JP6962042B2 - Simulation equipment and simulation method - Google Patents

Description

The present invention relates to a simulation device and a simulation method for predicting an operating state of a plant device.

Conventionally, an apparatus for simulating a process by a plant apparatus in various plants such as a chemical plant, a power generation plant, and a heat treatment plant has been used. For example, Patent Document 1 below discloses a process simulation based on a model of a target process. This process simulation is executed on a system such as a computer different from the equipment of the target plant (hereinafter referred to as "actual plant").

On the other hand, in recent years, there has been active research on simulation technology that operates a simulator in parallel with an actual process and adapts the operation of the simulator to the operation of an actual process in an actual plant (hereinafter referred to as "actual process"). Has been done. For example, in the following Patent Documents 2 and 3 and the following Non-Patent Document 1, tracking that operates so that the measurement data (temperature, pressure, flow rate, composition, etc.) of each physical quantity obtained from the actual process and the simulation result are synchronized with each other. It is disclosed that a simulator is used to simulate the future operating state of the plant equipment.

Japanese Patent No. 6043348 Japanese Patent No. 4789277 Japanese Patent No. 521290

Tetsuya Otani, "Plant Operation Innovation by Online Simulator", Measurement and Control, Vol. 47, No. 11, p. 927-932, November 2008

However, in the above-mentioned technology using the conventional tracking simulator, it is possible to predict the operating state of the plant equipment in the future in consideration of future changes in the operating conditions of the actual plant and changes in disturbance conditions such as future weather changes. It's difficult to do. In addition, it is difficult to predict the operating state of the plant equipment in the future in consideration of the deterioration of the characteristics of the actual plant.

Therefore, the present invention has been made in view of such a problem, and an object of the present invention is to provide a simulation device and a simulation method capable of accurately predicting the operating state of a plant device in the future.

In order to solve the above problems, the simulation device according to one aspect of the present invention is a simulation device that predicts the operating state of the plant device, and adjusts the simulation model set in tracking the operating state by an external tracking simulator device. Operation using a simulation model based on the parameter storage unit that stores parameters in association with the information on the operating conditions of the plant equipment obtained from the actual plant during tracking by the tracking simulator equipment and the operating conditions of the plant equipment to be predicted. It includes a prediction simulator unit that predicts the state and an adjustment parameter setting unit that dynamically sets the adjustment parameters of the simulation model in the prediction simulator unit, and the adjustment parameter setting unit has operating conditions similar to the operating conditions to be predicted. The adjustment parameters associated with the information are acquired from the parameter storage unit, and the adjustment parameters of the simulation model of the prediction simulator unit are set based on the acquired adjustment parameters.

The simulation method according to another aspect of the present invention is a simulation method for predicting the operating state of the plant device, and the parameter accumulating unit is the adjustment parameter of the simulation model set in the tracking of the operating state by the external tracking simulator device. Using a simulation model, the simulation device uses a simulation model based on the steps of accumulating information related to the operating condition information of the plant device obtained from the actual plant during tracking by the tracking simulator device and the operating conditions of the plant device to be predicted by the simulation device. The step of predicting the operating state and the step of dynamically setting the adjustment parameters of the simulation model by the simulation device are provided, and the step of dynamically setting the adjustment parameters is an operating condition similar to the operating condition to be predicted. The adjustment parameters associated with the information in are acquired from the parameter storage unit, and the adjustment parameters of the simulation model are set based on the acquired adjustment parameters.

According to the simulation device or simulation method having such a configuration, the history information of the adjustment parameters of the simulation model set in the past in the external tracking simulator device is associated with the information of the operating conditions of the plant device of the actual plant by the parameter storage means. The simulation device acquires the adjustment parameters associated with the information of the operation conditions similar to the operation conditions to be predicted, and the operation state of the plant equipment is used by using the simulation model in which the adjustment parameters are dynamically set. Is expected. As a result, adjustment parameters set under operating conditions similar to the section to be predicted are acquired from the execution results of the tracking simulator device in the past, and the operation state is predicted using the adjustment parameters. Therefore, the operation state of the plant device is predicted. Can be predicted accurately. As a result, it can be used to create an appropriate plant operation plan or an appropriate plant equipment maintenance plan.

The parameter storage unit stores the adjustment parameters in association with the disturbance information which is the information of the operating conditions related to the disturbance, and the adjustment parameter setting unit acquires the adjustment parameters associated with the disturbance information similar to the disturbance information to be predicted. It is also preferable.

By adopting such a configuration, it is possible to accurately predict the operating state of the plant equipment by performing a simulation in consideration of the operating area regarding the disturbance for each process of each plant.

Further, the parameter storage unit stores the adjustment parameter in association with the control amount information which is the information of the operating condition regarding the control amount for the plant device, and the adjustment parameter setting unit associates the adjustment parameter with the control amount information similar to the control amount to be predicted. It is also preferable to obtain the adjusted adjustment parameters.

If such a configuration is adopted, the operating state of the plant device can be accurately predicted by performing a simulation considering the operating area regarding the control state for each process of each plant.

Further, the parameter storage unit stores the adjustment parameter in association with the operation time information which is the information of the operation condition regarding the operation time of the plant device, and the adjustment parameter setting unit associates the adjustment parameter with the operation time information similar to the operation time of the prediction target. It is also preferable to obtain the adjusted adjustment parameters.

If such a configuration is adopted, the operating state of the plant equipment can be accurately predicted by performing a simulation in consideration of the operating area regarding the time factor such as the deterioration state for each process of each plant.

It is also preferable that the adjustment parameter setting unit determines the similarity by calculating the distance between the operating condition to be predicted and the information of the operating condition.

In this case, it is possible to perform a simulation that matches the operating area for each process of each plant.

Furthermore, the adjustment parameter setting unit acquires a plurality of adjustment parameters from the parameter storage unit, and uses machine learning based on the information on the operating conditions of the plant equipment associated with the plurality of adjustment parameters and the adjustment parameters of the simulation model. It is also preferable to set adjustment parameters.

In this case as well, it is possible to perform a simulation that matches the operating area for each process of each plant.

Furthermore, a control amount prediction unit that predicts the control amount by the controller for the plant device based on the output of the prediction simulator unit is further provided, and the prediction simulator unit operates based on the control amount predicted by the control amount prediction unit. It is also preferable to predict the state.

In this case, even when the control amount is controlled by the controller in the actual plant, the operating state of the plant apparatus can be accurately predicted by predicting the control amount by the controller.

Furthermore, it is also preferable that the adjustment parameter setting unit acquires the adjustment parameter associated with the information of the operating condition similar to the control amount predicted by the control amount prediction unit.

In this case, the operating state of the plant equipment can be predicted more accurately by performing a simulation in consideration of the operating area regarding the control amount of the process of each plant. can.

It is also preferable that the prediction simulator unit predicts the operating state based on the prediction result of the control amount prediction unit in which the algorithm or parameter for predicting the control amount is dynamically or statically changed.

By adopting such a configuration, it is possible to compare the operating states of the plant devices in a plurality of controlled states by performing the simulation while changing the control states of the processes of the individual plants.

According to the present invention, it is possible to accurately predict the operating state of the plant equipment in the future.

It is a structural schematic diagram of the simulation system which concerns on one preferred embodiment of this invention. It is a block diagram which shows the hardware composition of the computer 100 which constitutes each apparatus of FIG. It is a block diagram which shows the functional structure of the prediction simulator apparatus 7 of FIG. It is a graph which shows the data distribution of the setting history data which is searched by the adjustment parameter setting unit 7c of FIG. It is a graph which shows the data distribution of the setting history data which is searched by the adjustment parameter setting unit 7c of FIG. It is a flowchart which shows the operation procedure of the simulation of the real process by the simulation system 1 of FIG. It is a flowchart which shows the operation procedure of the tracking simulator apparatus 3 of FIG. It is a flowchart which shows the operation procedure of the prediction simulator apparatus 7 of FIG. It is a graph which shows the time change of the state value obtained by the synchronous simulation and the prediction simulation in the conventional method. It is a graph which shows the time change of the state value obtained by the synchronous simulation and the prediction simulation in the conventional method. It is a graph which shows another time change of the state value obtained by the synchronous simulation and the prediction simulation in the conventional method. It is a graph which shows another time change of the state value obtained by the synchronous simulation and the prediction simulation in the conventional method. In the present embodiment, the time change of the state value obtained by the synchronous simulation and the prediction simulation when the control amount of the past operation time is added to the search query point is shown. It is a block diagram which shows the functional structure of the prediction simulator apparatus 7A which concerns on a modification. It is a block diagram which shows the functional structure of the prediction simulator apparatus 7B which concerns on a modification. It is a flowchart which shows the operation procedure of the prediction simulator apparatus 7B of FIG. It is a flowchart which shows the modification of the operation procedure of the simulation of the real process by the simulation system 1. It is a flowchart which shows the modification of the operation procedure of the tracking simulator apparatus 3. It is a structural schematic diagram of the simulation system which concerns on the modification of this invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. In the description of the drawings, the same or equivalent elements are designated by the same reference numerals, and duplicate description will be omitted.

FIG. 1 is a schematic configuration diagram of a simulation system according to a preferred embodiment of the present invention. The simulation system 1 shown in the figure is a simulation apparatus according to the present embodiment, and is a process by a plant apparatus (hereinafter, referred to as “actual plant”) in various plants such as a chemical plant, a power generation plant, and a heat treatment plant (hereinafter, referred to as “actual plant”). It is a computer system that performs simulations related to (referred to as "real process"), and simulation processing is realized by transmitting and receiving data via a communication network by a plurality of devices. Specifically, the simulation system 1 may target a gas turbine, a boiler, a heat exchanger, a glass melting furnace, an aircraft engine, or the like as a plant device, but the plant device to be simulated is not limited to this. The simulation system 1 includes a tracking simulator device 3, a database device 5, and a prediction simulator device 7. The tracking simulator device 3, the database device 5, and the prediction simulator device 7 are configured to be capable of transmitting and receiving data to and from each other via a wired or wireless communication network such as a LAN. It should be noted that these devices may be integrated by a common device in any combination, or conversely, may be configured by a plurality of devices. Further, these devices do not necessarily have to be installed in the actual plant, and may be installed in a place away from the actual plant.

FIG. 2 is a block diagram showing a hardware configuration of a computer 100 constituting a tracking simulator device 3, a database device 5, and a prediction simulator device 7. As shown in the figure, the computer 100 physically outputs the CPU 101, the RAM 102 and ROM 103 which are the main storage devices, the input device 104 which is an input device such as an input key and a touch sensor, a touch panel display, and a liquid crystal display. It is configured as a computer system (information processing processor) including a device 105, a communication module 106 which is a data transmission / reception device, an auxiliary storage device 108 such as a semiconductor memory, and the like. The processing functions described later of each device are the input device 104 and the output device under the control of the CPU 101 by loading one or a plurality of predetermined computer software on the hardware such as the CPU 101 and the RAM 102 shown in FIG. This is realized by operating the 105 and the communication module 106, and reading and writing data in the RAM 102 and the auxiliary storage device 108. Each device may be composed of one computer 100 or a plurality of hierarchically connected computers 100.

The tracking simulator device 3 is configured to be capable of transmitting and receiving data via a communication network NW with an actual plant 9 including a plant device, a control device for controlling the plant device (control computer, control panel, etc.). It has a function to estimate the operating state of the plant equipment. For example, the tracking simulator device 3 changes over time such as the physical and chemical internal states of the actual process, for example, the temperature of a certain part in the actual plant 9, the pressure, the flow rate, and the composition of a certain part in the actual plant 9. Estimate the physical quantity to be used.

The tracking simulator device 3 acquires the current operating condition data (information on the operating conditions) and the current operating state data regarding the actual process from the actual plant 9. Examples of the operation condition data to be acquired include a control amount for controlling the plant apparatus, disturbance information regarding the disturbance in the actual plant 9, and the like. The control amount is a control amount set by being directly or indirectly operated by the user in the control device of the actual plant 9 (target temperature by heater control, flow rate of cooling water, input amount of raw material, etc.), and the control amount of the actual plant 9. It is the amount of control when automatically controlled by the control device. The disturbance information is information on the characteristics of the plant that cannot be operated by the user acquired by the sensor device in the actual plant 9, and is, for example, the temperature, atmospheric pressure, humidity, amount of solar radiation, and the characteristics (specific gravity) of the fuel in the power generation plant. Alternatively, it is information such as calorific value). The operation state data to be acquired is data showing the current physical and chemical internal states of the actual process of the actual plant 9, and for example, changes over time such as pressure, flow rate, composition, etc. of a certain part in the actual plant 9. It is the data which shows the physical quantity to do.

によって表される動的モデルであり、ｘは実プロセスの運転状態を示す値であり、ｕは実プロセスの制御量を示す値であり、ｄは実プロセスの外乱を示す値であり、ｐは実プロセスを記述するシミュレーションモデル内の物理的・化学的係数としての調整パラメータである。値ｘ，ｕ，ｄ，ｐは、それぞれ、１次元の変数あるいは多次元ベクトルである。この調整パラメータｐを調整することにより、実プラント９内の設備の性能劣化、性質変化等がシミュレーションモデルによって表現される。例えば、この調整パラメータｐの調整により、実プラント９の圧縮機性能の変化を表現することができる。ここで、上記式（１）のシミュレーションモデルは連続時間モデルとして微分方程式で記述されているが、離散時間モデルとして差分方程式で記述されてもよい。 The tracking simulator device 3 executes a tracking simulation (synchronous simulation) (hereinafter, this process is also simply referred to as "synchronization" or "tracking") based on the operating condition data and the operating state data obtained from the actual plant 9. do. The tracking simulator device 3 is equipped with a dynamic simulation model designed in advance based on the principles of physical and chemical phenomena of the target process, experimental results of the target process, and the like. This simulation model uses the nonlinear function f to formulate the following equation (1);

In the dynamic model represented by, x is a value indicating the operating state of the actual process, u is a value indicating the control amount of the actual process, d is a value indicating the disturbance of the actual process, and p is a value indicating the disturbance of the actual process. Adjustment parameters as physical and chemical coefficients in the simulation model that describes the actual process. The values x, u, d, and p are one-dimensional variables or multidimensional vectors, respectively. By adjusting this adjustment parameter p, the performance deterioration, property change, and the like of the equipment in the actual plant 9 are expressed by the simulation model. For example, by adjusting the adjustment parameter p, it is possible to express a change in the compressor performance of the actual plant 9. Here, the simulation model of the above equation (1) is described by a differential equation as a continuous-time model, but may be described by a difference equation as a discrete-time model.

Specifically, the tracking simulator device 3 acquired the operating condition data and the operating state data by using a simulation model in which the operating condition data obtained from the actual plant 9 was applied to the values u and d of the nonlinear function f. _{The estimated value x cal} of the quantity x indicating the operating state of the actual plant 9 at the time point is calculated. Further, a parameter adjusting unit 3a is mounted on the tracking simulator device 3, and in the parameter adjusting unit 3a, the estimated value x _{cal by simulation using a simulation model is the measured value y obs} indicated by the operating state data from the actual plant 9. The adjustment parameter p in the tracking simulator device 3 is set so as to follow (track). That is, the parameter adjusting unit 3a continuously acquires the operating condition data and the operating state data from the actual plant 9 in a timely manner, and continuously estimates the operating state value x using a simulation model that reflects the operating condition data. do. Further, the parameter adjusting unit 3a calculates the observed value y _{cal of the operating state value x cal} estimated by using the observation formula represented by the following equation (2). This observation formula is a non-linear function for expressing the value obtained as a result of observing the operating state with the sensor device of the actual plant 9.
y = g (x) ... (2)
Furthermore, the parameter adjustment portion 3a, while adjusting the adjustment parameter p of the simulation model repeated estimation value _{x cal,} observed value _{y cal} calculated from the value _{x cal} is operating condition data used to determine the value _{x cal} The adjustment parameter p is adjusted so as to match the _{measured value yobs} indicated by the operating state data corresponding to. At this time, as an adjustment method for the adjustment parameter p, a method for estimating the adjustment parameter using the output error method described in Japanese Patent No. 4789277, Japanese Patent Application Laid-Open No. 2009-163507, etc. For the method of estimating adjustment parameters using the PID regulator described in "Plant Operation Innovation by Simulator", Measurement and Control, Vol. 47, No. 11, p.927-932, published in November 2008, or using the Kalman filter. The method of estimating the adjustment parameters that has been used is adopted.

In addition, the tracking simulator device 3 acquires the setting history data of the adjustment parameter p set by repeatedly executing the tracking simulation based on the operation condition data and the operation state data from the actual plant 9 at the time of the tracking simulation. It is associated with the condition data and stored in the database device 5. Specifically, the tracking simulator device 3 uses the repeatedly set adjustment parameter p, the value x _cal obtained by the tracking simulation at the time of setting the adjustment parameter p, and the observed value y _cal for the operation. Control amount (control amount information) u _obs and disturbance value (disturbance information) _{dobs included in the condition data, measured value y obs} of the operating state obtained from the actual plant 9 corresponding to the operating condition data, and tracking. It is stored in the database device 5 in association with the operation time information (operation time information) of the actual plant 9 simulated by the simulation. This setting history data is stored in the parameter storage unit (parameter storage means) 5a of the database device 5 at the timing when the tracking simulation is executed by the tracking simulator device 3. Further, the database device 5 is provided with a data management unit 5b, and the data management unit 5b enables output and management of setting history data. That is, the data management unit 5b outputs the setting history data to the output device 105 (FIG. 2), and deletes a part of the setting history data in response to the instruction input by the user using the input device 104 (FIG. 2). do. By having such a function, the user can know the state of the process. It is also possible to delete the setting history data that the user considers unnecessary for predicting the operating state. For example, delete the data when synchronizing based on the apparently incorrect measured value due to a malfunction of the measurement system, or delete some of the similar data in the past due to the large amount of data in the database to secure the capacity of the database. The past data becomes useless due to the addition of new equipment to the actual plant 9, so the past data is deleted. The data with poor followability to the measured value (difference | _yobs- y _cal | is larger than the threshold value. Data) can be deleted.

Next, the functional configuration of the prediction simulator device 7 will be described in detail. FIG. 3 is a block diagram showing a functional configuration of the prediction simulator device 7. The prediction simulator device 7 includes a process simulator unit (prediction simulator unit) 7a, a control model unit (control amount prediction unit) 7b, and an adjustment parameter setting unit 7c as functional components.

The process simulator unit 7a uses a simulation model based on the operating conditions of the actual plant 9 to be predicted set by the user using the input device 104 (FIG. 2), and uses a simulation model for the future prediction period of the actual process. The process of predicting (prediction simulation) is repeatedly executed. Specifically, the process simulator unit 7a has a value d set by the user indicating the disturbance of the actual process, an adjustment parameter p set by the adjustment parameter setting unit 7c, and a control amount u output from the control model unit 7b. And, in the same manner as the tracking simulator device 3, the value x and the observed value y of the operating state of the actual process are predicted by using the simulation model mounted in advance. The initial value x of the operating state value x of the simulation model used by the process simulator unit 7a may be duplicated in advance from the tracking simulator device 3 into the prediction simulator device 7. Then, the process simulator unit 7a outputs the simulation result based on the predicted value x to the output device 105 (FIG. 2), and outputs the value x or the observed value y = g (x) to the control model unit 7b each time. Output to.

The control model unit 7b predicts the control amount u set in the actual plant 9 by simulating the control device in the actual plant 9. Specifically, a mathematical model relating to the operation of the control device is mounted on the control model unit 7b, and the control model unit 7b is based on the value x of the operating state of the actual process output from the process simulator unit 7a. The predicted value of the control amount u is calculated using a mathematical model. Then, the control model unit 7b feeds back the calculated predicted value of the control amount u to the process simulator unit 7a. This makes it possible to predict the operating state in consideration of the control of the actual control device by the process simulator unit 7a. The mathematical model mounted on the control model unit 7b is preferably a model that operates in the same manner as the actual control device in the actual plant 9, but what happens when the algorithm or parameters of the actual control device are changed? It may be a model in which algorithms or parameters different from those of the actual control device are set in order to know by simulation whether the driving tendency is different. Specifically, the predicted value of the control amount u calculated while changing the algorithm or parameter little by little (dynamic change) may be fed back to the process simulator unit 7a, or the algorithm or parameter may be fed back to a certain algorithm or a certain parameter. The predicted value of the control amount u calculated while being set to (static change) may be fed back to the process simulator unit 7a.

The adjustment parameter setting unit 7c has set history from the database device 5 based on the operating conditions in the prediction period of the prediction target actual plant 9 set by the user and the prediction period of the prediction target actual plant 9 set by the user. By acquiring and referring to the data, the adjustment parameter p of the simulation model of the process simulator unit 7a is dynamically set. The adjustment parameter setting by the adjustment parameter setting unit 7c is executed by switching to either the first setting method or the second setting method described below, depending on the user's selection.

First, the first setting method is as follows. That is, the adjustment parameter setting unit 7c searches for a combination _{of the disturbance value d 1} and the control amount u ₁ and the operation time t ₁ based on the operation conditions at the operation time during the prediction period of the set prediction target. The data closest to the search query point is searched and extracted from the setting history data stored in the parameter storage unit 5a of the database device 5. Then, the adjustment parameter setting unit 7c sets the adjustment parameter p included in the extracted nearest setting history data as the adjustment parameter p of the simulation model of the process simulator unit 7a. FIG. 4 shows the data distribution of the setting history data searched by the adjustment parameter setting unit 7c. In this way, when the _{outside air temperature is set as the disturbance value d 1} included in the search query point and the input amount of the raw material is set as the _{control amount u 1} included in the search query point, the search query point is set. setting history data P ₂ a distance to have the closest outside temperature and input amount is extracted. As a result, the adjustment parameter p associated with the operation condition data similar to the operation condition of the prediction target included in the search query point is set to the adjustment parameter of the simulation model of the process simulator unit 7a each time.

により計算することにより評価する。上記式（３）において、値ｕ_２，ｄ_２，ｔ_２は、それぞれ、設定履歴データＰ_２に含まれる制御量、外乱値、運転時刻であり、“||ｘ−ｙ||”は、２つのベクトルｘ，ｙから計算されるユークリッドノルム、すなわち、差分ベクトルの長さを表し、ａ，ｂ，ｃは非負パラメータを示す。つまり、上記式（３）において、第１項は制御量の違いに対応する距離を、第２項は外乱値の違いに対応する距離を、第３項は時間的な距離をそれぞれ表している。このような距離の演算式を用いることで、調整パラメータ設定部７ｃは、予測対象の運転条件である外乱値及び制御量に類似する運転条件データと、運転時刻が近い運転時間情報とに関連付けられた調整パラメータｐを、パラメータ蓄積部５ａから取得することができる。運転時刻が近い調整パラメータｐを取得するのは、将来の予測対象の運転時刻に近い最新の運転時刻に同期（トラッキング）により設定された調整パラメータを用いるほうが運転状態のシミュレーションの信頼性が向上するためである。 At this time, adjustment parameter setting unit 7c is searched when evaluating the distance of a query point P ₁ and the setting history data P ₂ (decision), the search range of the query point P ₁ and the setting history data P ₂ D ( P _{1, P 2)} of the following formula (3);

Evaluate by calculating by. In the above equation (3), the values u ₂ , d ₂ , and t ₂ are the control amount, the disturbance value, and the operation time included in the setting history data P _{2, respectively, and “|| xy ||” is} The Euclidean norm calculated from the two vectors x and y, that is, the length of the difference vector, and a, b, and c indicate non-negative parameters. That is, in the above equation (3), the first term represents the distance corresponding to the difference in the control amount, the second term represents the distance corresponding to the difference in the disturbance value, and the third term represents the temporal distance. .. By using such a distance calculation formula, the adjustment parameter setting unit 7c is associated with the operation condition data similar to the disturbance value and the control amount which are the operation conditions to be predicted, and the operation time information whose operation time is close. The adjustment parameter p can be obtained from the parameter storage unit 5a. To acquire the adjustment parameter p that is close to the operation time, the reliability of the simulation of the operation state is improved by using the adjustment parameter set by synchronization (tracking) with the latest operation time that is close to the operation time of the future prediction target. Because.

The second setting method is as follows. Since the first setting method described above refers only to the setting history data of the nearest neighbor, the setting value of the adjustment parameter may not be stable. For example, the set value of the adjustment parameter set may fluctuate greatly even though the position of the search query point changes only slightly. When such a case is assumed, the second setting method is selected by the user. That is, the adjustment parameter setting unit 7c extracts a plurality of data from the setting history data stored in the parameter storage unit 5a of the database device 5 based on the distance from the search query point, and includes the extracted data in the plurality of data. The adjustment parameter p of the simulation model of the process simulator unit 7a is set based on the adjustment parameter. FIG. 5 shows the data distribution of the setting history data searched by the adjustment parameter setting unit 7c. Thus, the distance from the search query point set P ₃ of a plurality of setting history data with the outside temperature and input amount within the predetermined value is extracted. The distance calculation method in this case is the same as the first setting method. Adjustment parameter setting unit 7c is a method of setting the adjustment parameter from the set P ₃ a setting history data, we calculate a weighted sum of average value of the adjustment parameters included in the data set P _3, the data set P ₃ On the other hand, a method such as applying machine learning by supervised learning is used. Machine learning methods include linear multiple regression by least squares regression, ridge regression, partial least squares regression (PLS: Partial Least Squares), principal component regression (PCR: Principal Components Regression), non-linear neural network, and deep learning. A method such as learning can be used.

In order to accurately predict the operating state, it may be better to consider the temporal dynamics of the adjustment parameters due to changes in operating conditions or disturbance conditions during the prediction period. In such a case, in the first and second setting methods, in addition to the control amount and the disturbance value at the operation time of the prediction target, the control amount and the disturbance value at the past operation time are added to the search query point. May be good. As a result, the adjustment parameter of the operation time of the prediction target can be appropriately set in consideration of the time of the adjustment parameter.

Further, in the first and second setting methods, as search query points, in addition to continuous values such as control amount, disturbance value, and time, depending on the type of the actual process to be predicted, the index value, discrete value, etc. are discrete. Value may be set in the search query point. For example, when there are a plurality of control modes in the actual process, the control modes may be added to the search query point as an operating condition. Further, in the case of a plant process in which the actual process can be generated by switching between a plurality of types of products, an index value indicating the types of products may be added to the search query point as an operating condition. However, if such a discrete value is added to the search query point, it becomes difficult to evaluate the distance between the search query point and the setting history data, so the setting history that matches the operating conditions indicated by the discrete value of the search query point. After the data is extracted from the parameter storage unit 5a, the adjustment parameters can be set by using the first or second setting method based on the extracted setting history data.

Next, the operation procedure will be described in the simulation system 1 described above, and the simulation method according to the present embodiment will be described in detail. FIG. 6 is a flowchart showing the operation procedure of the simulation of the actual process by the simulation system 1, FIG. 7 is a flowchart showing the operation procedure of the tracking simulator device 3, and FIG. 8 is a flowchart showing the operation procedure of the prediction simulator device 7. ..

First, with reference to FIG. 6, the user sets a period of synchronization with the tracking simulator device 3 (step S01). Accordingly, the tracking simulator device 3 executes the processing of the synchronization simulation during the synchronization period (step S02). After that, when the user instructs the prediction simulator device 7 to execute the prediction simulation (step S03; YES), the user further sets the operation conditions and the prediction period for the actual process to be predicted for the prediction simulator device 7. (Step S04). Then, the prediction simulator device 7 executes the prediction simulation (step S05). This prediction simulation is repeatedly executed according to the user's instruction. At this time, a single computer 100 constituting the prediction simulator device 7 may repeatedly execute the prediction simulation, or a plurality of computers 100 constituting the prediction simulator device 7 may perform the prediction simulation under individual operating conditions and prediction periods. It may be executed in parallel. For example, one computer 100 may perform a prediction simulation under an operating condition of an outside air temperature of TA degrees, and another computer 100 may perform a prediction simulation under an operating conditions of an outside air temperature of TB degrees.

The procedure of the synchronous simulation processing by the tracking simulator device 3 will be described with reference to FIG. 7. First, in the tracking simulator device 3, the current operating condition data and the current operating state data are acquired from the actual plant 9 (step S101). Then, in the tracking simulator device 3, a synchronous simulation using those data is executed (step S102). Next, the tracking simulator device 3 stores the synchronization result data including the setting history data in the database device 5 in association with the operating condition data and the operating time information acquired from the actual plant 9 (step S103). Further, the tracking simulator device 3 determines whether or not the synchronization simulation for the synchronization period set by the user has been completed (step S104), and if it is completed (step S104; NO), the synchronization simulation process is terminated and terminated. If not (step S104; YES), the process returns to step S101 and the synchronous simulation process is repeated.

Moving to FIG. 8, the procedure of the prediction simulation processing by the prediction simulator device 7 will be described. First, when the user sets the operating conditions and the prediction period for the actual process to be predicted for the prediction simulator device 7 (step S201), the user sets the adjustment parameters in the prediction simulator device 7 first. It is selected from the second setting method (step S202). Then, in the prediction simulator device 7, information on the value x of the current state quantity of the simulation model is acquired from the tracking simulator device 3, and is set to the initial value of the operating state value x of the prediction simulator (step S203). After that, the prediction simulator device 7 uses the selected first or second setting method to perform a simulation based on the operating conditions in the prediction period of the actual plant 9 to be predicted and the prediction period of the actual plant 9 to be predicted. The model adjustment parameter p is dynamically set (step S204). The set adjustment parameter p is presented to the user in the prediction simulator device 7 (step S205).

After that, the prediction simulator device 7 determines whether or not the user has instructed to execute the prediction simulation (step S206), and if not (step S206; NO), the process is returned to step S202. The adjustment parameters are set again. On the other hand, when the execution of the prediction simulation is instructed (step S206; YES), the prediction simulator device 7 executes the prediction simulation using the set adjustment parameters (step S207), and the predicted operating state value. The simulation result for is output (step S208). According to such a prediction simulation process, the user can know the adjustment parameters for the prediction simulation in advance.

According to the simulation system 1 and the simulation method using the simulation system 1 described above, the history information of the adjustment parameters of the simulation model set in the tracking simulator device 3 by the database device 5 in the past is the operating condition of the plant device of the actual plant 9. The simulation model, which is accumulated in association with the information, acquires the adjustment parameters associated with the information of the operating conditions similar to the operating conditions to be predicted by the prediction simulator device 7, and the adjustment parameters are dynamically set by using the simulation model. , The operating state of the actual plant 9 is predicted. As a result, adjustment parameters set under operating conditions similar to the section to be predicted are acquired from the execution results of the synchronization processing of the tracking simulator device 3 in the past, and the operating state of the actual plant 9 is predicted using the adjustment parameters. Therefore, the operating state of the actual plant 9 can be accurately predicted. As a result, the operating state of the plant equipment can be determined by performing a simulation that considers the operating area related to the disturbance of each process of each plant, the operating area related to the control state of each process, and the operating area related to time factors such as the deterioration state. It can be predicted accurately, and the prediction result can be used to create an appropriate plant operation plan or an appropriate plant equipment maintenance plan. For example, it is possible to accurately obtain a predicted value that can be evaluated as to how much the production efficiency of a chemical plant will be when it is operated for several months under specific operating conditions, and it depends on the evaluation result. You can create an operation plan for the plant or a maintenance plan for the plant equipment. Further, in the present embodiment, since the simulation model is set by using the first or second setting method, it is possible to perform the simulation that matches the operation area for each process of each plant.

Further, in the present embodiment, since the prediction simulator device 7 includes the control model unit 7b, even when the control amount is dynamically controlled by the control device in the actual plant 9, the control amount by the control device is large. By predicting, the operating state of the actual plant 9 can be accurately predicted. In addition, in the present embodiment, in the prediction simulator device 7, the operating state is predicted based on the prediction result of the control model unit 7b in which the algorithm or the parameter is dynamically or statically changed. With such a configuration, it is possible to compare the operating states of the actual process in a plurality of control states by performing the simulation while changing the control state of the actual process of the actual plant 9.

In the conventional simulation method for plant state prediction disclosed in Patent Document (Patent No. 4789277), state data such as parameters held by a device that executes synchronous simulation is used for prediction simulation for state prediction. After copying to the simulation device that executes, the prediction simulation is executed. However, in such a conventional method, the prediction accuracy tends to decrease when the operating conditions are changed within the prediction period of the prediction target, or when the disturbance conditions such as the weather change in the prediction target plant change. be. 9 and 10 show the time change of the state value obtained by the synchronous simulation and the prediction simulation in the conventional method. FIG. 9A shows a predicted change in temperature in the actual plant 9 when the synchronous simulation is stopped at the operation time T1 of the actual plant 9 and the prediction simulation is executed for the operation time after that time T1. 9 (b) shows the changes in the adjustment parameters set in the synchronous simulation and the prediction simulation at this time. It should be noted that the temperature is greatly reduced immediately before the time T1, which is assumed to be that the user has performed a certain operation on the actual plant 9. Further, FIG. 10A shows a solid line showing the change in the temperature actually measured in the actual plant 9 after the time T1, and FIG. 10B shows the case where the synchronous simulation is executed after the time T1. The changes in the adjustment parameters of are shown by the solid line. In this way, since the adjustment parameter is dynamically changed by the synchronous simulation until time T1, a temperature close to the measured value can be obtained, but after time T1, the adjustment parameter is set to a fixed value in the prediction simulation. , The predicted temperature value is lower than the measured value, and an error tends to occur. In order to reduce such an error, it is necessary to predict in advance the setting values of appropriate adjustment parameters due to changes in operating conditions or fluctuations in disturbance values.

Further, FIGS. 11 and 12 show another time change of the state value obtained by the synchronous simulation and the prediction simulation in the conventional method. FIG. 11A shows a predicted change in temperature in the actual plant 9 when the synchronous simulation is stopped at the operation time T1 of the actual plant 9 and the prediction simulation is executed for the operation time after that time T1. 11 (b) shows the changes in the adjustment parameters set in the synchronous simulation and the prediction simulation at this time. The adjustment parameters are gradually decreasing by time T1, but it is assumed that the deterioration of the actual process is gradually progressing. Further, FIG. 12 (a) shows the change of the temperature actually measured in the actual plant 9 after the time T1 with a solid line, and FIG. 12 (b) shows the case where the synchronous simulation is executed after the time T1. The changes in the adjustment parameters of are shown by the solid line. Even in such a case, since the adjustment parameter is set to a fixed value in the prediction simulation after the time T1, the predicted value of the temperature tends to be lower than the measured value and an error tends to occur. In order to reduce such an error, it is necessary to predict in advance the setting value of an appropriate adjustment parameter according to the deterioration of the actual process.

On the other hand, according to the simulation method disclosed in the conventional patent document (Japanese Unexamined Patent Publication No. 2009-15477), a method of substituting a parameter in a tracking simulator with a parameter function expression is adopted. However, the temporal dynamics of the simulation model cannot be considered. Therefore, when the parameter change has a long time constant with respect to the controlled variable or disturbance, the prediction accuracy in the steady state can be improved by this method, but the prediction accuracy in the transient state is low. Moreover, this method cannot cope with the deterioration of the actual process.

In contrast to the above-mentioned conventional method, in the present embodiment, a simulation model in which appropriate adjustment parameter setting values are predicted in advance in response to changes in operating conditions, fluctuations in disturbance values, or deterioration of the actual process. Predict the future operating conditions of the actual process using the adjustment parameters of. Therefore, a more accurate predicted value can be obtained as compared with the conventional method.

Further, in the present embodiment, in the first and second setting methods implemented in the prediction simulator device 7, in addition to the control amount and the disturbance value at the operation time of the prediction target, the control amount and the disturbance value of the past operation time Can be added to the search query point. FIG. 13 shows the time change of the state value obtained by the synchronous simulation and the prediction simulation when the control amount of the past operation time is added to the search query point in the prediction simulator device 7. FIG. 13A shows a controlled amount in the actual plant 9 predicted when the synchronous simulation is stopped at the operation time T1 of the actual plant 9 and the prediction simulation is executed for the operation time after the subsequent time T2. 13 (b) shows the change of the adjustment parameter set in the synchronous simulation and the prediction simulation at this time. Thus, when setting the adjustment parameter p of the simulation model to obtain a prediction value at operation time T2 (T2), the time T1 the previous value u _G of a plurality of past pump flow search query point Is included in. Therefore, the value of the adjustment parameter suitable for the actual process is set by predicting the change of the adjustment parameter corresponding to the tendency of the change of the pump flow rate based on the setting history of the plurality of adjustment parameters in the past. For example, as shown by the dotted line in FIG. 13B, after time T2, it is predicted that the value of the adjustment parameter will gradually increase from the value of the adjustment parameter predicted by the synchronous simulation immediately before that time.

The present invention is not limited to the above-described embodiment.

For example, the configuration of the prediction simulator device may be changed. FIG. 14 shows the functional configuration of the prediction simulator device 7A according to the modified example. The prediction simulator device 7A shown in FIG. 14 is different from the prediction simulator device 7 in that it does not include the control model unit 7b. In the prediction simulator device 7A, the process simulator unit 7a executes the prediction process of the operating state based on the control amount preset by the user. This modification can be applied when the output of the control device in the actual plant 9 or the control amount set in the actual plant 9 is predetermined, and is predicted without being connected to the control model simulating the control device. You can run simulations.

Further, the configuration of the prediction simulator device may be changed to the configuration shown in FIG. The prediction simulator device 7B shown in FIG. 15 is different from the prediction simulator device 7 in that the output of the control model unit 7b is also input to the adjustment parameter setting unit 7c each time. This modification is applied when the state of the future process cannot be predicted correctly if the adjustment parameters are not set based on the behavior of the control device in the actual plant 9. For example, if the actual process tends to deteriorate and a certain controlled variable tends to change according to the deterioration and the output of the control device tends to fluctuate, it is necessary to apply the configuration of this modification. be. Further, in the prediction simulator device 7B, since the adjustment parameter setting unit 7c is included in the processing loop of the prediction simulation, the adjustment parameter used in the prediction period cannot be known unless the prediction simulation is actually executed. Therefore, the prediction simulation is executed without presenting the set adjustment parameters to the user.

FIG. 16 shows a procedure for processing a prediction simulation by the prediction simulator device 7B. First, when the user sets the operating conditions and the prediction period for the actual process to be predicted for the prediction simulator device 7B (step S301), the user sets the adjustment parameters in the prediction simulator device 7B. It is selected from the second setting method (step S302). Then, in the prediction simulator device 7B, the information of the value x of the current state quantity of the simulation model is acquired from the tracking simulator device 3, and is set to the initial value of the value x of the operating state of the prediction simulator (step S303). Further, the prediction simulator device 7B determines whether or not the prediction simulation processing at all the operation times in the prediction period has been completed (step S304). When the processing of all the operating times is completed (step 304; YES), the simulation result regarding the predicted operating state values for all the operating times within the predicted period is output (step S308).

On the other hand, when the processing of all the operation times is not completed (step 304; NO), the actual plant 9 to be predicted is predicted by using the first or second setting method selected by the prediction simulator device 7. The adjustment parameter p of the simulation model is dynamically set based on the operation conditions in the prediction period and the operation time of the actual plant 9 to be predicted (step S305). After that, the prediction simulator device 7B executes the prediction simulation using the set adjustment parameters (step S306), and after the operation time of the prediction target is advanced to the next time step (step S307), the process proceeds to step S304. Be returned.

Further, the operation procedure shown in FIGS. 6 and 7 may be changed to the procedure shown in FIGS. 17 and 18.

In the procedure shown in FIG. 17, first, when the operation of the simulation system 1 is started, the operation condition data and the operation state data for a certain period related to the actual process are accumulated in the actual plant 9 by waiting for a certain period. (S401). After that, the tracking simulator device 3 executes a synchronous simulation process based on the operating condition data and the operating state data acquired from the actual plant 9 (step S402). Subsequently, the prediction simulator device 7 executes the prediction simulation (step S403). Based on the simulation result of this prediction simulation, the prediction simulator device 7 determines whether or not there is an abnormality in the operating state of the process (step S404). As a result, if it is determined that there is an abnormality (step S404; YES), the process is returned to step S401 after a warning is output by the prediction simulator device 7 (step S405). On the other hand, when it is determined that there is no abnormality (step S404; NO), the process is returned to step S401 by the prediction simulator device 7. According to the operation procedure of this modified example, after executing the synchronous simulation using the operation data of the process accumulated regularly, the operation tendency of the plant for a predetermined period ahead (future) is grasped by using the result of the synchronous simulation. can do. At this time, if there is an abnormal value in the predicted value (for example, when the NOx emission amount exceeds the threshold value in the gas turbine), an operation such as automatically sending a warning mail to the user becomes possible. A plurality of operating conditions that are the basis of the prediction simulation in this case may be set.

FIG. 18 shows the operation procedure of the synchronous simulation in step S402 of FIG. First, in the tracking simulator device 3, the operating condition data and the operating state data accumulated in the actual plant 9 are acquired (step S501). Then, in the tracking simulator device 3, the synchronous simulation using the accumulated data is continuously executed a plurality of times (step S502). Next, the tracking simulator device 3 stores the synchronization result data including the setting history data acquired as a result of the continuous synchronization simulation process in the database device 5 in association with the operation condition data and the operation time information corresponding to the synchronization result. (Step S503).

例えば、予測期間が仮に１年間である場合、１年分の動的モデルを用いた予測シミュレーションの計算コストは大きい。このような場合は、静的モデルを用いて１か月ごとの１年間の定常的な運転状態を求めることが好ましい。例えば、現在の実プラント９の運転時刻からＴ秒後の制御量ｕ_Ｔ、外乱値ｄ_Ｔ、及び調整パラメータｐ_Ｔを定めることにより、現在の運転時刻からＴ秒後のプロセスの運転状態値ｘ_Ｔが求められる。このような予測シミュレーションを、Ｔ＝１か月、２か月、…、１２か月と各時点を対象に繰り返し実行すれば、少ない計算コストで運転状態の長期のトレンド予測が可能となる。 In the above-described embodiment, the simulation model used in the prediction simulation uses a dynamic model represented by the same function as the simulation model used in the synchronous simulation. On the other hand, even if the synchronous simulation is a dynamic model that considers the temporal dynamics, if the purpose of the prediction simulation is to predict the steady state against the disturbance or deterioration of the process, the following equation (4) is used. A simulation model represented by the formula of the static model as shown may be used.

For example, if the prediction period is one year, the calculation cost of the prediction simulation using the dynamic model for one year is large. In such a case, it is preferable to use a static model to obtain a steady operation state for one year every month. _{For example, by defining the control amount u T} , the disturbance value d _T , and the adjustment parameter p _T T seconds after the current operation time of the actual plant 9, the operating state value x of the process T seconds after the current operation time. _T is required. If such a prediction simulation is repeatedly executed for each time point such as T = 1 month, 2 months, ..., 12 months, it is possible to predict a long-term trend of the operating state at a low calculation cost.

Further, in the above-described embodiment, the prediction simulation may be sequentially executed in parallel with the synchronous simulation, or may be executed at an arbitrary time set by the user.

In the above-described embodiment, the tracking simulator device 3 and the prediction simulator device 7 do not necessarily have to be installed at the same site as the actual plant 9. For example, the tracking simulator device 3 and the prediction simulator device 7 may be installed at different sites as long as the data of the actual plant 9 can be transmitted and received via a communication network such as an internet line or a LAN.

Further, the synchronous simulation process may be executed in real time in synchronization with the actual process, or may be executed in batch processing using the operating condition data and the operating state data acquired / accumulated in the actual plant 9.

In the above-described embodiment, the value of the disturbance, which is the operating condition to be predicted, is set by the user, but it may be acquired by referring to an external server device. FIG. 19 shows an outline of the configuration of the simulation system 1A according to the modified example of the present invention. In the simulation system 1A, the prediction simulator device 7 uses a communication network NW1 such as an Internet line to predict future temperature and pressure pressure as disturbance values that are operating conditions to be predicted from the external server device 11. It is configured so that weather data such as values can be acquired.

In addition, the disturbance value related to the meteorological conditions may not be actually measured in the actual plant 9. In such a case, instead of the measured value of the disturbance value acquired from the actual plant 9, data obtained from a public institution such as a meteorological organization or recorded data measured at a nearby point of the actual plant 9 is used instead. You may. Further, it is assumed that noise such as high frequency noise is mixed in the measured value obtained from the actual plant 9. In such a case, if the raw measurement value is input to the tracking simulator device as it is, the synchronization result may vibrate. Therefore, after removing noise from the measured value from the actual plant 9 by a noise removing means such as a low-pass filter, the synchronous simulation may be operated using the measured value from which the noise has been removed.

In the first and second setting methods implemented in the prediction simulator device 7, other values such as _{an estimated value x cal of a quantity x indicating the operating state of the actual plant 9 may be added to the search query point.} ..

1,1A simulation system (simulation device)
3 Tracking simulator device 5 Database device 5a Parameter storage unit 7,7A, 7B Prediction simulator device 7a Process simulator unit (prediction simulator unit)
7b Control model unit (Control amount prediction unit)
7c Adjustment parameter setting unit 9 Actual plant

Claims (11)

A simulation device that predicts the operating status of plant equipment.
The adjustment parameter set in the tracking of the operating state by the external tracking simulator device, and the adjustment parameter as a physical / chemical coefficient describing the actual process in the simulation model, is set at the time of tracking by the tracking simulator device. A parameter storage unit that stores information related to the operating condition information of the plant equipment obtained from the actual plant, and a parameter storage unit.
A prediction simulator unit that predicts the operating state using a simulation model based on the operating conditions of the plant device to be predicted.
An adjustment parameter setting unit that dynamically sets the adjustment parameters of the simulation model in the prediction simulator unit,
With
The adjustment parameter setting unit acquires the adjustment parameter associated with the information of the operation condition similar to the operation condition to be predicted from the parameter storage unit, and based on the acquired adjustment parameter, the prediction simulator unit of the prediction simulator unit. Set the adjustment parameters of the simulation model,
Simulation equipment. The parameter storage unit stores the adjustment parameter in association with the disturbance information which is the information of the operating condition regarding the disturbance, and stores the adjustment parameter.
The adjustment parameter setting unit acquires the adjustment parameter associated with the disturbance information similar to the disturbance information to be predicted.
The simulation apparatus according to claim 1. The parameter storage unit stores the adjustment parameter in association with the control amount information which is the information of the operation condition regarding the control amount for the plant device.
The adjustment parameter setting unit acquires the adjustment parameter associated with the control amount information similar to the control amount to be predicted.
The simulation apparatus according to claim 1 or 2. The parameter storage unit stores the adjustment parameters in association with the operation time information which is the information of the operation conditions regarding the operation time of the plant device.
The adjustment parameter setting unit acquires the adjustment parameter associated with the operation time information similar to the operation time to be predicted.
The simulation apparatus according to any one of claims 1 to 3. The adjustment parameter setting unit determines the similarity by calculating the distance between the operating condition to be predicted and the information of the operating condition.
The simulation apparatus according to any one of claims 1 to 4. The adjustment parameter setting unit acquires a plurality of the adjustment parameters from the parameter storage unit, and uses machine learning based on the plurality of the adjustment parameters and information on the operating conditions of the plant device associated with the adjustment parameters. To set the adjustment parameters of the simulation model,
The simulation apparatus according to any one of claims 1 to 5. A control amount prediction unit that predicts the control amount of the plant device by the controller based on the output of the prediction simulator unit is further provided.
The prediction simulator unit predicts the operating state based on the control amount predicted by the control amount prediction unit.
The simulation apparatus according to any one of claims 1 to 5. The adjustment parameter setting unit acquires the adjustment parameter associated with the information of the operating condition similar to the control amount predicted by the control amount prediction unit.
The simulation apparatus according to claim 7. The prediction simulator unit predicts the operating state based on the prediction result of the control amount prediction unit in which the algorithm or parameter for predicting the control amount is dynamically or statically changed.
The simulation apparatus according to claim 7 or 8. The prediction simulator unit predicts the future operating state for a predetermined period of time.
The simulation apparatus according to any one of claims 1 to 9. It is a simulation method that predicts the operating state of plant equipment.
The parameter storage unit tracks the adjustment parameters set in the tracking of the operating state by an external tracking simulator device, and is the adjustment parameters as physical and chemical coefficients that describe the actual process in the simulation model. A step of accumulating information related to the operating condition information of the plant device obtained from the actual plant during tracking by the simulator device, and a step of accumulating the information.
A step in which the simulation device predicts the operating state using a simulation model based on the operating conditions of the plant device to be predicted.
The step in which the simulation device dynamically sets the adjustment parameters of the simulation model,
With
In the step of dynamically setting the adjustment parameter, the adjustment parameter associated with the information of the operation condition similar to the operation condition to be predicted is acquired from the parameter storage unit, and based on the acquired adjustment parameter, the adjustment parameter is acquired. Set the adjustment parameters of the simulation model,
Simulation method.

Images