JPH0619706B2 - Block computing method of digital computing unit - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a simulation device and a control device using a digital computer, and more particularly to a digital arithmetic device suitable for efficiently and at high speed performing calculations based on block diagrams.

[Background of the Invention] When a blunt dynamic characteristic or a control algorithm as a control device is calculated using a digital computer, it can be classified into two types from the viewpoint of model description style. The first is a numerical expression description format in which FORTRAN is used and the dynamic characteristics of the simulated object are written by differential equations and four arithmetic expressions. The second is a block description style in which the simulation target is composed of several types of arithmetic element blocks such as an integration element and an addition element, and their connection relationship.

In principle, it is possible to obtain the same calculation result by any of the models only with different description modes of the model, but in general when the original equation is expressed by mathematical formulas, differential equations, etc. It seems natural to use the numerical formula description format. However, when the simulated object is originally represented by a block diagram like a plant control device (the control device is configured by combining integral elements, addition elements, etc.), the blunt dynamic characteristic has a transfer function block line. When it is expressed by combining figures, it is natural to use the block description style. Furthermore, the interlocking sequence of the plant, that is, the logical operation processing by the combination of AND, OR, etc., was originally developed for the purpose of numerical operation.
It is not suitable for N, and the block description style is suitable.

As described above, there is a strong need for the block description format, and there are many results that have been used in the past, but the calculation efficiency is poor compared to the numerical expression format, and especially in the online real-time system, the real-time performance is low. It was difficult to secure it, and the load factor was too high. An example of the arithmetic processing procedure according to the block description format of the prior art will be described below with reference to FIGS. 1 to 8.

This model has a harmonic vibration 1 defined by integral elements 12 and 13.
And the waveform generated by multiplication of the harmonic vibration 2 defined by the integration elements 14 and 15 is limited by a limiter. The waveform is shown in FIG. Figure 2 (a) shows harmonic vibration 1
, (B) shows the waveform of harmonic vibration 2, and (c) shows the limiter output.

The model-specific data describing the operation type (integration, multiplication, limiter, etc.), operation constants (integral constant, limiter value of limiter, etc.) of each of the operation elements 11 to 20 shown in FIG. It is stored as a fixed data table in the table 23, and a specific example thereof is as shown in FIG. Further, when the model of FIG. 1 is calculated by an analog device, all the processes are performed in parallel, but when it is calculated by a digital device, it is necessary to specify which element should be calculated and in what order. That is the order table 2
2 is stored as fixed data. Arithmetic control unit 21
Refers to the contents of the element description table 23 in the order specified by the order table 22 (203), passes the operation constant and the connection relation data to the subroutines 25a to 25n corresponding to the specified operation type, and jumps (201). The subroutine executes an operation corresponding to one operation element based on given data, writes the result in the result table 24 (202), and then returns the control to the operation control unit 21. The arithmetic control unit 21 terminates after executing the above processing up to the last element of the order table 22, and is periodically activated after a predetermined period to repeat the arithmetic operation.

According to the above procedure, you can easily create an arbitrary model,
In addition, the table description method is inevitably adopted because it can be added or modified, but as a result, the subroutine jump method can be incorporated as described above. Internally, it also uses the subroutine jump method. At the time of the subroutine jump, it is necessary to save the contents of various registers and deliver the operation constants and connection-related data from the operation control unit 21. The other control processing described above requires far more arithmetic steps than the integral operation and the limiter operation, which are essentially required, and the operation efficiency is poor.

FIG. 4 shows an element description table. (a) is a multiplication element table, which has a fixed-point representation.
(b) is a limiter element table, P1, P2 and P
Floating point representation is used for 3, and fixed point representation is used for others. (c) shows the relationship between P1, P2, and P3.

FIG. 5 shows an example of a concrete main program of the conventional method. Initialization processing and other processing are further added before the processing of FIG. 5, but the processing is not necessary for each computing element and has little effect on the processing time, and therefore is omitted here. The processing contents for each step are omitted because they are shown in the figure, but the processing in the figure requires 14 steps each time. From the figure, jumps to different operation execution subroutines for each operation type. Here, the processing contents of the multiplication execution subroutine are specifically shown in FIG. Also in this case, the processing content is referred to in the figure, but the processing requires 10 steps. From the figure, further to subroutine WTEM 1
7 times in RDEM and each in Fig. 7 (a),
As shown in FIG. 7 (b), processing of 9 steps and 6 steps is required. Therefore, a total of 45 steps of processing is required to execute the processing of one multiplication element. Of these, the operation that is essentially necessary is F in FIG.
M, LDD in FIG. 7 (a), and STD in FIG. 7 (b). In addition, from FIG.
FIG. 8 shows the names and functions of the mnemonics of the instruction words used in the figure. FIG. 8 shows only a part of the command words.

[Object of the Invention]

An object of the present invention is to realize a model to be operated by a digital device, or to easily perform maintenance such as addition / correction, adopt a table description method with few errors, and have a very good operation efficiency. To provide the equipment.

SUMMARY OF THE INVENTION As a result of performing a computation time analysis according to the prior art, the present invention essentially requires a small number of computations for simulating an object, and most of the rest is for processing for controlling the computations. Focusing on what is spent, it is intended to eliminate the control for calculation as much as possible.

Example of Invention

An embodiment of the present invention will be described with reference to FIG. Data necessary for calculation is installed in the card reader 31 and is input to a central processing unit (hereinafter abbreviated as CPU) 34 by an operation from a console input / output device 33, and necessary data files and processing programs are stored in a disk 35. The input data and the processing result are printed by the line printer 32. The above is the equipment required to create the calculation program,
The created calculation program simulates the dynamic characteristics of the plant, and equipment required as a plant simulator is added. That is, the analog input device 361 that reads the set value of the setting device 371 attached to the simulated control panel 37 that is actually operated by the operator, the digital input device 362 that reads the state of the operation switch 372, and the calculation program in the CPU 34. It is composed of an indicator 373 and a recorder 374 driven via the analog output device 363, and an annunciator 375 and a lamp 376 output via the digital output device 364.

Next, the contents of software processed by the CPU 34 and the disk 35 will be described with reference to FIG. In the present embodiment, the gist of the invention, that is, the calculation efficiency is made extremely high, but the descriptiveness of the model is good, that is, consideration is given to easily and easily create a calculation program from a given model. . In FIG. 10, broken line 4
03 means loading the created execution program. Also, arrows 401 and 402 indicate that the program is periodically started every 0.5 seconds.
One calculation element of the model can be described by one card as shown in FIG. 11, and also with a card containing necessary minimum information. FIG. 11 (a) shows a multiplication operation element (MU
L), and (b) shows a limiter element (LIM).
That is, a plant dynamics program of a nuclear power plant composed of 2000 calculation elements is
It is possible to create with a single card. These cards are input from the card reader 31 to the CPU 34,
It is stored in the operation block description table 41. Since the large model has a problem in terms of description space, it will be described with reference to the model in FIG. As shown in FIG. 12, the program for arithmetically processing the model shown in the same figure is an arithmetic block description table that prescribes the arithmetic type, connection relation, arithmetic constants, etc. of each arithmetic block, and the part that controls the arithmetic operations before and after that.
That is, it is configured by a process of determining whether or not the calculation is the first activation and the initial calculation is necessary, and executing the initial calculation if necessary. The former, that is, the arithmetic block description table is input as a source program in a language called a macro language, and is translated into an assembly language when processed by a processing program called a macro processor. On the other hand, the latter process is originally described in assembly language, and is processed in assembly together with the assembly language of the result processed by the macro processor and converted into an execution module. These processes are described in detail in books on system programs, but as an example, "System Programming" by John J. Donovan, McGraw-Hill.
See Chapters 2, 3, and 4 of the company.

In order to process a macro language, a process called macro definition is required in advance. That is, the macro processor converts a macro language into an assembly language, and the macro definition defines the conversion method. As a concrete example of the macro definition in this embodiment, FIG. 13 shows the definition of the multiplication element. This is the first
This is for creating an arithmetic block execution program based on the card input shown in FIG. 1 (a). In the present embodiment, the execution program creation program 43 is the 14th
As shown in the figure, it consists of the following processes. That is, the macro expansion process 431 sequentially inputs the contents of the operation block description table from the beginning, and outputs the contents described in the assembly language as they are. If the contents are described in the operation block description table, that is, from the 11th column of the input card (source program), "MUL", "LI"
If there is a description such as "M", it is output in the form described in the assembly language according to the definition table shown in Fig. 13. In the case of the multiplication element MUL, first the 19th column and thereafter, that is, the first operand as the operand is shown in Fig. 13. Input (multiplier) into area A, and if the content is a natural number, the next processing MUL1
If not, jump to ERR (final command).
Similarly, when the next input is (comma), the next processing MUL
Jump to 2 or else ERR. Thus, in the case of the input shown in FIG. 11 (a), areas A, B,
8, 7, and 4 are input to C, respectively. Next, in MUL5, the content 8 of A is shifted to the left by 1 bit, that is, 16 is assigned to & 0. This is because the operation result of the operation block is described in the floating point format, and each block requires an area of 2 words, so the operation result of element number 8 is stored 16th from the beginning. Is. After setting the doubled input for & 1 and & 2, set the table KEKA for storing the result in & C. The subsequent TEXT is a part for generating an actual assembly language instruction, and LDD, FM, STD (processing functions are shown in FIG. 8) and processing target data thereof are created based on the calculation result. FIG. 15 shows the contents described in assembly language generated as a result. As is clear from this figure, only the instructions essential to the operation are used, and an extremely efficient program is created.

After the above processing is executed for all the data in the operation block description table 41, the operation block execution program 44 is generated by the assembly 432 and the loader 433 and stored in the disk 35. The description of the above processing is ordinary and well known, and the details thereof are based on the above-mentioned books and will be omitted.

Calculation block execution program 44 created in this way
Is activated periodically in 0.5 seconds as shown in Fig. 10,
The arithmetic processing is executed based on the contents of the arithmetic result table 45, and the result is stored in the arithmetic result table 45. The process input / output processing program 46 drives the indicator 373, the recorder 374, the lamp 376 and the like via the process input / output device 36 based on the contents of the calculation result table 45, and also sets the state of the setter 371 and the operation switch 372. input.

According to the embodiment of the present invention, it is possible to program easily and without error from the operation block diagram, so that the productivity of the program is extremely high, and since the number of execution steps is small, a small C
It is possible to manufacture a plant simulator using PU.

The execution time of the model of FIG. 1 is about 1.153 according to the prior art.
Although it took milliseconds, according to the present embodiment, about 0.08
It can be completed in 9 milliseconds, and its execution efficiency has been improved about 13 times. This means that even if a large-scale simulation model is executed, for example, in the case where the conventional method requires 13 CPUs for the calculation of the model, only one CPU is used according to the present invention. Processing becomes possible, and its contribution to the improvement of economic efficiency is extremely large.

〔The invention's effect〕

According to the present invention, when a model to be calculated is realized by a digital device, the input of the model to the digital device can be simply and accurately executed, and the input model is extremely efficient, and the number of execution steps is Since it is small, the required hardware can be downsized and reduced in number. According to one experiment, CPU load factor is 13
It was possible to reduce the size by a factor of one.

[Brief description of drawings]

FIG. 1 is a block diagram of a concrete model, FIG. 2 is a diagram showing the calculation result waveform of FIG. 1, FIG. 3 is an explanatory diagram of a processing method according to the prior art, and FIG. 4 is an example of an element description table. Figure showing,
FIG. 5 is a conventional main program coding example, FIGS. 6 and 7 are conventional multiplication and loading / storing subroutine coding examples, and FIG. 8 is an instruction word used in the coding example. FIG. 9, FIG. 9 is an overall configuration diagram of an embodiment of the present invention, FIG. 10 is a diagram showing the software configuration, FIG. 11 is a diagram showing an arithmetic block input card, and FIG. 12 is an arithmetic block description. Figure showing a table,
FIG. 13 is a diagram showing a calculation block definition table, and FIG.
FIG. 15 is a diagram showing an execution program creation process, and FIG. 15 is a diagram showing a part of the created execution program. 31 ... Card reader, 32 ... Line printer, 33 ... Console input / output device, 34 ... CPU, 35 ... Disk, 36 ... Process input / output device, 37 ... Simulated control panel.

Claims (2)

[Claims] 1. A digital arithmetic unit comprising a central processing unit including a program for fetching a process signal of a plant through a process input device, executing a predetermined arithmetic processing, and outputting a result thereof, wherein The arithmetic processing corresponding to the process input signal executed in the processing device is composed of a plurality of arithmetic blocks, and the connection relation of the operated data necessary for executing the arithmetic type of each arithmetic block is used as a source program. Input,
Prior to executing this block arithmetic processing, a calculation method of an operand for designating an execution instruction type of the central processing unit and its processing target data is defined in advance for each arithmetic type of the source program, and then the arithmetic processing is performed. A digital operation characterized in that an execution program is created in advance by replacing the source program defining the above with the above defined contents, and the central processing unit executes the operation processing of the operation block based on the execution program. Block operation method of device. 2. A block operation method for a digital operation device according to claim 1, wherein the operands are calculated before the execution program is executed. Calculation method.