JPH05196793A - Supporting apparatus for preparation of operation plan - Google Patents

Abstract

(57) [Abstract] [Purpose] To support the restart plan by predicting the criticality of the core quickly and accurately after an unexpected reactor shutdown. [Configuration] At normal startup, the learning mode is entered and a specific relational expression (k = f (Xe, Sm, C, P, V, T, E), where k: neutron multiplication factor, Xe: xenon density, Sm: Samarium density, C:
Equivalent control rod insertion amount, P: output, V: equivalent void: core equivalent temperature, E: core average burnup, f: Xe, Sm, C, P, V, T, E
Neutron multiplication factor is calculated by a function relating to the operation parameter), and the relationship between the calculated value of the neutron multiplication factor and the actual value to the operation parameter is learned. When a restart plan is created, the prediction mode is entered, other parameters to be predicted for the set operating parameters are obtained from the above relational expression, and the error of the neutron multiplication factor predicted from the operating parameters by learning is converted using the above relational expression. Then, the operating parameters to be predicted are corrected. Accurate and high-speed predictive calculation is possible with the operating parameters corrected based on the actual results.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a support device and an operation training simulator which employ an algorithm for improving prediction accuracy and prediction calculation time in creating a reactor operation plan.

[0002]

2. Description of the Related Art Conventionally, a skilled core administrator has made trial and error using a three-dimensional core simulator to prepare a restart plan after an unexpected reactor shutdown. A system for supporting the preparation of a restart plan in recent years analyzes a core state based on a three-dimensional simulator or a two-dimensional diffusion calculation.

Regarding the system to be supported in recent years, "Atomic Energy Society of Japan 1987 Annual Meeting D57", "Atomic Energy Society of Japan 1988"
It came to be discussed in the "A13 Autumn Meeting". These all utilize the knowledge and inference of a core operation management expert, and analyze the core state based on a three-dimensional simulator or two-dimensional diffusion calculation. Further, in Japanese Patent Laid-Open No. 63-302318, an operation plan creation support device having a function of making a rough operation plan, correcting the difference from a detailed operation plan using a three-dimensional simulator, and re-planning while learning. Is being discussed. Further, Japanese Patent Laid-Open No. 55-70795 describes a method of calculating core operating parameters at high speed by a specific calculation formula instead of using such a simulator as a rough prediction method.

On the other hand, Japanese Laid-Open Patent Publication No. 63-200100 discusses an operation automation device that automatically corrects a subsequent plan while taking into account the difference between the operation plan and the actual result.

[0005]

In the above-mentioned conventional technique, since the simulator is used, the actual results of the core state of the nuclear reactor are not reflected, and an error is likely to occur in the restart plan.

When using a three-dimensional simulator or a two-dimensional diffusion calculation, iterative calculation takes time, and after an unexpected reactor shutdown, restarts as soon as possible in order to fill the blank of power supply. No consideration is given to what needs to be done.

An object of the present invention is to assist in planning a restart plan by predicting the criticality of the core quickly and accurately after an unexpected reactor shutdown.

[0008]

[Means for Solving the Problems] The above object is to calculate a difference between a calculated value of the physical quantity and a physical quantity of an operation record by calculating the physical quantity from the operation parameter of the core at the time of normal reactor startup. Learning means for learning the relationship, and learning means for storing the learning result, at the time of restart planning after the operation stop, predict an unknown operating parameter such that a physical quantity set from the input operating parameter is predicted, and from the learning result Prediction means for calculating the correction amount of the operation parameter and outputting the corrected operation parameter, and outputting the operation parameter to the prediction means, inputting the corrected operation parameter corrected by the operation record, and inputting the corrected operation parameter And the operation planning means for determining the operation parameter to be operated at the next point in time of the restart plan.

[0009]

[Operation] To improve the accuracy of the restart plan after an unexpected reactor shutdown, the operating parameters of the core (reactor output, flow rate, control rod insertion amount, xenon amount, samarium (Amount, burnup), learn the relationship between the error between the simple calculation value and the actual value of the neutron multiplication factor of the core (determine the weighting function of the neural net), and predict the means after an unexpected reactor shutdown. , Calculation of the neutron multiplication factor of the core using a simple calculation formula, and planning using a simple calculation formula using the relationship between the operating parameters and the simple calculation error of the neutron multiplication factor of the core (weight function of neural network) Prediction of neutron multiplication calculation error at points (points with different operating parameters) and by correcting it, the neutron multiplication factor of the core is accurately predicted according to actual results It is possible to support the planning.

In order to increase the speed of the restart plan, some of the coefficients of the simple calculation formula are determined in advance by a three-dimensional simulator or two-dimensional diffusion calculation, and the simple calculation formula is mainly used for the calculation. Since it is not necessary to use a three-dimensional simulator or a two-dimensional diffusion calculation, which requires a long calculation time by iterative calculation, the calculation can be performed quickly.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

First, the structure of this embodiment will be described.

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention. In the figure, 1 is a learning unit, which is a neutron multiplication calculation unit 3a, an error learning unit 4a, a weighting coefficient storage unit 4
It has c. Reference numeral 2 is a prediction unit, which includes an output prediction unit 3b, an error prediction unit 4b, and an output correction unit 3c.
Further, 5 is a plan preparation unit, 6 is an input / conversion unit, 7 is a result input unit, 8 is a switching unit, 9 is a detailed simulator such as three-dimensional or two-dimensional, and 10 is an operation plan preparation support device. Also, 3a to 3c are incorporated in a simplified simulator 3 not shown. The switching unit 8 in the figure does not perform a function of switching any input signal, and selects the mode in which the operation plan creation support device 10 is in the learning mode or the prediction mode.

Next, the operation of this embodiment will be described.

1. Learning mode In this learning, a weighting function is input so that the actual operation parameter is input from the actual result input unit 7 and the calculation error between the actual neutron multiplication factor obtained for the input and the simplified neutron multiplication factor is output. To fix. That is, in the weighting function, information about what kind of input causes a simplified calculation error with respect to the actual result is accumulated. By using this information when the prediction mode is entered, it is possible to make predictions in line with actual results.

1) Determination of the coefficient used in the simplified calculation formula.

If the operator is in the learning mode after switching by the switching unit 8, there is nothing to be input by the operator, and before the learning, the coefficients used in the simplified calculation formula by using the three-dimensional or two-dimensional detailed simulator 9 in advance. Need to be decided in advance. As input operation parameters, core output P, core flow rate F, control rod insertion notch number CRG, xenon density Xe,
It is assumed that the samarium density Sm, the burnup E, and the neutron multiplication factor k are used as physical quantities, and a core power P is planned at the time of restart. From the plant, the performance input section 7
The actual operating parameters and the neutron multiplication factor k shown in FIG. After inputting these data, the input and conversion unit 6 newly calculates the operating parameters such as the core equivalent void fraction V, the core equivalent temperature T, and the core equivalent control rod insertion amount CR.

FIG. 2 is an explanatory diagram showing detailed functions of the input / conversion unit 6. The state of this calculation will be described with reference to FIG. That is, using the operation parameters input from the performance input unit 7, the input and conversion unit 6

[0019]

## EQU1 ## CR = g (Nr (i), Wr (i), Wa (C
RG (i))) where CRG: number of control rod insertion notches belonging to group (i), Nr: number of control rods belonging to group (i), Wr: radial position of control rods belonging to group (i) weight for k, Wa: weight for k according to control rod insertion notch number CRG (i), g: function for Nr, Wr, Wa

[0020]

## EQU00002 ## V = h (F, P) where F: core flow rate, P: power, h: function relating to F, P

[0021]

## EQU00003 ## T = j (P) Here, the parameters of CR, V, and T are calculated by the following equation: P: output, j: function relating to P. Also,

[0022]

## EQU00004 ## Nr = Nr (i),

[0023]

## EQU00005 ## Wr = Wr (i),

[0024]

## EQU00006 ## The coefficient such as Wa = Wa (CRG (i)) is obtained in advance by performing a parameter survey with a detailed simulator 9 such as a two-dimensional or three-dimensional simulator. Further, the control rod group i can utilize the grouping or the like used in the control rod pull-out sequence.

Next, in the learning unit 1 shown in FIG. 1, the neutron multiplication calculation unit 3a receives the input operation parameters and the operation parameters calculated by the input and conversion unit 6 as inputs.
At, the neutron multiplication factor k is calculated.

[0026]

K = f (Xe, Sm, CR, P, V, T, E) where k: neutron multiplication factor, Xe: xenon density, Sm: samarium density, CR: insertion amount of equivalent control rod, P: Output, V: Equivalent void fraction, T: Core equivalent temperature, E: Core average burnup, f: Function relating to Xe, Sm, CR, P, V, T, E This state will be described with reference to FIG.

FIG. 3 is an explanatory diagram showing the detailed functions of the simplified simulator.

The simplified simulator 3 includes a neutron multiplication factor calculation unit 3a, an output prediction unit 3b, and an output correction unit 3c. In the neutron multiplication factor calculation unit 3a, the neutron multiplication factor k'is

[0029]

K ′ = f (Xe, Sm, CR, P, V, T, E) Xe: Xenon density, Sm: Samarium density, CR: Equivalent control rod insertion amount, P: Output, V: Equivalent void ratio , T: core equivalent temperature, E: core average burnup, f: Xe, Sm, CR, P, V, T, E are calculated using a relational expression of a function.

The form of the function f representing the above relational expression is determined in advance by the three-dimensional or two-dimensional detailed simulator 9. This state will be described with reference to FIGS. 4 to 7.

FIG. 4 shows the difference (X
_{3 is a} table showing the relationship between e ₀ -Xe) and the effective neutron multiplication factor keff. Effective neutron multiplication factor keff as neutron multiplication factor k
Is selected and the relationship between the xenon density Xe and keff calculated by the detailed simulator 9 is shown with the core power and the core flow rate as parameters. In this figure, Xe ₀ is the reference xenon density. At this time, other operating parameters are used as reference values. From this figure, it is understood that the difference (Xe ₀ −Xe) from the reference value of the xenon density and the neutron effective multiplication factor keff are in a proportional relationship. This coefficient can be determined by a detailed simulator for each core. Although not shown, the relationship of keff with respect to the samarium density Sm has the same proportional relationship as the relationship of keff with respect to the xenon density Xe.

FIG. 5 is a table showing the relationship between the equivalent core temperature T and the effective neutron multiplication factor keff. The equivalent core temperature T is calculated by the detailed simulator 9. In this figure, T ₀ is a reference equivalent core temperature. At this time, other operating parameters are used as reference values. From this figure, the difference of the square root of the equivalent core temperature from the square root of the standard value (√T
_It can be seen that ₀ −√T) and the effective neutron multiplication factor keff are in a proportional relationship. This coefficient is detailed for each core 9
Can be determined by

FIG. 6 is a table showing the relationship between the difference (CR ₀ -CR) from the reference value of the equivalent control rod insertion amount and the effective neutron multiplication factor keff. The equivalent control rod insertion amount CR is calculated by the detailed simulator 9. In this figure, CR ₀
Is a reference equivalent control rod insertion amount. At this time, other operating parameters are used as reference values. From this figure, it is understood that the difference (CR ₀ -CR) from the reference value of the equivalent control rod insertion amount and the effective neutron multiplication factor keff are in a proportional relationship. This coefficient can be determined by a detailed simulator for each core.

FIG. 7 is a table showing the relationship between the equivalent void fraction V and the effective neutron multiplication factor keff. The equivalent void ratio V is calculated by the detailed simulator 9. In this figure, V ₀ is a reference equivalent void fraction V. At this time, other operating parameters are used as reference values. From this figure, it can be seen that the difference (V ₀ −V) from the reference value of the equivalent void fraction and the effective neutron multiplication factor keff are in a proportional relationship. This coefficient depends on the core power and can be determined for each core by a detailed simulator.

2) Learning the difference between the calculated value of the neutron multiplication factor and the actual result for the operating parameter.

An error Δk between the neutron multiplication factor k'calculated as described above and the neutron multiplication factor k input from the actual result.
The error learning unit 4a shown in FIG. In the present embodiment, the learning function of the neural network is utilized in the error learning unit 4a. As described above, the error learning unit 4a uses the operating parameters as inputs and the neutron increase calculated by the simplified simulator 3a as the teacher signal by the simple calculation formula. By taking the error Δk between the magnification (k ′) and the actual neutron multiplication factor (k) passed from the input and conversion unit 6 as a teacher signal and updating the input weighting function for the output stored in 4c, Information representing the relationship of the error of the neutron multiplication factor with respect to the operating parameter is accumulated in the weighting function of 4c. As a result, it is possible to learn the relationship between the error of the simple calculation formula and the operating parameter, and it is possible to predict the error of the simple calculation formula for the operating parameter at the planned point. It is possible to accurately predict the neutron multiplication factor.

This process will be described in detail with reference to FIG.

FIG. 8 is a block diagram showing the detailed functions of the error learning unit of the present invention. First, when the actual neutron multiplication factor k is input from the input and conversion unit 6 and the neutron multiplication factor k ′ calculated using the simplified calculation formula is input from the neutron multiplication calculation unit 3a, this difference is taken and the teacher signal Δ Calculate kt. on the other hand,
Other operating parameters are input from the input and conversion unit 6 to the neutron effective multiplication error prediction unit 4a ₁ and the weight update unit 4a ₂ . The neutron effective multiplication error prediction unit 4a ₁ multiplies the weighting coefficient w ₀ accumulated in the weighting coefficient storage unit 4c so far to the input operation parameter and outputs the prediction error Δk ₀ . The weight updating unit 4a ₂ reduces the weight coefficient wn so that the difference δk between the output prediction error Δk ₀ and the teacher signal Δkt becomes small.
Is calculated and stored in the weighting coefficient storage unit 4c. By repeating this, the weighting coefficient can gradually represent the relationship with respect to the input of a certain operating parameter, what kind of output, that is, the calculation error due to the simplified calculation formula.

2. Prediction mode Next, the operation when a restart plan is created due to an unexpected reactor shutdown, etc. will be explained. At this time, the switching unit 8 is switched to enter the prediction mode. Switching is performed by the operator, but automatic switching requires means for determining whether the cause of the unexpected stop of the device has been eliminated and switching.

When the predictive mode is entered, the operator determines which operating parameter (here, core output P) is to be operated, and corrects and inputs the parameter into the startup plan creating section 5. At the beginning of the prediction mode, all the parameters have the values immediately before the mode switching. If the operation decision of the operation parameter is automatically made by, for example, an expert system, it is not necessary for the operator to be involved. This judgment is based on whether the flow rate is within the operable range of the core flow rate and core output, or a preset control rod sequence (control rod operation procedure).
It is done comprehensively according to whether it is in line with, or whether the thermal margin is satisfied. How will other operating parameters change in order to maintain criticality by manipulating this operating parameter?
Alternatively, the predicting unit 2 predicts how to change. FIG. 1 predicts what the critical core power will be for the operating parameters determined by the operator. For example, if it is determined that the core output obtained from the prediction unit 2 is not within the operable range for the planned flow rate, control rod, or the like, or that the output does not satisfy the thermal limit, the operation is performed. The staff makes a plan again for the flow rate, control rod, etc. in the plan making unit 5, and inputs it again. In the prior art, the prediction unit 2 of this embodiment corresponds to a three-dimensional or two-dimensional simulator, which has an error with respect to the actual core state. In the present embodiment, the prediction error can be corrected by learning the core performance and the calculation error in advance, and can be used as the input for the next plan creation, so the accuracy of the established plan is high.

The prediction mode shown in FIG. 1 shows the process of predicting and calculating the core power P when other operating parameters are planned as described above. The error predictor 4b inversely transforms the control rod insertion amount CRG, the core flow rate F, the xenon density Xe, and the samarium density Sm from the input and converter 6 as shown in FIG. 2 and the simplified calculation formula as shown in FIG. The core output P ′, the void fraction V, and the core temperature T, which are the functions, are input from the output predictor 3b. This becomes the operating parameter at the planned point, that is, the input of the neural network of the error prediction unit 4b. The error predictor 4b weights these inputs by 4c and obtains the neutron multiplication factor calculation error Δk ′ by the simple calculation formula at the operating point. The fact that there is Δk ′ means that the core power P ′ actually obtained by the inverse transformation of the simplified calculation formula in the power predicting unit 3b does not become critical, but becomes critical at the point shifted by the power prediction error ΔP. It will be. The output correction unit 3c corrects this, and the corrected core power P is displayed on the plan creation unit 5, and the corrected core power P is used for the next plan creation.

When the operation parameter to be predicted is input to the operation plan creation unit 5, the input and conversion unit 6 newly calculates the equivalent control rod insertion amount CR. Next, with the operating parameters transferred from the input and conversion unit 6 as inputs, the power prediction unit 3b calculates the core power P '. This process will be described with reference to FIG. In FIG. 3, Xe and S are added to the output prediction unit 3b.
When core parameters such as m and CR are input, other parameters are calculated by the following equation.

[0043]

## EQU9 ## P '= f' (Xe, Sm, CR, F, E; k
(= 1.0)) where: F: core flow rate f: function relating to Xe, Sm, CR, F, E

[0044]

## EQU10 ## V = h (F, P ')

[0045]

## EQU11 ## T = j (P ')

[0046]

The function f'of

In Equation (7), k is fixed, the core equivalent void fraction V is expressed using the core flow rate F and the core output P ′, and the core output P ′ is solved again.

Now, returning to FIG. 1, from the input / conversion unit 6, operation parameters such as Xe, Sm, CRG, and CR are exchanged as inputs to the error prediction unit 4b. And 4c form a neural network. The error prediction unit 4b, which is a neural network,
Using the weighting function stored in 4c, the neutron multiplication error Δk ′ when these operating parameters are input is output.

The output correction unit 3c uses this error Δk 'to calculate the correction value ΔP of the output P'.

The error learning unit 4a takes in an error between the neutron multiplication factor (k ') calculated by the simple calculation formula in the simplified simulator 3a as the teacher signal and the actual neutron multiplication factor (k), and this teacher signal. Is necessary to correct the weight so that is equal to the output (not shown) of the neural network, the error predictor 4b of the predictor 2 (simply weights the input and calculates the output) Although it is different, it may be considered that it is not a different thing as a neural net and is used in a different way. The error learning units 4a, 4b, and 4c are combined to form a neural network. The neural nets (4a, 4c) used in the learning unit 1 are the operating parameters (inputs) during the normal operation of the error between the calculated value (k ') of the neutron multiplication factor by the simple calculation formula and the actual value (k) during the normal operation. )
To learn the relationship to (store weight in 4c). The error predicting unit (4b, 4c) of the predicting unit 2 sets the operating parameter at the planned point as an input of the neural network, and
The error (Δk ′) due to the simple calculation at the planned point is predicted and output by multiplying the weight stored in c. By using this as a correction term,
The neutron multiplication factor can be predicted more accurately.

The weight storage unit 4c stores the relationship between the operating parameter and the simple calculation error of the neutron multiplication factor as a weight function during learning. Therefore, it is the simple calculation error of the neutron multiplication factor of the planned point that can be calculated with the operating parameter of the planned point as an input, and the simple calculation error of the output is not directly obtained. In order to do this, it is conceivable to use the output prediction error as a teacher signal during learning to create a weighting function, but during actual planning, not only the output prediction but also the output is given so that it becomes critical. It is also necessary to predict the insertion amount of the control rod and the core flow rate, and to directly predict these, it is necessary to learn and store different weighting functions,
It is not efficient in terms of calculation time and capacity. Than that 3
Since the prediction calculation and the correction calculation of b and 3c hardly cause a bottleneck in the calculation time and the capacity, it is preferable to provide the prediction unit and the correction unit for each operation parameter. This is the reason why only the calculation error of the core neutron multiplication factor is learned and the respective prediction units and correction units are provided for the prediction calculation of other operating parameters.

The process of calculating the above-mentioned correction value ΔP will be described with reference to FIG. In FIG. 3, the output correction unit 3c calculates the output correction amount by the following formula.

[0052]

[Equation 12]

Here, (∂k / ∂P) is the function f of the formula 1.
Can be saved when using the detailed simulator 9.

When the correction amount ΔP is obtained in this way, as shown in FIG. 1, by adding the correction amount ΔP and the output P ′ calculated by the output predicting unit 3b, the predicted output P based on the actual result is obtained. Can be calculated. By using the core power P to create the xenon density, the samarium density, etc. at the next time by the operation plan creation unit 5, the operation for obtaining the power at the next time can be continued. That is, an operation plan for restarting is set up.

The prediction in the case where the core power P is selected as the operation parameter to be predicted has been described above. However, although not shown, in addition to the power prediction unit 3b, the control rod operation amount prediction unit and the core flow rate prediction unit are also provided. If a control rod operation amount correction unit and a core flow rate correction unit are provided in addition to the output correction unit 3c,
In a similar manner, for example, when the xenon changes under a constant output, the predicting unit 2 can predict what the control rod operation amount should be, what the flow rate should be, and the like. Using the core flow rate F as the operating parameter to be predicted

If you choose

Instead of [Equation 9], use the relational expression solved for the core flow rate F

Yes,

Instead of (∂k / ∂P) in [Equation 12], (∂k / ∂F)
Similarly, a plan is made by using.

FIG. 9 is an explanatory diagram for explaining the operation of the predicting unit when predicting the flow rate. In the example shown in FIG. 1, the core flow rate F, the equivalent control rod insertion amount CR, etc. were planned and the critical core power P was predicted. The control rod insertion amount CR and the like are planned and passed to the flow rate predicting unit 3b ′ through the input / converting unit 6. The flow rate predictor 3b 'calculates the flow rate F', the equivalent void fraction V, the equivalent core temperature T, etc., and transfers it to the error predictor 4b. The error predicting unit 4b receives the output of the flow rate predicting unit 3b ′ and the input and the output of the converting unit 6 as input, and multiplies the weighting coefficient stored in 4c to calculate the prediction error Δk ′ of the neutron multiplication factor. The flow rate correction unit 3c 'has a neutron multiplication prediction error Δ
A correction amount ΔF of the flow rate F ′ predicted by the flow rate predicting unit 3b ′ is calculated from k ′, and this is corrected and displayed as the core flow rate F in the operation plan creating unit 5. With the core flow rate F, the operation plan creation unit 5 confirms that the plan is within the flow rate, the output limit value, and the thermal limit, and creates the next operation plan. If the core flow rate F does not satisfy the above condition, the operation plan is recreated and the above procedure is repeated.

When the equivalent control rod insertion amount CR is selected as the operation parameter to be predicted,

Also

Use the relational expression solved for the equivalent control rod insertion amount CR instead of

,

Instead of (∂k / ∂P) in (Equation 12), (∂k / ∂C)
A plan is similarly created by using R).

FIG. 10 is an explanatory view for explaining the operation of the predicting unit when predicting the equivalent control rod insertion amount CR. When predicting the equivalent control rod insertion amount CR, the operation planning unit 5 plans the core output P, the core flow rate F, and the like, and transfers it to the equivalent control rod insertion amount prediction unit 3b ″ through the input and conversion unit 6. In the rod insertion amount predicting unit 3b ″, the equivalent control rod insertion amount CR ′,
The equivalent void fraction V, the equivalent core temperature T, etc. are calculated and passed to the error prediction unit 4b. The error predictor 4b is the flow rate predictor 3b ″.
And the input and the output of the conversion unit 6 are input, and the weighting coefficient stored in 4c is multiplied to calculate the neutron multiplication prediction error Δk ′. The flow rate correction unit 3c ″ calculates the correction amount ΔCR of the equivalent control rod insertion amount CR ′ predicted by the equivalent control rod insertion amount prediction unit 3b ″ from the prediction error Δk ′ of the neutron multiplication factor, and corrects this to obtain the equivalent value. The control rod insertion amount CR is displayed on the operation plan creation unit 5. With the equivalent control rod insertion amount CR, the operation plan creation unit 5 confirms that the operation plan is within the preset control rod operating procedure and thermal limits, and creates the next operation plan. If the equivalent control rod insertion amount CR does not satisfy the above condition, the operation plan is recreated and the above procedure is repeated.

For flow rate prediction and equivalent control rod insertion amount prediction,
The simplified simulator 3 has the functions shown in FIG.
As shown in FIG. 1, the flow rate predicting unit 3b ′ and the equivalent control rod insertion amount predicting unit 3b ″ are provided.
a ′, equation 3 ″, and equation 4 are used to calculate the flow rate F ′, the equivalent void fraction V, the equivalent core temperature T, etc., and pass them to the error predictor 4b. Further, the equivalent control rod insertion amount predictor 3b ″ is Formula 1
a ′, equations 3 and 4 are used to calculate CR ′, V, T, etc., and the result is passed to the error predictor 4b. The error predictor 4b outputs signals 3b, 3b ′, 3b ″ as shown in FIG. There is a changeover switch 4d for changing the output destination to 3c, 3c ', and 3c "depending on the presence or absence of each.
Other operating parameters that are not input from b ″ are input and input from the conversion unit 6. When all operating parameters other than the neutron multiplication factor are input to the neutron multiplication error prediction unit 4a ₁ , 4a ₁ is stored in 4c. The weighting coefficient is taken in, multiplied by the input, and the prediction error Δk ′ is output to the output destination set by the changeover switch 4d.

According to the present embodiment, the past performance can be used for the operation plan at the time of restarting, and there is an effect that the prediction accuracy can be improved as compared with the case where only the conventional simulator or the simple calculation formula is used. Further, the three-dimensional or two-dimensional detailed simulator may be used only at the time of the first operation planning to determine the functional form relating to the operation parameter of the neutron multiplication factor, and it is not necessary to repeatedly use it when creating the operation plan. There is an effect that can be shortened.

Next, another embodiment of the present invention will be described.

FIG. 13 is a block diagram showing the structure of the learning section of another embodiment.

The point different from the learning section in the above embodiment is that the relationship between the operating parameter and the neutron multiplication factor, which has been previously determined by using the detailed simulator 9, is changed in time series of the operating parameter at the time of initial startup and operation.

[0068]

[Equation 13]

And the sensitivity coefficient creating means 11 which is obtained from the time series change of the neutron multiplication factor. The sensitivity coefficient creating means 11 may use a method of obtaining a sensitivity coefficient ((∂k / ∂P), (∂k / ∂Xe) in the figure) using a neural network. Also, if the functional form is determined,
Least-squares fitting formula etc. can be used. With such a configuration, an operation plan can be created without using a detailed simulator, and high-speed analysis becomes possible.

Further, another embodiment will be described.

FIG. 14 is a block diagram showing the construction of a performance input section of another embodiment.

According to the actual results, when the effective neutron multiplication factor is selected as the physical quantity, it is usually set to 1.0. Since FIG. 14 further uses the actual effective multiplication factor, the period obtained by the period meter 7a using the neutron flux φ as an input is converted using the period-neutron multiplication factor conversion unit 7b to obtain the neutron effective multiplication factor. It is what This makes it possible to more accurately capture the time change of the neutron multiplication factor and improve the accuracy.

The simplified simulator using the relational expression of the present invention shown in FIG. 3 can also be used when simulating the core for core operation training.

In the relational expression used in the simplified simulator of the present invention, operating parameters having a strong correlation with the neutron multiplication factor are selected as shown in FIGS. 4 to 7. It is also possible to select the one shown in Japanese Patent Publication No. 55-70795.

..

According to the present invention, the predicting means after unexpected reactor shutdown stops the criticality of the core by correcting the error learned at the time of normal reactor startup by reflecting the results of normal reactor startup. Because it makes a prediction, it is possible to support accurate restart planning.

[Brief description of drawings]

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

FIG. 2 is an explanatory diagram showing detailed functions of an input / conversion unit 6 according to the embodiment of this invention.

FIG. 3 is an explanatory diagram showing detailed functions of a simplified simulator according to an embodiment of the present invention.

FIG. 4 shows a difference (Xe ₀ −X) from a reference value of xenon density.
3 is a table showing the relationship between e) and the effective neutron multiplication factor keff.

FIG. 5 is a chart showing the relationship between the equivalent core temperature T and the effective neutron multiplication factor keff.

FIG. 6 shows the difference (CR ₀ −) from the reference value of the equivalent control rod insertion amount.
(CR) and a neutron effective multiplication factor keff.

FIG. 7 is a chart showing the relationship between the equivalent void fraction V and the effective neutron multiplication factor keff.

FIG. 8 is a block diagram showing detailed functions of an error learning unit according to the embodiment of this invention.

FIG. 9 is an explanatory diagram illustrating an operation of the predicting unit according to the embodiment of the present invention when predicting a flow rate.

FIG. 10 is an explanatory diagram illustrating an operation of the predicting unit according to the embodiment of the present invention when predicting the equivalent control rod insertion amount CR.

FIG. 11 is an explanatory diagram showing detailed functions of a simplified simulator according to another embodiment of the present invention.

FIG. 12 is an explanatory diagram showing detailed functions of a prediction unit according to another embodiment of the present invention.

FIG. 13 is a block diagram showing a configuration of a learning unit according to another embodiment of the present invention.

FIG. 14 is a block diagram showing a configuration of a performance input unit according to another embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Learning part 2 Prediction part 3 Simplified simulator 3a Neutron multiplication factor calculation part 3b Output prediction part 3c Output correction part 4a Error learning part 4b Error prediction part 4c Weighting factor storage part 5 Operation plan creation part 6 Input and conversion part 7 Actual result input part 7a Period meter 7b Period neutron multiplication conversion unit 8 Switching unit 9 Detailed simulator 10 Operation plan creation support device 11 Sensitivity coefficient creation unit

Claims (10)

[Claims] 1. At the time of normal reactor startup, a physical quantity is used to calculate from the operating parameters of the core, the relationship between the calculated value of the physical quantity and the physical quantity of the actual operation is learned with respect to the operating parameter, and the learning result is obtained. Learning means to be stored and, at the time of creating a restart plan after an operation stop, predict an unknown operation parameter such that a physical quantity set from the input operation parameter is obtained, and calculate a correction amount of the operation parameter from the learning result. Then, the predicting means for outputting the corrected operating parameter and the operating parameter are output to the predicting means, the corrected operating parameter corrected by the operation record is input, and the next operation of the restart plan is performed using the corrected operating parameter. An operation plan creation support device comprising: an operation plan creation means for determining an operation parameter to be operated at a time point. 2. The operation plan creation support device according to claim 1, wherein the physical quantity is a neutron multiplication factor. 3. The operation plan creation support apparatus according to claim 2, wherein the neutron multiplication factor is represented by the following relational expression. k = f (Xe, Sm, CR, P, V, T, E) where, k: neutron multiplication factor, Xe: xenon density, Sm: samarium density, CR: equivalent control rod insertion amount, P: output, V: Equivalent void fraction, T: core equivalent temperature, E: core average burnup, f: function relating to Xe, Sm, CR, P, V, T, E 4. The operation plan creation support device according to claim 3, wherein the equivalent control rod insertion amount is represented by the following relational expression. CR = g (Nr (i), Wr (i), Wa (CRG
(I))) where CRG: the number of control rod insertion notches belonging to group (i), Nr: the number of control rods belonging to group (i), Wr: k for the radial position of the control rods belonging to group (i) Weight, Wa: Weight for k by the control rod insertion notch number CRG (i), g: Function relating to Nr, Wr, Wa 5. The operation plan creation support device according to claim 3, wherein the equivalent void ratio is represented by the following relational expression. V = h (F, P) where F: core flow rate, P: power, h: function relating to F, P 6. The operation plan creation support apparatus according to claim 3, wherein the core equivalent temperature is represented by the following relational expression. T = j (P) where P: output, j: function related to P 7. The operation plan creation support apparatus according to claim 1, wherein a neural network is used as means for learning an error between the calculated value of the physical quantity and the physical quantity of the driving record. 8. The operation plan creation support apparatus according to claim 3, further comprising means for learning the sensitivity of the neutron multiplication factor to the operation parameter from the actual results. 9. The operation plan creation support device according to claim 8, wherein a neural network is used as the learning means. 10. A simulator for operation training, which simulates a core state using the relational expression according to claim 3.