JPH04201450A - Controller for multicolor press - Google Patents

Abstract

PURPOSE:To make it possible to keep a printing state in a multicolor press optimum all the time without sampling when simulation data are adjusted so as to coincide with reference data by inputting setting signals to the multicolor press and producing the simulation data corresponding to actual printing quality evaluation data to be obtained from a printed matter. CONSTITUTION:A PSS 4 supplies setting signals A of colors to a controller 6. Accordingly, the controller 6 converts the setting signals A of the colors into operation signals C which are supplied to a multicolor press 1 as setting signals D. On the other hand, the operation signals C from the controller 6 are supplied also to a simulator 7 as setting signals E simultaneously. The setting signals E, environment signals, and speed signals are inputted to the simulator 7 which generates simulator data F in accordance with the hierarchical structure. A comparator 8 compares print quality evaluation data G with simulation data F and its deviation data H is supplied to the simulator 7 as a coupling degree correction command.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a control device for a multicolor printing press that can always keep the printing state of the multicolor printing press in an optimal state.

[Conventional technology]

Conventionally, with multicolor printing machines, it has been difficult to quantitatively evaluate the quality of printed matter, and operators have had to adjust settings such as blade opening amount, ink feed amount, water feed amount, and printing pressure based on visual judgment. , the printing condition was to be adjusted to the optimum condition.

In recent years, inspection marks are printed in strips outside the image area of printed matter, and various print quality evaluation data such as density, dot gain, trapping, color balance, etc. are taken from these inspection marks, and based on these print quality evaluation data. Adjustment of setting values, that is, adjustment of setting signals, is beginning to be carried out to bring the printing state to an optimum state.

[Problem to be solved by the invention]

However, in such control methods in recent years, each print quality evaluation data is obtained only when a print sample is taken, and in other words, setting signals are adjusted only when a print sample is taken. However, it could not be said that the printing condition was always maintained at the optimum condition.

More specifically, there is a time lag between the timing when a print sample is taken and the adjustment timing of the setting signal, and as a result, the adjustment of the setting signal may be over or under.

Furthermore, the correlation between each print quality evaluation data could not be quantitatively grasped, and the data used for adjusting the setting signal was mainly density data, and other data were treated as auxiliary data.

[Means to solve the problem]

The present invention has been proposed to solve such problems, and the first invention (invention according to claim 1) is to input setting signals to a multicolor printing machine and to control the actual printed matter to be printed. The apparatus includes a simulation means for generating each simulation data corresponding to each print quality evaluation data.

Further, the second invention (invention according to claim 2) is that in the first invention, the simulation means is of a large force, multi-output, and variable structure type, and each generated simulation data matches each actual print quality evaluation data. In order to do so, the structure is automatically corrected.

Further, the third invention (invention according to claim 3) is the first invention or the second invention, in which setting signals to the multicolor printing press are adjusted based on each simulation data generated by the simulation means, and print quality is improved. It is equipped with a control means for continuously controlling the

Further, the fourth invention (invention according to claim 4) is the third invention, wherein the control means is of a large force/multiple output/structure variable type, and each simulation data generated by the simulation means is controlled in advance. The structure is automatically corrected to match each reference data set.

[Effect]

Therefore, in the first invention, if each simulated data is varied according to variations in the printing condition in a multicolor printing machine, each simulated data that fluctuates in the same way as each actual print quality evaluation data can be continuously generated. And it can be obtained in real time.

Furthermore, in the second invention, in the first invention, each of the simulated data always matches each of the actual print quality evaluation data.

In addition, in the third invention, in the first invention or the second invention, based on each simulation data (by appropriately adjusting setting signals to the multicolor printing machine, the printing state is optimized continuously and in real time). state.

In addition, in the fourth invention, in the third invention, the multicolor printing machine is controlled so that each simulated data matches each reference data, that is, each actual print quality evaluation data matches each reference data. The configuration signal is adjusted.

〔Example〕

Hereinafter, a control device for a multicolor printing press according to the present invention will be explained in detail.

FIG. 1 is a conceptual diagram showing an embodiment of a control device for this multicolor printing press. In the figure, 1 is a multicolor printing machine, 2 is a printed matter printed by this multicolor printing machine 1, 2-1 is an inspection mark printed in a strip shape outside the image area of this printed matter 2, and 3 is an inspection mark. A detector detects each print quality evaluation data (density, dot gain, trapping, color balance, etc.) from the printing mark 2-1 and converts it into an electrical signal. Approximate setting signals for each color (blade opening amount, ink feed amount).

A pattern area ratio measurement device (
(hereinafter referred to as PSS), 6 is a large-power, multi-output, variable structure type that receives each setting signal output from PSS4 and environmental signals such as temperature and humidity as input, and adjusts and outputs each setting signal according to the hierarchical structure. The control device 7 inputs each setting signal (control signal) and environment signal from the control device 6, as well as a speed signal indicating the printing speed in the multicolor printing press 1, and according to the hierarchical structure, prints the inspection marks 2-1 to 2-1. A large-power, multi-output, structurally variable simulator that generates each simulated data corresponding to each detected print quality evaluation data; 8 represents each print quality evaluation data detected by the detector 3 and each simulation generated by the simulator 7; A first comparator 9 compares each simulation data generated by the simulator and provides the deviation data to the simulator 7 as a coupling degree modification command.
This is a second comparator that compares the reference data obtained from K receipt 0 and supplies the deviation data to the control device 6 as a coupling degree correction command.

The hierarchical structure in the control device 6 and the simulator is as follows:
This is a so-called neural network, and in this embodiment, a three-layer neural network is used. Since the characteristics of the multicolor printing press 1 are generally fixed, the hierarchical structure in the simulator 7 has approximately known initial values, and is initially set by giving these initial values.

For more information on neural networks, please see "Introduction and Practical Neurocomputer: Published by Gijutsu Hyoronsha,"
I will briefly explain its basic operation here, leaving it to Shigeru Kiritani's author. FIG. 2 shows a model of a three-layer neural network, with various inputs entering the input layer 100. Each element 100-1 constituting the input layer 100 has an appropriate coupling degree (
Each element 200-1 constituting the intermediate layer 200 is connected to the output layer 300 in the same way.
It is connected to each element 5oo-i constituting the circuit with an appropriate degree of coupling (coupling coefficient). These degrees of coupling are variable and can take any value between -1 and l. input layer 100
The output from the output layer 300 obtained in that state should be made to match the training data (by modifying the degree of coupling between the input layer 100, the middle layer 200, and the output layer 300, that is, pack By modifying the hierarchical structure while learning using the propagation method, the output from the output layer 300 can be brought closer to the teacher data.

There are many combinations of combination methods that bring this output closer to the training data, and even if all the data matches the training data, it cannot necessarily be said that the structure of the printing machine is simulated.

Therefore, learning using a large amount of sample data is necessary. In the present invention, by selecting 30 types of printed matter and learning about 30 samples for each printed matter, for a total of 900 samples, using the pack propagation method,
It was possible to create a structure that was always the same as the input and output of a printing machine.

Next, the operation of the control device of this multicolor printing press will be explained.

First, the PS 84 gives each setting signal (A) for each color to the control device 6. As a result, the control device 6 converts each setting signal (A) for each color into each operation signal (C), and uses each operation signal (C) as each setting signal (D) to control the multicolor printing machine 1.
give to 1 for multicolor printing machine 1, each setting signal (D)
Based on the information, the blade opening amount, ink feed amount,
Settings such as the water feed location are made. Then, the printed matter 2 is printed in the set printing state.

On the other hand, each operation signal (C) from the control device 6 is simultaneously given to the simulator 7 as each setting signal (E).

The simulator 7 receives the setting signal (E), the environment signal, and the speed signal as input, and generates each simulated data (F) according to the hierarchical structure. Each simulated data (F) generated by this simulator 7 is given to a comparator 8.

Here, each print quality evaluation data is detected from the inspection mark 2-1 printed on the printed material 2 using the detector 3.

Each print quality evaluation data detected by detector 3 (G)
is applied to comparator 8. The comparator 8 compares each print quality evaluation data (G) with each simulation data (F), and uses the deviation data (H) as a coupling degree correction command to the simulator 7.
Donate to. The simulator 7 uses this coupling degree modification command (
According to H), make each simulated data (F) match each print quality evaluation data (G) as teacher data (change the degree of connection between each layer and modify the hierarchical structure.

Simulator 7 has been sufficiently trained as described above, and each print quality evaluation data (G) and each simulated data (F) show the same values, but due to slight deviations in conditions, they show slight differences. Sometimes. The correction here refers to correcting this slight deviation, and is more similar to calibration than learning, and learning is completed in a short period of time.

By the above-mentioned modification, each simulated data (F) from the simulator 7 can be accurately brought close to each print quality evaluation data that will actually be obtained, and the input/output characteristics of the simulator 7 can be adjusted to the multicolor printing machine 1. The input/output characteristics can be made close to that of .

Therefore, after the above-described learning is completed, when the printing condition of the multicolor printing press 1 changes due to a change in the environment (temperature, humidity, etc.) or print speed, the simulator 7 Each simulated data from .

That is, after the above-described learning is completed, the simulator 7 can continuously and in real time obtain simulated data that fluctuates equally to each actual print quality evaluation data.

Next, learning by the simulator is interrupted and learning by the control device 6 is started. The control device f16 inputs each setting signal (A) from the PSS4 and the environment signal, adjusts the setting signal (A) according to the hierarchical structure, and outputs the control signal (C).
Send out. This control signal (C) is given to the multicolor printing machine 1 as a setting signal (D), and the simulator 7
is given as a setting signal (E) to The simulator 7 receives the setting signal (E), the environment signal, and the speed signal as input, and generates each simulated data (F) according to the hierarchical structure. Each simulated data (F) generated by this simulator is given to the comparator 9 as each simulated data (1). The comparator 9 compares each reference data (J) and each simulated data (
1) and provides the deviation data (K) to the control device 6 as a coupling degree modification command. In response to this coupling degree correction command (K), the control device 6 makes each simulated data (I) match each reference data (J) as teacher data.
Modify the hierarchical structure by changing the degree of connectivity between each layer.

That is, by constantly modifying the hierarchical structure while learning by the backpropagation method in the control device 6, each simulated data (1) matches each reference data (J) continuously and in real time. This results in
Each actual print quality evaluation data matches each standard data (J) continuously and in real time, and the printing condition of the multicolor printing press 1 can always be kept in the optimum condition without taking print samples. It becomes like this.

This control device 6 also requires initial learning. Each setting signal (A) of PSS4 to each setting signal (D) to multicolor printing machine 1
The characteristics up to this point are well known, and the backpropagation method is used to sufficiently learn them based on these inputs and outputs.

Furthermore, according to this embodiment, each piece of actual print quality evaluation data matches each standard data at the same time, and data other than density data, which was conventionally treated as supplementary data, can be effectively utilized. .

In addition, in this embodiment, an environmental signal is sent to the control device 6.
Although the environment signal and the speed signal are given to the simulator 7 as variable elements, they may not be given. In other words, if the printing time on the multicolor printing press 1 is short, the hierarchical structure of the simulator 7 and the control device 6 may be modified as an initial setting before the start of operation, and the simulated data generated by the simulator 7 may be modified. After the simulated data matches the reference data, the multicolor printing press 1 may perform printing without changing the simulated data.

In other words, if the printing time on the multicolor printing machine 1 is short, it is assumed that there are no major changes in the environment or speed.
It is possible to deal with this by simply making the simulated data match the reference data in the initial settings before starting operation. In this case, it is possible to take samples occasionally and
If the hierarchical structure in the control device 6 is modified, it becomes possible to maintain the printing state in the multicolor printing press 1 in an optimal state while occasionally modifying it.

Furthermore, the accuracy of the simulator 7 may deteriorate due to insufficient learning of the simulator. In this case, control by the control device 6 may be temporarily suspended and the hierarchical structure in the simulator 7 may be corrected by learning.

〔Effect of the invention〕

As is clear from the above explanation, according to the present invention,
By varying each simulated data in response to changes in the printing conditions on a multicolor printing machine, it becomes possible to continuously obtain simulated data that fluctuates in the same way as the actual print quality evaluation data in real time. By adjusting each simulated data to match each reference data, it becomes possible to always maintain the printing condition of the multicolor printing machine in the optimum condition without taking print samples.

Further, according to the present invention, it is possible to match each actual print quality evaluation data with each standard data at the same time, and it is possible to effectively utilize data other than density data, which has conventionally been treated as supplementary data. It becomes possible.

Further, according to the present invention, it becomes possible to quantitatively manage printing quality, which conventionally relied on the operator's experience, continuously and in real time, and even an operator with little experience can obtain stable printing quality.

[Brief explanation of the drawing]

FIG. 1 is a conceptual diagram showing an embodiment of a control device for a multicolor printing press according to the present invention, and FIG. 2 is a diagram showing a model of a three-layer neural network. l...Multicolor printing machine, 2...Printed matter, 2-1...Inspection mark, 3...Detector, 4...Picture area ratio measuring device (PSS), 6...Control Apparatus, 7...Simulator, 8...First comparator, 9.Φ.Second comparator.

Claims (4)

[Claims] (1) Control of a multicolor printing press, which is equipped with a simulation means that inputs a setting signal to the multicolor printing press and generates each simulation data corresponding to each actual print quality evaluation data obtained from printed matter. Device. (2) In claim 1, the simulating means is multi-input, multi-output, and variable in structure, and the structure is automatically corrected so that each simulated data to be generated matches each actual print quality evaluation data. A control device for a multicolor printing press featuring: (3) Multicolor printing according to claim 1 or 2, comprising control means for continuously controlling print quality by adjusting setting signals to the multicolor printing machine based on each simulation data generated by the simulation means. Machine control device. (4) In claim 3, the control means is multi-input, multi-output, and variable in structure, and its structure is automatically corrected so that each simulated data generated by the simulated means conforms to each predetermined reference data. A control device for a multicolor printing press.