JPS6293731A - Arithmetic system for rule type system - Google Patents

Abstract

PURPOSE:To apply a logical describing system which uses a rule and has high readability, understandability and changeability, to even an object which requires a large quantity of states, by having both of two table data for executing a JOIN operation, on an associative storage CAM and fetching a retrieval data from the CAM whose data quantity is smaller. CONSTITUTION:An operation control part 517 compares the contents of a data quantity register in a CAMI501 and a data quantity register in a CAMII502, and writes the contents of a mask data register 524 in a mask data register in the cam whose data quantity is larger. That is to say, the operation control part 517 generates a data sending-out signal to a signal line 5035, and sends out the contents of the mask data register 524 to an internal data bus 504. A data is inputted into the CAMII by generating an address of the mask data register of the CAMII502, and transmitting a write signal to a signal line 5016 of the CAMII502. Subsequently, a work table data which has been stored in the CAMI501 whose data quantity is smaller is fetched successively by one line each from the head, and the processing is repeated by the designated number of times.

Description

[Detailed Description of the Invention] [Field of Application of the Invention] The present invention relates to an arithmetic device for a rule-based system, and in particular, it determines conditions according to the state of each piece of equipment in accordance with the control δ-principle described in the rules, The present invention relates to a method for deciding l directives, and a calculation method for a rule-based system suitable for production planning, etc., which allocates orders and materials from given order data and material data according to the production goods lff11M1 principle described in the rules. .

[Background of the invention]

In factory automation (FA) systems, changes in production lines and operations are frequently made as product life cycles become shorter and product types become more diverse. This requires rapid development and maintenance of control programs.

Conventionally, a method has been adopted in which a split tube program is developed in a general-purpose language such as 1. However, this method cannot satisfy the above requirements because it takes time to develop, understand, and change the control program, and still requires specialized program personnel. Therefore, the present inventors proposed an "equipment group control method" (see Japanese Patent Application No. 79341/1983) whose logic is easy to describe, understand, and change. The above method is called a rule-based system, and the logic is expressed as an rIF (condition) using any Japanese character string.
, THEN (bound, action)" type rule format to create a program (hereinafter,
IP-THEN rules, or simply rules). According to this method, program development is easy and cooking is easy to understand and change.

In order to improve the processing speed of the above-mentioned rule-based system, the present inventors further developed a parameter section (character string information describing conditions and conclusions) in evaluating the conditions of the IF section of the rule.
The variable (X
, Y, Z, etc.) values are stored in an associative memory (generally (:on
Tent Addressable M e m o r
It is called by the name y etc. "A calculation device for rule-based systems" that can be performed at high speed using CA Ai (hereinafter referred to as CA Ai)
(Patent Application No. 60-12079). However, in this method, the buffer that takes out the value of one of the variables to be evaluated is fixed, and no consideration is given to minimizing the number of times the variable value is evaluated. Therefore, there is a problem that processing efficiency is poor depending on the amount of data in the input buffer.

Hereinafter, after an overview of the operating principle of the rule-based system is explained, the operating principle and problems of the conventional "rule-based system modeling device" will be explained in detail.

FIG. 3 is a functional block diagram of the "equipment group control system" of the prior application.

11 is a rule storage section, 12 is a rule operation section, and 13 is a state storage section. Determining a control command using an IF-THEN rule involves sequentially repeating the process of using the conclusion of one rule to determine the conditions of another rule, and finally determining the action (control command) in the THEN part of the rule. It is performed on the principle that the conditions of the written rules are satisfied and the control command is determined. That is, in FIG. 3, the rule operation unit 12 retrieves the rules one by one from the rule storage unit 11 that stores the rules, and extracts the rules from the content of the associative storage unit 13 that stores the target state and the retrieved rules. It is determined whether or not the conditions are satisfied. If the conditions are satisfied, the conclusion of the THEN section of the extracted rule is added to the state storage section 13. Here, in the condition judgment of the IF section, if all the sentences (states) that match the sentences (conditions) of each IF stored in the IF section are in the state, the target storage section 13
It is determined whether the variable (x.

y, z, etc.), it is determined whether there is a value that simultaneously satisfies all the conditions described in the IF section (AND condition), and if it exists, that value is generated.

FIG. 4 is a detailed diagram of the process in the apparatus shown in FIG. 3 when variables are described in the parameter section of the rule.

21 to 25 are work tables within the rule operation section 12. In the rule storage section 11, only one rule currently being processed by the rule operation section 12 (J4!2 has been exported) is shown. The rule operation unit 12 first
The character string "''(A<X><
A match is determined between "Y>)" and the character string stored in the state storage unit 13. At this time, the match is determined using the part of the character string excluding the parameter part, that is, "(A)". If a matching character string exists in the state H memory section 3, write the value of the parameter part in the matching character string in the corresponding parameter part of the character string in ■F as the value of the variable. Import into work table 21 (■). Furthermore, the rule operation unit 12 performs the following 1. Character string of F” CB<Z>kuY>
)", and import the parameter part of the character string in the 4th storage part 13 into another work table 22 as the variable value of the character string in the IF part (■).
f. All combinations of variable values taken into both work tables 21 and 22, generate variable values that satisfy the AND condition (conditions in the IF section of the rule are treated as AND conditions), and store in another work table 23. (■). The process of generating a transgressive value that satisfies the AND condition means that one row in one work table and one row in another work table have a common variable (one shift is taken in both tables). Finds all combinations of rows where the pupils of the variables included in the variable) match, and for each combination, generates a new variable value with the value of the common variable unchanged and other variables set to the value of the variable that has the value. (This is called overlapping row data.) For example, in the example of FIG. 4, the work table 21
In the first row of
=34. Y=a, Z==o are generated. This process is a so-called JOIN operation of table data. If there are three or more character strings in the IF part, the variable value is imported (■) for the next character string in the IF part, and the variable value is added to the variable that satisfies the AND condition generated so far. The values (in this case, those stored in the work table 23) are combined with the variables 1-4 that have just been imported.
Then, a variable value that satisfies all AND conditions is generated and stored in the work table (■).

This process is returned for all IP character strings on the third star. After that, if there is a variable value that finally satisfies the A N D condition, that value is added to the corresponding variable of the THEN part by 1, and a new value is written to the state storage part 13 (■ ).

The earlier application's "arithmetic device for a rule-based system [is the above AND
CAM the condition satisfaction variable value generation process (J OI N calculation)
This is done quickly using the . The processing contents of the conventional JOIN operation will be explained below using FIG. W.5.

In FIG. 5(a), 41 is a normal storage device, 42.
43 is a CAM, 44 is a comparison section within the CAM, and 45 is a circuit dedicated to row data overlapping for performing row data overlapping processing.

First, one of the work tables on which the JOIN calculation is to be performed is stored in a normal storage device 41, and the other is stored in one CAM 42.
, respectively. Further, table data resulting from the JOIN operation is stored in another CAM 43. The operation is as follows.

That is, I! table data is stored one row at a time from the normal storage device 41. Z, and use this as the search data of CAM42 (at this time, data X, z other than the common number Y)
mask). In the CAM 42, the comparator 44 selects C
Data comparison processing is performed simultaneously on all rows of table data in AM42. The rows of the table in the CAM 42 where the values of the search data and the common data match are taken out one by one, superimposed on the search data by the dedicated circuit 45 for row data superimposition, and transferred to another CAM 43.
Stored next.

As shown above, the number of times data comparison processing is performed is C
Regardless of the number of rows of table data in AM42, it is determined by the number of rows of table data in normal storage f41, that is, the number of search data. Therefore, if the number of table data in the normal storage device 41 is greater than the number of table data in the CAM 42, the number of comparison processes will be reduced and the processing speed will be slowed down. In this example, the number of times the comparison process is performed is two times as shown in Figure 5(a).
As shown in the figure, even with the same data as in Figure 5, the table
If the data contents change, the comparison process will be required four times (FIG. 5(b) [F], ■, ■, ■). in this way,
In the conventional method, table data on the normal storage device 41 is always used as search data, so if the number of data in the normal storage device 41 is greater than the number of data in the CAM 42, the number of times of soft processing is reversed. There are many cases compared to the number of cases.
There was a problem with poor processing efficiency.

[Purpose of the invention]

The purpose of the present invention is to solve such conventional problems and to
A rule-based system that efficiently executes IN operations and makes it possible to apply a logical description method using rules that is highly readable, easy to understand, and easy to change, even to objects that need to handle a large amount of money. The objective is to provide a calculation method.

[Summary of the invention]

In order to achieve the above object, the arithmetic device of the rule-based system according to the present invention includes a rule storage section that stores rules consisting of conditions and conclusions of equipment groups, and a rule storage unit that stores rules consisting of conditions and conclusions of equipment groups; a state memory section for storing contents;
In the rule-based system, the rule operating unit stores information stored in the status storage unit and the conclusion of the rule stored in the rule storage unit in the status storage unit, and determines a control command for a group of equipment. For two sets of data consisting of 100 million pieces of information, a letter that matches two sets of multiple condition contents described in one rule is
Two sets of associative memory means each holding the above two sets of data for a JOIN operation that creates set data consisting of all combinations of the above two sets of data that have the same head corresponding to a common item among the condition contents of the sets. and an output buffer for holding the result of the JOIN operation, and a data sub-matching means for creating one data by changing the value of item I from both data to a meaningful value, 2 two associations 2d
Data is retrieved one by one from the memory means with less data, and the data is used as search data.The data in the other associative memory means is associatively searched using the above common items as search keys, and the retrieved data and search data are searched. %i lies in the repeated execution of the first operation of storing the superimposed results in the output buffer.

[Embodiments of the invention]

An embodiment of the present invention will be described below with reference to the drawings. 6th
The figure is an explanatory diagram showing the operating principle of the present invention.

JO in the previously proposed “arithmetic unit for rule-based systems”
In IN operation processing, one of the two work tables to be JOIN operated on is on the normal storage device, and the other is on the CA
M, and table data on a normal storage device was always used as search data. As a result, the number of search data may be greater than the number of table data in the CAM, and the number of comparison processes cannot always be minimized, creating a bottleneck in processing. In contrast, in the present invention, two work tables to be subjected to a JOIN operation are placed together on the CAM, and of the two work tables, the work table that stores the fewest rows of direct variables is used as the search data. By doing so, the number of times of comparison processing is always minimized, and JOIN calculations are carried out efficiently.

In FIG. 6, 41, 42, and 43 are CAM.

44, 45, and 46 are comparison sections in the CAM, 47 is a dedicated circuit that performs row data overlapping processing, and 48° and 49.50 are
This is a data amount register that stores the number of rows of each table data.

In the present invention, two work tables and data amounts for performing JOIN operations are stored in CAM41, CAM42, and
It is stored in the data amount register 48 and the data amount 1/register 49. The table data resulting from the JOIN operation is stored in the CAM 43, and the amount of data is stored in the data amount register 50. The operation is as follows. Compare the contents of the two data amount registers 48° and 490, and retrieve one search data from the CAM that stores the work table with a small amount of data.
Extract row by row (at this time, data X other than the common variable Y,
Z is masked). In addition, if the amount of data is the same, C
Retrieve the search data from AM41. In the example in Figure 6,
Retrieve data from CAM41. In the CA and M42, the comparison unit 55 simultaneously performs data comparison processing on all rows of table data in the CAM42.

The rows of the table in the CAM 42 in which the values of the search data and the common data match are taken out one by one, superimposed on the search data by the dedicated circuit 49 for row data superimposition, and stored in another CAM 43 in the 1p order. At this time, the data amount register 50 stores the number of data stored in the CAM 43. In this device, by using three CAMs, in the rules of the rule-based software, if there are three or more character strings describing the conditions in the IF section, then the variable value that satisfies the AND condition, and the following If you have to combine the imported new variable values with the IF conditions and return to the JOIN process again, you can import the new variable values into the CAM41.
By importing the information into the CAM 42 and changing the roles of the CAM 42 and CAM 43 as appropriate, it is possible to execute JOIN operations continuously without data transfer. In addition, by storing the amount of data stored in each CAM, when performing a JOIN operation, the search data can always be retrieved from the side with the least amount of data, and the number of soft processing operations can always be minimized. It is possible to

From the above, it becomes possible to efficiently execute rule processing of rule-based software.

In FIG. 1, 501 is CAM1%, 502 is CAMn, 503 is CAMIII, 504 is an internal data bus, 505 is an internal address bus, 506 is an external data bus, 507 is an external address bus, 508 is a bus switching control unit, 509 is an internal address generation unit, 510 is an external access control unit, 511 is a table specification flag, 512 is a mode flag, 521 is a CAMI internal data amount register, 5
22 is the data amount register in CAMII, 523 is CA
M[1 data amount register, 524 is a mask data register, 514 is a search data register, 515 is a search result register, 516 is a data superimposer, 517 is an operation control section, 518 rfi, a reset section. Further, the internal data bus 504 and the internal bus address bus 505 are a pass line for data signals and a pass line for address signals inside the device, respectively.

On the other hand, an external data bus 506 and an external address bus 507
are respectively inside this device (7) CAM I 501°C
AMII 502, CAMII503, other flags,
This is a pass line for data signals to access registers etc. from the outside. This device has two modes: an external access mode and an internal operating mode.

In the external access mode, an address is specified from the outside to the external address bus 507, and data on the external data bus 506 can be written into the device, or data inside the device can be read out onto the external data bus 506. The internal operation mode is a mode in which JOIN operations are automatically executed based on the principle shown in FIG. Mode switching is specified by mode 7 lag 512. That is, when the mode flag is "0", it indicates an external access mode, and when the mode flag is "'1", it indicates an internal operation mode. The mode flag 512 indicates the signal line 5
By latching the signal of the internal data bus 504 through the signal line 5003 by the latch signal of 002, '1
``'', and is reset to 0'' by a reset signal on signal line 5004. Further, the state of the mode flag is sent to the signal line 5001. Normally, the mode flag is in the state of "0", which is the external access mode.

The bus switching control unit 508 controls the external data bus 506, external address bus 5,
07 are connected to the internal data bus 504 and internal address bus 505, respectively, and are disconnected in the internal operation mode, and when an internal address is generated, the address signal line 509 is connected to the internal data bus 504 and the internal address bus 505.
At 06', an operation is performed to connect to the internal address bus 505.

Internal address generator 509 generates CA in internal operation mode.
This is a part that generates an address for reading the data of M1501, CAMII502, and CAMII[503.

This internal address generation section 509 generates CAM I 501, CAM II 502, CAM II on the signal line 5005.
When the operation control unit 517 generates a control signal specifying one of the registers and memory cells in the I 503, it generates addresses corresponding to these signals and sends them onto the signal line 5006. The address is generated by CAM
I501 memory cell, CAMn502 memory cell,
The memory cells of the CAM III 503 are accessed in order from the beginning of the address space.

FIG. 2 is a diagram showing the relationship between the addresses of the three CAMs in FIG. 1.

As shown in FIG. 2, the memory cell of CAM1501, C
Each time an access to a memory cell of the AMl 1502 or a memory cell of the CAMl 503 is specified, the next generated address is incremented by the required amount. Also, CAM1501, CAM[502, CAMII
For the registers in I503, addresses assigned to each are generated. In addition, the signal line 5007
By the reset signal of CA M I 501, CA
The generated address for the memory cell in MII 502, CAM]ll 503 is reset to the beginning of the memory space.

As shown in FIG. 6, CAMI 501 is for storing one work table for performing JOIN operations, and CAM II 502 and CAMll 1503 are for storing the sixth work table.
As shown in the figure, the other work area that performs the JOIN operation
table and the work result of the JOIN operation.
It is not meant to store tables. CAMI501
, CAMn502, CAMIII503, respond to the control signal on the CAM meme memory cell data control signal 5016 specified by address No. 16 on the address signal 5014,
It can be accessed through a data signal line 5015, and at the same time, an associative search function is controlled by a reset signal line 5017, a search signal line 5018, and a search result signal line 5019.

FIG. 7 is a layout diagram of each register in the CAM. CA
, M's associative search function is explained in FIG.

61 is a search data register, 62 is a mask data register, 63 is a memory cell, and 64 is a search result flag.

The memory cell 63 stores data to be associatively searched. Further, the search data register 61 stores data to be searched. The mask data register 62 is
This designates a mask for the portion of the data in the search data register 61 that is not considered during the search. The data in the search data register 61 that corresponds to the part where 0'' is stored in the mask data register 62 is masked and is not compared for data matching during a search (only the part where ``1'' is stored is compared). These registers 61, 62 and memory cell 63 are accessed using signal lines 5014, 5015, and 5016 in the same way as a normal memory.

After data is stored in these registers 61, 62 and memory cells 63, an associative search is performed by turning on the search signal line 5018. That is, among the data in the search data register 61 and the memory cell 63, which data is in the mask data register 62? A match comparison of the data in the portion where +111 is specified is performed simultaneously for all data in the memory cell 63, and 1'' is set in the search result flag 64 corresponding to the row of the memory cell 63 containing the matched data.Search result flag 6
If 1" is set for even one of the search results, this is indicated on the search result signal line 5019 and communicated to the outside. If no 1" is set, it means that there is no search result on the signal line 5019. is shown. Further, the data in the memory cell 63 of the first row for which 1" is set in the search result flag 64 is sent to the data signal line 5015. Next, when the search signal 115018 is turned on again, the search result flag At 64,
In the second scan, the data in the memory cell 63 in the row where 1'' is set is sent to the data signal line s015, and the flag in the first scan is set to γ angle.The reset signal %Jj15017
By turning on, the contents of all memory cells 63 are cleared.

Which of CAM[502 and CAMIII503 should be used as the work table for JOIN operation, and which should be used as JO
Whether or not each work table is used for concatenating IN operations and conjunctive operations is specified by a table designation flag 511. When this flag is '1', CAMII502
CAM as a work table for JOIN operation.
IIJ503 is a work area that stores the results of JOIN operations.
Use each as a table. When it is "0", the opposite is true. The table designation flag 511 is set by latching the signal on the internal data bus 504 through the entire signal line 5021 in response to the latch signal on the signal line 5020. ! , the contents are the signal line 5022
will be sent to.

The external access control unit 510 is the CAM described above.
I501. CAM11502, CAMIII503, mode flag 512, table specification flag 511, CAM
I internal data amount register 521, CAMII internal data amount register 522, CAMIII internal data amount register 523
, mask data register 524yFr: This is a part that controls access from the outside.

In this external access control section 510, the signal on the external address bus 507 is decoded, which of the above is to be accessed is determined, and depending on the read/write signal on the signal line 5023, the access target is a control signal, a latch signal, etc. Send out. Also, at the end of the JOIN operation (when the mode flag 512 changes from on to off), J
An interrupt signal indicating the end of the OIN calculation is sent onto signal line 5024.

The mask data register 524 is connected to the external address bus 50.
It is possible to store data by specifying the address assigned to this register 524 in 7 and generating the write signal t-signal +i5023. That is, the write signal causes the external access control unit 510 to switch the signal line 50
A write signal can be generated at 33 to write the contents of external data bus 506 to the register via signal line s034. Also, the contents of the register are stored on the signal line 503.
The read signal of No. 5 is sent to the internal bus 504.

The search data register 514 is a register that extracts and stores one row of data from two work tables that perform JIN operations and is stored in a CAM with low data weight.The search result register 515 is J
CAMk is the two work tables that perform OIN operations, and has a large amount of data.When an associative search is performed using the same data as the contents of this register as search data, the matched search result data is retrieved from the CAM and stored.

These registers transfer the signals on the internal data bus 504 to the signal line 5026' by the launch signal on the signal line 5025.
It is set by having lunch all day long. The contents are sent to signal lines 5027 and 5028, respectively.

Also, in the search data register 514, the signal 15028
Turn on 2, send its contents to signal line 50281
The entire data can be sent on the internal data bus 504.

The CAMI internal data amount register 521, the CAMII internal data violet data register 522, and the CAMIII internal data amount register 523 each indicate the number of rows in which data is actually stored in the work table stored in the CAM. These registers 521, 522, and 523 can be stored in the external access mode by specifying the address assigned to each register on the external address bus 507 and generating a write signal on all signal lines 5023. It is. That is, in response to the write signal, the external access control unit 510 can generate a write signal on the signal line 5011 and write the contents of the external data bus 506 to the register via the signal line 5012. Similarly, the contents of the register are transferred to the external data bus 5 via the above signal line.
It can also be read out on 06. 1, the data amount of the JOIN operation result is stored in these registers via the signal line 5032. Note that the contents of the register are shown on the signal line 501.
3 is sent up.

The data superposition unit 516 is a part that performs data superposition. Signals of +15027 and 5028 are combined, and the combined resultant signal is sent onto a signal line 5030 in response to a timing signal on a signal line 5029.

The operation control section 517 is a section that controls the operation timing of each section described above and advances the JOIN calculation in the internal operation mode. Mode flag 512, table specification 7 lag 511 status, CAM data amount register 521
, 522, 523, CAMI 501, CAMII 5
02, CAMI [l503, Control is performed by appropriately transmitting signals to other parts according to the result of the associative search (state of the search result signal line 5019).

The reset unit 518 is a part that sets the initial state of this JOIN arithmetic unit, and receives the p1 external reset signal 5sos.
x reset signal, mode flag 512, CAM
1501, CAMII502, and CAMll1503.

8 to 13 are data flow diagrams during operation of the present invention. Below, Fig. 1, Fig. 2, Fig. 7, and Fig. 8~
The operation of this embodiment will be explained with reference to FIG.

Symbol 1 numbers in Figures 8 to 13 refer to Figures 1 and 2.
Same as the figure.

Prior to starting the JOIN operation, the user first connects the signal line 5.
031 to set the initial state of this device. In response to the reset signal, the reset unit 518 generates an internal reset signal on the signal lines 5017.
503 internal registers 61, 62, memory cells 63 .
7 Clear the lag 64 and set the mode flag 512 to '0'' (■ in Figure 8).
”, the bus switching control unit 508 connects the external data bus 506 and internal data bus 504, and the external address bus 507 and internal address bus 505, making it possible to set data from the outside. .Next, the user selects two work pieces for which a JOIN operation is to be performed.
The data in one of the work tables is C.
It is set in the memory cell 63 of the AMl 1502 (■ in FIG. 8). In the following, data to be subjected to JOIN operation is set in CAMl[502, and JOIN operation is performed in CAMITI503.
Let me explain how to set the data resulting from the IN operation. Setting of data to the memory cell 63 of the CAMl 1502 is controlled by the external access control unit 51O. The user sets data on the external data bus 506, sends the address assigned to the memory cell 63 of the CAM 1502 in which the data is to be stored to the external address bus 507, and sends a write signal to the signal line 5023.

Data and addresses are communicated to an internal data bus 504 and an internal address bus 505,
02 signal lines 5015.5014. The external access control unit 510 generates an access control signal 1cAM[502 of the CAMII 502 on the signal line 5016, and stores the data in the memory cell 63 of the CAMII 502. At the same time, change the number of rows set in CAM1502 to
It is set in the AMn internal data amount register 522 (■ in FIG. 8). Setting this register 522 is CAMn5
02, the external access control unit 51
It is done by 0. That is, by sending the data to be transferred to the external data bus 50G, the data is transmitted to the internal data bus 504. Furthermore, CAMn
The address assigned to the internal data amount register 522 is set to the external address bus 507, and the signal line 5023
By transmitting a write signal to
011 and generates a latch signal on internal data bus 504.
The above data is taken in via the signal line 5012.

In the same way, CAM1501, CAMIII503
, CAMI internal data amount register 521, CAMIII internal data amount register 523, table specification flag 511,
The mode flag 512 and mask data register 524 are also accessed by the external access control unit 510 and data is stored therein. The user also CA the remaining data to perform the JOIN operation and the number of rows.
M1501 and CAMI internal data amount register 524, respectively (■ and ■ in FIG. 8).

After setting these data, the user sets data to mask variables other than the common variable values of the two sets of data to be joined (see X and z in FIG. 5) in the mask data register 524. ■) in Figure 9), the CAM (in this case, CAM11502) that stores the data to be subjected to the JOIN operation is used as the JOI of CAn1501.
The table specification 7 lag 511 is designated as the N operation partner (■ in FIG. 9), and the mode flag 512 is set to ``1'''' to start the JOIN operation (■ in FIG. 9).

When l'' is set in the mode flag 512, the bus switching control unit 508 disconnects the external data bus 506, internal data bus 504, external address path 507, and internal address bus 505, and sends the signal i[ to the internal address bus 505.
5006' and sends out the address signal generated by the internal address generation section 509. The operation control unit 517
When the signal on the signal line 5001 becomes 01'' (internal operation mode), the control operation of the JOIN calculation is started.
07 to reset the internal address generating section 509 (for details of the reset, refer to the above-mentioned explanation of the internal address generating section 509). After that, the operation-specific e4i unit 517 performs JOIN according to the operation principle explained in FIG.
Perform calculations.

In the following, the data on which the JOIN operation is performed is CAMI501
, set in CAM1502, CAM150
A case where the amount of data set to 1 is smaller than the amount of data set to CAMI 502 will be explained. First, the operation control unit 517 inputs the contents of the data amount register in the CAM 1501 and the f-sunset register in the CAM 502 via the signal line 5013, compares the contents, and selects the data amount register in the CAM with the larger amount of data. The contents of the mask data register 524 are written into the mask data register 62. That is, the operation control unit 517
5, and generates a data sending signal to mask data register 5.
The contents of 24 are sent to the internal data bus 5 via the signal line 5034.
Send on 04. Then, it generates the address of the mask data register 62 of CAM, 11502, and
By sending a write signal to the signal line 5016 of 502, data is imported into CAMII (Fig. 10 ■)
.

Next, select the CAM with the smaller amount of data, that is, CAM1.
The work table data stored in CAMI 501 is sequentially retrieved line by line from the beginning, and the next process is frozen the number of times specified in CAMI internal data amount register 521.

The operation control unit 517 first clears the CAM III internal data amount register 522 to zero. Next, the signal line 5005
A control signal specifying the memory cell 63 of CAMI 501 is sent to the internal address generator 509 to
generates an address for accessing the memory cell 63 (see the above-mentioned explanation of the internal address generation unit 509 for the method of generating the address). After that, the signal line 501
Generates a memory read control signal to CAMI50
The contents of one row of work table data stored in memory cell 63 of No. 1 are read onto internal data bus 504. The read data is stored in the search data register 514.
By sending a latch signal to the signal line 5025 of
It is latched into the search data register 514 via the signal line 5025 (@ in FIG. 10). Next, the operation control section 517
generates a data sending signal on signal [50282, and sends the contents of search data register 514 onto internal data bus 504 via signal line 50281. The sent data is stored in the search data register 61 in the CAMn502.
is read into. That is, the operation control unit 517
■By generating the address of the search data register 61 of the CAM II 502 and sending a write signal to the signal line 5016 of the CAM II 502, data is fetched into the CAMI [502 (Fig. 11 o cv o). ■5
The access procedures for the registers 02, CAMIII 503, and other registers are the same as described above, and therefore the description thereof will be omitted. After storing the data in the search data register 61 of the CAMn502, the operation control unit 517 stores the data in the search data register 61 of the CAMn502, and then stores the data in the signal line 5 of the CAM1502.
A search signal is generated at 018 to perform an associative search. The operation control unit 517 repeats the following operation and stores the JOIN calculation result in the CAMu 503 until the search result signal (signal m5ox9) of the CAMn 502 turns off.

That is, the search signal (signal line 5o1s)'lt+m is turned on, the search result data is sent onto the internal data bus 504, and is latched into the search result register 515 (11th
Figure @). Thereafter, the contents of the search data register 514 and the contents of the search result register 515 are transferred to the data superimposer 5.
16, and by sending a sending signal to the signal line 5029, the data superimposing result is sent to the internal data bus 504 via the signal Ivi15030. The sent data is stored in order from the beginning of the memory cell 63 of CAMl[503 (O in FIG. 11). Note that the storage address is generated by the internal address generation unit 509 according to the procedure described above. Further, when all the data is delivered, the value of the CAMII internal data amount register 521 is counted up by the signal line 5032. When the operation up to this point is repeated the number of times specified in the CAMI internal data amount register 521, the operation control unit 517 generates a reset signal to the signal line 5017 and the signal line 5004 of the CAM11502, and the CAMII5
02i is cleared and the mode flag 512 is reset to '0'' (see Figure 12).This makes it possible to access the inside of this device from the outside again.The end of the JOIN operation is , mode flag 512 is “1”
At the point when the value changes from 0 to 0, the external access control unit 510
by sending an interrupt signal to signal line 5024,
Be notified. If you want to retrieve the result of the JOIN operation as it is, read the contents of the memory cell 63 of the CAMIIr 503, which is determined by the contents specified by the table specification flag 511 from the outside, by the number of digits indicated in the CAMIII internal data amount register 522. good. Reading from the outside is
As described above, it is controlled by the external access control unit 510. In addition, if three or more conditions are specified in the IF part of the rule in rule-based software, JOIN will be executed again.
If calculations must be performed, the work table data captured for the third or more IFs must be CA
After setting M1501 (O in Figure 13) and the number of lines (■ in Figure 13) in the CAMI internal data amount register 521 (O in Figure 13), set the mask data in the mask data register 524. (O in Fig. 13), designate that CAM as the JOIN target (O in Fig. 13), and start the JOIN operation.

As described above, according to this embodiment, when performing a JOIN operation, it is possible to always minimize the number of comparison processes by starting from table data with a small amount of data and losing all search data. The processing speed of rules can be improved.

〔Effect of the invention〕

As explained above, according to the present invention, two sets of table data for performing a JOIN operation are both stored in the CAM, and the CAM with the smaller amount of data is
Since the search data is retrieved from , the number of comparison processes can be minimized, and rule processing can be executed at high speed. Therefore, in rule-based software, the processing speed of rules with variables can be greatly improved.
It becomes possible to apply a highly readable, understandable, and modifiable logical description method using rules to objects that require handling a large number of states.

[Brief explanation of drawings]

FIG. 1 is a block diagram of a JOIN calculation device showing an embodiment of the present invention, FIG. 2 is an address layout diagram of the memory and CAM in FIG. 1, and FIG. 3 is a functional explanatory diagram of rule-based software.
Figure 4 is an explanatory diagram of rule processing of rule-based software.
FIGS. 5(a) and 5(b) show a conventional "arithmetic device for a rule-based system", and FIGS. 8 to 13 show data flows showing embodiments of the present invention, respectively. Agent: Patent Attorney Katsuo Ogawa 5th (L)

Claims (1)

[Claims] 1. A rule storage unit that stores rules consisting of conditions and conclusions for equipment groups; a status storage unit that stores the status of the equipment group, the content of work to be performed, and the content concluded by the rules; and the status of the equipment group. It has a rule operating unit that compares the information stored in the storage unit with the conditions stored in the rule storage unit and stores the conclusion of the rule whose conditions are satisfied in the status storage unit, and issues control commands to equipment groups. In a rule-based system that determines the condition, for two sets of data consisting of state storage information that match two sets of condition contents described in one of the above rules, JOIN to create both of the above data with the same values corresponding to common items in the condition contents
2, which holds two sets of data for each operation.
the two associative memories, an output buffer for holding the result of the JOIN operation, and a data superimposition means for creating one data by replacing the item value from both data with a meaningful value; Among the means, data is retrieved one by one from the associative memory means with a small amount of data, and using the data as search data, the data in the other associative memory means is associatively searched using the above common item as a search key, and the retrieved data and the data are retrieved one by one. An arithmetic method for a rule-based system, characterized in that a series of operations in which search data is superimposed by the superimposing means and the result is stored in an output buffer is repeatedly executed. 2. The output buffer is an associative memory means, and continuously executes two or more JOIN operations while switching between the associative memory means and another associative memory means. The calculation method of the rule-based system described in Section 1.