JPH07107559B2 - Automatic operation method and automatic operation device for nuclear power plant - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to automation of operation of a nuclear power plant, and is particularly suitable for a case where an operation plan needs to be changed in order to satisfy an operation restriction condition of a nuclear reactor. The present invention relates to an automated method and an automated device for automatically obtaining a modified corrected load curve and setting an operation target.

[Conventional technology]

Japanese Patent Laid-Open No. 57-135397 discloses, for example, a known technique whose main task is to automate the operation plan creation, power control, core monitoring, and operation plan correction during operation of a nuclear power plant. In this publicly known example, (a) an operation plan is created by using the core simulator to automatically search for a load pattern that is closer to the required load pattern (equal to the basic load curve) from the central power supply command station and that complies with the operation restriction conditions. (B) Importing plant data from the reactor and outputting the operation guide request command when it is judged that the margins (core flow rate, linear power density, etc.) for the operation restriction conditions exceed the preset allowable values. Core condition monitoring device, and (c)
According to the operation plan creation device, the margin for the core state and the operation restriction condition after a short time of about 1 hour is predicted and calculated, and if it is determined that the operation restriction condition is violated if it is operated according to the operation plan, the prediction is performed. A reactor load follow-up control system is proposed which is composed of an operation guide device that formulates a tentative operation method by converting the ratio of a value and an operation limit value into an output and subtracting it from a planned value.

[Problems to be solved by the invention]

In the above-mentioned publicly known example, only the case where there is no margin of the operation restriction condition as a factor for correcting the load curve determined from the operation plan is taken up. However, even if there is a margin in the operational restriction conditions, when the reactor power is changed significantly, for example, at plant startup, due to the error of the core simulator used to predict the reactor state, the set control rod pattern is not always required. In some cases, the target value of reactor power / core flow rate may not be achieved. In other words, even if there is a difference in the degree, the simulator used always has an error, so it is necessary to recreate and calculate the detailed load curve.

Further, in the publicly known example, when it is determined that the operation limit condition is violated when operating according to the plan, a method of drafting a provisional operation method by converting the ratio of the predicted value and the operation limit value into an output and subtracting from the planned value However, as a result of the local change in the driving method, the subsequent operation plan also needs to be significantly modified, and as a result, there is a possibility that the required basic load curve will deviate significantly. To prevent this, when it is necessary to change the operation method, a modified load curve that reduces the deviation from the required reactor output due to the basic load curve over the entire load curve and increases the margin of operation restriction conditions Is desirable, and a new operation method is determined accordingly.

The purpose of the present invention is to cope with the above-mentioned two problems, namely, (i) error correction of the core simulator used for prediction, and (ii) load when there is a possibility of violating the prediction error of the simulator or the operational restriction condition. It is an object of the present invention to provide an operation automation method and an automation device for a nuclear power plant, which has a function of optimally correcting a curve and can flexibly respond to the actual operation status of a nuclear reactor.

[Means for solving problems]

For the above-mentioned purpose, the prediction error countermeasures can be achieved first by matching the measured values of the operating parameters of the plant with the predicted values by the core simulator every hour and performing statistical processing to create an error prediction model. Next, when the prediction error is larger than the value predicted in advance, or when it is predicted that the operation restriction conditions will be violated, the deviation from the required basic load curve and the margin of the operation restriction conditions are respectively output variable operation. By using the evaluation function that has the value integrated over the interval as a parameter and creating the optimum modified load curve, it is possible to achieve an appropriate change of the operation method in view of the entire operation plan.

[Action]

The operating parameters of the plant obtained by the prediction calculation using the core simulator are (A) those obtained by a lumped parameter model such as reactor average power and core average void fraction, and (B)
Some depend on the core distribution model such as power distribution and thermal limit values. The former (A) parameter locally shows an approximately linear fluctuation response to the output fluctuation according to the load curve, but the latter (B) parameter also depends on the change of the control rod pattern, so it is not always necessary. It does not show a continuous variation response that can be described by a linear model. However, when the core flow rate is adjusted and the output is changed under the condition that the control rod pattern can be regarded as constant or similar,
Local linearization can be realized. Therefore, if the difference between the actual measured value and the predicted value at that time, which was previously obtained by the simulator, is continuously obtained under the above-mentioned uniform operating conditions, the time series will be in an appropriate linear form or at most. It can be described by a quadratic equation. The error correction of the simulator can be realized by creating a statistical prediction model for such future variations of the time series by a method such as multivariate analysis.

On the other hand, as a function of the control parameters (core flow rate, control rod pattern, etc.) at each point of the modified load curve, the deviation from the required basic load curve and the margin of the operation restriction condition can be defined. The operation restriction condition can solve the optimization problem of the modified load curve even at the end as a restriction condition in the optimum search, but generally, the reciprocal of the difference up to the limit value of the standardized operation restriction condition is used as a parameter. Then, this can be incorporated into the evaluation function. The former non-linear optimization problem with constraints can be solved by using, for example, the effective permissible direction method. The latter generalized optimization problem is a multi-objective programming problem that minimizes the vector objective function under given constraints, but here the absolute value of the vector is the evaluation function for comprehensive evaluation. By doing so, it is possible to solve by a usual optimization method.

〔Example〕

An embodiment of the operation automation device for a nuclear power plant of the present invention will be described below. FIG. 1 is a diagram in which the basic functional configuration of the embodiment is killed, and is roughly divided into control amounts such as the plant 3, the reactor output of the plant 3, the reactor pressure, and the reactor water level, and becomes equal to the target value. To control the control rod withdrawal amount, turbine control valve opening, the feed water flow rate, and other manipulated variables, the sub-loop controller 4, the integrated controller 5 that outputs the target value to the sub-loop controller, and the integrated controller A state predicting unit 6, a state monitoring unit 7, and a simulator error correcting unit 8 which form the function of outputting a load curve as a whole.
A man who inputs the operation request command of the operator to the operation plan correction unit composed of the curve optimum correction unit 9 and the integrated control unit 5 and presents the operator with the operation state of the plant and the calculation result of each partial function of the operation automation unit. It is composed of Machine Eftace 2. Hereinafter, the operation of the operation plan correction unit, which is the main technical feature of the present invention, will be described in detail.

The state monitoring unit 7 takes in the plant data 11 and displays the operating parameters such as the core power distribution and the thermal limit value in a short time by the core monitoring device using the core performance calculator which is the technology already adopted in the current plant. It is calculated and compared with the operation limit value from the operation plan creation knowledge database 10 to obtain the margin of the operation limit condition. The state prediction unit 6 takes in the plant data 11 and uses the core simulator.
When the operation target of the output fluctuation is set and operated according to the load curve, the operation parameter predicted value at each point of the curve is obtained. The simulator error correction unit 8 takes in the measured values of the operating parameters obtained by the state monitoring unit 7 and the predicted values of the operating parameters that correspond to these measured values and have been previously obtained by the state prediction unit 6, The correction value of the simulator error is obtained using the time series of these differences. The curve optimum correction unit 9 takes in the margin of the operation restriction condition and the correction value of the simulator error, determines whether or not the load curve currently used needs to be corrected, and determines that the correction is necessary. In this case, the operation plan creation knowledge database 10 is used to create a modified load curve.

The basic calculation procedure of the operation plan correction unit described above is shown in FIG. The detailed calculation contents of the simulator error correction 8'for processing the calculation steps surrounded by the thick frame and the curve optimum correction 9'will be described below.

FIG. 3 shows a calculation procedure in the simulator error correction unit 8. The prediction error time series is calculated by the following formula.

δx (t) = (t) −x (t) (1) (t = t ₀ , t ₁ , ... t _n ) where x (t) = (x ₁ (t), x ₂ (t ), X _q (t)) is a vector whose elements are measured values x ₁ , x ₂ , ..., X _q of the operating parameter, (t) is a predicted value vector of the operating parameter obtained from the predicted value by the simulator, and δx (t ), (T =
t ₀ ,, t ₁ , ... t _n ) is a time series of prediction errors. The basic assumption adopted for the statistical model that represents the time series of this prediction error is that these time series include systematic fluctuation factors as represented by the linear form of some explanatory variables. is there. If a systematic fluctuation factor is not included, the magnitude of the fluctuation is approximately the variance of the stochastic fluctuation, so it can be ignored as a prediction error. When the multivariate multiple regression model is adopted as the statistical model, the following relational expression can be applied by taking the time series of the prediction error as the objective variable.

δx (t) = β · u (t) + ε (t) (2) (t = t ₀ , t ₁ , ... t _n ) where β is a partial regression coefficient matrix (p × q), u
(T) is an explanatory variable vector (p-dimensional), and ε (t) is a residual vector (q-dimensional). In this case, the explanatory variables to be adopted may be the manipulated variable changed when the plant is operated according to the load curve, for example, the core flow rate, the number of steps of the control rod drawing sequence, the Xe concentration and the like.
Therefore, the least-squares estimated value of the partial regression coefficient matrix β is calculated by the method of multivariate multiple regression analysis. Then, using this, the prediction result correction value * (t _f ) of the simulator at time t _{f which} is future than the present time t _n is represented by the following equation.

Here, the formula (3) giving the prediction result correction value represents the independent variable as the future time t _f , but this is expressed in this way in the representation form, and the actual independent variable is the explanatory variable vector. u. This is because the time parameter t ₁ in the model equation (2) simply indicates the numbering of individual data. Further, from this fact, the expression (3) is a relational expression that is established for each scalar component of the prediction result correction value vector, and accordingly, a load curve different from that used to create the model expression (3) is obtained. It can also be applied.

The simulator error correction method described above is an example of a statistical analysis method that does not rely on a physical model, but other than this, the physical model that has been often adopted in the past is used to correct the measured value by feedback back. And the method of tuning the input data of the simulator can be used at the same time.

FIG. 4 shows a detailed calculation block diagram of the curve optimum correction unit 9. In the modified load curve assumption unit 21, the state monitoring unit 7 determines the margin of the current operation restriction condition, and the simulator error correction unit 8 corrects the prediction error correction value. First, it is determined whether or not the road curve currently used needs to be newly modified.

As a criterion for the determination, it is assumed that the load curve correction calculation is started when there is no margin of the operation restriction condition or when the correction value of the prediction error of the simulator is larger than a certain set reference value.

Next, when entering the correction calculation, the load curve used at the moment is used as the initially assumed load curve, but if the correction has not been done until now, the input data from the outside The basic load curve 27 given as is used. The input processing unit 22 takes in the assumed modified load curve, and inputs the input data for the simulation for obtaining the predicted values of the operating parameters (output distribution, thermal limit value, etc.) for the main operating points that characterize the load curve. Create and process the data. In the prediction simulation 23, the input data for simulation is input to the core simulator 28, and the predicted values of the operating parameters at the main operating points of the load curve are obtained. At this time, the correction of the simulator model from the simulator error correction unit 8 and the information on the input data tuning are reflected. Next, the result extraction unit 24 extracts the predicted value of the operating parameter from the simulation result, and the predicted result correction value ((3) given from the simulator error correction unit 8).
Equation) is used to correct the predicted value of the simulator. The curve performance evaluation unit 25 takes in the operating parameter prediction values at each point of the load curve, and evaluates the adequacy of the load curve by using an evaluation function as shown in the following equation, for example.

However, f (u) = (f ₁ (u), f ₂ (u), ... f
_k (u)) is an evaluation function vector whose elements f ₁ ... F _k
Is a standardized and weighted parameter for evaluating the appropriateness of the load curve, such as the deviation of various operating parameters from the target values, the margin of thermal limit values, the goodness of output distribution, the number of changes in the control rod pattern, etc. It is a thing. g (u) =
(G ₁ (u), ... g _r (u)) is a constraint vector, u ∈ R ^P
Is a vector of the manipulated variable and the operating condition when the vehicle follows a load curve, and may be equal to the explanatory variable vector used in the predictive error model of the simulator (equation (2)). The optimum measure search unit 26 uses the load curve evaluation function created by the curve performance evaluation unit 25 to perform an optimum direction search for road curve correction. Whether the modified load curve is optimal can be determined by whether the evaluation function has stopped decreasing. When the optimum point is reached, the integrated control unit 5 outputs the optimum modified load curve and the information on the manipulated variable and the operating condition vector when the plant is operated according to the optimum modified load curve. If a new modified road curve is obtained in the optimum direction search, the modified road curve assumption unit 21
The data is sent to and the above-mentioned simulation and curve performance evaluation are repeated.

FIG. 5 shows an example of the simulation when the reactor power increase is performed by using the operation automation device of the nuclear power plant according to the present invention. In this example, an error is intentionally added to the core simulator. When the output is increased according to the initial basic load curve, the measured value of the thermal limit value shows a change curve with a margin smaller than the predicted value due to the error of the simulator. So B
At the point, when the simulator error was corrected and the thermal limit value was predicted, it was found that the operation limit condition was violated. Therefore, the operation plan correction unit created the optimum corrected load curve. In the optimum modified load curve, the output increase by pulling out the control rod is interrupted at point C ', the output is increased by increasing the core flow rate up to point C ", and the output is increased by pulling out the control rod again at points C" to C. There is. At this time, the target control rod pattern during the rated operation is also corrected by the prediction simulation in which the error is corrected so that the margin of the thermal limit value increases.

〔The invention's effect〕

According to the present invention, since the error of the predicted value of the core simulator is corrected based on the measured value, the prediction accuracy regarding various operating conditions at the time of creating the load curve is improved, and the case where the operating method is changed during the output operation occurs Is less. Also,
When changing the driving method, the optimum correction function of the load curve makes it possible to change the driving method appropriately by evaluating the entire operation plan.

[Brief description of drawings]

FIG. 1 is a block diagram showing the basic configuration of the embodiment of the present invention,
2 is a flow chart showing the calculation procedure of the operation plan correction unit, FIG. 3 is a flow chart showing the error correction procedure of the simulator, FIG. 4 is a detailed block diagram of the optimum correction unit of the load curve, and FIG. 5 is an atom. It is a chart which shows an example of the simulation when making a furnace output raise. 1 ... Operator, 2 ... Man-machine interface, 3 ... Plant, 4 ... Subloop control unit, 5 ...
Integrated control unit, 6 ... State prediction unit, 7 ... State monitoring unit, 8
…… Simulator error correction unit, 9 …… Curve optimum correction unit, 10 …… Operation plan creation knowledge database, 11 …… Plant data, 21 …… Modified load curve assumption unit, 22 …… Input processing unit, 23 …… Prediction Simulation, 24 …… Result retrieval section, 25 …… Curve performance evaluation section, 26 …… Optimal measure search section, 27 …… Basic load curve, 28 …… Core simulation, 29 …… Operation restriction conditions.

Claims (4)

[Claims] 1. A general control section gives a control target value to a sunloop control section based on a load curve given from an operation plan database, and a control amount detection signal of each component of a nuclear power plant is sent to the subloop control section. Input,
In a method of controlling a nuclear power plant constituent device so that the controlled variable approaches a target value, (a) based on operating parameters of the nuclear power plant, its fluctuation is predicted, and (b) based on the operating parameters. Determining whether or not the operation restriction condition is met, and (c) calculating an error correction value based on the actual measurement value and the predicted value of the operation parameter, and (d) determining the item (b). An automatic operation method for a nuclear power plant, characterized in that the load curve is modified based on the result and the error correction value in the item (c), and the modified load curve is fed back to the general control unit. 2. The calculation of the error correction value in the item (c) uses time series data of the deviation between the actual measurement value and the predicted value,
A statistical prediction model for predicting an erroneous deviation is created, and an error correction value is calculated based on a deviation prediction value using this prediction model. Automatic operation method of nuclear power plant in Japan. 3. The correction of the load curve in the item (d) is based on the load curve of the database using (i) the judgment result in the item (b) and the error correction value in the item (c). Assuming a modified load curve, (ii) using a core simulator, determining the reactor state at each point of the modified load curve and the margin from the operation restriction conditions, and (iii) the reactor power fluctuation curve determined from the modified load curve. And the deviation from the same required fluctuation curve, and the output fluctuation of the operating restriction condition allowance, an evaluation function that uses the integral value over the operating section as a parameter is searched for the optimum direction of the modified load curve, and (iv) the above search. The new modified load curve is assumed based on the result, and the same calculation is repeated until the optimum modified load curve is obtained by the evaluation function. Automatic method of operating a nuclear power plant. 4. A sub-loop control unit which takes in a control amount detection signal from each component of a nuclear power plant and controls the component to a control target value, and the sub-loop control unit based on a load curve given from an operation plan database. On the other hand, in the automatic operation device of the nuclear power plant provided with the integrated control unit that gives the control target value, (e) the control amount detection signal from each component of the nuclear power plant is taken in and the fluctuation of the control amount is predicted. A state predicting unit, (f) a state monitoring unit that takes in the control amount detection signal and determines whether or not the operating parameter matches the operating condition, and (g) based on the measured value and the predicted value of the control amount. Then, based on the simulator error correction unit that calculates the error correction value of the predicted value, (h) the judgment result of the above (f) term, and the error correction value of the above (g) term, the operation plan data base An automatic operation device for a nuclear power plant, which is provided with a curve optimum correction unit that corrects the load curve of the vehicle.