JPH08137146A - Image forming device - Google Patents

Abstract

PURPOSE: To reduce the cost and further to drastically decrease the man-hours for development by decreasing the number of sensors as small as possible. CONSTITUTION: A solid development patch and a highlight development patch are first formed as initial setting and are respectively measured by a development density sensor 10. The contents thereof are stored as a control case in a control case memory 25. Similarly, the control cases for two times are stored in the control case memory 25 while the scorotron set values and the LP set values are respectively changed. These control cases are supplied via a state quantity comparator 26 and a cluster memory 27 to a control loop calculator 28 where control rules are determined. Henceforth, the determined control rules are properly combined to obtain the fresh manipulated variable, i.e., the scorotron set values and LP set values.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image forming apparatus using an electrophotographic system, and in particular, control for always maintaining an image at a predetermined quality can be performed at low cost and with high precision, and further data for product development can be obtained. The present invention relates to an image forming apparatus capable of significantly reducing development man-hours related to collection and optimization design.

[0002]

2. Description of the Related Art Conventionally, in an image forming apparatus using an electrophotographic system, feedback control for keeping an optimum image density is generally used. This is because in the electrophotographic system using static electricity, the image output state of the device itself changes due to environmental conditions such as temperature and humidity of the day, or deterioration over time of the photoreceptor and developer, and the density reproducibility changes. The reason is that

The contents of the conventional feedback control will be described more specifically. The density reproduction situation is monitored by the density patch to obtain an error amount from the target density, and this is multiplied by the feedback gain to set the control actuator. The most general method is to calculate the value correction amount.

Here, the density patch is often a developed image patch. This is because the developed image is easier to create and erase than the transferred image and the fixed image created on the sheet, and the correlation with the fixed image density that the user holds is extremely high. Further, as the control actuator, a charging device applied voltage, an exposure amount, a developing bias, etc. that influence the developing characteristics are often used.

For example, in the techniques described in JP-A-63-177176, 63-177177 and 63-177178, the developing potential is varied to control the developing density to a desired value. . The method of changing the developing potential is valid for both the one-component and two-component developing methods.

However, the optimum developing potential is always affected by various uncontrollable external factors such as temperature, humidity, and the number of accumulated copies. The charging potential, the exposure amount, and the developing bias are set by these factors. It is difficult to always consider the conditions of. Moreover, the relationship between the quantity of state such as temperature and humidity and the amount of charging, the amount of exposure, and the setting value of the developing bias is complicated, and sufficient physical modeling has not been made at the current technical level.

Therefore, although quantified control using an approximate expression is performed, in electrophotographic technology, which is mainly an electrostatic process, usually, the optimum setting of charging, exposure dose, and bias with respect to state quantity is performed. Since the relationship between values is non-linear, sufficient control accuracy has not been obtained. Due to these circumstances, it is necessary to understand in advance various environmental conditions, such as the environmental effects under high temperature and high humidity conditions and low temperature and low humidity conditions, and the effects of deterioration over time. Since it is necessary to collect detailed data over the condition range, enormous development man-hours were required.

Moreover, the feedback gain determined by investing such a huge number of man-hours has not always been optimal because of the difference in each machine and the usage conditions of various users. In particular, the effect of deterioration over time on image density varies greatly depending on the degree of deterioration of the parts used for each device and the usage by the user, so long-term image density control performance after market launch is not always guaranteed. I couldn't say it was perfect.

Since the control method is as described above, a potential sensor for monitoring the charging potential and the exposure potential, which are intermediate parameters for obtaining the control accuracy, and a temperature for monitoring the environmental conditions. Many control methods require a sensor or a humidity sensor, which causes a problem of cost increase.

Further, recently, Japanese Patent Laid-Open No. 4-3199
As disclosed in Japanese Patent No. 71, 4-320278, etc., methods using fuzzy and neural networks have come to be used. These are mainly used as means for improving the control accuracy by utilizing the feature that fuzzy and neural networks can cope with the case where the relationship between the input and the output is complicated and nonlinear. For this reason, the above-mentioned problems, that is, the enormous development man-hours that must be invested in collecting a large amount of data, the cost increase due to the heavy use of sensors, and the long-term image of each device after being put on the market. It is hardly useful for solving problems such as not always ensuring the concentration control performance.

On the contrary, when the control accuracy is improved by using fuzzy or neural network, in order to take advantage of the feature that it is suitable for multi-input multi-output calculation, multi-input, that is, many sensors are often used. On the contrary, the cost is increasing.

Further, in fuzzy, it is necessary for an engineer to tune the membership function, and in neuro, the learning work itself can be automated, but the engineer must prepare in advance the training data for that, so that considerable development man-hours are required. It was the actual situation that needed.

Moreover, even when fuzzy or neural network is used in which time-dependent deterioration data is collected in advance and the fuzzy or neural network is used, the relationship between the input and the output itself is the actual time-dependent deterioration, machine difference, and parts replacement. However, there is a problem in that it is not possible to respond autonomously when it changes due to factors such as. That is, the long-term image density control performance of each device on the market cannot be guaranteed even if fuzzy or neural networks are used.

[0014]

SUMMARY OF THE INVENTION The present invention has been made in order to eliminate the above-mentioned drawbacks of the prior art. The purpose of the present invention is to reduce the number of sensors as much as possible, thereby reducing the cost. Especially. For example, when controlling the image density, there is a need to realize a cost reduction by eliminating the need for a potential sensor, a temperature sensor, a humidity sensor, etc. that have been conventionally used in addition to the image waste sensor.

Another object is to significantly reduce the development man-hours by enabling the engineer to automatically perform accurate control without knowing beforehand the influence of various environmental conditions and deterioration over time. To do.

Still another object is that even if a huge number of image forming apparatuses are put on the market and each of them is used in various ways and parts are replaced at any time, one image at a time It is to provide a technology capable of always automatically ensuring the concentration control performance.

Still another object is to make it possible to directly set the required control accuracy itself to the control device, and the control device itself is automatically adapted to meet the required accuracy. By doing so, it is possible to eliminate the cost increase and the development man-hour increase for accuracy improvement.

Still another object is to realize the above object with a limited memory capacity.

Still another object is to make it possible to realize optimum control more suitable for the current situation.

Still another object is to realize optimal control suitable for the current situation even if a clear abnormality such as noise occurs.

[0021]

In order to solve the above-mentioned problems, in the invention described in claim 1, an image quality changing means for changing the quality of an output image according to an operation amount, and an output image control example. And a control rule extraction unit that extracts a control rule from a plurality of control cases stored in the control case storage unit, and a detection that detects the output image quality and outputs the detection result as a control amount. Means and
Using the control rule extracted by the control rule extracting means, an operation amount calculating means for obtaining a new operation amount so that the control amount becomes a value corresponding to the target quality,
The operation amount calculated by the operation amount calculating means is supplied to the image quality varying means.

Further, in the invention described in claim 2,
An image quality varying means for changing the quality of the output image according to the command value, a cluster storage means for collecting similar ones of the output image control cases and storing them as a cluster, and the cluster storage means. The cluster-specific control rule extracting means for extracting the control rule for each cluster, the detecting means for detecting the output image quality and outputting it as a control amount, and the control rules extracted by the cluster-specific control rule extracting means Control amount calculating means for obtaining a new control amount so that the control amount becomes a value corresponding to the target quality, and the operation amount obtained by the control amount outputting means is supplied to the image quality varying means. It is characterized by doing.

Further, in the invention described in claim 3,
3. The image forming apparatus according to claim 2, wherein the control amount calculation means determines the suitability of each control rule extracted by the cluster-specific rule extraction means with respect to the immediately preceding control case, and determines the suitability of each control rule according to the suitability. It is characterized in that a new control amount is obtained by using the results obtained by averaging after weighting.

Further, in the invention according to claim 4,
The image forming apparatus according to claim 3, wherein the control amount calculation means is in a coordinate space in which each control rule is described,
The reciprocal of the distance between the n-dimensional plane indicating the control rule and the coordinate point indicating the immediately preceding control case is standardized for each control rule to obtain the conformity.

Further, in the invention described in claim 5 or 6, in the image forming apparatus according to claim 3 or 4, the control amount calculation means excludes a control rule whose conformity is equal to or less than a predetermined value. , And the degree of conformity is determined again with respect to the other control rules, and the result of averaging by performing weighting according to the degree of conformity with respect to these control rules is used.

Further, in the invention described in claim 7,
The image forming apparatus according to claim 1, wherein the control amount is compared with the target quality, and only when the comparison result exceeds a preset allowable value, the control amount is stored in the control case storage means. It is characterized in that it is provided with a comparison means to be used for the control after the next time.

Further, in the invention described in claim 8,
3. The image forming apparatus according to claim 2, wherein the control amount is compared with the target quality, and only when the comparison result exceeds a preset allowable value, the control amount is corresponded in the cluster storage means. It is characterized by comprising a comparison means for additionally storing the data in the cluster so that it can be used for the control on and after the next time.

Further, in the invention according to claim 9,
The image forming apparatus according to claim 7, wherein when the storage capacity is less than a predetermined amount as a result of additionally storing the control cases, the oldest control case in the control case storage means is erased. It is characterized by

Further, in the invention described in claim 10, in the image forming apparatus according to claim 8, when the storage capacity becomes less than a predetermined amount as a result of additionally storing the control case, The oldest cluster among the clusters in the cluster storage means is deleted.

Further, in the invention described in claim 11, in the image forming apparatus according to claim 3, the control rule is stored together with its creation time information, and the matching degree of each control rule is stored. When the control rule storage means for updating and storing the cumulative value is provided, and the remaining storage capacity of the control rule storage means is less than a predetermined amount, the control rules already stored are stored before the predetermined time. It is characterized in that one having the smallest cumulative value of the goodness of fit is selected and deleted from the storage means.

According to a twelfth aspect of the present invention, in the image forming apparatus according to the second aspect, there is provided control case storage means for storing a control case of an output image, and the cluster storage means carries out the control. Of the control cases in the case storage means, those having similar state quantities are collected and stored as a cluster, and when one cluster is completed, the cases that are the constituent elements of the cluster are stored in the control case storage means. It is characterized by erasing.

According to the thirteenth aspect of the present invention, in the image forming apparatus according to any one of the first to twelfth aspects, the control case includes an operation amount, a control amount, and an apparatus. It is characterized by being composed of three kinds of state quantities relating to states.

According to a fourteenth aspect of the present invention, in the image forming apparatus according to any one of the first to twelfth aspects, the control rule is n kinds which is set according to n control objects. Is extracted as a least square error n-th plane of a plurality of coordinate points indicating a control case in an (n + 1) -dimensional space composed of an axis indicating the operation amount and an axis indicating the control amount for the controlled object. And

According to the fifteenth aspect of the invention, in the image forming apparatus according to any of the first to twelfth aspects, the output image quality to be controlled is the image density. .

Further, in the invention according to claim 16, in the image forming apparatus according to claim 7 or 8, immediately after the previous comparison result by the comparison means exceeds a preset allowable value, Even when the comparison result of this time by the comparison unit falls within a preset allowable value, the control amount of this time and the operation amount used for the control are stored in the control case storage unit, and the next and subsequent times are stored. It is characterized in that it can be used for control.

According to a seventeenth aspect of the present invention, in the image forming apparatus according to the sixteenth aspect, the cluster storage means has a previous comparison result by the comparison means exceeding a preset allowable value. Immediately after that, when the comparison result of this time by the comparison means falls within a preset allowable value, the control case of this time is classified into the latest cluster.

In the image forming apparatus according to the eighteenth aspect, in the image forming apparatus according to the sixteenth aspect or the seventeenth aspect, the comparison result exceeds a preset allowable value based on the comparison result by the comparing means. Immediately after the above, again, a convergence detection means for detecting a change when it falls within a preset allowable value is provided.

Further, in the invention described in claim 19, in the image forming apparatus according to claim 7 or 8, the cluster storage means is a permissible value in which the previous comparison result by the comparison means is set in advance. Immediately after being set to, if the comparison result of this time by the comparison means exceeds a preset allowable value, the control case of this time is classified into a new cluster.

Further, in the invention according to claim 20, in the image forming apparatus according to claim 19, immediately after the comparison result falls within a preset allowable value based on the comparison result by the comparison means. In addition, it is characterized by further comprising non-convergence detection means for detecting a change when the preset allowable value is exceeded.

[0040]

In the invention described in claim 1, when the apparatus is in the operating state, the control cases are accumulated in the control case storing means, and the control rule extracting means controls the control rules based on the accumulated control cases. To extract. On the other hand, in the operation amount calculation means, the control amount detected by the detection means and the target quality are compared with each other, and the operation amount such that the control amount approaches the target quality is obtained. At this time, since the operation amount is calculated by referring to the extracted control rule, the operation amount is obtained by referring to past cases. Then, this operation amount is supplied to the image state changing means, and the image quality is operated.

Further, in the invention described in claim 2,
When storing the control cases in the cluster storage means, the control cases having similar state quantities are collected and stored as a cluster. Then, the control rule for each cluster is extracted by the cluster-specific rule extraction unit, and the control amount calculation unit calculates the control amount using each control rule. Therefore, the control rule required for the next control can be appropriately selected, and the control more suitable for the situation can be performed.

Further, in the invention described in claim 3,
The control amount calculation means, when using the control rule for each cluster, determines the degree of conformity to the immediately preceding control case,
Each control rule is weighted according to the goodness of fit and averaged, and the result is used to obtain a new control amount. Then, the control corresponding to the changing situation is performed.

Further, in the invention described in claim 4,
The control amount calculating means in the image forming apparatus according to claim 3, between the n-dimensional plane indicating the control rule and the coordinate point indicating the immediately preceding control case, in the coordinate space in which each control rule is described. Since the reciprocal of the distance is standardized for each control rule to obtain the goodness of fit, the goodness of fit can be calculated on the coordinates. Therefore, the calculation process can be performed at high speed.

Further, in the invention described in claim 5 or 6, in the image forming apparatus according to claim 3 or 4, except for the control rule whose conformity is equal to or less than a predetermined value, the other control rules are again described. Since the goodness of fit is determined, and the results of averaging by weighting these control rules according to the goodness of fit are used, clusters with few relations can be ignored, and control suitable for the situation can be performed with high accuracy. Can be done at.

Further, in the inventions according to claims 7 and 8, the control amount is compared with the target quality, and if the comparison result exceeds a preset allowable value, the control amount is changed. Since the comparison means is provided so as to be used for the control on and after the next time, the control case corresponding to the allowable value is fetched. That is, the accuracy of the device can be arbitrarily adjusted by appropriately setting the allowable value.

Further, in the inventions according to claims 9 and 10, when the storage of the control case becomes less than a predetermined amount as a result of additionally storing the control case, the control in the control case storage means is performed. Since the oldest case is deleted, the memory can be used efficiently.

According to the eleventh aspect of the present invention, the newly created control rules are sequentially stored in the control rule storage means. However, when the storage capacity becomes low, the control rules that have been created before the predetermined time and have the smallest cumulative fitness value, that is, the control rules of low importance are sequentially deleted.

Further, as in the invention described in claim 12,
Even when the case that is a constituent element of the cluster is deleted from the control case storage means when one cluster is completed, the memory can be used efficiently.

In the thirteenth aspect of the present invention, the control case in the image forming apparatus according to any one of the first to twelfth aspects is such that the operation amount, the control amount, and the device are set. Since it is composed of three types of state quantities regarding, it is possible to perform processing such as classifying control cases in units of state quantities that reflect the environment. For example, clustering can be performed according to the state quantity.

According to a fourteenth aspect of the present invention, in the image forming apparatus according to any one of the first to twelfth aspects, the control rule is n kinds which is set according to n control objects. In the n + 1 dimensional space composed of the axis indicating the manipulated variable and the axis indicating the controlled variable for the controlled object, the statistic is obtained as the least square error n-th plane of a plurality of coordinate points indicating the control case. It is possible to create a rule with little error.

According to the fifteenth aspect of the invention, in the image forming apparatus according to any of the first to twelfth aspects, since the image density is the output target, the image quality is determined in the copying machine or the like. Important factors are controlled according to past control cases.

According to the sixteenth aspect of the invention, in the image forming apparatus according to the seventh or eighth aspect, the comparison is performed immediately after the previous comparison result by the comparison means exceeds a preset allowable value. Even if the comparison result of this time by the means falls within the preset allowable value, the control amount of this time and the operation amount used for the control are stored in the control case storage means and used for the control of the next time and thereafter. Therefore, when the control rule is extracted, the operation amount can be optimally determined by using not only the control rule extracted in the past but also the latest control rule that grasps the current situation.

According to the seventeenth aspect of the invention, in the image forming apparatus according to the sixteenth aspect, the cluster storage means immediately after the previous comparison result by the comparison means exceeds a preset allowable value. If the comparison result of this time by the comparison means falls within the preset allowable value, the control case of this time is classified into the latest cluster, so that the number of control cases necessary for newly creating the cluster is collected. Moreover, even if the control amount falls within the allowable value, the control case having the information on the actual state change can be taken into the cluster. Therefore, the number of control cases taken in the same cluster increases, so the accuracy of the control rules extracted based on that cluster can be improved, and the operation amount can be optimized by using the latest control rules that grasp the current situation. You can decide.

According to the eighteenth aspect of the invention, in the image forming apparatus according to the sixteenth or seventeenth aspects, the comparison result exceeds a preset allowable value based on the comparison result by the comparison means. Immediately after that, since the convergence detection means for detecting the change when the value falls within the preset allowable value is provided again, the current control amount and the operation amount provided for the control are used for the next and subsequent controls. It is possible to improve the accuracy of the control rule.

Further, in the invention described in Item 19, in the image forming apparatus according to Item 7 or 8, in the cluster storage means, the previous comparison result by the comparison means falls within a preset allowable value. Immediately after that, if the comparison result of this time by the comparison means exceeds the preset allowable value, the control case of this time is classified into a new cluster, so that if a sufficient number of control cases are acquired, Changes in
It is possible to create a control rule that can respond more appropriately. In addition, even if a control amount temporarily exceeds the allowable value due to noise, etc., these abnormal control cases are classified into normal clusters that classify control cases consisting of the correct control amounts acquired and created so far. No need to add. further,
Since the control rule is not calculated from the cluster including the apparently abnormal control case due to noise etc., it is possible to secure a more correct control rule without the influence of the control case including noise, and to respond more appropriately to the current changes. Control can be performed based on a control rule that can be performed.

According to a twentieth aspect of the invention, in the image forming apparatus according to the nineteenth aspect, based on the comparison result by the comparison means, immediately after the comparison result falls within a preset allowable value. Again, since the non-convergence detecting means for detecting the change when the preset allowable value is exceeded is provided, an abnormal control case temporarily exceeding the allowable value due to noise or the like can be detected.

[0057]

EXAMPLES Specific examples according to the present invention will be described below. The basic configuration described below, and the development patch creating mechanism and its monitoring mechanism are common to the first to third embodiments described later. A: Configuration of First Embodiment (1) Basic Configuration First, the image output unit IOT of the image forming apparatus according to the present invention.
Figure 2 shows the outline of (Image Output Terminal). The image reading unit and the image processing unit are omitted in FIG. That is, only the electrophotographic image output unit IOT is shown.

The image forming procedure will be described with reference to FIG.
First, an image processing unit (not shown) is suitable for an original image signal obtained by reading an original document by an image reading unit (not shown) or created by an external computer (not shown). Performs various processing. The input image signal thus obtained is input to the laser output unit 1 and modulates the laser beam R. In this way, the laser beam R modulated by the input image signal is raster-irradiated onto the photoconductor 2.

On the other hand, the photoconductor 2 is uniformly charged by the scorotron charger 3, and when irradiated with the laser beam R, an electrostatic latent image corresponding to the input image signal is formed on the surface thereof. Next, the electrostatic latent image is developed with toner by the developing device 6, the developing toner is transferred onto a sheet (not shown) by the transfer device 7, and is fixed by the fixing device 8. After that, the photoconductor 2 is cleaned by the cleaner 11, and one image forming operation is completed. Also, 10
Is a development density sensor, which detects the density of a development patch (described later) formed outside the image area.

(2) Developing Patch Creating Mechanism and Monitoring Mechanism Thereof The developing patch and its monitoring mechanism in this embodiment will be described below. The development patch is for monitoring the output image density, and as shown in FIG. 3, two types of patches are a solid (halftone dot coverage 100%) density patch a1 and a highlight (halftone dot coverage 20%) density patch a2. Has been adopted. Then, these solid density patches a1,
As shown in FIG. 3, each of the highlight density patches a2 is set to have a size of about 2 to 3 cm square and is formed outside the image area of the photoconductor 2. That is, as shown in FIG. 4, after the latent image is formed in the image area 2a, the solid density patch a1 and the highlight density patch a2 are sequentially formed in the empty area 2b.

The density sensor 10 is composed of an LED irradiating section for irradiating the surface of the photoconductor 2 with light and a photosensor for receiving specularly reflected light or diffused light from the surface of the photoconductor 2. A line L1 shown in FIG. 3 is a detection line of the development density sensor 10. Therefore, the solid density patch a1 and the highlight density patch a2 are detected by the detection line L1.
The developing density sensor 10 is formed on the upper surface of the developing density sensor 10.
It passes in the vicinity of.

Here, FIG. 5 is a diagram showing an example of the output signal of the developing density sensor 10. As shown in the figure, first, a density detection signal corresponding to the image of the original is obtained, and then, density detection signals for the solid density patch a1 and the highlight density patch a2 are obtained. Since the solid density patch a1 and the highlight density patch a2 are formed outside the image area,
It is not transferred to the sheet and is erased when passing through the cleaner 11.

In this embodiment, the density of the developing patch is detected because it has a high correlation with the density of the fixed image (final image density) of the user's hand, and can be removed by the cleaner 11. Is. Further, the development patch may be formed in the image area at a timing other than when the image is formed.

(3) Structure of Control Unit According to First Embodiment Next, FIG. 1 is a block diagram showing the structure of the control unit 20 for controlling the scorotron charger 3 and the laser output unit 1. In the figure, reference numeral 21 is a density adjustment dial, and the operator sets a value according to a desired density. The set value of the density adjustment dial 21 is converted by the converter 22 into a value converted into the output of the developing density sensor 10 (a value between "0" and "255" in the case of the first embodiment). . The target concentration output from the converter 22 is held in the controlled variable memory 23. In this case, the control amount memory 23 also stores the allowable error amount.

On the other hand, the output signal of the developing density sensor 10 and the output signal of the memory 23 are compared in the density comparator 24. In this comparison, the allowable error amount stored in the memory 23 is referred to. Then, the output signal of the phenomenon concentration sensor 10 is supplied to the control rule retrieval unit 30 if the difference between the two is within the allowable value, and is supplied to the control case memory 25 if the difference between the two is within the allowable value.

The control case memory 25 is a memory for storing control cases, and stores three kinds of quantities of state quantity, operation quantity and control quantity as a set. The control cases are stored in this way in order to perform various controls based on the control cases stored in the past in the first embodiment. This is a control method based on a method called case-based reasoning.

Here, the state quantity stored in the control case memory 25 refers to the temperature and humidity that have a dominant influence on the electrophotographic process, or the deterioration amount over time, and there are these state quantities. Since it can be regarded as almost constant within a limited time, in the case of the first embodiment, the occurrence time (date and hour / minute / second) of the case is used as a substitute. However, if the occurrence times are within a predetermined time unit (a predetermined time unit such as 3 minutes, 5 minutes, or 10 minutes), the state quantities are treated as equal. this is,
This is because, if the occurrence times are close to each other, it can be expected that the two are under substantially similar temperature and humidity and the degree of deterioration with time is also the same. Further, the time data indicating the time of occurrence is shown in FIG.
The clock timer 40 shown in FIG.

Next, the manipulated variable refers to a parameter adjustment amount for changing the output value of the controlled object. In the case of the first embodiment, the grid voltage set value (0 to 255) of the scorotron charger 3 is set. , And a laser power setting value (0 to 255, hereinafter abbreviated as LP setting value). These two amounts are used as manipulated variables because the final image density to be controlled is the solid density part and the highlight density part, and the scoro setting value and the LP setting value are the solid density and the high density value. This is because the light density has a high correlation.

The scoro setting value and the LP setting value are
Each value is stored in the manipulated variable memory 32, and the value corresponding to the output signal of the manipulated variable correction calculator 31 is read out as appropriate. Then, the scoro setting value read from the manipulated variable memory 32 is supplied to the grid power supply 15, whereby the grid power supply 15 applies a voltage corresponding to the scoro setting value to the scorotron charger 3. Further, the LP set value read from the manipulated variable memory 32 is supplied to the light amount controller 16, whereby the light amount controller 1 is supplied.
Reference numeral 6 gives the laser output section 1 laser power according to the LP set value.

Next, the control amount supplied to the control case memory 25 is an output signal of the developing density sensor 10, and as a result of the above, the control cases such as those shown in the following table are stored in the control memory 25. To be done.

[0071]

[Table 1] In this table, for example, in case 1, the state quantity (occurrence time) is 12:00:10 on April 1, 1994, the LP set value is "83", the scoro set value is "130", and the control quantity (sensor The output value) is “185” for the solid part and “23” for the highlight part, and in case 4 the state quantity 19
On April 2, 1994 at 9:05, the LP setting value is "14.
8 ”, the scoro setting value is“ 115 ”, the control amount is“ 185 ”for the solid portion, and“ 30 ”for the highlight portion.

Next, the state quantity comparator 2 shown in FIG.
6, cluster memory 27 and control rule calculator 28
Has a function of extracting a control rule by referring to the control case stored in the control case memory 25. The operation of these blocks will be described in detail later.

The control rule memory 29 is a memory for storing a plurality of control rules calculated by the control rule calculator 28. When a request is issued from the control rule search unit 30, the control rule memory 29 returns the control rule corresponding to the request. . In this case, the control rule searcher 30 sets the control rule according to the density difference supplied from the density comparator 24 and the operation amount (that is, the LP set value and the scoro set value) supplied from the operation amount memory 32 to the control rule. The memory 29 is requested.

Next, the manipulated variable correction value calculator 31 finds the correction value for the manipulated variable using the control rule searched by the control rule search unit 30, and supplies the found correction value to the manipulated variable memory 32. To do. As a result, the operation amount memory 32
Supplies the operation amount corresponding to the operation amount correction value, that is, the LP setting value and the scoro setting value to the grid power supply 15 and the light amount controller 16, respectively.

On the other hand, the reference patch signal generator 42 is a circuit for instructing the creation of the solid density patch al and the highlight density patch a2, and outputs the calibration reference patch signal to the image output unit ITO at the patch creation timing. As a result, the solid density patch a1 and the highlight density patch a2 shown in FIG. 3 are created.

In this case, the operation timing of the reference patch signal generator 42 is controlled by the I / O adjusting section 41.
The I / O adjustment unit 41 monitors the time signal output by the clock timer 40, and operates the reference patch signal generator 42 with the operation timing so that the solid density patch a1 and the highlight density patch a2 are formed at predetermined positions. Supply a signal.

B: Operation of the First Embodiment (1) Initial Setting Operation Next, the operation of the first embodiment having the above configuration will be described. First, the initial setting process (so-called function startup) Processing) will be described. First, the engineer appropriately sets the scoro setting value and the LP setting value selected as the control parameters. Then, the control unit 20 creates the solid development patch a1 and the highlight development patch a2, measures each by the development density sensor 10, and stores the contents in the control case memory 25 as a control case. As a result, the control case memory 25 stores the first control case (control case 1). In reality, since the concentration cannot be directly measured, an optical sensor is used to measure the reflected light amount or diffused light amount of light as a potential. After this, although there is development density measurement, in reality, a physical quantity (a quantity of light is used as a potential) instead of the density is measured.

Similarly, while changing the scoro setting value and the LP setting value respectively, two more control cases are stored in the control case memory. That is, the engineer creates a total of three sets of control cases when the control device is started up (within unit time with the same amount of state), and stores the control cases in the control case memory 25.

Here, the number of three sets is equal to the number of controlled objects +
This means 1 and in the first embodiment, 1 is added to 2 (the solid density and the highlight density) which is the number of control targets. Note that it is acceptable to show more control cases than this. As described above, when three sets of control cases (the number of control targets + 1) at the time of initialization are stored in the control case memory 25, the stored contents are stored in the state quantity comparator 26.
And the control rule calculator 2 via the cluster memory 27
8 where the control rules are sought. The control rule in this case is extracted as a control case plane as shown in FIG.

In FIG. 6, P1, P2, and P3 are points indicating the combination of the scoro set value and the LP set value for the three sets of control cases in the initial setting. Where point P
The points indicating the highlight densities (detection densities of the highlight density patches) corresponding to 1, P2, P3 are defined as H1, H2, H3, and similarly, the solid densities corresponding to the points P1, P2, P3 (detection of the solid density patch). The points indicating density are B1, B2, B3
And A plane passing through the points B1, B2, B3 is a solid case plane BP, and a plane passing through the points H1, H2, H3 is a highlight case plane HP. Here, when the state quantity does not change, all points indicating the solid density obtained when the scoro setting value and the LP setting value are appropriately changed fall within the solid case plane BP. Similarly, when the state quantity does not change, all points indicating the highlight density obtained when the scoro setting value and the LP setting value are appropriately changed fall within the highlight case plane HP. Thus, the solid case plane BP and the highlight case plane HP
Indicates all cases in which the state quantity does not change, in other words, these planes indicate the control rules for the solid density and highlight density at the initial time. With the above processing, the initial setting processing in the first embodiment is completed.

The reason why three sets of cases are stored at the time of initial setting will be described below. First, in general, when the number of controlled objects is n, n + 1 control cases are required, and the surface showing the control cases is an nth-order plane in an n + 1-dimensional space. Therefore, n + 1 data points are needed to uniquely determine this nth plane. In the case of the first embodiment, since two control objects, solid density and highlight density, are set, n = 2,
Three sets of control cases are needed.

(2) Operation during operation Basic operation Next, the operation of the first embodiment during operation will be described. However, in the following operation, it is assumed that the actual operation control is started from the next day when the initial control rule is determined as described above.

First, when the image forming apparatus is powered on, the setup operation is automatically executed. In this setup operation, the solid density patch a1 and the highlight density patch a2 are created by using the previous setting values (for example, the last image output of the previous day) as they are this setting value, and these density values are developed. It is measured by the density sensor 10. Here, assuming that the LP set value is “98” and the scoro set value is “76”, the development density sensor 10
Plot the detected wave degrees in the control case space. now,
Assuming that the densities of the solid density patch al and the highlight density patch a2 are B4 and H4, respectively, the plot shown in FIG. 7 is performed, and the control content of this time corresponding to the stored control case is recognized. It

This plot is performed by the control rule searcher 30 shown in FIG. That is, the control rule searcher 30
The densities B4 and H4 transferred from the density comparator 25 and the LP set value “9” transferred from the manipulated variable memory 32.
8 ”and the scoro setting value“ 76 ”, the plot is made on the control case plane at the time of initialization stored in the control rule memory 29.

By the way, the control case plane is created by plotting the output values when a certain setting is made under a certain state. Therefore, some change occurs in the state and the same setting is made. Also, if the output values become different, it naturally does not match the control case plane in the state before the change occurs. That is, as in the above example, when the control contents at this setup are plotted on the control case plane created at the start-up of yesterday (effectively, without separating the distance), And the current state of the image forming apparatus (for example, the influence of all factors that affect the electrophotographic process, such as temperature, humidity, and the degree of change over time) can be considered to be virtually the same. Means Here, "effectively without separating a distance" means that the difference between the actually output image density and the target density is the allowable error amount as a result of the control assuming that they match on the control case plane. Say when it does not exceed.

Next, the initially set print density or the desired print density designated by the user is converted into a density sensor output to obtain a target density output value, and the target density plane is set in the control case space described above. Set as. The print density is an actual density that is directly measured, and the density sensor output is an alternative value of density (a quantity of light is a potential). This setting is performed as follows on the circuit.

First, the adjustment value of the density adjustment dial 21 is converted by the converter 22 and stored in the memory 22. Then, the density target value in the memory 22 is transferred to the control rule searcher 30 via the density comparator 24. The control rule retrieval unit 30 describes the plane of the target density value in the control case space (the plane parallel to the scoro set value axis-LP set value axis plane), and reads from the control rule memory 29 the solid case plane BP and the highlight. Overlay on the case plane HP.

By the above processing, in the control case space,
As shown in FIG. 7, a solid case plane BP relating to solid density
And highlight case plane HP regarding highlight density
That is, the solid target density plane BTP and the highlight target density plane HTP are configured, and the control contents at the time of setup described above are plotted therein.

As is apparent from FIG. 7, if the control content of this time is plotted on the solid target realization line BTL where the solid case plane BP and the solid target density plane BTP intersect, the solid target density can be realized. Will be. If the control content this time is not on the target realization line, if each set value is changed, that is, corrected, and a combination that is plotted on the solid target realization line BTL is selected, Image output can be predicted to achieve the solid target density.

Similarly, with respect to the highlight density, it can be inferred that the highlight target density can be realized at the next image output by selecting a combination of set values plotted on the highlight target realization line HTL. Therefore, in order to control both the solid density and the highlight density to the target density at the same time, the solid target realization line BTL and the highlight target realization line HTL are created by the LP setting value axis and the scoro setting axis. Then, the LP set value and the scoro set value at the intersection may be adopted. In the case of the example used in FIG. 7, it can be seen that if the set value (98, 76) of this time is corrected to (128, 115) and set next time, the respective target densities of solid and highlight can be realized at the same time. In this way, from the setup data,
The next LP setting value and scoro setting value for realizing the desired solid and highlight densities can be determined.

The above-described processing for calculating the next set value is performed by the manipulated variable correction calculator 31 and the calculation result is transferred to the manipulated variable memory 32. As a result, signals corresponding to the new scoro setting value and new LP setting value are output from the operation amount memory 32 and supplied to the grid power supply 15 and the light amount controller 16. Thereafter, similarly, the optimum LP set value and scoro set value for achieving the target density are set, and accurate image density control is performed.

Generation of Cluster The present invention basically realizes the target density as described above, but in reality, the control content at the time of its operation is always on the solid and highlight case planes ( Effectively, it is not always plotted (without separating the distance). Physically speaking, if the temperature and humidity change or the deterioration with time progresses, the toner charge amount and the charging characteristics of the photoconductor change, so the laser power and the scorotron grid voltage are the same set values. Even
The concentration will be significantly different. For example, when the temperature is high and the humidity is high, the concentration shifts to the high side, and when the temperature is low and the humidity is low, the concentration shifts to the low side. In other words, if the temperature and humidity at the time of control, the degree of deterioration over time, etc., differ from the control case groups that have already been sampled and stored to some extent, a coordinate space that is far away from the existing solid and highlight control case planes. It will be plotted above.

In such a case, if a certain control case plane is used as it is as the control rule of this time, the inference error becomes large. This is because the image density reproduction mechanism is physically affected as described above, and the control case plane is changed. Therefore, in the present invention, a control case when the state changes is additionally stored, and a new control case plane including a control case group adapted to the new state is created. As a result, the control case plane gradually increases from the state of only one plane at the time of start-up as needed. That is, for example, a control case group under the condition of A,
A group of control cases under another state of B, and so on. In the present invention, each of these is named a cluster. That is, cluster A, cluster B, and so on. The determination as to whether or not to add the control case is made based on the result of the determination, by using the development patch created after the control operation is performed to determine whether the control result is good or bad.

Specifically, the density difference between the target density and the actual solid and highlight development patches is read, and it is determined whether this value is within the allowable error. By the way, in the first embodiment, the allowable error of the solid density is defined to be within the color difference of 3 and the allowable error of the highlight density is defined to be within the color difference of 1. However, this value is arbitrarily determined according to the target accuracy of the system.

If both solid and highlight are within the permissible error, the next control operation is directly performed as described above, but either one of the solid and the highlight exceeds the permissible error amount. If there is a large error, its content, that is, the control case is additionally stored in the control case memory 25.

This additional storage is performed as follows. First, the density comparator 24 shown in FIG. 1 determines that the density is equal to or more than the allowable value, and the output signal of the development density sensor 10 at that time is transferred to the control case memory 25. The control case memory 25 stores the state quantity and the operation quantity as a set together with the newly supplied control quantity. Then, the state quantity comparator 26 compares the latest cluster with the time based on the case newly written in the control case memory 25, and determines whether the state similarity is apricot. That is, the time information of the control case in the latest cluster, which is the control case group, and the control case memory 25.
Compare with the time information of the control case newly written in
If it is within a predetermined time, it is determined that the states are similar, and if they are separated by a predetermined time or more, it is determined that the states are not similar.

When it is determined that the states are similar, the cluster memory 27 is written to add a control case for the latest cluster. At this time, the control rule calculator 28
Calculates a case plane that includes the newly added control case, and assigns a coefficient indicating the plane to the control rule memory 2
Transfer to 9.

Here, a method of correcting the control rule when the number of control cases increases will be described. As described above, if the number of controlled objects is n, an n-dimensional plane of an n + 1-dimensional space is required for the control, and in order to uniquely determine this, n + 1 planes are required. Data points are needed.
For this reason, in the first embodiment, three sets of control cases are necessary in the initial setting. On the contrary,
If n + 1 or more, a statistically more reliable case group can be obtained. Therefore, the control rule calculator 2
8 determines the plane using a calculation method such as the least-squares error method using the added control case and the control cases stored before that (that is, using n + 1 or more data). Of course, the calculation method is not limited to the least squares method, and another calculation method such as an average method may be used. In short, other methods may be used as long as the n-dimensional plane can be set based on the control case.

On the other hand, when the state quantity comparator 26 determines that the states of the control cases newly written in the control case memory 25 are not similar, a new cluster is created and classified. This new cluster is transferred to the cluster memory 27, and a new rule (plane) is generated by the control rule calculator 28.
Is calculated. By the way, in the control rule memory 29, only the coefficient of the equation indicating the plane calculated by the control rule calculator 28 is stored, and the increase of the storage capacity is suppressed as much as possible.

Memory Management As described above, when additional control cases are additionally stored and new clusters are created, the number of control cases and clusters stored is gradually increased, and eventually stored. It's time to fill the capacity. In view of this, in the first embodiment, the control case and the cluster are stored in separate storage areas, the collection or creation time of each is noted, and the oldest ones are deleted.

The reason for this is that the current state has changed over time, so it seems that the older the control case or the older cluster, the worse the fit. This is because the probability of needing is extremely low, and it can be judged that it is less valuable to store and store than others. Also,
As shown in the first embodiment, when judging the similarity of the state quantity by date, when a cluster is fixed after a certain fixed period of time, that is, a control rule corresponding to the cluster is extracted. When finished, the control cases belonging to that cluster can be erased.

As described above, if the control case included in the cluster is erased when the cluster is established, the memory capacity can be extremely reduced. For example, in the first embodiment, the control case is made up of three elements, and the cluster is also made up of three elements, the inclination in the set value axis direction and the density axis intercept, and assuming that the size of the storage area of each element is equal to n bits. ,
Since three control cases are required to form one cluster, storing all of them requires a storage area of 3 × 4 × n bits. In this case, if the control case is deleted when the cluster can be formed, 3 × 1 ×
Since the storage capacity of n bits is sufficient, it can be seen that the total storage capacity of 1/4 is sufficient. This method is particularly effective when storing a large number of clusters. Similarly, if one cluster includes 10 control cases, the storage capacity becomes 1/11, and if 100 control cases are included, 1/1
The storage capacity is 01. In this way, significant memory savings can be achieved.

Of course, even if the memory is saved, a large number of clusters may eventually run out of memory capacity.
In that case, as described above, it can be dealt with by deleting the oldest clusters first.

Control Using Complex Clusters As is apparent from the above, various clusters are created when the first embodiment operates under various states. However, when the state changes,
It is not always necessary to additionally store a new control case and create a new cluster. For example, if other conditions such as humidity are effectively the same, and there is already a cluster when the temperature is high and a cluster when the temperature is low, the image forming apparatus that will operate at medium temperature this time In many cases, sufficient control accuracy can be obtained by using a combination of a high temperature cluster and a low temperature cluster without creating a new cluster. In such a case, based on the current control content and the distances from each of the multiple control case planes, a new plane is constructed so that the current control content is included in that plane, and with this control The control is performed assuming that it is the case plane.

Now, plane construction by combining clusters will be described with reference to FIG. FIG. 8 shows a case where a solid case plane of cluster A and a solid case plane of cluster B are formed, and the newly plotted point B5 is not located on any plane. At this time, a point indicating the current control content in the coordinate space, that is,
Calculate the distance between point B5 and each case plane. Then, the reciprocal thereof is obtained and standardized. That is, the sum of the reciprocal of the distance from each control case plane is 1
So that The value thus standardized is defined as the goodness of fit, and the tilts of the respective case planes in the respective coordinate axis directions are weighted and summed by the goodness of fit. Then, the totaled amount is used as the inclination in each coordinate axis direction of the new control case plane that can be adapted to the current state, and is further adjusted to the height (intercept of the concentration axis) including the current control content on the plane. .

Such processing has a compatibility of about 100%.
This is performed when the control case plane that can be regarded as is not retrieved. Here, the case where the goodness of fit is about 100%,
As described above, the newly plotted points are synonymous with “when actually plotted on the control plane without a distance.”

The above processing is performed in the control rule search unit 30 as follows. First, points corresponding to the operation amount supplied from the operation amount memory 32 and the detection value of the development density sensor 10 transferred from the density comparator 24 are plotted in the coordinate space. Then, the control planes of the respective clusters stored in the control rule memory 29 are sequentially read out, and the distance to the newly plotted point is obtained. However, the "distance" here is the difference between the calculated control amount obtained by substituting the operation amount into the control rule formula and the actually measured control amount, and is not necessarily the shortest distance between the surface and the point. May not be distance. Then, the goodness of fit is calculated from the distance thus obtained, and the inclinations in the coordinate axis directions of the respective case planes are weighted and summed according to the goodness of fit. The plane having the inclinations of the respective axes thus summed is set as a new control case plane, and the height of the new control case plane (intercept of the concentration axis is set so that the plotted points are located on the plane). ) Is adjusted. Using the new control case plane created as described above, the following LP set value and scoro set value are obtained by the same procedure as that shown in FIG. 7.

Incidentally, in an image forming apparatus immediately after start-up, or when the operating time or the number of times of image formation is small, the control case plane naturally exists only on one side created at the time of start-up. In this embodiment, it can be handled exactly the same as when there are a plurality of control case planes. That is, when the control case plane is only one surface created at the time of start-up, the degree of conformity of that surface is 1 (100%), so the inclination of the surface does not change and the current control content is within the surface. The control case plane used at this time is the control case plane that was created at the time of start-up that was translated in the concentration axis direction to the position included in.

On the other hand, with the past control cases alone, it is not enough to virtually construct a new control case plane using the goodness of fit as described above. If it is predicted that the control accuracy from the next time will be insufficient unless the control rule is improved (that is, if the density comparator 24 determines that the accuracy is equal to or higher than the allowable value), a new cluster is set as described above. create.

C: Effects of the First Embodiment The above is the configuration and control procedure of the first embodiment. The effects of the first embodiment will be described below. (1) First, in the first embodiment, a control method that does not depend on physical quantity sensors other than the temperature sensor or prior data collection and analysis work by an engineer can be realized by using the above-described control case. That is, the number of sensors and development man-hours can be reduced, and the cost can be reduced. On the other hand, in the prior art, since the control method according to the physical mechanism is used, when the same control as in the first embodiment is attempted,
For example, a potential sensor is used to monitor physical quantities such as charging potential and exposure potential, and the developing potential (difference between the exposure potential and the developing bias) and the cleaning potential (difference between the charging potential and the developing bias) are calculated from this and the developing bias setting value. Calculate the optimum development potential to achieve the target solid density from the relationship between the solid density and the development potential that was collected in advance, and then calculate the change in highlight density by changing the development potential to the optimum development potential. Then, the highlight density error to be corrected in consideration of the change in the highlight density is calculated from the relationship between the highlight density and the cleaning potential that has been collected in advance, and the charging potential and the exposure potential are determined. . Then, according to the relationship between the charging potential and the scoro setting value that has been data-collected in advance and the relationship between the exposure potential and the LP setting value applicable to such a charging potential, the LP setting value to be set at the next image formation is set. And had to decide the scoro setting value. Moreover, the prior data collection had to be performed under various temperature and humidity environments because of the temperature and humidity dependency of the electrophotographic method.

(2) In the first embodiment, even when the image forming apparatus encounters, the substitute value (sampling time) of the state quantity can be used, so that it is not always necessary to sense the state quantity. This means that from the viewpoint of the physical mechanism of image formation, even a completely black box can be controlled, and the present invention is effective even in a method other than the electrophotographic method. Can be said.

As another function, any parameter can be adopted as the control actuator. That is, in the prior art, there is no suitable sensor,
According to the present invention, even if the sensor cannot be used in terms of cost, its set value can be directly handled, so that it can be arbitrarily selected and used for control.

(3) Further, as described above, the startup (initial setting) of the first embodiment can be performed only by inputting at least n + 1 cases. That is, there is an effect that no special technique or device for starting up the device is required. Moreover, even if these at least n + 1 cases happen to be largely deviated from the target densities, the control performance of the image forming apparatus thereafter is not adversely affected. That is, the image forming apparatus itself can always extract a new cluster, that is, a control rule that matches a new state, as needed.

Comparing this action with the conventional technique, for example, in the learning of the control rule by the conventional neural network, the inference performance of the neural network is impaired if the optimum data is not prepared as the teacher data to be learned, and the automatic learning is performed automatically. However, since additional learning and relearning are not performed, sufficient control performance cannot be obtained. Alternatively, in the conventional technique using the fuzzy inference, sufficient control performance cannot be obtained unless the engineer performs optimum trial and error tuning. From these comparisons, it can be seen how excellent the operation of the present invention is.

Furthermore, in the first embodiment, even if the image forming apparatus first encounters a state that it has not experienced before, it only needs to perform at least n + 1 print operations to suit the environment. New control rules can be extracted, and if similar environmental conditions are encountered after that, immediate and accurate control can be performed.
It is possible to cope with changes over time without collecting data in advance. That is, even if a change occurs over time, it is possible to always follow the changed situation.

Due to this effect, in the conventional development work, tens or hundreds of thousands of print jobs have to be actually executed to collect data in advance on the variation over time. , A big effect that it is not necessary at all.

Further, in the prior art, even the data on the variation with time collected in such a manner is not necessarily effective for all devices because there is a machine difference in each image forming device. This is different from the predicted change over time because the user used the image forming apparatus under a condition different from the condition at the time of data collection by the manufacturer in advance, that is, a condition that the manufacturer did not expect. There is a problem that the change occurs, the control rule becomes unsuitable, and the image density cannot be controlled to a desired value. However, according to the present invention, no special measures are taken in advance regarding data collection or machine difference. There is an effect that density fluctuations due to changes with time can be solved by completely corresponding individually to each image forming apparatus in any user environment without any need.

Due to this action, even if the element parts such as the photoconductor and the developer which have a great influence on the image density are replaced, the new element parts are automatically changed by performing the printing operation at least n + 1 times. A further effect is also obtained in that the desired image density is correspondingly obtained.

This is a work manually performed by a service engineer or the like in the prior art, and there is an effect of reducing labor costs in this respect. Even if you are not an expert such as a service engineer, even if a general user replaces a part, the best image formation state is automatically obtained.
There is also the effect that it is easy for anyone to handle.

Furthermore, according to the present invention, when a new situation is dealt with as described above, a new control case need not be stored by using the concept of goodness of fit for a plurality of past clusters. There is also the action. That is, it is an action that can immediately cope with a new situation without waiting for the end of the printing operation at least n + 1 times or without requiring a memory for additionally storing a new control case.

(4) Furthermore, the first embodiment has the effect that the control accuracy itself can be set arbitrarily. That is, the allowable error amount for the target concentration can be set directly, and the control rule can be improved based on the allowable error amount.
This is the effect that the necessary and sufficient control accuracy is automatically reached because corrections and new creations can be performed. Moreover, if necessary and sufficient control accuracy is obtained, more control cases than necessary are not stored and the memory capacity is not pressed.

(5) As yet another operation, since any state quantity or substitute value can be used in the present invention, various control methods are possible according to the characteristics and purpose of the image forming apparatus. become.

That is, when it is desired to control daily fluctuations (use the date for classifying state quantities) or to eliminate concentration fluctuations mainly caused by cycle down / cycle up (after pressing the start button for classifying state quantities, Even if the control content of interest is different, such as using the number of prints), there is an effect that the conventional cumbersome development work of constructing a separate control algorithm is completely unnecessary. Of course, the present invention can be applied as it is without any modification even when performing state recognition using a temperature sensor or a humidity sensor as in the conventional case.

(6) Further, according to the first embodiment,
It has the effect that the memory capacity can be used to the maximum extent. In other words, the data having the lowest degree of importance is automatically and selectively erased, and important data can be stored without fail even if it has a limited memory capacity.

D: Second Embodiment The second embodiment of the present invention will be described below. Book second
The embodiment is different from the first embodiment described above in the method of acquiring the control case based on the comparison result of the control amount and the manipulated variable which is the target quality. (1) Configuration of Control Unit According to Second Embodiment FIG. 9 is a block diagram showing the configuration of the control unit according to the second embodiment of the present invention. The parts corresponding to those in FIG. 1 are designated by the same reference numerals and the description thereof will be omitted. In FIG. 9, the convergence determination unit 43 detects the output of the density comparator 44 when the difference (density difference) between the control amount (density) of the previous print and the target quality (density setting value) is the allowable value or more. ,
When the change from the allowable value or more to the allowable value or less, that is, when it converges, the convergence detection signal is sent to the state quantity comparator 45.
Supply to Specifically, when the difference (density difference) is equal to or larger than the allowable value, that is, when there is an output from the density comparator 44, “0” is stored, and when it is within the allowable value, that is, the output of the density comparator 44. When there is not, "1" is stored, and when it changes from "0" to "1", it is determined that the convergence has occurred, and the convergence detection signal is supplied to the state quantity comparator 45. As described above, the convergence determination unit 43 in the second embodiment detects the change from the state of exceeding the allowable value to the state of falling within the allowable value.

Further, the density comparator 44 supplies the output (control amount) of the development density sensor 10 to the control case memory 25 even when the density comparator 44 changes from the allowable value to the allowable value. Therefore, the case (state quantity, operation quantity, and control quantity) at that time is stored in the control case memory 25 even when the value exceeds the allowable value, that is, when the value changes from the allowable value to the allowable value. It is supposed to be remembered.

Further, the state quantity comparator 45 compares the latest cluster with the time based on the case newly written in the control case memory 25 as in the case of the first embodiment described above, and determines whether or not there is a state similarity. If it is determined that they are dissimilar, a new cluster is created and classified, while if a convergence detection signal is supplied from the convergence determination unit 43, the case at that time is set to the latest cluster (previous cluster). It is written in the cluster memory 27 to be added as a control case. At this time, the control rule calculator 28 calculates a case plane that includes the newly added control case, and transfers the coefficient indicating the plane to the control rule memory 29.

Differences between the first embodiment and the second embodiment will be described with reference to FIGS. 10 to 13. In the first embodiment, as shown in FIG.
The density measurement value (control amount) is compared with the density set value (manipulation amount) by the density comparator 24, and only when the difference between the two exceeds an allowable value, it is added as a control case of the latest cluster, or the number of cases is increased. Accordingly, it is classified as a new cluster case. For example, in the example shown in FIG. 11, since the first two cases exceed the allowable value, if the number of cases is sufficient (for example, 3), they are classified into a new cluster. In addition, it is not created when there are two cases.

On the other hand, in the second embodiment, as shown in FIG.
As shown in 2, in addition to the case where the difference between the density measurement value (control amount) and the density set value (manipulation amount) exceeds the allowable value based on the comparison result by the density comparator 44, the difference becomes equal to or larger than the allowable value. When it is within the allowable value, the case is added to the latest cluster. For example, FIG.
In the example shown in FIG. 3, since the first two cases exceed the allowable value, “0” is stored for each case, and the third case is within the allowable value range, so “1” is stored. Is stored, and in this third case, when it changes from “0” to “1”, the case (third) at that time is added to the latest cluster.

As described above, in the first embodiment, since the control case is not acquired unless the control amount exceeds the allowable value, the control case necessary for creating a new cluster, especially immediately after the state is changed. If the controlled variable falls within the allowable value before the points are gathered, the control case having the information on the actual state change is not immediately classified as a cluster. In such a case, since the control rule is not extracted, the control rule experienced in the past must be used. On the contrary,
In the second embodiment, even in the case as described above,
A control case having information about the actual state change can be acquired in the cluster.

E: Operation of the Second Embodiment Next, the operation of the second embodiment having the above configuration will be described. The following description will be made in comparison with the first embodiment described above. FIG. 14 is a conceptual diagram showing an acquisition example of a control case according to the first embodiment, and FIG.
It is a conceptual diagram which shows the example of acquisition of the control case by the 2nd Example. In the first embodiment, as shown in FIG. 14, the control case is acquired only when the comparison result between the control amount and the target quality (concentration setting value) exceeds a preset allowable value.
The ones enclosed by the circles indicate the control quantities that were control cases.

Here, the first to third print sheets are control cases, are classified into clusters, and when the third control case is acquired, the control rule is extracted from this cluster. Now, after the 7th print shown in the figure, if the state of the environment changes,
The ninth sheet is naturally acquired as a control case and classified into clusters. However, on the 10th sheet, the comparison result between the control amount and the target quality does not exceed the permissible value, so that it is not taken into the cluster as a control case. Therefore,
At the time of printing the tenth sheet, the control rule cannot be extracted yet, and the operation amount for the next control is determined from the existing control rules stored so far.

Then, like the eleventh print, when the comparison result between the control amount and the target quality again exceeds the preset allowable value, this is taken into the cluster as a control case, and finally, Control rules can be extracted. Here, even in the 11th and subsequent prints, if the comparison result between the control amount and the target quality does not exceed the preset allowable value, the state change such as the environment immediately after the 7th print has occurred. On the other hand, since the control rule cannot be immediately extracted, the 8th and 9th data cannot be used for the control immediately thereafter.

On the other hand, in the second embodiment, when the control amount exceeds the preset allowable value, when the control amount of a certain (previous) print exceeds the allowable value, The control cases are acquired both when the print control amount falls within the allowable value. Specifically, in FIG. 15, the convergence determination unit 43 stores “0” because the preset allowable value is exceeded for the first to third prints. Since the state quantity comparator 45 exceeds the allowable value for the first to third sheets, it is acquired as a control case and classified as a cluster, as in the first embodiment.
The control rule calculator 28 extracts a control rule.

Then, the convergence judgment unit 43 stores "1" for the fourth sheet, since it is within the allowable value. Furthermore, the convergence determination unit 43 detects the change from “0” to “1” and supplies a detection signal to the state quantity comparator 45. On the other hand, the state quantity comparator 45 also acquires the control amount of the fourth sheet, which is within the allowable value, as a control case and takes it into the cluster. As described above, in the second embodiment, as compared with the first embodiment, the amount of information for determining the operation amount is large, and even if the number of control rules is the same, the control amount (that is, the control amount forming the cluster extracted from it) Since there are many control cases), the precision of the control rules will be higher.

Then, the convergence determining section 43 stores "1" for the fifth to sixth sheets because the control amount is within the allowable value. Also, when the state of the environment or the like changes after the 7th sheet, since the allowable value is exceeded for the 8th and 9th sheets, "0" is stored and the same value is stored as in the first embodiment. The control amount is also acquired as a control case and taken into the cluster. Then, for the tenth sheet, the control amount exceeds the allowable value, so the convergence determination unit 43 stores “1” and detects the change from “0” to “1”.
A detection signal is supplied to the state quantity comparator 45. Therefore, the state quantity comparator 26 also acquires the control amount of the 10th sheet, which is within the allowable value, as a control case and takes it into the cluster. In this way, since the control amount of the 10th print is taken in as a control case, at this point,
That is, when the three control cases are acquired, the control rule can be extracted and can immediately respond to the change of the state such as the environment.

Similarly, the control amount for the twelfth sheet is also acquired as a control case and taken into the cluster. The number of clusters is the same as that of the first embodiment from the 8th to 12th sheets,
The number of control cases incorporated in the cluster is large.
Therefore, the control rules extracted from the 8th to 12th sheets are
More accurate. In the above description, the control rule is extracted when there are three control cases, but the present invention is not limited to this, and any number of control cases may be used.

F: Effect of Second Embodiment The above is the configuration and control procedure of the second embodiment. The operation and effect of the second embodiment will be described below. (1) Even if the control amount falls within the allowable value immediately after the control case is acquired, especially after the control cases are acquired and before the number of control cases required to newly create a cluster are collected, the actual state Control cases with change information can be captured in the cluster. (2) Therefore, when the control rule is extracted, not only the control rule extracted in the past but also the latest control rule that grasps the current situation can be used to optimally determine the operation amount. (3) Furthermore, since the number of control cases taken in the same cluster increases, the accuracy of the control rule extracted based on the cluster can be improved.

G: Third Example Hereinafter, a third example of the present invention will be described. Book Third
This embodiment is different from the above-described first embodiment in the method of cluster formation based on the result of comparison between the control amount and the manipulated variable which is the target quality. (1) Configuration of Control Unit According to Third Embodiment FIG. 16 is a block diagram showing the configuration of the control unit according to the third embodiment of the present invention. The parts corresponding to those in FIG. 1 are designated by the same reference numerals and the description thereof will be omitted. In FIG.
The non-convergence determination unit 46 determines the control amount (density) of the previous print.
The output of the density comparator 24 is detected when the difference (density difference) between the target quality (density setting value) and the allowable value is greater than or equal to the allowable value. It is supplied to the state quantity comparator 47. Specifically, when the difference (density difference) is equal to or larger than the allowable value, that is, when the density comparator 24 outputs, “0” is stored, and when the difference is within the allowable value, that is, the output of the density comparator 24. When there is not, "1" is stored, and when the state changes from "1" to "0", the non-convergence detection signal is supplied to the state quantity comparator 47. In this way, the non-convergence determination unit 46 in the third embodiment detects the change from the state within the allowable value to the state exceeding the allowable value, contrary to the second embodiment described above. It has become.

When the non-convergence detection signal is supplied, the state quantity comparator 47 creates and classifies a new cluster based on the case newly written in the control case memory 25. At this time, the control rule calculator 28 calculates a case plane that includes the newly added control case, and transfers the coefficient indicating the plane to the control rule memory 29.

Differences between the first embodiment and the third embodiment will be described with reference to FIGS. 17 to 23. In the first embodiment, as shown in FIG.
The density comparator 24 compares the density measurement value (control amount) with the density set value (manipulation amount), and only when the difference between them exceeds the allowable value, the control case of the same latest cluster immediately before that is added. For example, in the example shown in FIG. 18, since the first two cases exceed the allowable value, they are classified into a new cluster. Then, when the third control amount converges and the fourth control amount exceeds the allowable value again, it is classified into the latest cluster created immediately before. Then, in this state, that is, when the number of control cases becomes three, a new control rule is created.

Further, in the first embodiment, as shown in FIG. 19, the number of prints and the operating time are used as the state quantity so that the accumulated number of sheets corresponding to the density comparator 24 can be dealt with so as to cope with the state change of the environment and the like. When the number of prints exceeds the preset number of prints by the comparator 24a and the time comparator 24b, or when the operating time exceeds the preset time, a control case to be obtained thereafter is obtained. A new cluster is created so as to be stored as a different state from the cluster in which the control case that has been taken in is stored, and the control cases acquired after that are taken in to this new cluster.

For example, in the example shown in FIG. 20, the first and second
The second control case is classified into one cluster. In the figure, as the third control case, the state where the control converges, that is, the state where the control falls within the allowable value is shown.
Even if this has not converged, the following operation is the same. If it is assumed that either the number of prints or the operating time exceeds the preset value (given a value that is likely to change the state during printing in advance) in the third case, the state It is judged that has changed. Therefore, a new cluster is created in the fourth print when the control amount exceeds the allowable value for the target quality, and the fourth control case is classified into the above-mentioned first and second control cases. The new cluster that is different from the existing cluster is classified.

As described above, more clusters are created, control rules are extracted from each cluster, the degree of conformity between each control rule and the current state is determined, the control rule is weighted, and the control amount is set to the target. A new and more accurate control amount is calculated so that the value corresponds to the quality. However, with this method, the values (print count, operating time) that are likely to change during printing in advance.
Must be given in advance as a set value.

On the other hand, in the third embodiment, the non-convergence judging section 46, as shown in FIG. 21, determines the density measurement value (control amount) and the density setting based on the comparison result by the density comparator 24. The difference from the value (manipulation amount) is within the allowable value, and thereafter, it is determined that the allowable value is exceeded, and the case is added to the newly created new cluster. For example, in the example shown in FIG. 22, since the first two cases exceed the allowable value, “0” is set for each case.
Is stored. Since the third case is within the allowable value range, “1” is stored. Then, when the state changes after the third print and the control amount exceeds the allowable value for the target quality in the fourth print, "0" is stored and the state changes from "1" to "0". Is detected. As a result, a new cluster is created and the fourth control case is classified into the new cluster different from the cluster in which the first and second control cases are classified.

In this case, assuming that a control rule can be created by collecting three control cases, in the example shown in FIG.
No control rules are created in either cluster. Even after this, if the control amount exceeds the allowable value, the control rule is created assuming that the fourth control amount actually occurred due to the actual state of the printer at that time. On the other hand, after that, when the control amount does not exceed the allowable value, the fourth control amount does not occur due to a change in the state of the printer at that time, but is temporarily caused by noise or the like. It is regarded as exceeding the allowable value and the new control rule is not created. That is, in the subsequent control, it is determined that the control should be performed according to the control rules so far.

As described above, in the first embodiment, since a new cluster is not created until the number of prints or the time reaches a predetermined set value, for example, the state changes within that range and the control amount becomes the target quality. Even if it becomes possible to acquire a new control case beyond the above, the subsequent control cases are classified into the same cluster as before. Therefore, it is not possible to successively cope with a state that changes moment by moment. In addition, if the factors that become control cases (in particular, the controlled variable) such as the controlled variable and the manipulated variable temporarily become abnormal values due to noise, or if the controlled variable does not continuously exceed the allowable value. If (often
In the first embodiment, due to noise, the clusters are classified into the same clusters as the control cases so far. Therefore, noise-containing forms (noise components are statistically diluted according to the number of control cases). However, if the number of control cases is small, it will cause a serious error) and the control rules will be updated. Or, even if the control rule is not changed, even if the control value slightly exceeds the allowable value after it converges, the control rule is updated even if the control rules up to now conform to the current situation. I will end up.

On the other hand, in the third embodiment, when the value converges once, that is, when it falls within the allowable value, another control case in which the allowable value is exceeded is newly created. By classifying into clusters, it is possible to create a control rule that can more appropriately respond to changes in the current situation, and to obtain abnormal control cases due to noise against temporary changes in state due to noise. It is possible to prevent the accuracy of the control rule from decreasing by not adding the control cases that have been performed to the classified cluster (latest cluster).

H: Operation of Third Embodiment Next, the operation of the third embodiment having the above configuration will be described. The following description will be made in comparison with the first embodiment described above. (1) When state changes due to environment, etc. FIG. 23 is a conceptual diagram showing an example of acquisition of a control case according to the first embodiment, and FIG. 24 is an example of acquisition of a control case according to the third embodiment. It is a conceptual diagram shown. First, in the first embodiment, as shown in FIG. 23, the first cluster is created when the number of prints is 1 to 3, and three control cases are acquired up to the third print. A control rule is created based on the first cluster. In the 4th to 7th prints, the result of comparison between the control amount and the target quality does not exceed the allowable value, and therefore no cluster is created.

Next, assuming that the environment or the like changes immediately after the printing of the 7th sheet and the state changes from the printing of the 8th sheet, naturally the control amount so far is the correction of the operation amount. Cannot be performed accurately, and the comparison result between the control amount and the target quality exceeds the allowable value. The state quantity comparator 26 compares the time information of the control case in the latest cluster, which is the control case group, with the time information of the control case newly written in the control case memory 25, and if it is within a predetermined time, It is determined that the states are similar, and if they are separated by a predetermined time or more, it is determined that the states are not similar. In this case, assuming that the time from the 3rd sheet to the 8th sheet has not elapsed so much, the above determination results are similar to each other, and the control example of the 8th sheet is 1 to 3 sheets as shown in FIG. It is classified into the first cluster created by eyes. Similarly, since the 9th and 10th prints also exceed the allowable value, those control cases are classified into the first cluster. As described above, in the first embodiment, even if a state change such as an environment occurs, the control cases after that are classified into the latest cluster, so the control corresponding to after the state change (8th sheet or later) is performed. No rules are created.

On the other hand, in the third embodiment, when the control amount exceeds the preset allowable value, when the control amount of a certain (previous) print exceeds the allowable value, The control cases are acquired both when the print control amount falls within the allowable value. In FIG. 24, the non-convergence determination unit 46 stores “0” for the first to third sheets because the preset allowable value is exceeded. Since the state quantity comparator 47 exceeds the allowable value for the first to third sheets, it is acquired as a control case and classified as a cluster, as in the first embodiment. Then, the control rule is extracted when the three control cases are acquired. Therefore, in both cases of the first and third embodiments, the first cluster is created when the number of printed sheets is 1 to 3, and 3
Since three control cases have been acquired by the time of the first print, a control rule is created based on the first cluster.

Next, the non-convergence judging section 46 stores "1" because the value is within the permissible value for the fourth print. The same applies to the fifth to seventh prints. Next, as described above, when the state of the environment or the like changes immediately after the printing of the seventh sheet and the state changes from the printing of the eighth sheet, it goes without saying that the control rules so far accurately correct the operation amount. Therefore, the comparison result between the control amount and the target quality exceeds the allowable value. At this time, the non-convergence determination unit 4
Since 6 is equal to or more than the allowable value, it is stored as "0", and the change from "1" to "0" is detected and a non-detection signal is supplied to the state quantity comparator 26. On the other hand, the state quantity comparator 47 creates a new second cluster different from the first cluster, and classifies the control quantity for the eighth sheet into the second cluster.

Similarly, in the subsequent 9th and 10th prints, the optimum control rule cannot be created for this situation and the existing control rule is used. It exceeds the allowable value. Therefore,
The control cases in the 9th and 10th prints are sequentially classified into the second cluster. Then, by collecting three control cases, a new control rule is created based on the second cluster. As a result, the control rule created in the second cluster can be effectively used for subsequent printing, so that the comparison result between the control amount and the target quality falls within the allowable value. As described above, in the third embodiment, when the current control amount exceeds the allowable value, and when the immediately preceding control amount is within the allowable value, the control case in which the allowable value is exceeded is changed to another case. Since it is classified into the newly created cluster, it is possible to create a control rule that can appropriately respond to changes in the current situation.

(2) Case of Temporary Change FIG. 25 is a conceptual diagram showing an example of acquisition of control cases according to the first embodiment, and FIG. 26 is an example of acquisition of control cases according to the third embodiment. It is a conceptual diagram which shows. First, in FIG. 25, as in the case of FIG. 24 described above, a first cluster is created when the number of prints is 1 to 7, and control is performed based on the first cluster up to the first to seventh prints. A rule is created. In the 4th to 7th prints, the result of comparison between the control amount and the target quality does not exceed the allowable value, and therefore no cluster is created.

Next, when the eighth print exceeds the allowable value due to noise or the like, the state quantity comparator 47 causes the time information of the control case in the latest cluster which is the control case group,
The time information of the control case newly written in the control case memory 25 is compared, and if it is within a predetermined time, it is determined that the state is similar, and if the time is more than the predetermined time, it is determined that the state is not similar. In this case, assuming that a short time has not elapsed from the printing of the third sheet to the printing of the eighth sheet, the above determination result is similar to the state, and the control example of the eighth sheet is 1 to 3 sheets as shown in FIG. It is classified into the first cluster created by eyes.

In the ninth print, the operation amount is calculated based on the control rule created in the first cluster and the other control rules stored so far, but the control amount is still affected by noise, Since the allowable value is exceeded without being overwhelmed, it is classified into the first cluster. However, the cause of exceeding the allowable value is an abnormality due to noise, etc.
The control example of the 10th print is within the allowable value. If the 11th print also exceeds the allowable value due to an abnormality due to noise or the like, the 11th control case is also classified into the first cluster. As described above, in the first embodiment, even when an abnormal control amount is temporarily generated due to noise or the like, those control cases are classified into the first cluster, and thus an error is included. .

On the other hand, in the third embodiment as well,
Similar to FIG. 24 described above, up to the seventh sheet, the first cluster is created when the number of printed sheets is the first to third sheets, and the control rule is created based on the first cluster. And 4
In the 7th print, the cluster is not created because the comparison result between the control amount and the target quality does not exceed the allowable value. Further, as shown in FIG. 26, the non-convergence determining unit 46
Stores "0" for the 1st to 3rd sheets because it exceeds the preset allowable value, and stores "1" for the 4th to 7th sheets because it is within the allowable value. To do.

Next, as described above, when the eighth print exceeds the permissible value due to noise or the like, the non-convergence determination unit 4
6 stores “0”, detects the change from “1” to “0”, and supplies a non-detection signal to the state quantity comparator 47. On the other hand, the state quantity comparator 4
Upon receiving the non-detection signal, 7 creates a new second cluster different from the first cluster, and classifies the control amount for the eighth sheet into the second cluster. Then, in the ninth print, the operation amount is calculated based on the control rule created in the first cluster and the other control rules stored so far, but there is still the influence of noise, and the control cannot be completed. It exceeds the allowable value. Therefore, the non-convergence determination unit 46 stores “0”.

Next, in the tenth print, the non-convergence judging section 46 stores "1" because it falls within the allowable value as described above. Then, when the permissible value is exceeded again due to noise in the eleventh print, the non-convergence determination unit 46
Stores “0”, detects the change from “1” to “0”, and supplies a non-detection signal to the state quantity comparator 47. On the other hand, the state quantity comparator 26
Receives a non-detection signal to create a new third cluster different from the second cluster, and classifies the 11th control amount into the third cluster. Since the second and third clusters described above are clusters created due to noise or the like, there are few control cases (three or more) and no control rule is created. Thus, the third book
In the embodiment, even if the control amount temporarily exceeds the allowable value due to noise or the like, these are not classified into the cluster (first cluster) created by the state change.
It is not necessary to add an abnormal control case to a normal cluster. Further, since the control rule is not calculated from the cluster including the apparently abnormal control case due to noise or the like, it is possible to secure a more correct control rule that is not affected by the control case including the noise.

(3) When the state changes due to the environment, etc. FIG. 27 is a conceptual diagram showing an example of acquisition of the control case according to the first embodiment, and FIG. 28 shows the control case according to the third embodiment. It is a conceptual diagram which shows an acquisition example. 27 and 28,
The change in the control amount for each number of prints is the same, but the case where the state such as the environment changes is shown. First,
In the first embodiment, as shown in FIG. 27, the state has changed immediately after the printing of the seventh sheet due to the environment,
The control amounts for the eighth, ninth, and eleventh sheets are classified into the first cluster created for the first to third sheets. Therefore, as described above, the control rule created based on the first cluster is changed by the control cases of the eighth, ninth, and eleventh sheets.

On the other hand, in the third embodiment, as shown in FIG.
As shown in, the convergence judgment by the non-convergence judgment unit 46 is 7
Since the first print changes to "0" in the eighth print, a new second cluster is created at this point, and the control amount for the eighth print is added to the second cluster. being classified. Next, the control amount for the ninth sheet exceeds the allowable value, and is therefore classified into the second cluster. At this time, the non-convergence determining unit 46 stores “0”. And 1
The non-convergence determination unit 46 stores “1” because the 0th print fits within the allowable value. Furthermore, the 11th print exceeds the allowable value again, so the non-convergence determination unit 46 stores "1". Also, from the 10th print to 11
Since "1" is changed to "0" in the print of the first sheet, a new third cluster is created at this point, and the control amount of the eleventh sheet is classified into the third cluster. Second above
And in the third cluster, there are two acquired control cases,
Or, since there is only one, no new control rule is created.

As described above, in the third embodiment, compared to the first embodiment, when the state changes due to the environment or the like, even if a control case exceeding the allowable value occurs thereafter, those control cases are Since it is not classified into the first cluster before the state change, it is not necessary to add an abnormal control case to a normal cluster.

I: Effect of Third Embodiment The above is the configuration and control procedure of the third embodiment. The operation and effect of the third embodiment will be described below. (1) When the controlled variable exceeds the allowable value for the target quality,
If the immediately preceding control amount has converged, a new cluster is created, and the control amount exceeding the above allowable value is classified into the new cluster, and a sufficient number (3) of control cases is acquired. , It is possible to create a control rule that can appropriately respond to changes in the current situation. (2) In addition, even if a control amount temporarily exceeds the allowable value due to noise or the like, these abnormal control cases are categorized as control cases consisting of the correct control amounts acquired and created so far. Need not be added to another cluster. (3) Furthermore, it is considered that only apparently abnormal control cases due to noise and the like are classified into a new cluster, and noise (sudden ones) does not continue to occur thereafter. A sufficient number of control cases (for example, three) to calculate the control rule are not collected. Therefore, the newly created cluster is not stored here. As a result, since the control rule is calculated only from the clusters that have been stored so far, it is possible to obtain a more correct control rule that is not affected by noise, and control based on the control rule that can respond more appropriately to the current changes. It can be performed.

J: Modifications In the above-described first to third embodiments, various modifications as described below are possible. (1) In the above-described first to third embodiments, the image output unit IOT is an example of a single color laser printer, but the application of the present invention is not limited to this, and a multicolor laser is used. Whether it is a printer or an analog type copying machine, it is possible to achieve the same effect. Furthermore, the present invention is applicable not only to the electrophotographic system but also to an image output unit such as an inkjet system.

(2) The sensors used in the first to third embodiments are merely examples, and in order to obtain the effects of the present invention, any sensor can be used as long as it can accurately measure the density of the developing patch. A system type is also acceptable. Also, the object to be monitored may be any object as long as it has a high correlation with the final image density. For example, even if any of the developed image, the transferred image, and the fixed image is monitored, It may be associated with the image density.

(3) In the first to third embodiments, the density of the development patch is solid (halftone dot coverage 10
0%) density patch and highlight (halftone coverage 20
%) Two types of density patches are adopted, but this is not limited to these two types, and for example, only the density corresponding to 50% halftone dot coverage may be controlled,
More gradation points may be controlled by using more kinds of patches. However, when it is desired to control each gradation point independently, it is necessary to prepare the number of kinds of control parameters corresponding to the number of gradation points.

(4) Although the density of the development patch is monitored in the first to third embodiments, the reproduced image density (actual density) may be directly monitored, and other alternatives may be used. You may monitor a physical quantity.

(5) In the first to third embodiments, the developing bias set value is fixed or, for example, the laser power set value is fixed, and the grid voltage set value of the scorotron charger and the developing bias are control parameters. Can also be adopted as. This is because the developing bias also has a high correlation between the solid density and the highlight density. Therefore, as another combination, the grid voltage setting value of the scorotron charger can be fixed and the laser power setting value and the developing bias can be adopted as the control parameters.

Alternatively, it is also possible to control three gradation points by using three of the laser power setting value, the developing bias setting value and the scorotron charger's crid voltage setting value. That is, for example, the dot coverage is 100%, 50%, 20%.

(6) In the first to third embodiments, description has been made on the assumption that the two-component developing system is used. In this case, the toner concentration in the developer, that is, the mixing ratio of the toner and the carrier is related to the development concentration. For this, the toner supply amount is proportional to the number of pixels of the image to be output. The concentration is controlled. Alternatively, as another method, the toner density may be controlled to a substantially constant value by monitoring with a commercially available magnetic sensor or optical sensor that has been generally used conventionally.

In the first to third embodiments, the toner density is not positively variably controlled to control the image density to a desired value. Therefore, the toner density is kept substantially constant. Is enough. This is because even if the toner density changes to some extent, it can be absorbed by the setting of the above-mentioned control parameters (scoro setting value, LP setting value).

On the other hand, if the one-component developing method is used, since the toner density is always 100%, it does not directly affect the image density. Therefore, conventional toner amount management such as empty developer detection is possible. It is enough.

(7) The control case in the first embodiment is
It consists of three elements: state quantity, set value (operation quantity), output value (control quantity), and since time is used as the state quantity, it is a temperature sensor or humidity sensor for monitoring temperature and humidity. Need not be used.

Further, as described above, as another example, the number of prints after the power is turned on that day, the cumulative number of prints after the image forming apparatus is operated, or the continuous print after the print button is pressed. The number of prints or the like may be used as a substitute value for the state quantity. This is because in the case of the electrophotographic system, the characteristics of the photosensitive member are extremely dependent on the number of prints. For example, in a series of image forming operations, the image density of the first few images and the image density after that greatly differ. This is because there is a forming device. In such a case, setting the substitute value of the state quantity as the number of prints is very effective. If the third embodiment described above is applied to this example, it is not necessary to preset the number of prints, the cumulative number of prints, and the number of continuous prints.

As described above, as the state quantity relating to the image density, it is not always necessary to sense some physical quantity, and an arbitrary state quantity or a substitute of the state quantity is used in accordance with the assumed characteristics of the image forming apparatus. Values can be used.

Of course, if there is a margin in cost and sensor mounting space and it is desired to more accurately grasp the state quantity and improve the control performance by using the temperature sensor and the humidity sensor, other states can be appropriately used. It goes without saying that a sensor for sensing the amount may be provided. Even in such a case, there is no need to add any special processing or change to the present invention.

According to the experiment, the control cases corresponding to the temperature change and the humidity change can be classified as a result without directly using the temperature sensor and the humidity sensor, and the control rules corresponding to the respective cases are automatically created. Therefore, it is not necessary to use a temperature sensor or a humidity sensor in the case of a normal image forming apparatus.

(8) Next, a method of acquiring a control rule different from those of the first to third embodiments will be described. That is, it is possible to use a “curved surface” having a higher degree than the “planar surface” according to the control case as described above.

Comparing the case of the plane and the case of the curved surface,
If it is a plane, the number of control cases should be at least three, and if there are more cases, statistically averaging can reduce the influence of measurement errors and the like. Moreover, it should be noted that according to the present invention, since the control rules themselves are created in a complementary manner according to the accuracy of the control rules, there is an advantage that the desired overall control accuracy can be obtained. This state is shown in FIG. 29 as a schematic diagram.

That is, the plane corresponds only to the area that can be covered by a certain control case plane, and another control case plane is newly generated for the area beyond that plane. The generation of this control rule is as described above.
The control is automatically continued until the desired control accuracy is obtained.

Further, in FIG. 30, the dimension is lowered by one for easy understanding (that is, the plane is a straight line and the curved surface is a curved line).
Since it is expressed, or between one control rule and an adjacent control rule, synthesis is performed according to the “fitness” of both control rules, so for example, at a position just in the middle of both, the fitness of both Is 50% each, a plane having an average inclination of the two is virtually created, and is translated so as to coincide with the actual physical phenomenon. Therefore, it is shown that the control can be performed in exactly the same state as the presence of the smooth curved surface.

On the contrary, when a high-order approximate curved surface is used, one control rule can cover a wide area, but many control rules are required to set up one control rule. Therefore, the response speed becomes slower.

Therefore, what kind of policy, that is, the control rule is quickly determined by a simple plane, and a large number of planes are combined as needed, a highly accurate control rule is expressed by a high-order curved surface from the beginning. However, whether or not the number of curved surfaces is small may be determined depending on what control characteristics are desired for the image forming apparatus or the user, and the present invention is applicable to both cases. is there.

(9) The development patch creation and its sensing may be performed in exactly the same manner as the patch creation and detection performed in the prior art, and there are no special restrictions in constructing the present invention. That is, a patch may be created each time an image is formed, as in the conventional example, or a patch may be created only before or after a series of jobs. Alternatively, a patch may be created for each fixed number of sheets and for each fixed time.

Generally, in the patch creation and its detection, the higher the frequency, the more accurate the grasping of the reproduced state of the image density is, but the disadvantage is that the toner is consumed to that extent, so that the image The optimum patch creation frequency may be adopted according to the specifications and purpose of the forming apparatus.

(10) State quantity Whether the error that exceeds the permissible error amount is caused by an essential physical change, or the measurement accuracy of the past cases stored so far is not accurate. You have to judge whether it was because it was insufficient. And if it is due to an intrinsic physical change, the control rule itself must be newly created. -On the other hand, if the control rule was not well applied due to the lack of accuracy of the past cases (measurement error was large, etc.), there was no substantial physical change. It is more effective to use the cases together to statistically reduce the effects of measurement errors that individual cases have. For example, as in the start-up of the first to third embodiments, rather than determining the control case plane with only a minimum of three cases, a statistical calculation method such as the least square method is used from a larger number of control cases. It can be expected that a more accurate control rule can be obtained by reducing the error.

Therefore, in the present invention, in order to make such a distinction, the state quantity which is an element of the control case is used. In the first embodiment described above, the physical quantity sensing is required as the state quantity. No control time is used. Therefore, it is determined whether or not they are considered to be in the same state based on the degree of coincidence of control times in each case, that is, if the dates are different, the states such as temperature and humidity are considered to be different, and control cases that are close in time are the same. Considered to be under condition.

Here, the setting of the range "close in time" may be optimally made in consideration of the required specifications of the image forming apparatus, the assumed usage environment of the user, and the like. For example, in an office in an area where the outside temperature fluctuates greatly depending on the season, even in the same day, there is an electrostatic charge between the early morning when the air conditioning starts to work and the day when the air conditioning is fully functioning. For an electrophotographic image forming apparatus that creates an image by various mechanisms, the environmental conditions are greatly different.

In order to deal with such a situation, for example, cases within 1 hour in the morning and within 3 hours in the afternoon are considered to be in the same state, or before 10 am and after 10 am. You can deal with it by treating it as another state.
Such classification can be set as finely as necessary. Alternatively, if such a division becomes too fine and too complicated, the temperature / humidity sensor is used as described above, and those measured values are also included as one of the state quantities of the control case. It is also possible to do so.

Here, for the sake of understanding, a relatively simple first embodiment corresponding to only daily fluctuations will be described. That is, control cases having the same date can be classified by assuming that they are in the same state.

Now, if the control result exceeds the allowable error amount,
As described above, the control case is taken in, and the state quantity of the control case group in which the control rule used for the control is extracted, that is, the control time is examined. In order to deal with daily fluctuations, whether or not the dates are the same is the criterion. That is, if the current control case and the control case group from which the control rule is extracted have the same date, the control rule is improved by adding the present control case to the control case group.

On the contrary, if the dates are different, it is considered that the control rule itself has essentially changed, and the creation of a new control rule is started. That is, n + 1 or more control cases are stored from that time point, and a new control case plane, that is, a new control rule is acquired when n + 1 or more new control cases are obtained. It should be noted that, at each control time when the number of new control cases is n + 1 or less, even if the control rule is updated by tentatively substituting the most recent one of the past control cases. Good.

Whether or not n + 1 or more control cases are to be sampled is whether or not the result of controlling a new control case plane created from n + 1 control cases as a control rule satisfies the above-mentioned tolerance. Can be used to make a unique decision. In other words, if the tolerance is satisfied, the control rule is obtained to enable sufficiently accurate control.If the tolerance is not satisfied, the number of control cases is increased and the control cases are increased. Judge that it is necessary to improve the accuracy of the plane. At this time, as described above, the state quantities of the (n + 1) th and later control cases and the (n + 1) th control cases, that is, the dates are compared, and it is determined whether or not they are on the same day.

Needless to say, such improvement of the control rule or acquisition of a new control rule does not depend on the criterion for determining the same state quantity. That is,
Whether you manage by date or by hour,
Even if it is managed in a certain unit of temperature change width, or even in the unit of the number of prints, it is possible to unambiguously determine whether or not to regard the same state quantity as a unit. You can improve the rules or obtain new control rules.

(11) In the memory management in the first to third embodiments, the old data is erased, but the control rule accumulates the conformance degree every time it is used for control, The presence or absence of erasure may be determined using the cumulative fitness obtained as a result. That is, since the control rule with a small cumulative conformity is a control rule with a low degree of use, it may be deleted in order from such a rule.

Next, this erasing method will be described more forgivingly. First of all, it should be noted that when determining the importance of a control rule, it is not always good to perform it based only on old and new. This is because the image forming apparatus using the electrophotographic process is greatly affected by the temperature and humidity environment.For example, when considering the relationship with the four seasons in Japan, when performing control in summer, extraction in winter, which is six months ago, is performed. This is because the control rule extracted in the summer one year ago may be more effective than the control rule described above.

Therefore, it is meaningful to make a more accurate judgment, not to judge the importance only by the old and new data. Therefore, each time it is used for control, its fitness is accumulated, and the cumulative fitness obtained as a result is used to judge the importance.

However, in this case, a problem arises when it is attempted to make a judgment only based on the cumulative goodness of fit. This is because the newer (newer-to-be-made) control rule has a smaller cumulative fitness, and the older control rule has a smaller fitness each time (thus, even if the importance is low), but the cumulative count is high. Therefore, there is a high possibility that the cumulative fitness will increase.

On the other hand, the external environment such as temperature and humidity, which greatly affects the image forming apparatus, changes greatly depending on the four seasons, especially in Japan. Therefore, it is desirable to consider the accumulative situation of the goodness of fit over the period in which the similar environmental condition of the season in which the control rule is extracted continues.

For example, it is assumed that the environment is similar to the time (season) when the control rule was extracted for three months ago,
The control rule with the lowest cumulative fitness is deleted as the rule with the least importance among the rules that have passed three months or more. In this way, the newly extracted rule will not be deleted unnecessarily. That is, if it is within three months after the extraction, it is common that the frequency of application is low and the cumulative fitness is low, but such new rules will not be unnecessarily deleted.

A more detailed description will be given. If the control rule is expressed by the first-order approximation by the least squares method, the elements of the control rule (or cluster) are stored as shown in Table 2.

[0202]

[Table 2]

That is, the control rule of Table 2 is: solid density = al × LP set value + a2 × scoro set value + a
3 Highlight density = bl × LP setting value + b2 × scoro setting value + b3 Coefficients al, a2, a3, bl, b
2, b3 and the created year, month, day, minute, second, that is, the last (newest) in the control case group that extracted the rule
Each time the control case is generated, it is composed of the respective elements of the cumulative fitness and the cumulative fitness is added every time the control rule is used for control and the fitness is calculated.

Incidentally, Table 2 corresponds to Table 1, and Case 1
3 to control rule 1 and cases 4 to 6 to control rule 2
An example of the extracted data is shown.

Since the control rule (or cluster) is described by the above-described method, when the remaining memory capacity is insufficient, first, three months or more based on the created year / month / day / hour / minute / second. It can be judged whether or not it has passed, and then when three or more power months have passed, it is possible to judge the importance by comparing with the cumulative fitness of other rules that have passed more than three power months. It has become.

If the memory capacity prepared in advance is small, or if new control rules are created very frequently, the remaining memory capacity may run short before the three power months. obtain. In that case, as described above, the oldest data is simply deleted. Therefore, in any case, a memory for storing the latest data can be secured.

(12) In the control rule searcher, when the degree of conformity of the control rules is obtained and combined, the degree of conformity smaller than a predetermined value (10% or 20%) is ignored, and the remaining control rules are The fitness may be obtained again, and these may be combined for control. By performing such control, it is possible to avoid the influence of a control rule that is not related, so that it is possible to perform more precise control.

(13) In the first to third embodiments, the image density is controlled, but instead of this, for example, line width, sharpness, gradation, etc. may be controlled.

[0209]

As described above, according to the invention described in claims 1 to 13, the number of sensors can be reduced as much as possible, and the cost can be reduced. Further, the engineer can automatically perform accurate control without grasping the influence of various environmental conditions and deterioration with time in advance, and the man-hours for development can be greatly reduced.

Further, even when a huge number of image forming apparatuses are put on the market and each is used in various ways or parts are replaced at any time, the image density control performance of each one is increased. Can always be secured automatically.

Further, in the inventions according to claims 5 and 6, it is possible to directly instruct and set the required control precision itself to the control device, and the control device itself automatically sets so as to satisfy the required precision. By configuring so as to correspond to (1), it is possible to eliminate the cost increase for the accuracy improvement and the increase of the development man-hours at all.

Further, in the invention described in claims 8, 9 and 10, while effectively utilizing the limited memory capacity,
Control can be performed.

Further, in the invention described in claims 16 and 17, when the control rule is extracted, not only the control rule extracted in the past but also the latest control rule which grasps the present situation is used to improve the operation amount. Can be determined optimally.
Further, even if the control amount falls within the allowable value before the number of control cases necessary for newly creating the cluster are collected, the control case having the information of the actual state change can be taken into the cluster. Therefore, since the number of control cases taken in the same cluster increases, the accuracy of the control rule extracted based on the cluster can be further improved.

Further, in the invention described in claims 19 and 20, if a sufficient number of control cases are acquired, it is possible to create a control rule that can more appropriately respond to the current change. In addition, even if a control amount temporarily exceeds the allowable value due to noise, etc., these abnormal control cases are classified into normal clusters that classify control cases consisting of the correct control amounts acquired and created so far. No need to add. Further, since a sufficient number of control cases for calculating the control rule do not gather in a cluster including a clearly abnormal control case due to noise or the like, the control rule is not calculated from the cluster and the control including the noise occurs. It is possible to secure a more correct control rule without the influence of the case, and it is possible to perform control based on the control rule capable of more appropriately responding to the current change.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a control unit in a first embodiment of the present invention.

FIG. 2 is a configuration diagram showing an outline of an image output unit of the first embodiment.

FIG. 3 is a schematic diagram showing a density patch in the first embodiment.

FIG. 4 is a schematic view showing a position where a density patch is created in the first embodiment.

FIG. 5 is a waveform diagram showing an example of an output signal of the developing density sensor in the first embodiment.

FIG. 6 is a conceptual diagram showing a case plane at the time of start-up of the first embodiment.

FIG. 7 is a conceptual diagram showing an inference method for controlling solid and highlight densities in the first embodiment.

FIG. 8 is a conceptual diagram showing how a new cluster is created from a plurality of past clusters by using the goodness of fit in the first embodiment.

FIG. 9 is a block diagram showing a configuration of a control unit according to a second exemplary embodiment of the present invention.

FIG. 10 is a block diagram showing an outline of allowable value comparison according to the first embodiment.

FIG. 11 is a conceptual diagram for explaining control case acquisition and cluster creation according to the first embodiment.

FIG. 12 is a block diagram showing an outline of allowable value comparison according to the second embodiment.

FIG. 13 is a conceptual diagram for explaining control case acquisition and cluster creation according to the second embodiment.

FIG. 14 is a conceptual diagram showing an example of acquisition of a control case according to the first embodiment.

FIG. 15 is a conceptual diagram showing an example of acquisition of a control case according to the second embodiment.

FIG. 16 is a block diagram showing a configuration of a control unit according to a third embodiment of the present invention.

FIG. 17 is a block diagram showing an outline of allowable value comparison according to the first embodiment.

FIG. 18 is a conceptual diagram for explaining control case acquisition and cluster creation according to the first embodiment.

FIG. 19 is a block diagram showing an outline of allowable value comparison when the number of prints and operating time are used as state quantities.

FIG. 20 is a conceptual diagram for explaining control case acquisition and cluster creation when the number of prints and operating time are used as state quantities.

FIG. 21 is a block diagram showing an outline of allowable value comparison according to the third embodiment.

FIG. 22 is a conceptual diagram for explaining control case acquisition and cluster creation according to the third embodiment.

FIG. 23 is a conceptual diagram for explaining control case acquisition and cluster creation according to the first embodiment.

FIG. 24 is a conceptual diagram for explaining control case acquisition and cluster creation according to the third embodiment.

FIG. 25 is a conceptual diagram for explaining control case acquisition and cluster creation according to the first embodiment.

FIG. 26 is a conceptual diagram for explaining control case acquisition and cluster creation according to the third embodiment.

FIG. 27 is a conceptual diagram for explaining control case acquisition and cluster creation according to the first embodiment.

FIG. 28 is a conceptual diagram for explaining control case acquisition and cluster creation according to the third embodiment.

FIG. 29 is a schematic diagram showing how a control rule consisting of an arbitrary curved surface can be approximately represented by a plurality of control rules extracted by a plane in the present invention.

[Fig. 30] Fig. 30 is a schematic diagram showing a state in which approximation accuracy is increased as much as necessary by substituting a straight line for a control rule with a straight line and a curved surface with a curve and connecting adjacent clusters with a goodness of fit. FIG. 11 is a functional block diagram for explaining the operation based on the allowable value comparison according to the first embodiment.
FIG. 12 is a conceptual diagram showing the relationship between the control case and the cluster according to the first embodiment.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 Laser output part (image state changing means) 3 Charger (image state changing means) 10 Development density sensor (detection means) 15 Grid power source (image state changing means) 16 Light intensity controller (image state changing means) 24,44 Density comparator (Comparison means) 25 Control case memory (control case storage means) 26, 45, 47 State quantity comparator (cluster storage means) 27 Cluster memory (cluster storage means) 28 Control rule calculator (control rule extraction means: cluster-specific control rules) Extraction means) 29 Control rule memory (operation amount calculation means: control rule storage means) 30 Control rule retrieval device (operation amount calculation means) 31 Operation amount correction calculator (operation amount calculation means) 32 Operation amount memory (operation amount calculation means) ) 43 convergence determination unit 46 non-convergence determination unit

Claims (20)

[Claims] 1. An image quality changing means for changing the quality of an output image according to an operation amount, a control case storing means for storing a control case of an output image, and a plurality of control cases stored in the control case storing means. A control rule extracting means for extracting a control rule from the detection means, a detecting means for detecting the output image quality and outputting the detection result as a control amount, and a control rule extracted by the control rule extracting means. And an operation amount calculating unit for calculating a new operation amount so as to obtain a value corresponding to the target quality, and the operation amount calculated by the operation amount calculating unit is supplied to the image quality changing unit. Image forming apparatus. 2. An image quality changing means for changing the quality of an output image according to a command value, and control cases of the output image, which have similar state quantities of the image forming apparatus, are collected and stored as a cluster. Cluster storage means, cluster-specific control rule extraction means for extracting control rules for each cluster stored in the cluster storage means, detection means for detecting output image quality and outputting it as a control amount, and for each cluster Using each control rule extracted by the control rule extraction means, a control amount calculation means for obtaining a new control amount so that the control amount becomes a value corresponding to the target quality, and the control amount output means An image forming apparatus which supplies the manipulated variable thus obtained to the image quality varying means. 3. The control amount calculation means determines the suitability of each control rule extracted by the cluster-specific rule extraction means with respect to the immediately preceding control case, and weights each control rule according to the suitability. Averaged,
The image forming apparatus according to claim 2, wherein a new control amount is obtained using the result. 4. The control amount calculation means calculates a reciprocal of a distance between an n-dimensional plane indicating a control rule and a coordinate point indicating a control case immediately before in a coordinate space in which each control rule is described. 4. The image forming apparatus according to claim 3, wherein each control rule is standardized to obtain the compatibility. 5. The control amount calculation means again determines the conformance for other control rules except for the control rules whose conformity is less than or equal to a predetermined value, and weights these control rules according to the conformity. The image forming apparatus according to claim 3, wherein the averaged result is used. 6. The control amount calculating means again determines the conformance for other control rules except for the control rules whose conformity is less than or equal to a predetermined value, and weights these control rules according to the conformity. The image forming apparatus according to claim 4, wherein the averaged result is used. 7. The control amount and the target quality are compared, and only when the comparison result exceeds a preset allowable value,
The image forming apparatus according to claim 1, further comprising a comparison unit configured to store the control amount in the control case storage unit and use the control amount in the control from the next time onward. 8. The control amount is compared with the target quality, and only when the comparison result exceeds a preset allowable value,
3. The image forming apparatus according to claim 2, further comprising a comparison unit configured to additionally store the control amount in a corresponding cluster in the cluster storage unit so that the control amount can be used for subsequent control. 9. When the remaining storage capacity becomes less than a predetermined amount as a result of additionally storing the control cases, the oldest control case in the control case storage means is erased. The image forming apparatus according to claim 7. 10. When the control case is additionally stored and the remaining storage capacity is less than a predetermined amount, the oldest one of the clusters in the cluster storage means is erased. 9. The image forming apparatus according to item 9. 11. A control rule storage unit for storing the control rule together with its creation time information and updating and storing the cumulative value of the goodness of fit of each control rule, the storage capacity of the control rule storage unit When the remaining amount is less than the predetermined amount, among the control rules already stored, the one which is created before the predetermined time and has the smallest cumulative value of the fitness is selected from the storage means. The image forming apparatus according to claim 3, wherein the image is erased. 12. A control case storage unit for storing a control case of an output image, wherein the cluster storage unit is similar to the control cases in the control case storage unit and has a similar state quantity of the image forming apparatus. 3. The image forming apparatus according to claim 2, wherein the control case storage means deletes the cases that are constituent elements of the cluster when the clusters are collected and stored as a cluster. 13. The control case according to claim 1, wherein the control case is made up of three types of an operation amount, a control amount, and a state amount relating to a state in which the device is placed. Image forming apparatus. 14. The control rule is composed of n + 1 axes indicating the manipulated variables set according to n controlled objects and n + 1 axes indicating the controlled variables for the controlled objects.
The image forming apparatus according to claim 1, wherein the image forming apparatus is extracted as a least square error nth plane of a plurality of coordinate points indicating a control case in a dimensional space. 15. The image forming apparatus according to claim 1, wherein the output image quality to be controlled is image density. 16. Even when the comparison result of this time by the comparing means falls within the preset allowable value immediately after the comparison result of the previous time by the comparing means exceeds the preset allowable value, 9. The image forming apparatus according to claim 7, wherein a control amount and an operation amount used for the control are stored in the control case storage unit so that the control case storage unit can use the control amount for the next and subsequent control. 17. The cluster storage means stores the current comparison result by the comparison means within a preset tolerance value immediately after the previous comparison result by the comparison means exceeds a preset tolerance value. In this case, the control case of this time is classified into the latest cluster.
The image forming apparatus described. 18. A convergence detection unit that detects a change when the comparison result falls within a preset tolerance value immediately after the comparison result exceeds a preset tolerance value, based on the comparison result by the comparison unit. The image forming apparatus according to claim 16, further comprising: 19. The cluster storage means, immediately after the previous comparison result by the comparison means falls within a preset allowable value, the current comparison result by the comparison means exceeds the preset allowable value. In this case, the control case of this time is classified into a new cluster.
Alternatively, the image forming apparatus according to item 8. 20. Non-convergence detection for detecting a change when the comparison result exceeds the preset tolerance value immediately after the comparison result falls within the preset tolerance value, based on the comparison result by the comparison means. 20. The image forming apparatus according to claim 19, further comprising means.