JPH0424743A - Learning method for prewire neuro of rule part - Google Patents

Abstract

PURPOSE:To execute the readjustment of a whole rule part prewire neuro which fits the learning result of weighing by learning the whole rule part prewire neuro after learning the weight of at least one part of an antecedent membership function realization part, rule part, and a consequent membership function realization part. CONSTITUTION:A weight initial value with preliminarily held knowledge intro duced are set for the antecedent membership function realization part 10, a rule part 12 and the consequent membership function realization part 14, and it is constituted so that the weight of the consequent membership function realization part 14 is finally learned after at least one of the weights of the antecedent membership function realization part 10, the rule part 12 and the consequent membership function realization part 14 is learned. Thus, the readjust ment of the weight of the whole rule part prewire neuro which fits the learning result of the weight of at least one of the membership function realization part 10, rule part 12 and the consequent membership function realization part 14 learned in the previous step.

Description

[Detailed Description of the Invention] [Summary] A control output corresponding to an input is generated according to a hierarchical network structure that realizes a fuzzy control rule, which is composed of an antecedent membership function implementation unit, a rule unit, and a consequent membership function implementation unit. Regarding the learning method of the rule part pre-wired neuro, we have developed an antecedent membership function realization part, a rule part and a consequent part membership function realization part, with the aim of efficiently learning to perform control suitable for the target system. After setting an initial weight value that introduces the knowledge held in advance for each of , first learning the weight of at least one of the antecedent part membership function realization part, the rule part, and the consequent part membership function realization part, The system is configured to finally learn the weights of the consequent membership function realization unit.

Incidentally, the initial setting of the weight values for the part whose weights are to be learned first may be performed using random numbers, or may be performed using a combination of knowledge held in advance and random numbers.

The present invention relates to a learning method for a pre-wired neuron in the rule section that combines fuzzy control and a neural network.

In information processing using neurons, learning data consisting of a set of input patterns of the target system and desired output patterns for that input pattern is used in a hierarchical format.
It is presented to the physical network to learn and perform adaptive processing. In particular, a processing method called pack propagation method is attracting attention due to its high practicality in this learning process.

On the other hand, fuzzy theory was proposed by 2adeh in 1960, and is a theory that models the process of thinking and judgment based on ambiguity that humans make, such as whether the temperature is "low" or "high." This fuzzy theory is mathematically based on fuzzy set theory, which introduces membership functions to represent ambiguity.

As an application of fuzzy theory, in 1974 Maqmda
Various types of fuzzy control have been used since ni was first used to control steam engines.

Since the system that performs this fuzzy control handles uncertain things, it is necessary to have a configuration that can adaptively respond to adjustments and changes in the control algorithm.

[Prior Art] Generally, the following procedure is taken to construct a fuzzy control system.

■Obtain ambiguous control rules such as ``If the temperature is high, reduce the fire'' held by experienced operators.

■ Quantify the meaning of the words (propositions) in the control rules in the form of membership functions, and convert the control rules into rlF-THEN~
” Described using a type rule expression.

■Conduct inspections through simulations and on-site tests.

■Evaluate inspection results and improve membership functions and rules.

To explain in more detail, the fuzzy control rule is, for example, [item 1, is big and! 2 is sma
It is written as ll thenT+ is bigJ. Here, the IF section is called an antecedent section, and is a section that describes conditions regarding control state quantities (control inputs) such as temperature. Further, the THEN section is called a consequent section, and is a section that describes the conditions regarding the control operation amount (control output) of the operating terminal, etc.

Fuzzy control manages multiple fuzzy control rules prepared as the control logic of the target system, and also manages the meaning of ambiguous linguistic expressions such as "large" or "small" written in each fuzzy control rule. Quantify and manage it as a ship function.

The control procedure of the completed fuzzy control system is as follows.

■When a controlled state quantity such as temperature is input from the controlled object, the truth value is calculated according to the membership function of the antecedent part that has been set.

(2) Next, in accordance with the antecedent computation such as selecting the minimum value, a process is executed to determine the value to be applied to the consequent in each fuzzy control rule. For example, r If is b
Truth value of igJ; 0.8[12is smxllJ
When the truth value of = 0.5, the truth value = 0.5 is output as the applied value to the consequent part of the fuzzy control rule according to the antecedent part operation that selects the minimum value. Process.

(2) Next, a process is executed to determine an output application value from the input value (applied value) from the antecedent part to the membership function of the consequent part, which has the same amount of control operation.

For example, with two fuzzy control rules, r y+
Application value of is blgJ = 0.5 "y1+s bigJ
When the applied value=0.6 is obtained, the applied value=0.6 is selected and determined according to the consequent operation that selects the maximum value.

■Subsequently, the membership function of the control operation amount is reduced (or expanded) according to the determined applied value, and the center of gravity of the figure of the function part of the membership function for the same control operation amount is calculated. Then, fuzzy inference is used to calculate and output the control operation amount.

By the way, fuzzy control rules are planned to be generated according to the knowledge possessed by a skilled operator, and it is difficult to establish fuzzy control rules that can realize optimal control from the beginning. Therefore, it is necessary to improve the generated fuzzy control rules through trial and error while evaluating them through simulations and field tests.

Therefore, the inventors of the present application developed a hierarchical network structure that realizes fuzzy control rules by applying neurology.
That is, we have proposed an adaptive fuzzy control system using prewired neuro in the rule section (Patent Application No. 2-602).
5'6, "Hierarchical Network Configuration Data Processing Device and Data Processing System"). That is, the rule part prewire neuro has a hierarchical network configuration that realizes a fuzzy control rule consisting of an antecedent part membership function realization part, a rule part, and a consequent part membership function.

Considering the automatic extraction of fuzzy control rules by the pre-wired neuron of the rule section, the following procedure may be taken.

■Set the membership function of the rule section prewire neuro.

■The rule section prewire neuro is presented with the input pattern of the target system and a set of desirable output patterns for that input pattern, and is made to learn.

■Extract the optimal fuzzy control rule for the target system by analyzing the weights of connections in the rule section.

[Problems to be Solved by the Invention] However, the learning performed for extracting fuzzy control rules for the rule section pre-wired neuron is
The learning method for learning, including the weights of the antecedent membership function part, rule part, and consequent part membership function realization part, has not been determined, and the performance of the roll part prewire neuron has not yet been determined. There is a need to develop efficient learning methods that can bring out the best in students.

The present invention has been made in view of the above situation, and an object of the present invention is to provide a learning method for a rule section pre-wired neuro that can efficiently perform learning for controlling a target system. .

[Means to solve the problem] FIG. 1 is a diagram explaining the principle of the present invention.

First, the present invention provides antecedent membership function realization unit 10.1
Or, it has a hierarchical network structure that realizes a fuzzy control rule consisting of a multi-layered rule section 12 and a consequent membership function realization section 14, where the rule section 12 is an antecedent membership function realization section 10 in the preceding stage and/or a subsequent stage. The consequent part membership function realization unit 14 does not internally connect all the units, but internally connects some of them according to the control rule, and input control state quantities (X+, X2, . . .
The target is the rule section prewire neuro+00 that outputs one or more control operation amounts (Y) corresponding to Xo).

In the present invention, the learning method of the rule section pre-wired neuron 100 is comprised of the following 10 learning methods.

■Claim 1; [First step] Initial setting of weight values based on knowledge that the antecedent membership function realization unit 10 has in advance (shown as "prior knowledge" in the figure) or using random numbers. At the same time,
Initialize weight values based on knowledge held in advance in the rule unit 12 and consequent membership function realization unit 14; [Second process] Initialize weight values in the antecedent membership function realization unit 10 using learning data. Learn the weights; [Third process: Antecedent membership function realization unit 10 using learning data;
The overall weights of the rule section 12 and the consequent membership function realization section 14 are learned.

■Claim 2 [The first process corrule unit 12 initializes the weight values based on knowledge held in advance or using random numbers, and the antecedent membership function realization unit 10 and the consequent membership function realization unit Initialize the weight values based on knowledge held in advance in Step 14; [Second process] Learning the weights of the rule section 12 using the learning data; [Third process] Setting the antecedent using the learning data The overall weights of the part membership function realization part 10, the rule part 12, and the consequent part membership function realization part 14 are learned.

■Claim 3 [The weight values are initialized based on the knowledge held in advance in the first process corrule unit 12 or using random numbers, and the antecedent membership function realization unit 10 and the consequent membership function realization unit Initialize the weight values based on knowledge held in advance in step 14; [Second process learning data to learn the weights of the rule section 12; Third process learning data to set the antecedent. When the weights of the part membership function realization part 10 are learned;
, rule part 12 and consequent part membership function realization part 14
Learn the overall weight of: Configure as follows.

■Claim 4 [First step] Initial setting of weight values based on knowledge held in advance in the antecedent membership function realizing unit 10 or using random numbers, and realizing the rule unit 12 and the consequent membership function. Initialize weight values based on knowledge held in advance in unit 14. [Second process] Learning the weights of antecedent membership function realization unit 10 using learning data;
[Third step: Learning the weights of the rule section 12 using the learning data; [Fourth step] Using the learning data, learn the antecedent section membership function realization section 10, the rule section 12, and the consequent section membership. The entire weight of the function realization unit 14 is learned; it is configured as follows.

■Claim 5 [First step] Initializing weight values based on knowledge held in advance in the antecedent membership function realizing unit 10 and the rule unit 12 or using random numbers, and realizing the consequent membership function. Initialize weight values based on knowledge held in advance in section 14; [Second process] Using learning data, the antecedent membership function realization section 10 and rule section 12
[Third process] Learning the overall weights of the antecedent membership function realizing unit 10, the rule unit 12, and the consequent membership function realizing unit 14 using the learning data; Configure it as follows.

■Claim 6; [The first step is to initialize the weight values based on the knowledge held in advance in the antecedent membership function realizing unit 10 or using random numbers, and also to set the weight values in the rule unit 12 and the consequent membership function. Initialize the weight values based on the knowledge that the realization unit 14 has in advance; [Second process] Learning the weights of the antecedent membership function realization unit 10 using the learning data;
[Third step: Use the learning data to learn the weights of the consequent membership function realization unit 14; [Fourth process] Use the learning data to learn the weights of the antecedent membership function realization unit 10;
The overall weights of the rule section 12 and the consequent membership function realization section 14 are learned.

■Claim 7 [First step] Initial setting of weight values based on knowledge previously held in the rule unit 12 or using random numbers, and realization of the antecedent membership function realization unit 10 and the consequent membership function. Initialize the weight values based on the knowledge that the section 14 has in advance; [Second step] Learning the weights of the rule section 12 using the learning data; [Third step] After using the learning data. Learning the weights of the subject membership function realization unit 14 [Fourth step] Using the learning data, the antecedent membership function realization unit 10. The overall weights of the rule unit 12 and the consequent membership function realization unit 14 are learned: The configuration is as follows.

■Claim 8 U first step] Initial setting of weight values based on knowledge held in advance in the rule unit 12 or using random numbers, and realizing the antecedent membership function realization unit 10 and the consequent membership function. Learning the weights of the weight value realization unit 10 based on the knowledge that the unit 14 has in advance; [Third process] Learning the weights of the rule unit 12 using learning data; [Fourth process] Learning data learning the weights of the consequent membership function realization unit 14 using the learning data; [Claim 9] Learning the overall weight of the function realizing unit 14 [Claim 9] [First step antecedent membership function realizing unit 10 and rule unit 12 to initialize the weight values based on knowledge held in advance or using random numbers] At the same time, the weight values are initialized based on the knowledge that the consequent membership function realization unit 14 has in advance.

[Second process] The weights of the antecedent membership function realization unit 10 and the rule unit 12 are simultaneously learned using the learning data; [Third process: The weights of the rule unit 12 are learned using the learning data. [Fourth process] Learning the weights of the consequent membership function realizing unit 14 using the learning data; [Fifth process) Learning the weights of the consequent membership function realizing unit 10 using the learning data;
The overall weights of the rule section 12 and the consequent membership function realization section 14 are learned.

■Claim 1.0 [First step] The weight values are initialized based on the knowledge held in advance in the antecedent membership function realization unit 10 and the rule unit 12 or using random numbers, and the consequent membership Initialize weight values based on knowledge held in advance in the function realizing unit 14; [Second process] The antecedent membership function realizing unit 10 and the rule unit 12 using learning data
Simultaneously learn the weights of the consequent membership function realization unit 14 using the learning data; [4th process] learn the antecedent membership using the learning data; Function realization unit 10,
The overall weights of the rule section 12 and the consequent membership function realization section 14 are learned.

In the above learning method (① to 0), the initial weight values in the first process of learning the weights are set by initializing the weights that can be set based on the knowledge held in advance,
Others may be initialized using random numbers. Further, at the start of learning of each process after initial setting of the weight values of the first process, the antecedent membership function realization unit 10 and the rule unit 12
and the consequent membership function realization unit 14, turns on the learning flag provided for each connection in the part that is the learning target, and turns on the learning flag provided for each connection in the part that is not the learning target. is turned off, and only the weight values of connections with learning flags in the on state are optimized through learning processing.

[Operations] According to the learning method of the rule section pre-wired neuron of the present invention having such a configuration, the following operations can be obtained.

First, among the antecedent membership function realization part of the rule part pre-wired neuron, the rule part and the consequent part membership function realization part, the part that learns the weights first is based on the weight value based on the knowledge held in advance. or initial setting of weight values using random numbers, and initial setting of weight values based on knowledge held in advance is performed for parts where weights are learned from the second time onwards.

Next, the learning process is started, the input cover turn for learning is presented to the rule section pre-wire neuron, and the output cover turn from the rule section pre-wire neuro is a learning ratio cover turn (the teacher The weights are learned by, for example, the pack propagation method, in which the initially set weight values are corrected so as to substantially match the signal).

This weight learning is to ultimately learn the weight of the entire pre-wired neuron, and prior to the final stage weight learning, the following weight learning from 1 to 0 is performed in the previous stage.

■Perform only the antecedent membership function realization part;■Perform only the rule part; ■Carry out in the order of rule part → antecedent membership function realization part; ■Carry out in the order of antecedent membership function realization part → rule part ; ■ Perform the antecedent part and the rule part at the same time; ■ Perform the antecedent part membership function realization part - the consequent part membership function realization part in the order; ■ The rule part → the consequent part membership function realization part in the order Execute; ■Rule part - Antecedent part membership function realization part - Consequent part Membership function realization part in order; ■Antecedent part membership function realization part - Rule part - Consequent part membership function realization part Perform in order; [phase] Perform the antecedent membership function realization part and the rule part at the same time → then perform the consequent membership function realization part; In this way, the antecedent membership function realization part, the rule part, and After learning the weight of at least one part of the consequent part membership function realization part, the whole rule part pre-wire neuron is finally learned, so that the antecedent part membership function realization part learned in the previous stage, The weight of the entire rule section prewire neuron can be readjusted in accordance with the learning result of at least one weight of the rule section and the consequent section membership function realization section.

In addition, the initial setting of weights that incorporates pre-existing knowledge facilitates learning and improves learning efficiency.

Furthermore, by initializing the weights using random numbers, it is possible to maximize the neural learning function.

[Embodiment 2] Figs. 2A to 2J are learning operation flowcharts showing each processing procedure of the first to tenth embodiments of the learning method of the rule section pre-wire neuro according to the present invention. This second A
The rule section pre-wired neuron that is the target of the learning method according to the procedure shown in FIG. It has a hierarchical network structure for realizing fuzzy control rules, including a realization unit 18).

Here, the rule section pre-wire neuro means the rule section 12.
has a network structure in which all units are not internally connected with the antecedent membership function realizing unit 10 in the previous stage and/or the consequent membership function realizing unit 14 in the subsequent stage, but some are internally connected according to the control rule. , is compared to a rule-part fully connected neuron that internally connects everything between hierarchical units.

In the learning process of the first to 1oth embodiments shown in FIGS. 2A to 21, the following steps 81 to S6 are performed.

[Step SL] Weight Cab of the antecedent membership function realization unit 1o.

Initialize Cbt to implement the given membership function. Note that Cab and Cbc are connection group numbers of connections whose weights are initially set, as will be described later.

[Step S2] Initialize the weights Ccd and Cde of the rule section 12 so as to realize the given fuzzy rule.

[Step S3] The weight Cel of the consequent membership function realization unit 14 is initialized so as to realize the given membership function.

[Step S4] The weight Cfg of the center of gravity calculation implementation unit 18 is initialized to realize center of gravity calculation.

[Step S5] A phase/group correspondence table indicating the neuron groups for each phase in which the learning process is performed and a group/connection correspondence table indicating the connection groups belonging to each neuron group are set.

[Step S6] Start the learning scheduler and perform learning processing using the pack propagation method.

In the learning process consisting of the above steps 81 to S6,
The initial setting of the weight values in steps 81 to S3 is performed as follows for each embodiment. First, by starting the learning scheduler in step S6, for the part where weight learning is first performed, that is, the part where weights are learned in phase 1, weight values are initialized based on the introduction of pre-existing knowledge, or random numbers are used. Initialize random weight values. On the other hand, for the part where the weights are learned from the second time onwards (second phase and onward), only the initial setting of the weight values is performed based on the introduction of knowledge held in advance.

The initial setting of weight values for the center of gravity calculation implementation unit 18 in step S4 is the same for all embodiments.

Further, regarding the setting of the phase-group correspondence table in step 5, a unique phase-group correspondence table is set according to the change of the learning phase in step s6 in each embodiment. On the other hand, the settings of the group/connection correspondence table are set in common for all embodiments.

Furthermore, in executing the weight learning in step S6, in the final stage of weight learning, that is, in the last phase, weight learning of the entire neuron is performed, that is, the antecedent membership function realization unit 1o1 rule unit 12 and the post They are common in that they learn the overall weight of the subject membership function realization unit 14. In contrast to the weight learning of the entire neuron performed in this final phase, the weight learning in the previous stage is performed in a manner unique to each embodiment.

Hereinafter, the learning method of the present invention will be explained in detail by taking the learning operation flow of the first embodiment shown in FIG. 2A as an example and dividing it into steps S1 to S6.

[Configuration of Rule Section Prewire Neuron] The rule section prewire neuron to which the present invention is applied has a hierarchical network structure capable of realizing the fuzzy control rules shown in FIG. 3, for example.

In FIG. 3, each node shown in each part represents one basic unit, and two types of basic units are used: a sigmoid function unit shown by a black circle, and a linear function unit shown by a white circle.

FIG. 4 shows the configuration of the basic unit shown in FIG. 3.

In FIG. 4, the basic unit 20 constitutes a multi-power single output system, and includes a multiplication section 22 that multiplies a plurality of inputs by the weight value of each internal connection, and a multiplication section 22 that adds all the multiplication results of the multiplication section 22. It is composed of an adding section 24 and a threshold processing section 26 that performs threshold processing on the output of the adding section 24.

Here, if the first layer is the h layer and the second layer is the i layer, the multiplier 22
The addition unit 24 executes the calculation of equation (1) below, and the threshold processing unit 26 executes the calculation of equation (2) below.

x1, = ΣVphW+h (1
)y 1, == 1 / (1+ e-CxpI-θ”
) (2) However, h: unit number of h layer i: unit number of i layer p: input signal pattern number θl: threshold value W1 of the first unit of i layer,: weight value of internal connection between h-i layer x1, : Sum of products of inputs from each unit of h layer to the 1st unit of i layer Vph: h of h layer for input signal of I)th pattern
Output from unit No. y This is threshold processing.

On the other hand, a linear function unit that satisfies V p 1 = X p is set in the threshold processing unit 26, and becomes a linear function unit indicated by a white circle in FIG. However, the input section, which is the first unit of the antecedent membership function realization section 10, does not necessarily have to be a linear function unit. The hardware of the basic unit shown in Fig. 4 is described in Japanese Patent Application No. 63-216865.
"Network Configuration Data Processing Device".

Referring again to FIG. 3, the configuration of each part of the rule section pre-wire neuro will be explained.

First, the antecedent membership function realization unit 10 processes each input variable X1, X2 . ...Xn is input, input variable X,
Output the grade value of the membership function of ~Xn. In the figure, each membership function is a one-variable function for simplicity, but it may be a multi-variable function. In other words, it may be a membership function of X, &X2 (SA) with variables X1 and X2 as inputs.

The rule section 12 includes an antecedent section rule section 12-1 and a consequent section rule section 12-2. The antecedent part rule part 12-1 is a membership function X, (SA) of input variables x1 to Xn.
The grade value of ~Xn (LA) is input, and the grade value of each rule is output.

Further, the consequent part rule part 12-2 contains the grade value L of the rule.
H8-1 to LH8-n are input, and grade values of membership functions y (SA) to y (LA) of output variables are output.

Further, the consequent membership function realization unit 14 (including the centroid calculation realization unit 16) calculates the membership function y (
Input the grade values of SA) to y (LA) and output the value of the output variable Y.

Here, the antecedent part rule part 12-1 and the consequent part rule part 12
Since both the input and output of the rule unit 12 including -2 are the grade values of the membership function, these can be cascaded and multi-stage inference is possible.

Further, the neurons connecting each module 10.12-1.12-2.14 can be shared. For example, the output neuron of the antecedent membership function implementation unit 10 and the input neuron of the antecedent rule unit 12-1 may actually be the same.

Further, in FIG. 3, each unit is described as one neuron, but if the function of the unit can be specified, it may be used as a gate circuit, an arithmetic unit, etc. that realizes the unit function instead of a neuron.

FIG. 5A shows a specific embodiment of the rule section pre-wired neuron of the present invention, taking as an example a case where two inputs, X1 and X2, are given to obtain a control output Y.
The configuration of each part will be explained in detail with reference to the drawings as follows.

The input section 16 is composed of separation units 28-1 and 28-2, which divides the inputs X1 and X2 into two and supplies them to the antecedent membership function realization section 10. The antecedent membership function realization unit 10 includes four sigmoid function units 30-1 to 30.
-4 and the same (four linear function units 32-1 to 32
-4.

The rule section 12 includes, for example, sigmoid function units 34-1 to 34-8 that realize eight fuzzy rules, and the antecedent membership function realization section 10 connects each fuzzy rule according to the descriptions 1 to L8. It is being done. The outputs of the sigmoid function units 34-1 to 34-8 are connected according to fuzzy rules to two linear function units 36-1 and 36-2 which perform consequent operations according to fuzzy rules.

The consequent membership function realization unit 14 is composed of, for example, nine linear function units 38-1 to 38-9, and determines membership based on two unit output values (grade values) output from the rule unit 12. The function is reduced (or enlarged) and the function sum of the two reduced (or enlarged) membership functions is calculated.

Furthermore, the center of gravity calculation implementation unit 18 is composed of a center of gravity determining element output device 40 and a center of gravity calculation device 42. The linear function units 40-1 and 40-2 of the center-of-gravity determining element output device 40 calculate the weight of the unit output from the consequent membership function realization unit 14 by calculating the sum and calculating the two center-of-gravity calculation elements Y1 and Y2. Finally, the center of gravity calculation device 42 calculates the two center of gravity calculation elements Y1.
, Y2, and generates a control output Y as a fuzzy inference value.

FIG. 5B shows neuron groups and connection groups for realizing the learning method according to the present invention for the rule section pre-wire neuron of FIG. 5A.

In Figure 5B, the neuron group is Ga-Gg.
However, two of these, the neuron group Ga in the input section and the neuron group Gg in the center of gravity calculation implementation section 18, are simply separated and combined, so the weights are calculated using the pack propagation method. are excluded from the study subjects. Therefore, there are five neuron groups that are subject to weight learning by the pack propagation method: Gb, Gc, Gd, Ge, and Gf.

On the other hand, there are six connection groups, Cab-ClH, representing the input connections of each two-neelon group, and the weight values of the input connections located in the previous stage of the neuron group to be learned are learned by the pack propagation method.

Regarding the learning operation flow in Figure 2, which targets the rule section Briwire neuron with the structure shown in Figures 5A and 5B,
First, the initial setting of weight values for each part will be explained as follows.

[Initial setting of the antecedent membership function realizing unit] The initial setting of the weight value for the antecedent membership function realizing unit 10 is to set the weight of each connection included in the connection groups Cab and Cbc to the given membership function. Initial settings are made to achieve this.

FIG. 6 shows the antecedent membership function realization unit 10 of FIG. 5A.
The control human power x1 side in is extracted and shown. In FIG. 6, the output value y output from the sigmoid function unit 30-1.30-2 is '! = 1/ (1+ e-
""-") (3).

In the initial setting of the weight value for the antecedent membership function realizing unit 10 of the present invention, based on the knowledge that is held in advance, for example, as the weight value W of the input connection of the sigmoid function unit 30-1, W=- 20 is set, and θ=-10 is set as the threshold value θ in the sigmoid function unit 30-1.

Further, the weight value W of the input connection of the other sigmoid function unit 30-2 is set to W=20, and at the same time, the threshold value θ of the sigmoid function unit 302 is set to θ=10. In addition, linear function unit) 32-1. 32-
The weight value of the input connection No. 2 is set to the initial weight value=1.

In this way, in the sigmoid function unit 30-1, W〈0
When the weight value and threshold value are initially set such that θ<0, the unit output value x, (SS) for the control output X shown in the characteristic curve 44 in FIG. 7, that is, the unit output value It is possible to set a truth value calculation function for a membership function having a function shape such that X and (SS) become small.

Furthermore, in the sigmoid function unit 30-2, W〉0,
By initially setting the weight value and threshold value such that θ〉0, the unit output value Xl (LA) for the control human power X1 shown in the characteristic curve 46 in FIG. ) can be realized to calculate the truth value of a membership function with a function shape in which the function becomes large.

Sigmoid function unit 30-1.3 shown in FIG.
The initial setting of the weight value and threshold for 0-2 is such that the desired antecedent membership function is also realized for the remaining sigmoid function units 30-3 and 30-4 in the antecedent membership function realization unit 10. Initial settings are made based on prior knowledge.

[Initial Setting of Rule Section] The initial setting of the weight value for the rule section 12 in FIG. 5A is as follows.
The weight values of the connections belonging to the connection groups Ccd and Cde shown in FIG. 5B are initialized by introducing knowledge held in advance so as to realize a given fuzzy rule.

FIG. 8 is an explanatory diagram showing a specific example of initial setting of weight values for the rule section 12 together with fuzzy rules.

In FIG. 8, eight sigmoid function units 34-
1 to 34-8 correspond to fuzzy rules 1 to L8 shown on the right side, respectively. For each of the sigmoid function units 34-1 to 34-8, fuzzy rules are calculated from the linear function units 32-1 to 32-4 of the antecedent membership function realization section 10 located at the previous stage.
Input connections are made in accordance with ~L8. Furthermore, fuzzy rules are applied to the two linear function units 36-1 and 36-2 provided following the eight sigmoid function units 34-1 to 34-8, and the consequent parts of 1 to L8 (T HE
Connections are made in accordance with N).

Here, based on the weight values initially set to the input connections of the sigmoid function units 31-1 to 34-8 in the rule section 12 of FIG. become that way.

FIG. 9 shows one sigmoid function unit 34 of the rule section 12, in which truth values produces an output value y.

The output value y output from the sigmoid function unit 34 in FIG. 9 is as follows: 5=Wx1・X1+Wxl−X2−θ (4)y=
It is calculated as 1/(1+e-s) (5). The unit output value y given by the above equation (5) is the weight value Wa of the input connection.

By arbitrarily setting Wx7 and the threshold value θ, an appropriate relationship shadow shape can be realized.

Specifically, the sigmoid function units 34-1 to 34-1 in FIG.
The numerical value shown in the input connection of 34-8 is set as the initial weight value, and the sigmoid function units 34-1 to 34-8
The numerical value shown below indicates the threshold value θ.

For example, in the sigmoid function unit 34-1 that realizes the fuzzy rule L1, the following is set, and by setting these initial values, the sigmoid function unit 34-1 that realizes the fuzzy rule L1
will perform threshold processing of the function shape shown by the three-dimensional coordinates in FIG.

Hereinafter, similarly, sigmoid function units 34-2 to 34-
By initializing the weight values and threshold values for each of the 8 fuzzy rules 2 to L8, the function shapes shown in FIGS. 11 to 17 can be obtained.

Note that the linear function unit 36 provided at the output stage in FIG.
The weight value of the connection for -1.36-2 is initially set to 1, and the threshold processing is a linear function, and θ
=0 is set.

[Initial Setting of Consequent Part Membership Function Realization Unit] Initial setting of weight values for the consequent part membership function realization unit 14 in FIG. 5A is as follows: Initial setting of weight values for the consequent part membership function realization unit 14 shown in FIG. Initial settings are made to realize the given membership function based on the knowledge of the given membership function.

FIG. 18 is an explanatory diagram specifically showing the initial setting of weight values for the consequent membership function realization unit 14.

In FIG. 18, the consequent membership function realization unit 14
Nine linear function units 38-1 to 38- provided in
9, a connection input is performed from the linear function unit 36-1, 36-2 of the rule section 12 at the previous stage as shown in the figure. These nine units are 38-1 to 38-9
correspond to points in the range of 0 to 1 on the training line of the control output y at equal intervals.

The initial setting of the weight values of the input connections for the linear function units 38-1 to 38-9 is performed by the unit 3 of the rule section 12.
Regarding the weight value of the connection of the output y (SS) from the unit 38-1, the weight value of the connection of the output y (SS) from the unit 38-1 is determined according to the characteristic curve 46 shown on the right side of the graph in which the horizontal axis is the control output Y1 and the vertical axis is the unit output value.
Weight values from the side to the unit 38-9 side, 1, 1. 1. Q, 75.0.5. Q, 25.
0. 0. Initialize 0.

Further, regarding the weight values of the input connections from the unit output y (LA) of the rule section 12, the weight values are as follows from the unit 38-1 side to the unit 38-9 side according to the characteristic curve 48: 0, O, 0, 0, 25,0,5,0,75,1,1
, ].

Initialize.

Furthermore, the threshold value θ of linear function unit) 38-1 to 38-9 is θ
= 0.

Due to the initial setting of the weight values in the consequent membership function realization unit 14, the unit output values y (SS) and y (LA) output from the linear function unit 36-1.36-2 of the rule unit 12 are are reduced by multiplication by initial weight values, added, and output to the next center-of-gravity calculation implementation unit 18.

In FIG. 18, a weight value of 1 or less is initially set so as to reduce the unit output, but it is also possible to set a weight value of 1 or more to enlarge it.

[Initial Setting of Center of Gravity Calculation Implementation Unit] The weight values for the center of gravity calculation implementation unit 18 in FIG. 5A are initialized so as to realize center of gravity calculation for the connection group C1g in FIG. 5B.

FIG. 19 shows a specific example of initial setting of weight values for realizing the center of gravity calculation performed on the center of gravity determining element output device 40 of the center of gravity calculation implementation unit 18.

For the two linear function units 40-1 and 40-2 provided in the centroid determining element output device 40, the nine linear function units 38-1 provided in the consequent membership function realization unit 14 of the preceding stage ~389 are connected.

Then, as shown in the figure, weight values for calculating two gravity center derived values Y1 and Y2 used in the gravity center calculation are set for this input connection.

In fuzzy control, a method is generally adopted in which the control output Y, which is a fuzzy inference value, is calculated by finding the center of gravity of a figure consisting of the sum of functions of reduced membership functions for the same control output position according to the following formula: do.

Y= f grade(y)ydy/f grade
e(y)dy (6) In order to realize this formula (6), the linear function unit 40-1 is converted from the linear function units 38-1 to 38-9 of the consequent membership function realization unit 14.
As the connection weight value for the connection, the minimum value Y of the control output Y =
Starting from 0, weight values are assigned that increase from 0 at equal intervals up to the maximum value Y=1. On the other hand, the linear function unit 40
-2 input connection, maximum value of control output Y = 1
From 0 to the minimum value Y=0, weight values that decrease from 0 to -1, for example, are assigned at the same equal intervals as above.

By weighting the input connections to the linear function unit 40-1, 40-2, if the unit output values of the preceding stage linear function units 38-1 to 38-9 are 01 to C9, the linear function unit 40-1 becomes , Y 1 = I C1+0.875 C2+ 0.75
C3 〇, 625 C4+0.5 C, +0.375
C6+0.25C7+0.125 Cs+OC9(
7), and the linear function unit 40-2 outputs Y1
=OC+ 0.125 C20,25C30,375
Ca O,5C20,625Cb-0,75C fu -
Outputs 0.875C, +IC4 (8).

The centroid derived values Y1 and Y2 output from the linear function unit) 40-1. A control output Y, which is an inferred value, is calculated.

By substituting the above equations (7) and (8) into the above equation (9), it can be seen that Y in the above equation (9) matches Y in the above equation (6).

With the above processing, the initial setting of the weight value for the rule section prewire neuron shown in steps 81 to S4 in FIG. 2 is completed, and the process proceeds to step S5.

[Setting of learning schedule] In step S5 of FIG. 2A, following the completion of the initial settings, a learning plan, ie, a learning schedule, for causing the rule section pre-wire neuro to perform weight learning is set. This learning schedule involves setting up a phase/group correspondence table as shown in FIG. 20(a) and setting a group/connection correspondence table as shown in FIG. 20(b).

First, the phase-group correspondence table specifies the neuron groups to be learned in each phase as the learning phase progresses. In the first embodiment of the present invention, since the antecedent membership function realization unit 10 and the entire neuron are trained in order after initial setting of the weight values, phase 1
The neuron groups Gb and G in FIG. 5A belonging to the antecedent membership function realization unit 1° are the learning target groups
C, and in phase 2, the weights of the entire neuron are learned, resulting in neuron groups Gb and Gb in Figure 5.
Set five values: Gc, Gd, Ge, and Gf.

On the other hand, the group/connection correspondence table shows the correspondence between each of the two knee long groups Ga to Gg in the rule section prewire neuron and the input connection groups Cab to Cfg in Figure 20 (k
).

When the setting of the phase-group correspondence table and the group-connection correspondence table is completed, the process proceeds to step S6 in FIG. 2A, starts the learning scheduler, and actually gives the learning data to the rule section prewire neuro to perform the pack propagation method. Execute weight learning by

[Processing and Configuration of Learning Scheduler] FIG. 21 shows a flowchart of the learning process of the rule section pre-wire neuro of the present invention. This processing flow diagram is, for example, the 22nd
First, the device configuration for learning processing that can be realized by the device configuration shown in the figure will be explained.
In the figure, a learning instruction unit 50 and a learning unit 70 are provided for the rule section pre-wired neuron 100.

The learning instruction unit 50 has a learning scheduler 52,
For the learning scheduler 52, there is a learning signal storage unit 54 that stores learning signals consisting of pairs of control inputs and desired control outputs for the control inputs, and a phase group storage unit 54 that stores the phase group correspondence table shown in FIG. A correspondence tape 56, a group/connection correspondence table 58 storing a group/connection correspondence table, and a learning convergence determination section 60 are provided.

On the other hand, the learning unit 70 is provided with a learning instruction reading section 72, a learning flag setting section 74, and a weight changing section 76 that changes weight values using the pack propagation method.

Furthermore, a feature of the device configuration in FIG. 22 is that a learning adjuster 80 is provided in the connection connecting each unit in the rule section prewire neuron 100.

The learning adjuster 80 has the configuration shown in FIG.

Note that the weight change unit 76 is shown together with the learning adjuster 80.

The learning adjuster 80 is provided with a flag storage section 82 , a weight change information reading section 84 , and a weight change amount adjustment section 86 .

On the other hand, the weight change unit 76 includes a weight calculation unit 88 that performs weight calculation based on the pack propagation method, and a weight storage unit 9.
0 and a weight change amount storage section 92 are provided.

In this way, by providing the learning adjuster 80 for each connection of the rule section pre-wire neuro 100, it is possible to determine whether or not the learning scheduler 52 of the learning instruction unit 50 changes the weight value of the connection for the rule section pre-wire neuro 100. It is possible to set a learning flag that determines the weight, and it is possible to perform weight learning only for connections for which the learning flag is valid, that is, the flag is turned on.

Furthermore, the flag set for the learning coordinator 80 sets the antecedent membership that can execute fuzzy rules for a general hierarchical network that does not have a network structure that can implement fuzzy rules as shown in FIG. It is also used when specifying a hierarchical network structure consisting of a function realization section, a rule section, and a consequent membership function realization section.

When a hierarchical network conforming to fuzzy rules is configured, the weight values are adjusted as follows when actually learning is performed by the pack propagation method.

In FIG. 23, the weight change information reading unit 84 of the learning adjuster 80 monitors whether the weight change amount has been written by the weight calculation unit 88 to the weight change amount storage unit 92 of the weight change unit 76. There is. When the weight change information reading unit 84 detects that the weight has changed, it notifies the weight change amount adjustment unit 86 of the fact. The weight change amount adjustment unit 86 checks the learning flag in the flag storage unit 82, and does nothing if the learning flag is on. That is, writing to the weight change amount storage section 92 is made valid. On the other hand, if the learning flag is off, the weight change amount adjustment section 86 changes the weight change amount storage section 9 of the weight change section 76.
2 is set to zero to disable the weight change. The hardware of this learning scheduler is disclosed in Japanese Patent Application No. 63-227825.
``Learning method for network configuration data processing device'' in the issue.

[Processing details of the learning scheduler] Next, refer to the learning process flow in Figure 21 and see Section 22.23.
The learning process of the rule section pre-wired neuron in the first embodiment of the present invention according to the device configuration shown in the figure will be explained by dividing it into phase 1 and phase 2.

■Phase 1 First, when the learning scheduler of the learning instruction unit 50 is started, in step Sl (hereinafter, "step" is omitted), the phase counter i indicating the progress of the learning phase is set as i=l.
, and the process proceeds to S2, where the neuron group numbers Gb and Gc corresponding to the learning phase i=1 are read out with reference to the phase/group correspondence table 56.

Next, proceed to S3 and enter neuron group numbers Gb and Gc.
The connection group numbers Cab and Cbc corresponding to the group and connection correspondence table 58 are read out. Next, the process advances to S4 and the connection group number C11b is entered.

Output Cbc and connect group number Cab, Cbc
The learning flag of the learning adjuster 80 provided in the connection belonging to the connection is turned on.

Specifically, the learning scheduler 5 of the learning instruction unit 50
2, the connection group numbers Cab and Cbc are output to the learning instruction reading section 72 of the learning unit 70 and read. The learning flag setting unit 74 that receives the read result from the learning instruction reading unit 72 instructs the flag storage unit 82 of the learning adjuster 80 provided in the connections belonging to the connection group numbers Cab and Cbc to turn on the flag, and sets the learning flag. Set.

Next, the process proceeds to S5, where the learning scheduler 52 issues a learning execution command to start the learning process. In this learning process, the control human power X prepared in the learning signal storage section 54 and the teacher signal d that is a desired control output for the control human power X are read out, the control input is given to the rule section prewire neuro 100, and the teacher signal is weighted. unit 76 and learning convergence determination unit 60
give to Then, the control output Y from the rule unit prewire neuron 100 that receives the learning input is taken into the weight change unit 76 and the learning convergence determination unit 60, and the weight change unit 76 changes the weight value according to the pack propagation method. The learning convergence determination unit 60 determines the end of phase 1 learning when the error of the control output Y with respect to the teacher signal d becomes equal to or less than a specified value.

When the phase 1 learning process based on the learning execution command in S5 is completed, the phase counter i is incremented by one in S6, and the end of the learning phase is checked in S7. In the learning process of the present invention, since it ends at phase 2, the process returns to S2 and enters the next phase 2 learning process.

(2) Phase 2 In the phase 2 learning process, first in S2, the phase/group correspondence table 56 is referred to and five neuron group numbers cb to Cf are read out. Next, in step 83, the group/connection correspondence table 58 is referred to and the connection group number Cab-Ce corresponding to the neuron group Gb-Gf is determined.
f is read and the connection group number Cab- is read in step S4.
Cef is output and the learning flag provided in the connection is set.

Next, weight learning in phase 2 is performed by the pack propagation method by giving a learning signal, and when weight learning for the entire neuron is completed, the phase counter i is incremented by 1 in S6, and the learning phase ends in S7. Check and complete the series of processing.

[Principle of learning by pack propagation method 1 Next, the principle of weight learning by the pack propagation method of the present invention will be explained together with the hierarchical network.

FIG. 24 shows a general hierarchical neural network, which is composed of a type of node called a unit and internal connections with weights.

In Figure 24, the h layer is the input device, and the i layer is the intermediate layer.
layer, and the j layer as the output layer. Of course, there is only one i-layer as the intermediate layer, but the same is true even if there are a plurality of intermediate layers. Each unit has the configuration shown in FIG. 4, and the calculations performed in the unit are as shown in equations (1) and (2) above for the basic unit 20. In the pack propagation method, the weight value Wih and threshold value θi of the hierarchical network are adaptively and automatically adjusted and learned using error feedback. As is clear from equations (1) and (2) above, it is necessary to adjust the weight value Wih and the threshold value θi at the same time, but this work is a difficult work that interferes with each other.

Therefore, in addition to the normal unit, the applicant of this application has
A unit whose output value is always 1 is provided in the h layer on the input side,
Threshold value θi for input connection from this unit with output value 1
By assigning as a weight value, the threshold θi becomes the weight value W
A pack propagation method has been proposed in which the threshold value θi can be treated as a weight value by incorporating it into i h (Japanese Patent Application No. 62-333484).

By doing this, the above equations (1) and (2) become x p,
=ΣythW1h (II)71
, = 1/ (1+cxp(-X1,) )
(12), and the threshold value θi does not appear.

By expanding from equations (11) and (12) above, we get the following (13)
) (14) is obtained.

Xp+=Photo3'p+W1+ (1
3) 71, = 1/ (1+exp(xp+))
(14) However, j : unit number Wl of layer j, : weight value X1 of the internal connection between i-j layers, + is the product sum y1 of inputs from each unit of the layer to the 1st unit of layer j, = p Output from the first unit of the j layer for the input signal of the th pattern The weight change unit 76 in FIG.
When receiving the teacher pattern apl, which is the signal that should be taken by the output cover turns y and I output from the output layer j and the output cover turn Vat when Calculate δp+=Vp+ (I Vp+). Then, (d pi y t+) (15) (I6) where ε: Learning constant ζ: Momentum t: The weight value between the i layer and the j layer according to the number of learning times. Update amount ΔW.

Calculate (1). Here, the reason for adding the weight value update amount determined in the previous update cycle in the second term on the right side of equation (16) is to speed up learning.

Subsequently, the weight value changing unit 76 uses Δδ,l calculated by the above equation (15) to calculate δ1,=y pl (I V pi)Σδp +W
+ + (+-1) (17) is calculated, and then the update amount Δw1h(+) of the weight values of the h layer and the i layer is calculated.

Subsequently, the weight change unit 76 determines the weight value for the next update cycle according to the update amounts calculated by the equations (16) and (18). To update the above weight values, the output pattern Vp+ for the input pattern pattern for learning is the teacher pattern ap.
The learning process is repeated so that a weight value matching l is obtained.

On the other hand, when the units constituting the hierarchical network in FIG. 24 are linear function units, the above equation (14) becomes 'J pi = X jl
(20). The above equations (15) and (17) in this linear function unit are δ−+= a pl ”j pi
(21) δ, 1=ΣδplWII (t 1
) (22).

[Weight Learning Process of First Embodiment] FIG. 25 is a process flow diagram of weight learning performed by the pack propagation method for the rule section pre-wired neuron of the present invention shown in FIG. 5A.

The parameters of each part in this learning process flow are the 26th
It is defined as shown in the figure.

FIG. 26 schematically shows the hierarchical structure of the rule section pre-wire neuron in FIG. 5A. As shown, it has a six-layer structure from the first layer to the sixth layer.
The layer is made up of sigmoid function units, and the rest are made up of linear function units. Also, the weights to be learned using the pack propagation method are from the first layer to the fifth layer.
The sixth layer of the center of gravity calculation implementation unit 18 at the final stage is excluded from the weight learning target.

Here, the fuzzy inference value output by the realization of the fuzzy rule by the pre-wired neuron in the rule section, that is, the control output value, is y6, and the learning process in one phase is performed from the 6th layer to the 5th layer, the 4th layer, and the 3rd layer. Layer, second layer, first layer, and so on.

The progress of learning for each layer is indicated by an i counter.

That is, the i counter is set to i=6 in the initial state.

To further simplify the explanation, Wl, l-1 and ΔWWl represent a matrix whose size is (number of neurons in the i-th layer) x (number of neurons in the i-1th layer). Furthermore, δi and yi represent vectors whose size is the same as the number of neurons in the i-th layer.

The weight learning process according to the first embodiment of the present invention will be explained in FIG. 25 based on the parameters of each part shown in FIG. 26 as follows.

In the first embodiment of the present invention, the weights of the first layer and the second layer belonging to the antecedent membership function realization unit 10 are learned in phase 1, and the weights of the entire neuron, that is, the first All weights of layers 5 to 5 are learned.

First, in the phase 1 learning process, a counter i for setting a learning target layer is initially set to i=6 in Sl.

With this initial setting, the unit in the sixth layer is designated as a learning target.

Next, the process proceeds to 82 to determine whether or not it is a sigmoid function unit, and since the sixth layer is a linear function unit, the process proceeds to 814.

In S14, it is determined whether or not the layer is the final stage, and since the sixth layer is the final stage, the process proceeds to 315, and uses the teacher signal d and control output y obtained at that time to perform the same process as in equation (21) above. The difference value δ6 is determined according to the following formula.

Next, the process proceeds to 86, where it is determined whether the unit is to learn weights. At this time, since the sixth layer is excluded from the learning target, specifically, the learning flag of the learning adjuster 80 provided in the connection is off, so the process proceeds to S8 and the weight update amount ΔW65=0 As a result, the process advances to S9 and the weight value W65 is not updated. Next, the process proceeds to S12, and it is determined whether the counter i has reached E=1, which indicates the final layer of the learning process in phase 1. At this time, since i=6, the process proceeds to 813, and the counter i is decremented by one. Then, the process returns to S2 with i=5 and starts learning the next fifth layer.

Since the fifth layer is also a linear function unit, 82
Proceeds to 314, but since it is not the final layer, S1
Proceed to step 6, and calculate the difference value δ according to a formula similar to formula (22) above.
, is calculated. Next, the process proceeds to 86 to determine whether or not it is a learning target, and since the fifth layer was excluded from the learning target in phase 1, the weight update amount ΔW, 4=0
As such, the update of the weight value W, 4 in S9 is invalidated. The process is repeated in the same manner up to the final unit of the fifth layer, and when the process of the fifth layer is completed, the process proceeds to the fourth layer.

The fourth layer is also a linear function unit and is excluded from the learning target in phase 1, so the same process as the fifth layer is repeated. Further, processing of the third layer proceeds from 82 to S3 since the third layer is a sigmoid function unit, and since it is not the final stage, six difference values unique to the sigmoid function unit are calculated using the above equation (17) at 85. Calculate according to the same formula. Since the third layer is also not a learning target, the process proceeds from 86 to 88 and sets the weight value update amount ΔW3□=0 to invalidate the update of the weight value W3□ in S9.

When the third layer processing is completed, the process proceeds to the second layer, and since the second layer is a linear function unit, the difference value δ2 is obtained in S16 via 82 to 314, and the second layer is antecedent part member. Since it belongs to the ship function realization unit 10 and is the learning target of phase 1, the process proceeds from S6 to 87 and the above (16
) Weight value update amount ΔW2 according to a formula similar to formula (18)
．． is calculated, and in step S19, the weight value W2. Update. Thereafter, the process is repeated in the same manner up to the final unit of the second layer, and the process moves on to the first layer.

Since the first layer is a sigmoid function unit and is not the final stage layer, the process proceeds from 82 to S5 via 83, and after calculating the difference value δ1, the process proceeds from S6 to 87 to calculate the weight value update amount ΔW1o, and in S9 Update the weight value W1o. The same process is repeated until the final unit of the first layer.
When the processing of the final unit of the first layer is completed, the process proceeds from 810 to 312, and since the i counter at this time matches the final layer E;1 of phase 1, it is determined that the learning process of phase 1 has ended. Finish the series of processing.

Next, in phase 2 in which the weights of the entire neuron are learned, the processing in the sixth layer only calculates the difference value δ6, and the weight value W6. is not updated, but when proceeding to the fifth layer processing, the weight value ΔW54 is effectively calculated and the weight value W,4
will be updated. Similarly, weight learning processing is performed in the order of the remaining 4th layer, 3rd layer, 2nd layer, and 1st layer, and when it is determined whether the final unit of the 1st layer has been processed or terminated, a series of learning processing is performed. end.

[Learning process of second embodiment] Weight learning of the second embodiment of the present invention is performed in step S in FIG. 2B.
As shown in 6, ①Rule part 12 ②The neuron as a whole is performed. As shown in step S2 corresponding to the order of weight learning, the initial setting of the weight values for the rule unit 12 is the initial setting of the weight values based on the introduction of knowledge held in advance or the random weight values using random numbers. Initial setting of values and further initial setting of weight values using both are performed.

On the other hand, the antecedent membership function realization unit 10 and the consequent membership function realization unit 14 that perform weight learning in Phase 2 from the second time onwards use the knowledge they have in advance as shown in steps SL and S3. Only the initial setting of the introduced weight values is performed.

Furthermore, in the setting of the phase-group correspondence table in step S5, as shown in FIG. Gb-G
Five f are set.

In the weight learning of this second embodiment, by learning the entire rule section pre-wire neuron after learning the weight of the rule section 12, the entire rule section pre-wire neuron is adjusted to match the result of the weight learning of the rule section. The weights can be readjusted.

[Learning Process of Third Embodiment] In the third embodiment of the present invention, as shown in step S6 in FIG. Learn. The rule section 12 that learns the weights first in accordance with the order of learning the weights is configured to initialize the weight values based on the introduction of pre-existing knowledge or to perform random settings using random numbers, as shown in step S2. Initial setting of weight values and initial setting of weight values using both are performed.

On the other hand, for the antecedent membership function realization unit 10 and the consequent membership function realization unit 14 that learn the weights from the second time onwards, as shown in steps SL and S3,
Only the initial setting of weight values is performed based on the introduction of knowledge held in advance.

In addition, in the setting of the phase-group correspondence table in step S5, as shown in FIG. The neuron group numbers Gb and Gc belonging to the function realization unit 10 are set, and in phase 3, the group number Gb for the entire neuron is set.
- Perform five settings for Gf.

According to the third embodiment of the present invention, after learning the weights of the rule section 12 and the antecedent section membership function realization section 10, the entire rule section prewire neuron is trained. Membership function realization unit 10
It is possible to readjust the weights of the entire rule section pre-wire neuron according to the learning results of the weights.

[Learning Process of Fourth Embodiment] In the fourth embodiment of the present invention, step S in FIG.
As shown in 6, the weights are learned in the following order: (1) Antecedent membership function realization unit 10 (2) Rule unit 12 (2) Weights are learned in the order of the entire neuron. Corresponding to the order of weight learning, the antecedent membership function realizing unit 10 that learns the weights first sets the weight values based on the introduction of pre-existing knowledge as shown in step S1, or Initial setting of random weight values using random numbers and further initial setting of weight values using both are performed. On the other hand, 2
For the rule section 12 and the consequent membership function realization section 14 that learn the weights from the time onwards, only the initial setting of the weight values is performed based on the introduction of knowledge held in advance, as shown in step S2.

In addition, the setting of the phase-group correspondence table in step S5 is as shown in FIG. Neuron group numbers Gd and Ge belonging to the rule section 12 are set, and further, in phase 3, five neuron group numbers Gb to Gf for the entire neuron are set.

According to the fourth embodiment of the present invention, after learning the weights of the antecedent membership function realizing unit 10 and the rule unit 12, the entire rule unit prewire neuron is trained, thereby realizing the antecedent membership function. The weights of the entire rule section pre-wire neuron can be readjusted in accordance with the learning results of the weights of the rule section 10 and the rule section 12.

[Learning Process of Fifth Embodiment] In the fifth embodiment of the present invention, step S in FIG.
As shown in 6, the weights are learned in the following order: (1) the membership function realization part 1o of the antecedent part, and (2) the whole neuron. Corresponding to the order of this learning, the weights of the antecedent membership function realization unit 10 and the rule unit 12 are simultaneously adjusted;
As shown in S2, for the antecedent meshibaship function realization unit 10 and the rule unit 12, the initial setting of weight values based on the introduction of prior knowledge or the initial setting of random weight values using random numbers, or both. Initialize the weight values using . On the other hand, for the consequent messibaship function realizing unit 14 that performs the second weight learning, as shown in S3, only the initial setting of the weight values is performed based on the introduction of knowledge that it has in advance.

Further, the setting of the phase-group correspondence table in step S5 is as shown in FIG.

Gc, Gd, and Ge are set, and in phase 2, the neuron group number Gb-Gf for the entire neuron is set to 5.
Set one.

According to the fifth embodiment of the present invention, after learning the weights of the antecedent membership function realization unit 10 and the rule unit 12 at the same time, the antecedent membership function is learned by learning the entire rule unit prewire neuron. The weights of the entire rule section pre-wire neuron can be readjusted in accordance with the result of simultaneous learning of the weights of the function realization section 10 and the rule section 12.

[Learning process of the sixth embodiment] In the sixth embodiment of the present invention, step S in FIG.
As shown in FIG. 6, 1. Antecedent membership function realization section 1.2 Consequent membership function realization section 14. 2. Weights are learned in the order of the entire neuron. The antecedent membership function realization unit 10, which learns the weights first in accordance with this learning order, uses initial setting of weight values based on the introduction of pre-existing knowledge or random numbers as shown in step s1. Initial setting of random weight values is performed, and initial setting of weight values using both of them is performed. On the other hand, for the rule part 12 and the consequent part membership function realization part 14 that undergo weight learning from the second time onward, the initial setting of the weight value is based on the introduction of the knowledge that they have in advance, as shown in step 82.33. Only do the following.

Further, the setting of the phase-group correspondence table in step S5 is as shown in FIG. In phase 2, the neuron group number G belonging to the consequent membership function realization unit 14
f is set, and further, in phase 3, five neuron group numbers Gb-Gf for the entire neuron are set.

According to the sixth embodiment of the present invention, after learning the weights of the antecedent membership function realizing unit 10 and the consequent membership function realizing unit 14 in sequence, the entire rule unit prewire neuron is trained. As a result, the weights of the entire rule section prewire neuron can be readjusted in accordance with the learning results of the weights of the antecedent membership function realization section 10 and the consequent membership function realization section 14.

[Learning Process of Seventh Embodiment] In the seventh embodiment of the present invention, step S in FIG.
As shown in 6, 1) the rule section 12 2) the consequent membership function realization section 14 2) learns the weights in the order of the entire neuron. As for the rule unit 12 that learns weights first in accordance with this learning order, as shown in step S2, initial setting of weight values based on the introduction of pre-existing knowledge or random weight values using random numbers is performed. Initial setting of weight values is performed using both of them. On the other hand, for the antecedent membership function realization unit 10 and the consequent membership function realization unit 14 that perform weight learning from the second time onwards, steps SL and S3 are performed.
As shown in Figure 2, only the initial setting of the weight values is performed based on the introduction of knowledge that is previously held.

In addition, as shown in FIG. 20(g), the setting of the phase-group correspondence table in step S5 is as follows: In phase 1, neuron group numbers Gd and Ge belonging to the rule part 12 are set, and in phase 2, the consequent part membership is set. A neuron group number Gf belonging to the function realization unit 14 is set, and further, in phase 3, five neuron group numbers Gb to Gf for the entire neuron are set.

According to the seventh embodiment of the present invention, the rule section 12
After sequentially learning the weights of the consequent membership function realizing unit 14, and learning the entire rule unit prewire neuron, the result of weight learning of the rule unit 12 and the consequent membership function realizing unit 14 is It is possible to readjust the weight of the entire combined rule section prewire neuron.

[Learning process of the eighth embodiment] In the eighth embodiment of the present invention, step S in Fig. 2H
As shown in 6, 1) the rule unit 12 2) the antecedent membership function implementation unit 10 2) the consequent membership function implementation unit 14 2) learns the weights in the order of the entire neuron. As for the rule unit 12 that learns weights first in accordance with this learning order, as shown in step S2, initial setting of weight values based on the introduction of pre-existing knowledge or random weighting using random numbers is performed. Initial setting of values and further initial setting of weight values using both are performed.

On the other hand, for the antecedent membership function realization unit 10 and the consequent membership function realization unit 14 that learn the weights from the second time onwards, as shown in steps Sl and S3,
Only the initial setting of weight values is performed based on the introduction of knowledge held in advance.

In addition, as shown in FIG. 20(h), the setting of the phase-group correspondence table in step S5 is as follows: In phase 1, neuron group numbers Gd and Ge belonging to the rule part 12 are set, and in phase 2, the antecedent part membership is set. The neuron group numbers Gb and Gc belonging to the function realization unit 10 are set, the neuron group number Gf belonging to the consequent membership function realization unit 14 is set in phase 3, and the neuron group number for the entire neuron is set in phase 4. Gb
- Set five Gf.

According to the eighth embodiment of the present invention, after learning the weights in the order of the rule section 12, the antecedent membership function realization section 10, and the consequent membership function realization section 14, the rule section pre-wires By learning the weights of the entire neuron, the antecedent membership function realization unit 10 and the rule unit 1
The weights of the entire rule section pre-wire neuron can be readjusted in accordance with the learning results of the weights of 2 and the consequent membership function realization section 14.

[Learning process of the ninth embodiment] In the ninth embodiment of the present invention, step S in FIG.
6, the antecedent membership function realizing unit 10 ■ the rule unit 12 ■ the consequent membership function realizing unit 14 ■ learns the weights in the order of the entire neuron. Antecedent membership function realization unit 1 learns the weights first in accordance with this learning order.
For 0, as shown in step S1, the initial setting of the weight value is based on the introduction of prior knowledge, the initial setting of the random weight value using random numbers, or the initial setting of the weight value using a combination of both. Do this. On the other hand, regarding the rule section 12 and the consequent section membership function realization section 14 that undergo weight learning from the second time onward, step S2. As shown in S3, only the initial setting of weight values based on the introduction of knowledge held in advance is performed.

In addition, the setting of the phase-group correspondence table in step S4 is as shown in FIG. In phase 3, the neuron group numbers Gd and Ge belonging to the rule section 12 are set, and in phase 3, the neuron group number Gf belonging to the consequent membership function realization section 14 is set, and in the final phase 4, the neuron group that becomes the entire neuron is set. Number G
Set five b-Gf.

In the ninth embodiment, after sequentially learning the weights of the antecedent membership function realizing unit 10, the rule unit 12, and the consequent membership function realizing unit 14, the entire rule unit prewire neuron is learned. By learning the weights, the weights of the entire rule section prewire neuron can be readjusted in accordance with the learning results of the weights of the antecedent membership function realization section 10, the rule section 12, and the consequent membership function realization section 14. can be done.

[Learning Process of 10th Embodiment] In the 10th embodiment of the present invention, as shown in step S6 of FIG. Ship function realization unit 14 ■Learns weights in the order of the entire neuron. Corresponding to this order, the antecedent membership function realization unit 10 and the rule unit 12, which learn the weights first, calculate the weight values based on the introduction of knowledge they have in advance, as shown in steps SL and S2. Initial settings, initial settings of random weight values using random numbers, and initial settings of weight values using a combination of both are performed. On the other hand, for the consequent membership function realization unit 14 that performs weight learning from the second time onward, only the initial setting of the weight values is performed based on the introduction of knowledge held in advance, as shown in step S3.

Further, as shown in FIG. 20(j), the setting of the phase/group correspondence table in step S5 is such that in phase 1, neuron group numbers Gb, Gc, Gd and Ge are set, and in phase 2, consequent membership function realization part 1
4 is set, and in the final phase 3, five neuron group numbers Gb-Gf corresponding to the entire neuron are set.

In the tenth embodiment of the present invention, the weights of the antecedent membership function realizing unit 10 and the rule unit 12 are learned simultaneously, and then the weights of the consequent membership function realizing unit 14 are learned simultaneously.
The weights of the antecedent membership function realizing unit 10, the rule unit 12, and the consequent membership function realizing unit 14 can be changed by learning the weights of the antecedent membership function realizing unit 10, the rule unit 12, and the consequent membership function realizing unit 14 by finally learning the weights of the entire rule unit prewire neuron. It is possible to readjust the weights of the entire rule section pre-wired neuron according to the learning results.

[Others] In the above embodiment, as shown in FIG. 5A, two control human forces X1. Although the rule section pre-wired neuro that implements fuzzy rules targeting X2 has been taken as an example, the number of control inputs may be one or three or more as appropriate.

Further, in the above embodiment, the membership function shown in FIG. 6.7 was taken as an example, but as shown in FIG. 28, two sigmoid function units 94-1.9
The unit output of 4-2 may be applied to the subtraction unit 96 to realize the function shape shown in FIG. 4(b). In this case, the unit output y is 3'=1/(1-e-cho!+#1) l/(1,-
W2X-02).

[Effects of the Invention] As described above, according to the present invention, after learning the weights of at least one of the antecedent membership function realization part, the rule part, and the consequent membership function realization part,
By learning the entire rule part pre-wired neuron, the rule part pre-wired neuron is adapted to the learning result of the weight of at least one part of the antecedent membership function realization part, the rule part, and the consequent part membership function realization part. The overall weights can be readjusted.

In addition, weight learning can be easily performed by initially setting weight values based on the introduction of knowledge held in advance.

Furthermore, the performance of the neuron network can be maximized by randomly initializing weight values using random numbers or by using a combination of pre-existing knowledge and random numbers.

[Brief explanation of the drawing]

1st A, IB, IC, ID, IE, IF, I
G. IH, I1, and IJ diagrams are explanatory diagrams of the principle of the present invention; 2A,
2B, 2C, 2D, 2E, 2F, 2G. Figures 2H, 21, and 2J are learning operation flow diagrams showing the first to tenth embodiments of the present invention; Figure 3 is a configuration diagram of an embodiment of the rule section prewire neuro of the present invention; Figure 4 is a basic unit explanation Figure; Figure 5A is a configuration diagram of a specific embodiment of the rule section prewire neuron of the present invention; Figure 5B is an explanatory diagram of the neuron group and connection group of Figure 5A; Figure 6 is an antecedent section of the present invention An explanatory diagram of the initial settings of the membership function realization section; FIG. 7 is an explanatory diagram of the membership function of the antecedent section; FIG. 8 is an explanatory diagram of the initial settings of the rule section of the present invention; FIG. 9 is an explanation of the rule section unit of the present invention Figure; Figure 10 is an explanatory diagram of fuzzy rule L1 in Figure 9; Figure 11 is an illustration of fuzzy rule L2 in Figure 9.
Explanatory diagram; FIG. 12 is an explanatory diagram of fuzzy rule L3 in FIG. 9; FIG. 13 is an explanatory diagram of fuzzy rule L4 in FIG. 9;
4 is an explanatory diagram of fuzzy rule L5 in FIG. 9; FIG. 15 is an explanatory diagram of fuzzy rule L6 in FIG. 9; FIG. 16 is an explanatory diagram of fuzzy rule L7 in FIG. 9; An explanatory diagram of rule L8; FIG. 18 is an explanatory diagram of the initial settings of the consequent membership function implementation unit of the present invention; FIG. 19 is an explanatory diagram of the initial settings of the center of gravity calculation implementation unit of the invention; An explanatory diagram of the settings of the phase/group correspondence table and the group/connection correspondence table; Fig. 21 is a learning processing flow diagram of the present invention; Fig. 22 is a device configuration diagram of the learning processing according to the present invention; Fig. 23 is the wiring diagram of Fig. 22. FIG. 24 is an explanatory diagram of a general hierarchical neural network; FIG. 25 is a flowchart of the weight learning process of the present invention; FIG. 26 is an explanatory diagram of parameters of each part of FIG. 25. FIG. 27 is an explanatory diagram of another membership function used in the present invention. In the figure, 10: Antecedent membership function realization unit 12, Rule unit 14: Consequent membership function realization unit 16, Input unit 18: Center of gravity calculation realization unit 20: Basic unit 22: Multiplication unit 24: Addition unit 26: Threshold processing unit 2B-1, 28-2, 32-1 to 32-4.38-1
~38-9.40-1.40-2: Linear function unit 30-1 ~30-2.34-1 ~34-8.94-1.
94-2: Sigmoid function unit 40: Center of gravity determining element output device 42: Center of gravity calculation device 50: Learning instruction unit 52: Learning scheduler 54: Learning signal storage unit 56: Phase/group correspondence table 58 Niger loop φ phase correspondence table 60: Learning Convergence determination section 70: Learning unit 72: Learning current events reading section 74: Learning flag setting section 76: Weight changing section 80: Learning adjuster 82: Flag storage section 84: Weight change information reading section 86 Duplex change amount adjustment section 88: Weight calculation section 90: Weight storage section 92: Weight change amount storage section 96: Subtraction unit 100: Rule section pre-wire neuro

Claims (12)

[Claims] (1) It has a hierarchical network consisting of an antecedent part membership function realization part (10), a rule part (12) consisting of one or more layers, and a consequent part membership function realization part (14), and the rule part (12) does not internally connect all units with the antecedent membership function realization unit (10) in the previous stage and/or the consequent membership function realization unit (14) in the subsequent stage, but only some parts according to the control rule. In a learning method of a rule section pre-wired neuro that is internally coupled and outputs one or more control operation quantities (Y) corresponding to input control state quantities (X_1, X_2,...X_n), The subject membership function realization unit (10) initializes the weight values based on the knowledge it has in advance or using random numbers, and the rule unit (12) and the consequent membership function realization unit (14) a first step of initializing weight values based on knowledge held in advance; a second step of learning the weights of the antecedent membership function realization unit (10) using learning data; Using the antecedent membership function realization unit (10)
, a third step of learning the overall weights of the rule section (12) and the consequent membership function realization section (14); and a method for learning a rule section prewire neuro. (2) It has a hierarchical network consisting of an antecedent membership function realization unit (10), a rule unit (12) having one or more layers, and a consequent membership function realization unit (14), and the rule unit (12) does not internally connect all units with the antecedent membership function realization unit (10) in the previous stage and/or the consequent membership function realization unit (14) in the subsequent stage, but only some parts according to the control rule. In a learning method of a rule section pre-wired neuro that is internally coupled and outputs one or more control operation quantities (Y) corresponding to input control state quantities (X_1, X_2,...X_n), the rule The weight values are initialized in the section (12) based on knowledge held in advance or using random numbers, and the antecedent membership function realization section (10) and the consequent membership function realization section (14) a first step of initially setting weight values based on knowledge held in advance; a second step of learning the weights of the rule section (12) using learning data; and a second step of learning the weights of the rule section (12) using learning data; A third step of learning the overall weights of the part membership function realization part (10), the rule part (12), and the consequent part membership function realization part (14); How to learn wire neuro. (3) It has a hierarchical network consisting of an antecedent part membership function realization part (10), a rule part (12) consisting of one or more layers, and a consequent part membership function realization part (14), and the rule part (12) does not internally connect all units with the antecedent membership function realization unit (10) in the previous stage and/or the consequent membership function realization unit (14) in the subsequent stage, but only some parts according to the control rule. In a learning method of a rule section pre-wired neuro that is internally coupled and outputs one or more control operation quantities (Y) corresponding to input control state quantities (X_1, X_2,...X_n), the rule The weight values are initialized in the section (12) based on knowledge held in advance or using random numbers, and the antecedent membership function realization section (10) and the consequent membership function realization section (14) a first step of initially setting weight values based on knowledge held in advance; a second step of learning the weights of the rule section (12) using learning data; and a second step of learning the weights of the rule section (12) using learning data; a third step of learning the weights of the part membership function realization part (10); and the antecedent part membership function realization part (10) using the learning data.
, a fourth step of learning the overall weights of the rule section (12) and the consequent membership function realization section (14); and a method for learning a rule section prewire neuro. (4) It has a hierarchical network consisting of an antecedent membership function realization unit (10), a rule unit (12) having one or more layers, and a consequent membership function realization unit (14), and the rule unit (12) does not internally connect all units with the antecedent membership function realization unit (10) in the previous stage and/or the consequent membership function realization unit (14) in the subsequent stage, but only some parts according to the control rule. In a learning method of a rule section pre-wired neuro that is internally coupled and outputs one or more control operation quantities (Y) corresponding to input control state quantities (X_1, X_2,...X_n), The subject membership function realization unit (10) initializes the weight values based on the knowledge it has in advance or using random numbers, and the rule unit (12) and the consequent membership function realization unit (14) a first step of initializing weight values based on knowledge held in advance; a second step of learning the weights of the antecedent membership function realization unit (10) using learning data; a third step of learning the weights of the rule part (12) using the learning data; A learning method for a rule section pre-wired neuro, comprising: a fourth step of learning the overall weight of the realization section (14); (5) It has a hierarchical network consisting of an antecedent part membership function realization part (10), a rule part (12) consisting of one or more layers, and a consequent part membership function realization part (14), and the rule part (12) does not internally connect all units with the antecedent membership function realization unit (10) in the previous stage and/or the consequent membership function realization unit (14) in the subsequent stage, but only some parts according to the control rule. In a learning method of a rule section pre-wired neuro that is internally coupled and outputs one or more control operation quantities (Y) corresponding to input control state quantities (X_1, X_2,...X_n), The weight values are initialized based on the knowledge that the subject membership function realization unit (10) and the rule unit (12) have in advance or using random numbers, and the weight values are initialized in the consequent membership function realization unit (14). a first step of initially setting weight values based on knowledge held in advance; a first step of simultaneously learning the weights of the antecedent membership function realization unit (10) and the rule unit (12) using learning data; a third step of learning the overall weights of the antecedent membership function realization unit (10), the rule unit (12), and the consequent membership function realization unit (14) using learning data; A learning method for a rule section pre-wired neuro, characterized by comprising: and; (6) It has a hierarchical network consisting of an antecedent membership function realization unit (10), a rule unit (12) having one or more layers, and a consequent membership function realization unit (14), and the rule unit (12) does not internally connect all units with the antecedent membership function realization unit (10) in the previous stage and/or the consequent membership function realization unit (14) in the subsequent stage, but only some parts according to the control rule. In a learning method of a rule section pre-wired neuro that is internally coupled and outputs one or more control operation quantities (Y) corresponding to input control state quantities (X_1, X_2,...X_n), The subject membership function realization unit (10) initializes the weight values based on the knowledge it has in advance or using random numbers, and the rule unit (12) and the consequent membership function realization unit (14) a first step of initializing weight values based on knowledge held in advance; a second step of learning the weights of the antecedent membership function realization unit (10) using learning data; Using the consequent membership function realization unit (14)
a third step of learning the weights of the antecedent membership function realization unit (10), rule unit (12), and consequent membership function realization unit (14) using the learning data; A method for learning a rule section pre-wired neuro, comprising: a fourth step of learning; (7) It has a hierarchical network consisting of an antecedent part membership function realization part (10), a rule part (12) consisting of one or more layers, and a consequent part membership function realization part (14), and the rule part (12) does not internally connect all units with the antecedent membership function realization unit (10) in the previous stage and/or the consequent membership function realization unit (14) in the subsequent stage, but only some parts according to the control rule. In a learning method of a rule section pre-wired neuro that is internally coupled and outputs one or more control operation quantities (Y) corresponding to input control state quantities (X_1, X_2,...X_n), the rule The weight values are initialized in the section (12) based on knowledge held in advance or using random numbers, and the antecedent membership function realization section (10) and the consequent membership function realization section (14) a first step of initially setting weight values based on knowledge held in advance; a second step of learning the weights of the rule section (12) using learning data; and a second step of learning the weights of the rule section (12) using learning data; a third step of learning the weights of the part membership function realization part (14); and the antecedent part membership function realization part (10) using the learning data.
, a fourth step of learning the overall weights of the rule section (12) and the consequent membership function realization section (14); and a method for learning a rule section prewire neuro. (8) It has a hierarchical network consisting of an antecedent part membership function realization part (10), a rule part (12) consisting of one or more layers, and a consequent part membership function realization part (14), and the rule part (12) does not internally connect all units with the antecedent membership function realization unit (10) in the previous stage and/or the consequent membership function realization unit (14) in the subsequent stage, but only some parts according to the control rule. In a learning method of a rule section pre-wired neuro that is internally coupled and outputs one or more control operation quantities (Y) corresponding to input control state quantities (X_1, X_2,...X_n), the rule The weight values are initialized in the section (12) based on knowledge held in advance or using random numbers, and the antecedent membership function realization section (10) and the consequent membership function realization section (14) a first step of initially setting weight values based on knowledge held in advance; a second step of learning the weights of the rule section (12) using learning data; and a second step of learning the weights of the rule section (12) using learning data; a third step of learning the weights of the part membership function realization part (10); and the consequent part membership function realization part (14) using the learning data.
a fourth step of learning the weights of the antecedent membership function realization unit (10), the rule unit (12), and the consequent membership function realization unit (14) using the learning data; A learning method for a rule section pre-wired neuro, comprising: a fifth step of learning; (9) It has a hierarchical network consisting of an antecedent membership function realization unit (10), a rule unit (12) having one or more layers, and a consequent membership function realization unit (14), and the rule unit (12) does not internally connect all units with the antecedent membership function realization unit (10) in the previous stage and/or the consequent membership function realization unit (14) in the subsequent stage, but only some parts according to the control rule. In a learning method of a rule section pre-wired neuro that is internally coupled and outputs one or more control operation quantities (Y) corresponding to input control state quantities (X_1, X_2,...X_n), The subject membership function realization unit (10) initializes the weight values based on the knowledge it has in advance or using random numbers, and the rule unit (12) and the consequent membership function realization unit (14) a first step of initializing weight values based on knowledge held in advance; a second step of learning the weights of the antecedent membership function realization unit (10) using learning data; a third step of learning the weights of the rule section (12) using the learning data; a fourth step of learning the weights of the consequent membership function realizing section (14) using the learning data; Using the antecedent membership function realization unit (10)
, a fifth step of learning the overall weights of the rule section (12) and the consequent membership function realization section (14); and a method for learning a rule section prewire neuro. (10) It has a hierarchical network consisting of an antecedent membership function realization unit (10), a rule unit (12) having one or more layers, and a consequent membership function realization unit (14), and the rule unit (12) does not internally connect all units with the antecedent membership function realization unit (10) in the previous stage and/or the consequent membership function realization unit (14) in the subsequent stage, but only some parts according to the control rule. In a learning method of a rule section pre-wired neuro that is internally coupled and outputs one or more control operation quantities (Y) corresponding to input control state quantities (X_1, X_2,...X_n), The weight values are initialized based on the knowledge that the subject membership function realization unit (10) and the rule unit (12) have in advance or using random numbers, and the weight values are initialized in the consequent membership function realization unit (14). a first step of initially setting weight values based on knowledge held in advance; a first step of simultaneously learning the weights of the antecedent membership function realization unit (10) and the rule unit (12) using learning data; a third step of learning the weights of the consequent membership function realizing unit (14) using the learning data; and a third process of learning the weights of the consequent membership function realizing unit (14) using the learning data;
, a fourth step of learning the overall weights of the rule section (12) and the consequent membership function realization section (14); and a method for learning a rule section prewire neuro. (11) In setting the initial weight values in the first process of the part where weights are first learned, weights that can be set based on knowledge held in advance are initialized using this knowledge, and other weights are initialized using random numbers. 11. A learning method for a rule section pre-wired neuron according to any one of claims 1 to 10. (12) At the start of learning of each process performed after the initial setting of the weight values of the first process, the antecedent membership function realization unit (10), the rule unit (12) and the consequent membership function realization unit ( 14), the learning flag set for each connection in the part to be learned is turned on, and the learning flag set for each connection in the part not to be learned is turned off and kept on. 11. A learning method for a rule section pre-wired neuron according to claim 1, wherein only weight values of connections having learning flags are optimized through learning processing.