JPH0822305A - Process controller - Google Patents

Abstract

PURPOSE:To process optimization control operation in parallel to real-time control operation and further to shorten the generation time of optimization control operation application as to a digital controller for plant control. CONSTITUTION:A control variable, a state variable, and an event variable are stored in a two-way memory 7 that a CPU system for control operation and a CPU system 17 for system analysis can share, and then the control variable for control operation which is performed on the CPU system 16 for control operation is optimized from the CPU system 17 for system analysis. Further, the CPU system 17 for system analysis can model and analyze a controlled system with the state variable obtained from the two-way memory 7 and can operate in parallel to the CPU system 16 for control operation.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a digital controller with a system analysis function used for process control of petroleum, chemicals, steel, etc., and in particular, a huge amount of calculation time required for system analysis such as modeling and system identification in real time. The present invention relates to a process control device that does not affect control calculation.

[0002]

2. Description of the Related Art In order to meet the demands for sophistication and diversification of control systems in the field of process control, process control systems have become larger and more complex, and extremely precise and accurate process control is required. There is. Therefore, in recent years, a digital control device has emerged that can perform process control and process identification in a process control system while performing process modeling and system identification to follow changes in process dynamic characteristics and perform predictive control, adaptive control, and other optimizing control calculations. ing.

In the prior art, a fixed period (several hundred ms)
Within a few seconds), the process control amount and target value are sampled, and the control calculation CPU system in the digital control device performs control calculation based on these signal values, while controlling with the same control calculation CPU system. The control variables of the arithmetic unit were optimized. Behind this, there is an improvement in the processing capacity of the microprocessor, and the control device has been able to process control operations at high speeds accordingly, and processing other than control operations has become possible. However, the introduction of optimization control operations applying modeling and system identification has advanced. Most of these modeling and system identification are methods for constructing a mathematical model of a plant. Therefore, there is a procedure for solving a matrix operation with a large degree at the mathematical model creating stage, resulting in a huge amount of calculation and sometimes giving a restriction to the control calculation. There is a constraint that the digital controller has to finish the process of the optimized control technique.
In the case of a compact optimization control algorithm that satisfies this constraint, there is no problem with the conventional method, but when modeling and system identification must be finished with high performance, the processing may not end within the sampling period. In such a case, the control system designer tries to extend the sampling period or divide the optimization control operation into several sampling periods. However, the former method requires careful attention because it cannot control the flow rate process with high response speed well. In the latter case, the calculation algorithm is complicated because the division processing is performed. Also, only a part for performing the optimization control may be provided as one digital control device for systemization, but this is disadvantageous in terms of installation space and cost.

[0004]

In the conventional method, since the application is constructed by combining the optimization control calculation and the control calculation, the processing time of the optimization control calculation is limited to the sampling cycle of the control calculation. A problem arises. This problem can be avoided if various measures are taken as described above. However, it is very troublesome for the control system designer, and the total system cost becomes high.

At the stage of constructing an optimization control calculation algorithm, modeling and selection of a system identification algorithm may be performed by trial and error. Various types of algorithms such as the least squares method, the maximum likelihood estimation method, and the stochastic approximation method have been proposed as modeling and system identification algorithms, and they may be mixed and used according to the characteristics of the process. Therefore, it is convenient if the modeling and system identification algorithms can be described in a high-level language (C language, BASIC language, etc.) so that the user can freely change them. On the other hand, the control calculation part of the real-time control application is in the form of a subroutine such as a control block and a calculation block as is implemented in a conventional control device, and a control calculation program is made by combining these blocks. Once the control calculation program is constructed, the quality of the software is high because it is rarely changed and it is clear, whereas the modeling and system identification programs open the software to users and let them create it freely. Quality tends to be low.

An object of the present invention is to allow a user to freely incorporate a huge amount of calculation modeling or system identification algorithm without being aware of the influence on the real-time control application.
This is to reduce the load on the CPU system for control calculation for executing the real-time control application by operating the optimization control calculations including modeling and system identification in parallel. Further, an object of the present invention is to allow an optimized control calculation program constructed by trial and error to be created even during execution of control calculation, thereby shortening the application creation time.

[0007]

A CPU system for control calculation inputs a process amount from an external sensor or actuator in a constant cycle, performs a control calculation so as to approach a target value aimed by a digital control device, and operates a manipulated variable. Decide
The result is output to the outside every cycle. Since the dynamic characteristics of the process can be known from the state variables such as the process amount, manipulated variable and target value, these state variables are stored in the bidirectional memory and passed to the system analysis CPU system. The system analysis CPU system performs modeling and system identification to obtain the dynamic characteristics of the process. If modeling that characterizes the process dynamics can be performed, it is possible to predict and analyze the change trend of the process, and to find the optimum control variable for the control calculation based on it. By storing the obtained control variable in the bidirectional memory, the optimum control variable can be reflected in the control calculation in the control calculation CPU system. Event variables are prepared in the bidirectional memory so that the variable information can be fetched from the bidirectional memory. When the conditions on the sender side are satisfied, the event variable is set to true and the receiver side is notified that the variable information may be retrieved. The receiving side clears the event variable after extracting the information.

By equipping the digital controller with the means described above, real-time control calculation and optimization control calculation can be performed in parallel.

[0009]

According to the present invention, a modeling or system identification program requiring enormous calculation can be incorporated as long as the memory capacity of the system analysis CPU system allows, and does not affect the control calculation CPU system. Since the system analysis CPU system transmits only control variables to the control calculation CPU system, unless an event variable that becomes a condition for passing the control variable to the control calculation CPU system is activated or the control variable is changed, Even if the system analysis CPU system does not operate, it does not affect the real-time control calculation. Therefore, the operation program of the system analysis CPU system can be freely changed even during real-time control. In addition, since the optimized control operation can change the program contents freely by the user, memory area destruction and program loops and numerical area overflows are likely to occur, but it is independent of the control operation CPU system that performs the control operation. Because
It is possible to minimize adverse effects on the control operation due to these soft abnormalities.

[0010]

Embodiments of the present invention will now be described in detail with reference to the drawings.

FIG. 1 is a block diagram of a digital controller according to the present invention. In the figure, the control calculation CPU system 16 includes an A / D conversion unit 11, a D / A conversion unit 10, a digital input unit 12, a digital output unit 13, a cycle timer unit 8, a real-time control memory on a bus of the control calculation processor. The parts 3, 4, the bidirectional memory 7, the display controller 14, and the key input controller 15 are connected. Further, in the system analysis CPU system 17, the cycle timer unit 9, the system analysis memory units 5, 6, and the bidirectional memory 7 are connected to the bus of the system analysis processor 2.

The real-time control memory units 3 and 4 include
A real-time control application is stored, and the control arithmetic processor 1 decodes this application to perform a process control operation. The analog amount taken from the sensor or actuator in the process is converted into the A / D conversion unit 1
The signal is converted into a digital quantity via 1 and the digital signal from the process is directly taken into the control calculation CPU system via the digital input section 12. These digital quantities are controlled by the cycle timer unit 8 and sampled at a constant cycle. Here, the analog quantity obtained from the process is called a control quantity (PV). Key input control unit 1
Reference numeral 5 captures a set value (SV) and the like. CP for control calculation
The U system derives an operation amount (MV) by performing a control calculation based on the set value (SV) and the control amount (PV). The calculated operation amount is converted into an analog amount by the D / A converter 10 and periodically output to an actuator or the like. The set value, controlled variable, and manipulated variable are displayed on the display or bar graph indicator via the indicator control unit 14.

The conventional digital control device has a control calculation C
The PU system has an adaptive calculation function that is an optimization control calculation as a part of the real-time control application, and the process control is performed with the configuration shown in FIG. Therefore, since the adaptive calculation unit 18 is stored in the real-time control memory unit 3, the control calculation must be stopped when changing the adaptive calculation algorithm.

In the present invention, the optimization control calculation performed by this adaptive calculation unit is assigned to the system analysis CPU system 17 shown in FIG. The system analysis CPU system performs the optimization control operation stored in the system analysis memory units 5 and 6 based on the state variable received from the bidirectional memory 7, and returns the operation result to the bidirectional memory 7. The cycle timer unit 9 controls the timing of taking out a state variable from the bidirectional memory and storing the calculation result in the bidirectional memory.

The real-time control application includes alarm monitoring, digital input / output processing, communication control, etc. in addition to control calculation, and the control calculation among them is determined according to the process characteristics and system configuration. The control calculation is created by a system engineer who constructs the process control system. Therefore, for easy programming, the control operation blocks can be easily assembled by the combination shown in the right diagram of FIG. 3, and this diagram is called a functional connection diagram. The functional connection diagram can be written in a simple language format as shown on the left side of FIG. For example, the fourth line of the function connection program in the left diagram of FIG. 3 is a subroutine for calling the PID operation, and it is not necessary to incorporate the PID operation in detail. In the case of this program, the control amount of the process is stored in AI1, is sent to the PID operation block together with the target value SV fetched from the key input control unit 15, and the result is output to AO1 as the manipulated variable MV. There is. The fifth line, "OPT CTL01 P, I,
D ″ is a control variable P of the PID operation block CTL01,
It is a descriptive statement for fetching I and D from the bidirectional memory and setting them in the control parameters of the PID operation block CTL01. The first and second lines are declaration statements that declare the variables to be stored in the bidirectional memory, and DL (data read) is an area for taking out the 2-byte control variables P, I, and D from the bidirectional memory. Secure. DS (data save) writes 2-byte state variables AI1, SV, AO1 in the bidirectional memory. Since the transfer time is not described for each variable, AI1, SV, AO1 are written to the bidirectional memory every cycle at the control cycle of the control calculation CPU system, and the P, I, and D variables are taken out from the bidirectional memory every cycle. It is reflected in the PID calculation block CTL01. If it is desired to access the bidirectional memory at every specified time interval t, a time variable may be attached to the state variable. That is, AI1 (t), SV
If (t) and AO1 (t) are written, the bidirectional memory is accessed at time interval t. Further, when it is desired to take out the control variable from the bidirectional memory when a certain start condition is satisfied, P (s), I (s) and D (s) are set. Here, s is an event variable and can be taken in when s is true, and is not taken from the bidirectional memory when s is false. When using an event variable, you must reserve a storage area in the bidirectional memory by using a declaration statement like a state variable or control variable.

The function connection program is executed by a device called a function connection proclaimer. After the program is finished, it is written in the real-time control memory units 3 and 4 of the control calculation CPU system in the machine language format. At this time
Create the data definition table shown in as a file. By including this data definition table file in the optimization control calculation program operated by the system analysis CPU system, it becomes possible to connect variables on the bidirectional memory. The data definition table describes a variable name, a type for identifying whether the variable is ASCII or a hexadecimal integer, and an address indicating in which bidirectional memory the variable is stored. A handshake code (HCODE) is 1 byte at the beginning of the area indicated by the address, and an area for storing data specified by the type is secured at the next address. The HCODE automatically sets 1 on the writer side when writing data, and clears 0 when the reader side reads data. If HCODE is not cleared to 0, the writer cannot rewrite. Figure 5 shows how this series of handshakes work,
It is a flowchart showing how the optimum control variable PAopt is replaced with the control variable PAset under the event variable ssw.

As described above, when the system structure shown in FIG. 1 is prepared for the digital controller, the optimized control operation can be performed while performing the input processing, the output processing, and the control operation processing, as shown in FIG. As described above, the load on the control calculation CPU system can be reduced as compared with the conventional method.

[0018]

According to the present invention, not only the processing capacity for modeling and system identification, which requires a huge amount of calculation, is increased, but also the control calculation load is reduced. High-level language modeling and system identification algorithms can be created, and users can easily build optimization control functions, making it easy to introduce system analysis, predictive control, and adaptive control. Also, since the optimization control function using modeling and system identification can be changed during real-time control, application creation time can be shortened and software creation man-hours can be reduced.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing an embodiment of the present invention.

FIG. 2 is a diagram illustrating a flow of state variables in the digital control device.

FIG. 3 is a diagram showing an example of a function connection program.

FIG. 4 is an explanatory diagram of securing a variable area in a bidirectional memory.

FIG. 5 is a flowchart showing an example of a handshake operation.

FIG. 6 is a time chart showing the difference between the conventional method and the present method.

[Explanation of symbols]

1 ... Control operation processor, 2 ... System analysis processor, 3, 4 ... Real-time control memory section, 5, 6 ... System analysis memory section, 7 ... Bidirectional memory, 8, 9 ... Period timer section, 10 ... D / A conversion section, 11 ... A / D conversion section, 12 ... Digital input section, 13 ... Digital output section, 14 ... Display control section, 15 ... Key input control section, 16
... CPU system for control calculation, 17 ... C for system analysis
PU system, 18 ... Adaptive computing unit, 19 ... Control computing unit,
20 ... Process, 21 ... Digital control device.

Claims (2)

[Claims] 1. A digital control device for performing process control, wherein a process control amount and a target value are fetched at regular intervals and control calculations such as constant value control and cascade control are performed to determine an operation amount to a plant. , A control calculation CPU system that operates a real-time control application that monitors a plant, and a process calculation and system identification from the process measurement value and the result calculated by the control calculation CPU system, and an optimized control calculation And a bidirectional memory connected between both CPU systems for communicating information and allowing the CPU systems to operate independently. Is a state variable that contains the target value or process control amount, a control variable used for control calculation, An operation in which an event variable for timing of receiving information between the systems is stored, and a control variable of a control calculation operating in the control calculation CPU system is derived from a result analyzed by the system analysis CPU system from a state variable. And the process control operation can be performed in parallel. 2. A C for control operation is provided by providing a handshake code for each variable of the bidirectional memory.
2. The process control device according to claim 1, wherein a plurality of state variables can be collectively passed to a partner system between a PU system and the system analysis CPU system.