JP2014102558A - Method for preparing liquid level vibration control cum curve - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily obtain a plurality of cum curves high in liquid level vibration control capacity by selecting a liquid level vibration control cum curve from among a plurality of parent cum curves.SOLUTION: After a plurality of kinds of initial cum curves are selected, an individual estimation 12 for ranking each of the initial cum curves in the ascending degree of good performance is performed by CFD simulation. Next, the plurality of kinds of initial cum curves having good performance are selected, and a parent individual selection 13 for selecting a plurality of sets of first generation parent cum curves is performed with two different initial cum curves as a pair. Next, after a crossing 14 in a genetic operation is executed for each set to generate a first generation child cum curve, a loop composed of the individual estimation, the parent individual selection, and the crossing is repeated on the first generation child cum curve so as to select the parent cum curve having the excellent achievement. While repeating the loop as needed, a mutation in the genetic operation is executed, thereby generating a plurality of kinds of child cum curves.

Description

  The present invention relates to a method of creating a liquid level damping cam curve, and more specifically, a liquid level damping cam curve that can be conveyed so as to reduce the fluctuation of the liquid level when a container containing liquid is conveyed. The present invention relates to a method for creating a liquid level damping cam curve for creating a liquid level.

In a filling apparatus that fills beverages and medicines, it is necessary to transport the filling liquid without overflowing the filled container.
As a conventional cam curve in intermittent conveyance, a modified sine curve, a universal cam curve, a B-spline curve, etc. are used. Even if each cam curve is used, the liquid overflows or does not overflow depending on the conditions. However, the shaking of the liquid level may become large.
Conventionally, a creation method has been proposed in which a cam curve capable of obtaining a smooth motion can be easily created (Patent Document 1).
Conventionally, genetic algorithms are already known, and a freight logistics cost optimization method using genetic algorithms has been proposed as an example (Patent Document 2).

Japanese Patent No. 3726880 JP 2009-223552 A

When using the above-mentioned modified sine curve, universal cam curve, B-spline curve, etc. as the liquid level damping cam curve, the optimal cam curve is selected by examining the fluctuation of the liquid level for each cam curve. The operation was complicated.
According to the above-mentioned patent document 1, it may be possible to facilitate the work when creating each cam curve, but which cam curve is most used even if this patent document 1 is used. It was not possible to immediately determine whether a favorable result was obtained.
In view of such circumstances, the present invention provides a method for creating a liquid level damping cam curve that can easily create a suitable liquid level damping cam curve from a plurality of cam curves by using a genetic algorithm. Is to provide.

That is, the invention of claim 1 is a method of creating a liquid level damping cam curve for obtaining a cam curve so that the fluctuation of the liquid level stored in the container satisfies a required criterion when the container is transported. And
When a plurality of types of initial cam curves are selected in advance, “individual evaluation” is performed in which the liquid surface state of each initial cam curve is simulated by CFD simulation, and each initial cam curve is ranked based on the result.
Next, a plurality of types of initial cam curves with good results are selected, and “parent individual selection” is performed in which two different initial cam curves are paired to select a plurality of first generation parent cam curves,
Next, “crossover” in genetic operation is performed for each group to generate a plurality of types of first generation child cam curves,
Next, for each of the multiple types of first-generation child cam curves, when the liquid surface state of each child cam curve is simulated by CFD simulation, “individual evaluation” is performed to rank each child cam curve based on the result. And
Next, a plurality of types of first-generation child cam curves with good results are selected, and each of them is used as a second-generation parent cam curve, and a plurality of sets of two different second-generation parent cam curves are paired. Execute `` Parent individual selection '' to select the parent cam curve,
Next, “crossover” in genetic operation is performed for each group to generate a plurality of types of second generation child cam curves,
In the same manner, the loop consisting of “individual evaluation”, “parent individual selection”, and “crossover” is repeated to select the parent cam curve with the best results as the liquid level damping cam curve. A method for creating a swing cam curve is provided.
Further, in the invention of claim 2, in the invention of claim 1, when “crossover” in the genetic operation is executed to generate a plurality of types of child cam curves, the “individual evaluation” is performed on these child cam curves. Of the liquid surface damping cam curve, wherein a plurality of types of child cam curves are generated by executing “mutation” in genetic operation at a predetermined rate during the repetition of the above loop. It provides a creation method.

According to the first aspect of the present invention, it is possible to rank the results for each initial cam curve by simply selecting a plurality of types of initial cam curves in advance, as well as “crossover” of genetic operations for these initial cam curves. Can generate a new cam curve and rank the new cam curve as well, thereby creating the cam curve with the best results while generating many cam curves. Can be selected as a curve.
Therefore, the liquid level damping cam curve can be selected very easily compared to the conventional one, and the liquid level damping cam curve can be selected from a number of cam curves. A cam curve with high vibration capability can be obtained.
Moreover, in the apparatus employing the cam curve selected according to the present invention, the tact time can be shortened and the production efficiency can be improved.
According to the second aspect of the present invention, the “mutation” in the genetic operation is executed at a predetermined rate while the loop is repeated, so that a cam curve with higher liquid surface damping capability is obtained. The possibility can be increased.

The schematic plan view which shows the filling apparatus with which the liquid level damping cam curve based on this invention is used. The step figure explaining the creation method of the liquid level damping cam curve concerning the present invention.

Hereinafter, the present invention will be described with reference to the illustrated embodiment. FIG. 1 is a schematic plan view of a filling device 1 that is intermittently driven to rotate according to a preset liquid level damping cam curve. A turntable 3 having pockets 3a for accommodating containers 2 at equidistant positions on the outer periphery is provided.
In the illustrated embodiment, 18 pockets 3a are provided at equal intervals, and the rotary table 3 has a pitch of 1 between the adjacent pockets 3a and 3a according to an electronic cam curve preset by a servo motor (not shown). It is designed to rotate intermittently counterclockwise.

The supply conveyor 4 for supplying the containers 2 to the pockets 3 a is arranged in the diameter direction of the rotary table 3, and the first container 2 among the containers conveyed by the supply conveyor 4 is the pocket 3 a of the rotary table 3. It is hold | maintained in the position contact | abutted to outer peripheral surfaces other than.
And while the turntable 3 advances one pitch, one container 2 is accommodated in one pocket 3a, and the container 2 is pocketed by a circular fixed guide 5 arranged along the outer peripheral surface of the turntable 3. In a state where the dropout from the inside of 3a is prevented, the rotary table 3 is moved forward.

The filling mechanism 7 for filling the filling liquid into the container 2 includes three filling nozzles 7a, and each nozzle 7a is inserted into each of the adjacent three containers 2 that are intermittently stopped. It has become.
The filling mechanism 7 is configured to intermittently advance the three filling nozzles 7a in synchronism with the operation of the container 2 that is intermittently advanced, while each container 2 is advanced three pitches. The filling of the filling liquid into each container is finished.
The nozzles 7a that have finished filling the filling liquid into the containers 2 are quickly returned to be inserted into the subsequent three new containers 2, respectively.

A capper 8 is provided on the downstream side of the filling mechanism 7, and the capper 8 can cap a cap (not shown) on the container 2 stopped intermittently.
The container 2 with the cap capped is moved onto the discharge conveyor 9 arranged adjacent to the supply conveyor 4 by intermittent advancement, and is carried out of the filling device 1 by the discharge conveyor 9 to the outside. Become.

By the way, the liquid level of the filling liquid in the container 2 in which the filling of the filling liquid is completed and the cap is not capped is 1 pitch from the intermittently stopped position state until it is intermittently stopped again. During the cycle, it will be shaken by the acceleration and deceleration of its advance, and depending on the conditions, the filling liquid may overflow.
For this reason, the turntable 3 needs to be advanced intermittently according to a suitable both-stop cam curve that does not allow the filling liquid to overflow from the container 2 and that the liquid level is small and smooth.

Hereinafter, a method of creating an electronic cam curve of a servo motor that intermittently rotates the rotary table 3 by one pitch will be described.
FIG. 2 is a step diagram for explaining a method for creating a liquid level damping cam curve according to the present invention. In this embodiment, an electronic cam curve is created for a servo motor that intermittently rotates the rotary table 3 as described above. Be able to.
In creating the electronic cam curve, first, as the “initial individual population” 11, 16 types of conventionally known stationary cam curves are selected as initial cam curves. The number of initial cam curves is not limited to 16, but a plurality of types of initial cam curves that are considered suitable may be selected.
Examples of the 16 types of stationary cam curves include a quintic curve, an isokinetic curve, a cycloid curve, a synthetic sine curve, a synthetic sine curve F-5, a modified trapezoidal curve, a modified sine curve, a modified isokinetic curve, and Kubota- A parabolic sine curve, a Jimbo-exponential curve, an asymmetric quartic curve, an asymmetric cubic curve, an asymmetric cubic curve, an asymmetric cycloid curve, an asymmetric deformed trapezoid curve, and a trapecoid curve can be selected.
As an example, the quintic curve can be represented by a function of S (T) = 6T ⁵ -15T ⁴ + 10T ³ (T = time). Since the other cam curves are already known in the prior art, the description of each specific function is omitted.

When the 16 types of initial cam curves are selected, a loop based on a genetic algorithm is repeated as described below.
That is, first, “individual evaluation” 12 is performed for each of the 16 types of initial cam curves selected as the “initial individual population” 11.
In this “individual evaluation” 12, the state of the liquid level is simulated by CFD simulation, whether or not the simulated result satisfies a predetermined determination criterion, and the ranking of the 16 types of initial cam curves. The attachment is done.
The CFD (Computational Fluid Dynamics) simulation is a simulation method that realizes thermofluid analysis by numerically solving a partial differential equation related to thermal fluid motion such as the Navier-Stokes equation using a computer. . By using this CFD simulator, an accurate simulation of the fluid behavior phenomenon is performed. For example, even if the filling apparatus 1 is intermittently operated at a container processing number of 40 per minute, the filling liquid does not overflow. It is possible to derive a cam curve that can be conveyed smoothly. Such fluid analysis software is commercially available, for example, as FLOW-3D (trade name) manufactured by Flow Science, and further description thereof is omitted.
In addition, as the determination criterion, the maximum liquid level height E1 generated during one cycle from the intermittently stopped position state by one pitch to the intermittent stop, and the maximum liquid due to residual vibration after the stop. The surface height E2 is set, and in “individual evaluation” 12, it is determined whether or not each value is within a predetermined value.

When the 16 individual initial cam curves are ranked according to the “individual evaluation” 12, the operation “parent individual selection” 13 is performed.
In this “parent individual selection” 13, a predetermined number of initial cam curves with poor results, for example, eight initial cam curves which are half of the above 16 types are discarded, and the remaining eight types of initial cam curves with good results are the first. The first generation parent cam curve is selected.
For example, 28 types of cam curves are selected from the selected 8 types of parent cam curves.
At this time, the same combination is excluded and a pair of different cam curves is selected. Further, the eight kinds of parent cam curves are considered to be incorporated at least once in the above 28 kinds of pairs.

When 28 kinds of parent cam curves are selected in this way, next, “crossover” 14 as a genetic operation is performed for each pair of groups, and thereby 28 kinds of child cam curves S′c (T ) Is created.
The “crossover” as the genetic operation is performed using the following formulas 1 and 2.
S′c (T) = fc (T) * S _p1 (T) + (1−fc (T)) * S _p2 (T) Equation 1
fc (T) = (R _c1 +0.5) T + (R _c2 +0.5) (1-T)...
S _p1 (T) represents a function of one parent cam curve of each pair, and S _p2 (T) represents a function of the other parent cam curve.
R _c1 and R _c2 are normal random numbers.
By the “crossover” as the genetic operation described above, each of the 28 kinds of child cam curves S′c (T) is collected together with the above-mentioned eight first generation parent cam curves having good ranks, A total of 36 types of first generation child cam curves are created as the “generation child individual group” 15.

When 36 types of first generation child cam curves are obtained by the genetic algorithm in the first loop in this way, “individual evaluation” 12 is performed in the second loop for each child cam curve.
Also in the “individual evaluation” 12 in the second loop, the liquid level state is simulated by CFD simulation in the same manner as described above, and whether or not the simulated result satisfies a predetermined determination criterion. And the 36 kinds of first generation child cam curves are ranked.

When the 36 individual first generation child cam curves are ranked by the “individual evaluation” 12 in the second loop, a predetermined number of cam curves having poor rankings, for example, half, are selected by the “parent individual selection” 13. The remaining 18 types of cam curves are discarded, and the remaining 18 types of cam curves with good results are randomly selected as two pairs, and 18 sets of second generation parent cam curves are newly selected.
In this case as well, the same combination is excluded and different cam curve pairs are selected as in the previous case. Further, the remaining 18 types of cam curves having good ranks are considered to be incorporated in one of the 18 pairs.

  Thereafter, in the same manner as in the first loop described above, each of the 18 sets of second generation parent cam curves selected is subjected to 18 types of second generation children by “crossover” 13 as the genetic operation described above. A cam curve S′c (T) is created. Then, a total of 36 types of second-generation child cam curves are selected by combining the 18 types of second-generation parent cam curves with good results described above.

When 36 second-generation child cam curves are obtained by the second genetic algorithm in this way, “individual evaluation” 12 is performed in the third loop for each child cam curve.
Thereafter, a total of 36 types of third-generation parent cam curves were created in the same manner as in the second loop described above, and the third-generation child cam curves obtained based on the third-generation parent cam curves were created. In the fourth loop, “individual evaluation” 12 is performed.
Then, after a predetermined number of times, for example, 100 loops are repeated and the final “individual evaluation” 12 is performed, the parent cam curve with the best result is selected as the liquid level damping cam curve. .

By the way, in this embodiment, after executing the “crossover” 14 in the genetic operation with a predetermined probability of repeating the loop, for example, 3%, the “mutation” 16 in the genetic operation is executed. It is.
That is, when “crossover” 14 as a genetic operation in a number of loops is performed and a plurality of child cam curves S′c (T) are created, “individual evaluation” 12 described above is executed for each child cam curve. Before performing the “mutation” 16 as a genetic operation, a plurality of child cam curves Sc (T) are created.
The “mutation” as the genetic operation is performed using the following Equation 3.
Sc (T) = Rm * Si (T) + (1−Rm) * S′c (T) Equation 3
However, Si (T) is one parent cam curve selected at random from the parent cam curves with good results, and Rm is a normal random number. Note that when selecting the Si (T), it may be selected at random from the 16 types of initial cam curves described above.
Of course, “individual evaluation” 12 is executed in the next loop for a plurality of child cam curves Sc (T) obtained by the above-described “mutation” operation.
If the “mutation” 16 in the genetic operation is executed with a predetermined probability of repeating the loop as described above, it becomes possible to greatly change the properties of the cam curve, thereby improving the liquid level vibration control. The possibility of obtaining a cam curve with capacity can be increased.

In the above embodiment, the loop is repeated a predetermined number of times, and the parent cam curve with the best result is finally selected as the liquid level damping cam curve. When a satisfying cam curve is obtained, the operation may be stopped at that time and the cam curve may be selected as the liquid level damping cam curve.
Moreover, although the said Example obtains the cam curve about the filling apparatus 1 rotationally driven intermittently, it is not limited to this, For example, in order to obtain the cam curve used for a linear conveyance apparatus, it is this invention. Of course, can be applied.

12 Individual evaluation 13 Parent individual selection 14 Crossover 16 Mutation

Claims (6)

A method of creating a liquid level damping cam curve for obtaining a cam curve such that the fluctuation of the liquid level contained in the container when the container is conveyed satisfies a required criterion,
When a plurality of types of initial cam curves are selected in advance, “individual evaluation” is performed in which the liquid surface state of each initial cam curve is simulated by CFD simulation, and each initial cam curve is ranked based on the result.
Next, a plurality of types of initial cam curves with good results are selected, and “parent individual selection” is performed in which two different initial cam curves are paired to select a plurality of first generation parent cam curves,
Next, “crossover” in genetic operation is performed for each group to generate a plurality of types of first generation child cam curves,
Next, for each of the multiple types of first-generation child cam curves, when the liquid surface state of each child cam curve is simulated by CFD simulation, “individual evaluation” is performed to rank each child cam curve based on the result. And
Next, a plurality of types of first-generation child cam curves with good results are selected, and each of them is used as a second-generation parent cam curve, and a plurality of sets of two different second-generation parent cam curves are paired. Execute `` Parent individual selection '' to select the parent cam curve,
Next, “crossover” in genetic operation is performed for each group to generate a plurality of types of second generation child cam curves,
In the same manner, the loop consisting of “individual evaluation”, “parent individual selection”, and “crossover” is repeated to select the parent cam curve with the best results as the liquid level damping cam curve. How to create a swing cam curve.   After executing “crossover” in the genetic operation to generate a plurality of types of child cam curves, before executing the “individual evaluation” for these child cam curves, the genetic loop is repeated with a predetermined probability of repeating the loop. 2. The method for creating a liquid level damping cam curve according to claim 1, wherein a plurality of types of child cam curves are generated by executing "mutation" in a manual operation. The method of creating a liquid level damping cam curve according to claim 1 or 2, wherein the "crossover" as the genetic operation is executed using the following formulas 1 and 2.
S′c (T) = fc (T) * S _p1 (T) + (1−fc (T)) * S _p2 (T) Equation 1
fc (T) = (R _c1 +0.5) T + (R _c2 +0.5) (1-T)...
Where S′c (T) is a function of a child cam curve, S _p1 (T) is a function of one parent cam curve of each pair, S _p2 (T) is a function of the other parent cam curve, R _c1 , R Each _c2 is a normal random number. The method for creating a liquid level damping cam curve according to claim 2, wherein the "mutation" as the genetic operation is executed using the following equation (3).
Sc (T) = Rm * Si (T) + (1−Rm) * S′c (T) Equation 3
However, Si (T) is a function of one parent cam curve selected at random from the parent cam curves used for the child cam curve, and Rm is a normal random number.   5. The loop according to any one of claims 1 to 4, wherein the loop is repeated a predetermined number of times, and the parent cam curve having the best result is selected as the liquid level damping cam curve. To create a liquid level damping cam curve.   5. If the parent cam curve satisfying the determination criterion is obtained, the loop is interrupted and the parent cam curve is selected as a liquid level damping cam curve. Method for creating a liquid level damping cam curve as described in 1.

Images