, minima, or first or second derivatives may be used as filling characteristics. Within the framework of an optimized filling parameter, these characteristics can then be stored and used for use in a filling process when filling a beverage with a filling machine.

The invention also covers a computer program product stored on a computer-readable data memory suitable for

Carrying out the abovementioned method according to the invention, wherein the computer program product is stored in the working memory of a computer which is connected to or connectable to a filling machine control. For example, the data model may be loaded into a cloud from the computer of a laboratory filling plant performing optimization tests, from which the beverage manufacturer or bottler then uses the correspondingly optimized filling parameters for the filling machines in connection with a serial filling of the corresponding products.

In this way, the filling process is standardized and optimized, resulting in a more reproducible beverage quality and a time-optimized filling process.

The invention can also be characterized in the following way:

With the help of search algorithms (evolutionary algorithms) the

Filling speed or the filling speed profile in the filling process during a learning phase (training) optimized. The controlled variable is

preferably the foaming during or at completion of filling, the sensory via a camera and a corresponding Image evaluation or one or more electrical contacts or in any other suitable manner.

The aim is to quickly fill different types of drinks with the least possible foaming by the possibility of varying the filling speeds during the filling process. Optimizing the

Filling speed control with the target value of the life of a filling valve may also be useful, since repair and / or the replacement of components such as bellows or complete filling valves are complex and expensive and also cause high costs due to the necessary for the measures mentioned production interruptions.

The described procedure can be of great use, in particular in filling operations with a multiplicity of different beverages and container shapes, and / or in the filling of new types of beverages / container shapes. Determining an optimized velocity profile of the

Volume flow during the filling process is preferably done automatically, e.g. via artificial neural networks or reference trajectory or other known control engineering methods. A learning algorithm approximates e.g. a predetermined optimum, e.g. by changing the weighting and adjusting the thresholds.

For example, the learning phase is divided into 5 steps:

- 1) forecast

- 2) filling / testing

- 3) Measure / Recognize

- 4) Analyze

- 5) Changing / adjusting the parameters

The learning phase is preferably repeated several times. Step 1 )

The prediction can be derived from similar beverages that are filled into similar containers with similar filling valves. It is conceivable that the process data (speed profiles, pressure, temperature, etc.), beverage parameters (CO ₂ content, viscosity, density, ingredients, etc.) of the filling valves and the filling valve type are stored worldwide in a cloud, for the new setting can be used.

Characteristic of beverages to be filled are in particular temperature, CO ₂ content, pressure and viscosity. With similar values, a database can be used anywhere in the world, and so on

Bottling behavior can be predicted. The database can be in

Context of the beverage and filling parameters are constantly supplemented. The database may be the property of a beverage manufacturer and bottler.

Step 1 is not absolutely necessary, but it does make the optimization of the speed profile faster. If there is no database with stored filling parameters of similar products, step 2 is started.

Step 2)

With the set filling parameters (either from step 1 or random or fixed starting value) the filling process is carried out.

Step 3)

The target value, e.g. the filling time or foaming is measured, e.g. via camera or electrical contacts.

Step 4) The relationships are analyzed. Step 5) The full parameters are changed or adjusted based on the data stored in the database or arbitrarily Revolutionary). If there is an improvement to previous fillings, the settings will be retained or by repeating steps 2-5 with light ones

Changes in the velocity profile were performed. Following is a data model with assigned target values and optimized

Filling parameters formed (each for a type of beverage in a container type and a Füllventilart).

The beverage-specific data could be stored in the described cloud database, in particular in the context of the data model, in order to have fill parameters of existing products quickly at hand, or to be able to quickly optimize those of new products. The data model is preferably a data-based modeling of correlating product and fill parameters of different products, if appropriate also considering different ones

Fill valve or filling machine types. These model data can be used for a self-learning filling valve, in particular with a control valve.

In the case of the use of a control valve, the speed of the filling material can be incorporated in a loop in such a way that it can

is regulated depending on the viscosity, or by means of self-learning

Algorithm, the filling speeds of different products are set automatically during the filling process.

As a result of the invention, characteristics such as certain inflection point as well as times to reach the maximum speed (as described above) may be considered as essential points of a filling parameter in the

Optimization method can be used to complete the filling process

Optimize dependence on beverage and filling valve in particular to accelerate, preferably by minimizing the foaming. It is obvious to the skilled person that the above

Embodiments of the invention can be combined in any way with each other. The following terms are used synonymously: drink - product;

Container - Bottle Can Keg; Flow meter - MID;

The invention will be further described by way of example in the schematic drawing. In this show:

1 shows a schematic representation of the filling valve of a filling machine, as used in modern filling machines for bottling beverages, FIG. 2 shows a schematic block diagram of a self-learning algorithm with the setting and control variables, which are optimized in the algorithm,

Fig. 3 shows the representation of a flow profile when starting a

Filling process in the product channel of a filling element,

4 shows an optimized filling speed profile, which theoretically can not be derived in this way, FIG. 5 shows a time profile of the opening position of the filling valve in an analogous manner to the filling speed profile according to FIG. 3, FIG.

6 shows a profile of the time sequence of the filled product quantity, and FIG. 7 shows the detection of characteristic values from the first and second derivation of the time profile of the filled product quantity for characterizing the filling parameter. Fig. 1 shows schematically a filling valve 10 which is used in a filling machine for filling a container, such as a bottle 12. The filling element 10 contains a product channel 14, in the course of which a magnetically inductive flow meter 16 (MID) for determining the amount of product flowing through the product channel 14 during filling

(Volume) is designed as well as a filling valve 20 is arranged, which is controllable with respect to its opening stroke, in order to be able to control the filling flow over the time course of the filling process in this way. Furthermore, in the filling area, an optical sensor 22 for detecting the

Foaming formed during filling of the product. The arrangement of MID 16 and filling valve 20 may be arbitrary. The fill valve 20 may be an on / off valve, a 2-stage valve or a control valve. Instead of a flow meter 16 can also be a load cell and / or a

Level sensor can be used.

According to FIG. 2, the data model in which product-specifically at least one, preferably a plurality of filling parameters, such as the time course of the filling speed, are optimized, has a self-learning algorithm 30 to which target values 32 are input, such as the filling time of a filling process, for example. the

Foaming or the level of the beverage in the container. Further, the self-learning algorithm 30 is supplied with product parameters 34, such as the viscosity of the product, the temperature of the product, the CO2 content, the beverage ingredients. Output or control variable is a control signal for the filling / control valve 20. It can thus over the time course of the control signal, a time course of the

Filling / regulating valve position can be set during the filling process, which leads to a minimal foaming or to a short total filling time.

In addition, the self-learning algorithm, for example, also known or similar filling parameters 36, for example, as start values supplied, such as filling speed, level in the container, foaming, filling time. The self-learning algorithm 30 is part of a data model 40 in which the correlated product parameters, fill parameters and target values of a product are stored in correlation with the self-learning algorithm 30 and maintained as a model. This model allows a quick optimization of the fill parameters that are essential for batch filling, in particular the position of the filling valve during the filling process for a defined product. This timing of Füllventilstellung (if this is a control valve or has) can be used for example via a data cloud in all filling machines used by the beverage manufacturer and / or bottler, with the result that all filling machines work with the same optimized Füllparametern and globally a reproducible quality is achieved when filling the product. Alternatively and / or additionally, the corresponding parameters of at least some of the machines manufactured by the machine manufacturer can also be stored in the data cloud, these parameters preferably also

Can be recalled product specific.

2 thus shows the mode of operation of the learning control with the control valve as an actuator and in particular the measurement of the foam by at least one sensor. The influences for the filling of beverages are dependent, in particular, on the CO ₂ content, viscosity, temperature, pressure and the ingredients of the beverage. The measurable target is the

Foaming behavior (amount and / or amount of the foam) is selected upon reaching the level or during the filling process. This can be realized by means of a camera or by an electrical contact (or several). As a result of the optimized filling parameters, it is thus possible to obtain directly a time-related control signal for the filling valves / control valves 20 of a filling machine. 3a to d show the product level 50 in the product channel 14 of a filling element 10 in Fig. 1 during the start of the filling process,

laboratory-technically modeled on the basis of a single

Filling. In Fig. 3a before the beginning of the filling process, the product level is flat, that is, it extends transversely to the wall of the product channel 14. In Fig. 3b, the filling valve 20 has just opened, so that the product in

Product channel accelerated and product level 50b slightly convex

is cursed. The indentation depends on the one from the

Acceleration and speed of the product in the product channel 14 and on the other hand the viscosity and adhesion of the product to the walls of the product channel 14. In Fig. 3c and 3d can be clearly seen how the product mirror 50c, 50d with increasing

Flow rate in the product channel 14 in the form of an ellipse convex.

FIG. 3 illustrates how product parameters, such as density [g / l], shroud shear wall (fill valve), viscosity of the product, adhesion of the product to product channel 14, interact with the fill parameters, as well the diameter of the product channel 14 and the shape of the product channel and the filling valve 20 count. The determined flow characteristic values (correlating to the viscosity) of the beginning filling by the acceleration in the

Fill valves can be used to optimize the self-learning control described above for a speed profile of the filling process.

Preferably, in the optimization of the filling process analogous to

Method of bionic development of technical solutions or bionic algorithms determines a beverage and container-dependent velocity profile, which, as an example, can look like this: Filling speed

[ml] [ml / s]

0-2 VO = 50

2-5 V1 = 2

5-10 V2 = 250

10-20 V3 = 100

20-25 ... V4 = 322

25-45 V5 = 127

45-60 V6 = 322

60-420 V7 = 255

420-425 V8 = 3

425-440 V9 = 1 1

440-500 V10 = 42

Such velocity profiles are shown in FIGS. 4 and 5. Possibly results in the optimization according to the invention such a Füllgeschwindigkeitsprofil that could not be derived from the current state of the present methods.

In the optimization, for example, the reproduced in Fig. 4

Füllgeschwindigkeitsprofil obtained in which the filling rate or volume flow, calculated from the measured data of the MID (electromagnetic flowmeter) is plotted over time. FIG. 4 shows a region A in which the filling speed during the

Start-up after opening the filling valve varies greatly. Thereafter, in a phase B, the filling speed is kept constant until closing filling valve at point C. Comparing the profiles of different

Liquids, then becomes an individual characteristic

Give velocity profile plotted as volume flow over time (e.g., in Fig. 6).

While Fig. 4 shows the time course of the filling speed or the product flow through the product channel 14, this can

Füllparameter also in a stroke of the filling or control valve 20th

be reproduced, as shown in Fig. 5. It is thus possible, analogous to the example above, instead of the volume or mass flow, the stroke of the control valve as a function of the filling time as to

optimizing fill parameters.

It is not absolutely necessary to determine the volume flow. So it is conceivable, by using a level control (using a camera or contact) to use a flow meter. Likewise, it is also possible to use a load cell instead of the flow meter. Different filling parameters can also be derived from one another by calculation.

For example, Fig. 6 shows a first filling curve P1 of a first product and a filling curve P2 for a second product. The filling curve represents the filled product quantity over time. Turning points 60a, b as well as end points 62a, b can be seen from the curves as well as the time until reaching the maximum filling speed, which represent essential parameters of the filling curve.

The turning points 60a, b of these curves indicate, for example, the

Transition from increasing filling speed to decreasing

Filling speed, while the end points 62a, b the end of

Mark filling process. By using first and second derivatives according to FIG. 7, the first derivative curve 70, which has its maximum at the point of inflection 60c of the filling curve P3, is obtained, for example, from a filling curve P3 by means of a first derivative. The first derivative curve 70 basically designates

taken marked the product channel 14 continuous product amount per unit time. Thus, at the inflection point 60c of the filling curve P3, the largest product flow flows through the product channel 14 of the filling valve 20, which then steadily decreases toward the end. The second derivative of the filling curve P3 is characterized by the second discharge curve 72. This curve

designates e.g. the control pulses for actuating the filling valve in the opening or closing direction, wherein the positive direction, the operation in the opening direction and the negative direction, the operation of the variable filling valve 20 in the closing direction. The point 74 denotes that

Maximum of the first derivative, i. the time of the biggest

Volume flow or the largest filling speed. For the second

This represents the zero crossing 78. The minimum 76 of the second discharge curve 72 indicates the time in which the control of the valve 20 in the closing direction is greatest. Thus, from the different discharges 70, 72 of the filling curve P3 with the innocuous characteristics 74, 76, 78 both relevant quantities for the actuation of the filling and control valve 20 can be derived as well as information about the

Product flow obtained during the filling process.

The invention is not limited to the accompanying embodiment, but may be within the scope of the following

Claims are varied. LIST OF REFERENCE NUMBERS

10 filling element

12 containers - bottle

14 Product channel of the filling element

16 flowmeters - MID

20 filling / regulating valve

22 optical sensor for foam detection and / or level 30 self-learning algorithm

32 target values

34 product parameters

36 filling parameters

40 Data model of different products with associated optimized filling parameters (possibly filling machine or

valve dependent)

50a-d product level in laboratory test Venturi nozzle

P1 -P3 filling curves of different products

60a-c turning point

62a-c end points (closing the filling valve)

70 first derivative of the filling curve

72 second derivative of the filling curve

74 maximum of the first derivative

76 minimum of the second derivative

78 zero crossing of the second derivative

Claims

claims:

1 . A method for optimizing the filling of a container with a water-containing product, wherein the method, the product is supplied to the container via an adjustable control valve whose

Passage cross section for controlling the filling rate during filling is adjustable, which method in a learning phase:

a) setting at least one filling parameter, in particular the filling speed, to a starting value, in particular the value of the filling parameter of a similar product,

b) the product is filled into the container with the set filling parameter, the foaming associated with the set filling parameter and / or the total filling time of the filling process being measured as the target value,

c) records and stores correlations between the filling parameter and the target value in a data model simulating the behavior of the filling machine for different products,

d) the stored target value is evaluated by means of an evaluation function for determining a best target value,

e) is jumped to step g) on reaching a termination criterion. f) the set filling parameter is varied based on an algorithm, in particular using the data model, and jumping back to step b),

g) the filling parameter assigned to the best target value as optimal

Filling parameter for the product is stored in the data model, and h) the optimal filling parameter when filling the product in one

Filling machine is used.

2. The method of claim 1, wherein the optimum filling parameter of the product, together with its product parameters, is stored in the data model and correlated

Derive filling parameters for products with similar product parameters.

3. The method of claim 2, wherein as an algorithm

self-learning algorithm is used.

4. The method of claim 2 or 3, wherein the product parameters include viscosity, beverage concentrate content, temperature, beverage kekonzentrattyp and / or CO2 content.

5. The method according to any one of claims 2 to 4, wherein in the

Algorithm the product parameters of a product to determine its filling properties with the influence on the filling behavior

reflecting weighting factors.

6. The method according to any one of claims 2 to 5, wherein for determining the optimum filling parameter for a new product of the associated optimum filling parameter of at least one similar product is queried and used as a starting point in step a).

7. The method according to any one of the preceding claims, wherein for determining a further filling parameter, the method after step g) jumps back to step a) and the further filling parameters in steps a) to g) is optimized.

8. The method according to claim 1, wherein the filling parameters include at least one or more of the following filling parameters: filling speed, product flow, position of the regulating valve, filling time (operating period of the filling valve), temperature, filling pressure,

Fill level.

9. The method according to any one of the preceding claims, wherein the variation of the filling parameter includes the size and / or the time course of the filling parameter.

10. The method according to any one of the preceding claims, wherein the filling speed is varied as a filling parameter.

1 1. The method of claim 10, wherein the time course of the filling speed is varied.

12. The method of claim 1 1, wherein first and / or second

Derivatives and maxima in the time course of the filling rate can be detected as a characteristic in the data model.

13. The method according to any one of the preceding claims, wherein in step g) the optimal filling parameter for the product is set in a data cloud.

A computer program product stored on a computer-readable data memory suitable for carrying out the method according to any one of the preceding claims, wherein the

Computer program product is stored in the main memory of a computer, which is connected to a filling machine control / connectable.