CN104678780A - Ontology-construction-model-based control method of chemical production process - Google Patents

Abstract

The invention discloses a chemical production process control method for constructing a DEVS model based on an ontology, comprising the following steps: using the ontology to describe the basic constituent elements of the DEVS model and the relationship between the basic constituent elements to obtain the DEVS ontology model; constructing the DEVS model according to the chemical production process Produce the model, and use the production model to instantiate the DEVS ontology model to obtain the initial DEVS instance; perform reasoning verification on the initial DEVS instance, and then correct the obtained conflicts to obtain the DEVS instance; map the DEVS instance to the DEVS model code; Apply the DEVS model code to the simulation of chemical production process, and control the chemical production process according to the simulation results. The invention utilizes the strong expression ability of the ontology to establish a DEVS ontology model, and utilizes the powerful reasoning ability of the ontology to verify the DEVS ontology model to ensure the safety of the chemical production process.

Description

A chemical production process control method based on ontology construction model

technical field

The invention relates to the field of chemical production control, in particular to a chemical production process control method based on an ontology construction model.

Background technique

In modern industry, since the production process is usually relatively complicated, modeling and simulation is generally carried out before production to guide the smooth progress of the production process.

Modeling and simulation technology analyzes the behavior and principle of the system by observing and analyzing the behavior and operating principle of the system, establishing an abstract model of the system on a computer or entity, and conducting system experiments on the model. Modeling and simulation technology has become one of the most commonly used technical means to design, analyze and study various systems.

In years of research and development, various technical methods with different characteristics have been produced in the field of modeling and simulation, such as continuous time simulation, discrete time simulation, discrete event simulation, etc. Among them, discrete event system simulation is a very important category. Modeling and simulation methods have been widely used in a large number of man-made systems, such as queuing systems and traffic systems.

Compared with other simulation techniques, discrete event system simulation has two outstanding features:

(1) The state of the system only changes at discrete event points, and discrete event points are generally uncertain;

(2) The state changes in the system often cannot be expressed by mathematical formulas, and are usually described in a way close to natural language such as graphs and tables.

DEVS (Discrete Event Simulation Paradigm) is a method to formally describe the simulation of discrete event systems. In the DEVS modeling method, the modeling object is first decomposed into simple modules, and then multiple simple modules can be aggregated together. In this way, all simple modules can be aggregated layer by layer. These simple modules are DEVS atomic models, and the new modules formed by aggregation are DEVS coupled models.

The input ports and output ports of the DEVS coupling model and the DEVS atomic model are similar, so when a new module is aggregated, the coupling model and the atomic model can be treated equally.

The basic DEVS atomic model is defined as follows:

AtomicDEVS＝<X,s ₀ ,S,Y,δ _int ,δ _ext ,λ,ta>

X: collection of input events;

Y: a collection of output events;

s ₀ : the initial state of the system;

S: a collection of state sequences;

δ _int : S→S, the internal state transition function of the model;

δ _ext : Q×X→S, the external state transition function of the model, where

Q={(s,e)|s∈S,0≤e≤ta(s)};

λ: S→Y, output function;

ta: time advance function;

The basic DEVS coupling model is defined as follows:

coupledDEVS＝<X,Y,D,{M _d },{I _d },{Z _i,d },Select>

X: collection of input events;

Y: a collection of output events;

D: collection of module indexes;

right M _d is a DEVS model;

right I _d is the set of modules that have an influence on module d, i.e.

right Z _i,d is an output transition function describing from i to d;

Z _i,d :X→X _d ,if i=N; represents the connection between the external input of the coupling model and the input of other modules;

Z _i,d :Y _i →Y,if d=N; represents the connection between the output of the module and the output of the coupling model;

Z _i,d :Y _i →X _d ,if d≠N and i≠N; indicates the connection between modules;

The above definitions of the DEVS model are formal descriptions, and many aspects of the DEVS model are difficult to express with mathematical equations. In actual use, the DEVS model exists in the form of code or documents. If there is an error in the development of the DEVS model, it is only possible to find it when the simulation system starts to run.

In the field of artificial intelligence and computing, ontologies are often used to solve knowledge-related problems. There are many definitions of ontology, and one of the better definitions is "ontology is a clear specification of a concept". Ontology is often used to describe entities, concepts, relationships between concepts and the rules of these relationships in a knowledge domain. Ontology has a powerful expressive ability. Ontology uses triples such as "resource", "attribute" and "attribute value" to represent a binary relationship, and any complex relationship can be decomposed by this binary relationship. express.

Contents of the invention

The invention provides a chemical production process control method based on ontology to build a DEVS model. The DEVS ontology model is established by using the ontology's powerful expressive ability, and the DEVS ontology model is verified by using the ontology's powerful reasoning ability to avoid errors from being substituted into the simulation process. , to ensure the safety of the chemical production process.

A kind of chemical production process control method that builds DEVS model based on ontology, comprises the following steps:

(1) Use the ontology to describe the basic components of the DEVS model and the relationship between the basic components to obtain the DEVS ontology model.

The DEVS Ontology Model (DEVS Modeling Ontology, referred to as DEVSMO) is a description of the DEVS model built using ontology.

The DEVS model includes three multivariate sets and four functions, wherein the three multivariate sets are: input event set, output event set, and state set, and the four functions are: external state transition function, internal state transition function , time-advancing function, and output function.

The input event collection, output event collection, and state collection can all be empty, or contain multiple variables, each variable has a value range, and each value range corresponds to a mathematical set.

Variables have different data types, and the data types can be floating-point numbers (float), integers (integer), strings (string), Boolean values (boolean), or more complex data types, such as enumeration (enumeration), queue (queue), etc.

According to the different functions of variables in the DEVS model, these variables can be divided into input variables (InputVariable), output variables (OutputVariable) and state variables (StateVariable).

The DEVS model (that is, the atomic model of DEVS) has many different ports to receive input events, and each port is bound to a variable to save the value of the input port.

The state collection of the DEVS model contains at least a special state variable named "Phase" and a time advancement variable, and most DEVS models contain many user-defined state variables.

In the DEVS model, the internal state transition function contains multiple internal state update behaviors, the external state transition function contains multiple external state update behaviors, and the output function contains multiple output variable assignment behaviors.

In DEVS modeling, a modeling object can define the input event set according to the received external information; define the output event set according to the output information; define the input port according to the channel receiving the external information; define the output port according to the output channel; according to The system state defines the state variables of the system; the process of internal event transfer and external event transfer is defined according to the transition of system state variables.

The DEVS ontology model includes three types of ontology and four functions, wherein the three types of ontology are: input set, output set and state set, and the four functions are: external event transfer function, internal event transfer function, time advancement function and output function.

After defining each part of the DEVS model, find the corresponding classes in the DEVS ontology model for the input event set, output event set, input port, output port, state variable, internal event transfer, and external event transfer.

In the DEVS ontology model, each input set (InputSet) contains several input variables (InputVariable) and input port sets, and the input port set contains several input ports, and each input port (InputPort) is associated with an input variable; each The output set (OutputSet) includes a set of output variables (OutputVariable) and output ports (OutputPort). The set of output ports includes a set of output ports, and each output port is associated with an output variable.

In the DEVS ontology model, the actions of the model may be state initialization (StateInitialization), state update (StateUpdate), input variable assignment (InputVariableAssignment), output variable assignment (OutputVariableAssignment), parameter assignment (ParameterAssignment) and model output (SendOutput), etc. Some of these actions are associated with corresponding operational variables, and these actions express model behavior by changing the values of various variables in the DEVS ontology model. The external event transfer function, internal event transfer function, and output function in the DEVS ontology model all contain many different model behaviors.

(2) Construct a production model according to the chemical production process, and use the production model to instantiate the DEVS ontology model to obtain the initial DEVS instance.

Add the corresponding instance in the corresponding class of DEVSMO to form a DEVSMO instance, which is also the ontology form of the DEVS model.

(3) Perform inference verification on the initial DEVS instance, and then correct the obtained conflicts to obtain the DEVS instance.

Before the model is coded, the ontology reasoning is used to verify the initial DEVS instance, and the conflict information in the DEVS instance in the form of ontology can be found, which is convenient for finding errors, making corrections, and improving the efficiency and accuracy of modeling.

(4) Map DEVS instance to DEVS model code.

According to the DEVS simulation environment, adjust the DEVS code form generated by the mapping as needed, and convert the verified DEVS instance into the DEVS model code, and the DEVS model code can be executed in a specific simulation environment.

Correspond the InputPort, OutputPort, InputVariable, OutputVariable, and StateVariable in the DEVS instance to the input port, output port, input variable, output variable, and state variable in the DEVS model code, respectively, and stateInitialization, StateUpdate, InputVariableAssignment, OutputVariableAssignment, etc. The process of state initialization, state update, input variable assignment, and output variable assignment in the model code corresponds to the final DEVS model code.

At present, the commonly used DEVS simulation environments include CD++, DEVSJAVA, etc. The logic of mapping from DEVS instance to DEVS model code in each simulation environment may be different, but the logic of mapping is roughly the same.

The invention takes the CD++ simulation environment as an example, and maps the DEVS instance to the executable model code in the CD++ environment. In the CD++ environment, a DEVS model is a C++ type. Such a DEVS model contains two files: header file and definition file. These two files are generally named after the name of the class. The header file mainly contains the class. , function function, and data interface declaration carrier file, while the definition file is mainly used to save the implementation of the program. According to the model syntax in CD++, the information in the DEVS instance can be reorganized into header files and definition files.

(5) Apply the DEVS model code to the simulation of chemical production process, and control the chemical production process according to the simulation results.

The present invention builds a DEVS model-based chemical production process control method based on an ontology, and uses the ontology to build a DEVS model modeling ontology DEVSMO, which enhances the visualization of modeling and improves modeling efficiency; uses the ontology reasoning engine to verify the DEVS instance, making up for It overcomes the defect that the traditional DEVS modeling cannot verify the obtained model, and can find errors in the modeling stage to ensure the smooth progress of the chemical production process.

Description of drawings

Fig. 1 is the flowchart of modeling in the chemical production process control method based on ontology construction DEVS model of the present invention;

Fig. 2 is a structural schematic diagram of a tank farm in an oil refinery;

Figure 3 is a schematic diagram of the basic components of a DEVS model built using an ontology;

Figure 4 is a schematic diagram of using the ontology reasoning engine to verify the initial DEVS instance;

Fig. 5 is a schematic diagram of the simulation results of the tank farm (the abscissa is the tank capacity, and the ordinate is time).

Detailed ways

The chemical production process control method based on ontology-based DEVS model construction of the present invention will be described in detail below in conjunction with the accompanying drawings.

As shown in Figure 1, a chemical production process control method based on ontology-based DEVS model construction includes the following steps:

(1) Use the ontology to describe the basic components of the DEVS model and the relationship between the basic components to obtain the DEVS ontology model. The relationship structure between the DEVS ontology model and the DEVS model is shown in Table 1 and Figure 3.

Table 1

DEVS model DEVS ontology model input event collection input collection output event collection output collection

state set state set external state transition function External Event Transfer Function internal state transition function internal event transfer function time advance function time advance function output function output function

(2) Construct a production model according to the chemical production process, and use the production model to instantiate the DEVS ontology model to obtain the initial DEVS instance.

Taking a tank farm of an oil refinery shown in Figure 2 as an example of a production model, construct an initial DEVS instance. In the tank farm shown in Figure 2, tank Tk02601 and tank Tk02602 are merged into tank Tk02603 through the confluence point Jc00009. The basic goal of tank farm scheduling is to control the discharge of the three tanks so that the liquid level of each tank in the tank farm at a safe level.

In the tank area shown in Figure 2, there are two types of chemical devices: tanks and confluence points. DEVSMO is used to establish the DEVS models of tanks and confluence points respectively.

Each tank assembly contains a feed side draw and an output side draw, with corresponding feed flow ports and output flow ports present in the original DEVS instance. In the initial DEVS example, there are corresponding control signals for the flow rate of the feed line and the flow rate of the discharge side line. The state of the tank device includes three parts: the flow rate of the feed side line, the flow rate of the discharge side line and the tank capacity.

Events in the initial DEVS instance of the tank unit include a change in the feed flow control signal and a change in the output flow control signal. When the feed flow control signal changes, the external event transfer is triggered, the feed flow in the state variable is updated, and enters a transient process; after a period of time, the internal event transfer is triggered, at this time, the tank capacity in the state variable is updated, into a steady state process. When the output flow control signal changes, it also triggers the transfer of external events and outputs the new output flow. At this time, it is in a transient process. After a period of time, the internal event transfer is triggered to update the tank capacity in the state variable, and then It is in the steady state process and is output with the new output flow.

When using the DEVS ontology model to create the initial DEVS instance of the tank, it is necessary to add a feed side line port "port_in_1", a discharge side line port "port_out_1" and a discharge side line flow control port "port_control_1", the initial DEVS instance of the tank device Also contains an external state transition function, an internal state transition function and an output function.

The external state transfer function is mainly used to deal with the change of the input flow change control signal and the change of the output flow control signal; the internal state transfer function is mainly used to switch the transient process of the tank device into a steady state process; the output function passes through the discharge side line port Output the output flow rate.

A confluence contains multiple feed sidings and an effluent siding, and the initial DEVS instance for a confluence also contains multiple feed flow ports and an effluent flow port. When the feed flow rate changes, an external event transfer is triggered to update the input flow rate. In the output function, the new flow is used to calculate the outflow, which is the sum of all inflows.

The initial DEVS instance of the confluence point is very simple, including multiple feed side ports and one discharge side port. Similarly, the initial DEVS instance of the confluence point also includes external state transition functions, internal state transition functions, and output functions, but in There is no transition process from transient state to steady state in the confluence point, so only the external state transition function is needed to deal with the change of input flow. There is no need for state transition in the internal state transition, and the output function can output the instantaneous output flow.

(3) Perform inference verification on the initial DEVS instance, and then correct the obtained conflict to obtain the DEVS instance;

The present invention uses the ontology to describe the DEVS model to obtain the DEVSMO. In DEVSMO, each basic component element in DEVS and the relationship between each basic component element are described. If an error occurs when the initial DEVS instance is constructed, it will be used for ontology reasoning. error.

Figure 4 shows the error message prompted by the inference engine when the input variable "in_value_1" does not point to the input port "port_in_1". In the error message, the place circled by box a is the inconsistency of the inference engine during the inference process . The line "in_value_1Type InputVariable" is circled, which means that there is a reasoning conflict here, that is, in the modeling process, the type of in_value_1 is InputVariable, but according to the relationship between in_value_1 and other basic components, it is inferred that the type of in_value_1 is not InputVariable . In DEVSMO, InputVariable is a Variable and must point to an input port, that is, the meaning expressed by the line "InputVariable EquivalentToVariable and(referToPort exactly 1InputPort)". In the case shown in Figure 4, it is only necessary to add a "referToPort" relationship pointing to "port_in_1" in "in_value_1".

(4) Map DEVS instance to DEVS model code;

(5) Apply the DEVS model code to the simulation of chemical production process, and control the chemical production process according to the simulation results.

Put the DEVS model code obtained in step (4) into the simulation engine, input the on-site control commands of the tank farm into the simulation environment, and obtain the simulation results shown in Figure 5. After comparing with the field results, the simulation results can be better of fitting.

In Figure 5, it can be seen that the tank capacities of the three tank devices are all within the range of 20,000, which verifies the correctness of the control command, and the control command can be issued.

In traditional simulation modeling, DEVS models are often directly given in the form of codes, which are prone to errors. These errors are often not easily detected during model checking, resulting in serious problems when using simulation results to guide the actual production process. Accidents, when using DEVSMO to model, the model can be verified in advance, avoiding errors being brought into the simulation process, so that the simulation results have better guidance for actual production.

Claims (6)

1. A chemical production process control method based on ontology construction DEVS model, is characterized in that, comprises the following steps: (1) Use the ontology to describe the basic components of the DEVS model and the relationship between the basic components to obtain the DEVS ontology model; (2) Construct a production model according to the chemical production process, and use the production model to instantiate the DEVS ontology model to obtain the initial DEVS instance; (3) Perform inference verification on the initial DEVS instance, and then correct the obtained conflict to obtain the DEVS instance; (4) Map DEVS instance to DEVS model code; (5) Apply the DEVS model code to the simulation of chemical production process, and control the chemical production process according to the simulation results. 2. the chemical production process control method based on ontology construction DEVS model as claimed in claim 1, is characterized in that, described DEVS model comprises three multi-variable sets and four functions, wherein three multi-variable sets are respectively: input Event collection, output event collection and state collection, the four functions are: external state transition function, internal state transition function, time advance function and output function. 3. the chemical production process control method based on ontology construction DEVS model as claimed in claim 2, is characterized in that, described DEVS ontology model comprises three types of ontology and four functions, wherein three types of ontology are respectively: input set, output Collection and state collection, the four functions are: external event transfer function, internal event transfer function, time advancement function and output function. 4. the chemical production process control method based on ontology construction DEVS model as claimed in claim 3, it is characterized in that, in DEVS ontology model, each input collection comprises some input variables and input port collection, includes several in the input port collection Input ports, each of which is associated with an input variable. 5. the chemical production process control method based on ontology construction DEVS model as claimed in claim 4, it is characterized in that, in DEVS ontology model, each output collection comprises some output variables and output port collection, includes several output port collections Output ports, each output port is associated with an output variable. 6. The chemical production process control method for constructing a DEVS model based on an ontology as claimed in claim 5, wherein the state set of the DEVS model contains at least a special state variable named "Phase" and a time advancement variable.