Startup / shutdown control method and apparatus for metal reactor, computer device, and storage medium

Abstract

The present application relates to a startup / shutdown control method and apparatus for a metal reactor, a computer device, and a storage medium. The method comprises: according to the operation data of a unit to be controlled in a first historical period, acquiring a first multi-dimensional timing series corresponding to target parameters and acquiring a modal component corresponding to the first multi-dimensional timing series; according to the modal component corresponding to the first multi-dimensional timing series, constructing a first prediction model for primary loop temperature prediction and a second prediction model for reactor power prediction; acquiring a real-time timing series corresponding to the target parameters, and separately inputting the real-time timing series into the first prediction model and the second prediction model so as to acquire a primary loop temperature prediction series and a reactor power prediction series; and, according to the primary loop temperature prediction series and the reactor power prediction series, acquiring a predicted working condition of a startup separator and, according to the predicted working condition, performing startup / shutdown control on the unit to be controlled. Using the method can reduce startup / shutdown control costs of metal reactors.

Description

Metal stack start-stop control method, device, computer equipment and storage medium

[0001] Cross-references

[0002] This application refers to Chinese Patent Application No. 2023114202120, filed on October 30, 2023, entitled “Method, device, computer equipment and storage medium for start-stop control of metal stack”, which is incorporated into this application in its entirety by reference. Technical Field

[0003] The present application relates to the field of metal stack control technology, and in particular to a start-stop control method, device, computer equipment, storage medium, and computer program product for a metal stack. Background Art

[0004] The startup separator is a crucial device during the startup and shutdown phases of a metal stack. Its primary function is to absorb the secondary circuit fluid and maintain its pressure near a reference value. During the startup and shutdown process, the startup separator's operating conditions must be identified in real time to provide decision-making information for the stack's startup and shutdown control.

[0005] In the related art, expensive hardware instruments are usually relied upon to identify the operating conditions of the start-up separator, which makes the start-stop control cost of the metal stack high.

[0006] Summary of the Invention

[0007] Based on this, it is necessary to provide a start-stop control method, device, computer equipment, computer-readable storage medium and computer program product for a metal stack that can reduce the start-stop control cost of the metal stack in order to address the above technical problems.

[0008] In a first aspect, the present application provides a method for starting and stopping a metal stack, comprising:

[0009] Obtaining, based on operating data of the unit to be controlled in a first historical period, a first multidimensional time series sequence corresponding to target parameters and obtaining modal components corresponding to the first multidimensional time series sequence; the target parameters include primary circuit temperature and reactor power;

[0010] A first prediction model for primary circuit temperature prediction is constructed based on the modal components corresponding to the first multidimensional time series, and a second prediction model for reactor power prediction is constructed based on the modal components corresponding to the first multidimensional time series; the first prediction model and the second prediction model are obtained by combining multiple single models;

[0011] Obtaining a real-time time series sequence corresponding to the target parameter, inputting the real-time time series sequence into the first prediction model and the second prediction model respectively, to obtain a primary circuit temperature prediction sequence and a reactor power prediction sequence;

[0012] According to the primary circuit temperature prediction sequence and the reactor power prediction sequence, the predicted operating conditions of the start-up separator are obtained, and the start and stop control of the control unit is performed according to the predicted operating conditions.

[0013] In one embodiment, the target parameters also include at least one of pressurizer pressure, pressurizer water level, primary circuit pressure, reactor coolant outlet temperature, reactor coolant inlet temperature, coolant flow time, primary circuit flow, primary circuit boron concentration, steam generator pressure, steam generator water level, high-pressure cylinder exhaust pressure, first-stage reheater extraction pressure, low-pressure cylinder inlet and outlet superheated steam pressure, low-pressure cylinder inlet and outlet superheated steam temperature, condenser operating pressure, feed water temperature and steam reheat degree.

[0014] In one embodiment, obtaining a first multi-dimensional time series sequence corresponding to a target parameter includes:

[0015] Obtain the initial time series corresponding to each target parameter;

[0016] Perform data preprocessing on the initial time series corresponding to each target parameter to obtain the sub-time series corresponding to each target parameter; the data preprocessing operation includes missing value filling and outlier replacement;

[0017] A first multidimensional time series sequence is obtained according to the sub-time series sequences corresponding to each target parameter.

[0018] In one embodiment, the single model includes a long short-term memory (LSTM) model and an extreme learning machine (ELM) model; a first prediction model for primary circuit temperature prediction is constructed based on the modal components corresponding to the first multidimensional time series, including:

[0019] Iteratively training the initial LSTM model and the initial ELM model according to the modal component corresponding to the first multidimensional time series sequence to obtain a trained LSTM model and a trained ELM model;

[0020] Get the initial weighting coefficient;

[0021] According to the initial weighting coefficient, the trained LSTM model and the trained ELM model are weightedly fused to obtain an initial prediction model, which is used as the first prediction model.

[0022] In one embodiment, constructing a first prediction model for primary circuit temperature prediction based on the modal components corresponding to the first multidimensional time series further includes:

[0023] Obtaining a second multi-dimensional time series corresponding to the target parameter based on the operating data of the unit to be controlled in the second historical period;

[0024] A sliding window mechanism is adopted to iteratively update the initial prediction model according to the second multidimensional time series and the fusion error threshold to obtain a target prediction model, which is used as the first prediction model.

[0025] In one embodiment, starting and stopping the controlled unit according to the predicted operating conditions includes:

[0026] When the predicted operating conditions are the same as the current operating conditions of the start-up separator, the target control parameters are determined according to the reactor outlet temperature deviation, the once-through steam generator steam pressure deviation, and the primary circuit outlet flow deviation.

[0027] In the case where the predicted operating condition is different from the current operating condition of the start-up separator, the preset control parameter corresponding to the predicted operating condition is used as the target control parameter;

[0028] According to the target control parameters, the load regulating valve, the start-up separator outlet isolation valve, the main feed water pump speed, the main feed water low load regulating valve, the main feed water pump return water regulating valve and the start-up separator steam pipeline regulating valve are adjusted.

[0029] In a second aspect, the present application further provides a start-stop control device for a metal stack, comprising:

[0030] an acquisition module, configured to acquire, based on operating data of the unit to be controlled in a first historical period, a first multidimensional time series sequence corresponding to a target parameter and a modal component corresponding to the first multidimensional time series sequence; the target parameters include a primary circuit temperature and a reactor power;

[0031] a construction module, configured to construct a first prediction model for primary circuit temperature prediction based on the modal components corresponding to the first multidimensional time series, and to construct a second prediction model for reactor power prediction based on the modal components corresponding to the first multidimensional time series; the first prediction model and the second prediction model are obtained by combining multiple single models;

[0032] A prediction module is used to obtain a real-time time series sequence corresponding to the target parameter, input the real-time time series sequence into the first prediction model and the second prediction model respectively, and obtain a primary circuit temperature prediction sequence and a reactor power prediction sequence;

[0033] The control module is used to obtain the predicted operating conditions of the start-up separator according to the primary circuit temperature prediction sequence and the reactor power prediction sequence, and to start and stop the controlled unit according to the predicted operating conditions.

[0034] In a third aspect, the present application further provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0035] Obtaining, based on operating data of the unit to be controlled in a first historical period, a first multidimensional time series sequence corresponding to target parameters and obtaining modal components corresponding to the first multidimensional time series sequence; the target parameters include primary circuit temperature and reactor power;

[0036] A first prediction model for primary circuit temperature prediction is constructed based on the modal components corresponding to the first multidimensional time series, and a second prediction model for reactor power prediction is constructed based on the modal components corresponding to the first multidimensional time series; the first prediction model and the second prediction model are obtained by combining multiple single models;

[0037] Obtaining a real-time time series sequence corresponding to the target parameter, inputting the real-time time series sequence into the first prediction model and the second prediction model respectively, to obtain a primary circuit temperature prediction sequence and a reactor power prediction sequence;

[0038] According to the primary circuit temperature prediction sequence and the reactor power prediction sequence, the predicted operating conditions of the start-up separator are obtained, and the start and stop control of the control unit is performed according to the predicted operating conditions.

[0039] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0040] Obtaining, based on operating data of the unit to be controlled in a first historical period, a first multidimensional time series sequence corresponding to target parameters and obtaining modal components corresponding to the first multidimensional time series sequence; the target parameters include primary circuit temperature and reactor power;

[0041] A first prediction model for primary circuit temperature prediction is constructed based on the modal components corresponding to the first multidimensional time series, and a second prediction model for reactor power prediction is constructed based on the modal components corresponding to the first multidimensional time series; the first prediction model and the second prediction model are obtained by combining multiple single models;

[0042] Obtaining a real-time time series sequence corresponding to the target parameter, inputting the real-time time series sequence into the first prediction model and the second prediction model respectively, to obtain a primary circuit temperature prediction sequence and a reactor power prediction sequence;

[0043] According to the primary circuit temperature prediction sequence and the reactor power prediction sequence, the predicted operating conditions of the start-up separator are obtained, and the start and stop control of the control unit is performed according to the predicted operating conditions.

[0044] In a fifth aspect, the present application further provides a computer program product, comprising a computer program, which, when executed by a processor, implements the following steps:

[0045] Obtaining, based on operating data of the unit to be controlled in a first historical period, a first multidimensional time series sequence corresponding to target parameters and obtaining modal components corresponding to the first multidimensional time series sequence; the target parameters include primary circuit temperature and reactor power;

[0046] A first prediction model for primary circuit temperature prediction is constructed based on the modal components corresponding to the first multidimensional time series, and a second prediction model for reactor power prediction is constructed based on the modal components corresponding to the first multidimensional time series; the first prediction model and the second prediction model are obtained by combining multiple single models;

[0047] Obtaining a real-time time series sequence corresponding to the target parameter, inputting the real-time time series sequence into the first prediction model and the second prediction model respectively, to obtain a primary circuit temperature prediction sequence and a reactor power prediction sequence;

[0048] According to the primary circuit temperature prediction sequence and the reactor power prediction sequence, the predicted operating conditions of the start-up separator are obtained, and the start and stop control of the control unit is performed according to the predicted operating conditions.

[0049] The above-mentioned start-stop control method, device, computer equipment, storage medium and computer program product of the metal stack first obtain the first multidimensional time series sequence corresponding to the target parameter based on the operating data of the unit to be controlled in the first historical period, and obtain the modal components corresponding to the first multidimensional time series sequence, and construct a first prediction model for primary circuit temperature prediction and a second prediction model for reactor power prediction based on the modal components corresponding to the first multidimensional time series sequence, and then obtain the real-time time series sequence corresponding to the target parameter, input the real-time time series sequence into the first prediction model and the second prediction model respectively, obtain the primary circuit temperature prediction sequence and the reactor power prediction sequence, obtain the predicted operating conditions of the start separator based on the primary circuit temperature prediction sequence and the reactor power prediction sequence, and start-stop control of the unit to be controlled based on the predicted operating conditions. In this way, the operating conditions of the start separator can be obtained based on the constructed prediction model, and then the start-stop control of the unit to be controlled can be performed, without relying on hardware instruments for operating condition identification, thereby reducing the start-stop control cost of the metal stack.

[0050] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the embodiments below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present application. The same reference numerals are used throughout the drawings to denote the same components. In the drawings:

[0052] FIG1 is a diagram illustrating an application environment of a start-stop control method for a metal stack according to an embodiment;

[0053] FIG2 is a schematic flow chart of a method for controlling the start and stop of a metal stack according to an embodiment;

[0054] FIG3 is a schematic diagram of a process for optimizing an initial prediction model according to an embodiment;

[0055] FIG4 is a schematic flow chart of a method for controlling the start and stop of a metal stack in another embodiment;

[0056] FIG5 is a structural block diagram of a start-stop control device for a metal stack according to one embodiment;

[0057] FIG6 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0058] The following embodiments of the technical solution of the present application will be described in detail with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present application and are therefore only examples and are not intended to limit the scope of protection of the present application.

[0059] The start-stop control method for a metal stack provided in an embodiment of the present application can be applied in an application environment as shown in FIG1 . In which, the terminal 102 communicates with the server 104 via a network. The server 104 can construct a primary circuit temperature prediction model and a reactor power prediction model based on the historical operating data of the unit to be controlled, and obtain a primary circuit temperature prediction sequence and a reactor power prediction sequence through the primary circuit temperature prediction model and the reactor power prediction model during the real-time operation of the unit to be controlled. Then, based on the primary circuit temperature prediction sequence and the reactor power prediction sequence, the predicted operating conditions for starting the separator are obtained, and the unit to be controlled is started and stopped according to the predicted operating conditions. The terminal 102 can be, but is not limited to, an equipment parameter acquisition and monitoring device in a nuclear power plant, and the server 104 can be implemented as an independent server or a server cluster consisting of multiple servers.

[0060] In an exemplary embodiment, as shown in FIG2 , a method for controlling the start and stop of a metal stack is provided. The method is described by taking the server 104 in FIG1 as an example, and includes the following steps:

[0061] S202: Obtain a first multidimensional time series sequence corresponding to target parameters based on the operating data of the unit to be controlled in a first historical period, and obtain modal components corresponding to the first multidimensional time series sequence; the target parameters include primary circuit temperature and reactor power.

[0062] The unit to be controlled is a metal pile generator unit. The first historical period can be any historical operating period of the unit to be controlled, and the period length can be set according to the actual operating conditions of the unit to be controlled, for example, 24 hours as one operating period.

[0063] Optionally, before performing start-stop control on the unit to be controlled, the server needs to build a primary circuit temperature prediction model and a reactor power prediction model.

[0064] During the model building process, the first step is to obtain model training data. The server will obtain the initial time series corresponding to each target parameter based on the operating data of the controlled unit in the first historical period, and then generate a multi-dimensional time series based on the initial time series corresponding to each target parameter.

[0065] In addition to the two measured parameters, primary circuit temperature and reactor power, target parameters can also include one or more related influencing parameters. These include pressurizer pressure, pressurizer water level, primary circuit pressure, reactor coolant outlet temperature, reactor coolant inlet temperature, coolant flow time, primary circuit flow, primary circuit boron concentration, steam generator pressure, steam generator water level, high-pressure cylinder exhaust pressure, first-stage reheater extraction pressure, low-pressure cylinder inlet and outlet superheated steam pressure, low-pressure cylinder inlet and outlet superheated steam temperature, condenser operating pressure, feedwater temperature, and steam reheat degree.

[0066] After obtaining the multidimensional time series sequence, the server uses the empirical mode decomposition (EMD) algorithm to decompose the time series sequence of each dimension in the multidimensional time series sequence separately, obtain the modal components of the time series sequence of each dimension, and recombine the modal components of the time series sequence of each dimension to obtain multiple modal components corresponding to the multidimensional time series sequence.

[0067] S204: Constructing a first prediction model for primary circuit temperature prediction based on the modal components corresponding to the first multidimensional time series, and constructing a second prediction model for reactor power prediction based on the modal components corresponding to the first multidimensional time series; the first prediction model and the second prediction model are obtained based on a combination of multiple single models.

[0068] Among them, the single model may include but is not limited to a long short-term memory network (LSTM) model, an extreme learning machine (ELM) model, a back propagation neural network (BPNN) model, and a support vector regression (SVR) model.

[0069] Among them, the ways of combining multiple single models include but are not limited to voting, weighted averaging, stacking, and weighted fusion.

[0070] Optionally, after obtaining the modal components corresponding to the first multidimensional time series sequence, the server first iteratively trains multiple single models using the modal components corresponding to the first multidimensional time series sequence as input and the primary circuit temperature prediction sequence as output until the prediction error of each single model is less than a set threshold, thereby obtaining multiple trained primary circuit temperature prediction single models. The multiple trained primary circuit temperature prediction single models are then combined to obtain a first prediction model.

[0071] Furthermore, the server needs to iteratively train multiple single models using the modal components corresponding to the first multidimensional time series as input and the reactor power prediction sequence as output, until the prediction error of each single model is less than a set threshold, thereby obtaining multiple trained reactor power prediction single models. The multiple trained reactor power single models are then combined to obtain a second prediction model.

[0072] By combining the single models to obtain a combined prediction model, and performing parameter prediction based on the combined model, the prediction accuracy of the primary circuit temperature and reactor power can be improved.

[0073] S206: Obtain a real-time time series sequence corresponding to the target parameter, input the real-time time series sequence into the first prediction model and the second prediction model respectively, and obtain a primary circuit temperature prediction sequence and a reactor power prediction sequence.

[0074] Optionally, after completing the model training, the server obtains the real-time multi-dimensional time series sequence corresponding to the target parameters based on the operating data of the unit to be controlled during real-time operation, inputs the real-time multi-dimensional time series sequence into the first prediction model, obtains the first-loop temperature prediction sequence, and inputs the real-time multi-dimensional time series sequence into the second prediction model to obtain the reactor power prediction sequence.

[0075] S208: According to the primary circuit temperature prediction sequence and the reactor power prediction sequence, the predicted operating condition of the start-up separator is obtained, and the start-up and shutdown control of the unit to be controlled is performed according to the predicted operating condition.

[0076] Optionally, the server pre-configures multiple start-stop conditions and corresponding determination criteria for each start-stop condition. After obtaining the primary circuit temperature prediction sequence and the reactor power prediction sequence, the server determines the predicted operating conditions for the start-up separator in the future time period based on the primary circuit temperature prediction sequence, the reactor power prediction sequence, and the determination criteria for each start-stop condition. The server then performs start-stop control on the controlled unit based on the predicted operating conditions.

[0077] In an optional embodiment, the preset startup and shutdown conditions include startup condition 1, startup condition 2, startup condition 3, startup condition 4, and shutdown condition 1. The startup and shutdown conditions are distinguished as follows:

[0078] (1) Startup condition 1: The primary circuit temperature rises to 0% FP power of the reactor

[0079] (2) Startup condition 2: Reactor power increases from 0% FP to 8% FP

[0080] (3) Startup Condition 3: Reactor power increases from 8% FP to 16% FP

[0081] (4) Startup Condition 4: Reactor power increases from 16% FP to 20% FP

[0082] (5) Shutdown condition 1: Reactor power reduced from 20% FP to 0% FP

[0083] In the above-mentioned start-stop control method of the metal stack, the first multidimensional time series sequence corresponding to the target parameter is obtained based on the operating data of the unit to be controlled in the first historical period, and the modal components corresponding to the first multidimensional time series sequence are obtained. According to the modal components corresponding to the first multidimensional time series sequence, a first prediction model for single-loop temperature prediction and a second prediction model for reactor power prediction are constructed. Then, the real-time time series sequence corresponding to the target parameter is obtained, and the real-time time series sequence is input into the first prediction model and the second prediction model respectively to obtain the single-loop temperature prediction sequence and the reactor power prediction sequence. According to the single-loop temperature prediction sequence and the reactor power prediction sequence, the predicted operating conditions of the start separator are obtained, and the unit to be controlled is started and stopped according to the predicted operating conditions. In this way, the operating conditions of the start separator can be obtained according to the constructed prediction model, and then the unit to be controlled is started and stopped, without relying on hardware instruments for operating condition identification, thereby reducing the start-stop control cost of the metal stack.

[0084] In one embodiment, obtaining a first multidimensional time series sequence corresponding to a target parameter includes: obtaining an initial time series sequence corresponding to each target parameter; performing data preprocessing operations on the initial time series sequence corresponding to each target parameter to obtain a sub-time series sequence corresponding to each target parameter; the data preprocessing operations include missing value filling and outlier replacement; and obtaining a first multidimensional time series sequence based on the sub-time series sequence corresponding to each target parameter.

[0085] Optionally, during the collection and recording of historical operating data, missing values ​​and outliers may exist in the initial time series corresponding to the target parameters due to reasons such as equipment failure, sudden failure of communication equipment, or human outage. Therefore, the server needs to perform data preprocessing on the initial time series to meet the input data requirements of the model.

[0086] For missing values, since the data usually have certain similarities under similar operating conditions in different historical operating cycles, we can obtain the corresponding data in multiple different historical cycles before and after the first historical cycle and calculate the average value, and use the obtained average value to fill the missing values ​​in the initial time series.

[0087] For outliers, horizontal and vertical comparison corrections can be performed, as shown below:

[0088] (1) Horizontal comparison correction: Using the adjacent time data in the initial time series as a reference, set the first change threshold. If the data at time t in the initial time series exceeds the first change threshold compared with the adjacent time data, that is, max{|y(d,t)-y(d,t-1)|,|y(d,t)-y(d,t+1)|}>ε

[0089] The data is stabilized by the average value of the adjacent time data, that is:

[0090] Where ε is the first change threshold, y(d,t) is the value of the initial time series at time t, y(d,t+1) is the value of the initial time series at time t+1, and y(d,t-1) is the value of the initial time series at time t-1.

[0091] (2) Vertical comparison correction: Using the data in different historical operation cycles as a reference, set the second change threshold. If the data of the initial time series at time t exceeds the second change threshold compared with the data of different historical operation cycles at time t, that is: y(d,t)-m(d,t)>r

[0092] The data in the initial time series are corrected using the data at the same time in different historical operation cycles, that is: y(d,t)=m(d,t)±r

[0093] Where r is the second change threshold, y(d, t) is the value of the initial time series at time t, and m(d, t) is the value of the data in different historical operation cycles at time t.

[0094] After completing the preprocessing operation on the initial time series sequence corresponding to each target parameter and obtaining the sub-time series sequence corresponding to each target parameter, the server constructs a multidimensional time series sequence based on the sub-time series sequence corresponding to each target parameter.

[0095] In this embodiment, by obtaining the initial time series sequence corresponding to each target parameter and performing data preprocessing operations on the initial time series sequence corresponding to each target parameter, a sub-time series sequence corresponding to each target parameter is obtained. Then, based on the sub-time series sequence corresponding to each target parameter, a first multidimensional time series sequence is obtained, which can improve the data integrity and validity of the multidimensional time series sequence.

[0096] In one embodiment, a single model includes a long short-term memory (LSTM) model and an extreme learning machine (ELM) model; based on the modal components corresponding to the first multidimensional time series sequence, a first prediction model for single-loop temperature prediction is constructed, including: iteratively training the initial LSTM model and the initial ELM model according to the modal components corresponding to the first multidimensional time series sequence to obtain a trained LSTM model and a trained ELM model; obtaining an initial weighting coefficient; and based on the initial weighting coefficient, weightedly fusing the trained LSTM model and the trained ELM model to obtain an initial prediction model, and using the initial prediction model as the first prediction model.

[0097] Alternatively, considering the advantages of the LSTM model in better capturing long-term dependencies in sequence data through cell states and gating mechanisms, and the advantages of the ELM model in fast learning speed and strong generalization ability, the LSTM model and the ELM model can be selected as single models for weighted fusion.

[0098] After obtaining the modal components corresponding to the first multidimensional time series, the server iteratively trains the initial LSTM model and initial ELM using the modal components corresponding to the first multidimensional time series as input and the first-loop temperature prediction sequence as output until the prediction error is less than a set threshold, resulting in a trained LSTM model and a trained ELM model. Initial weighting coefficients are obtained, and the trained LSTM model and the trained ELM model are weightedly fused to obtain a first initial prediction model corresponding to the first-loop temperature. In an optional embodiment, the first initial prediction model corresponding to the first-loop temperature can be directly used as the first prediction model.

[0099] The training method of the initial prediction model of the reactor power can refer to the training method of the initial prediction model of the first loop temperature. Specifically, in an optional embodiment, the single model includes an LSTM model and an ELM model, and according to the modal components corresponding to the first multidimensional time series sequence, a second prediction model for reactor power prediction is constructed, including: according to the modal components corresponding to the first multidimensional time series sequence, the initial LSTM model and the initial ELM model are iteratively trained, the trained LSTM model and the trained ELM model are obtained, the initial weighting coefficient is obtained, and according to the initial weighting coefficient, the trained LSTM model and the trained ELM model are weightedly fused to obtain a second initial prediction model, and the second initial prediction model is used as the second prediction model. The specific implementation method is the same as the specific implementation method in the above embodiment and will not be repeated here.

[0100] The initial weighting coefficient can be obtained in the following way:

[0101] (1) Arithmetic mean method

[0102] Among them, l i represents the weight coefficient of the i-th single model, and m represents the number of single models.

[0103] (2) Inverse sum of squares of prediction errors method

[0104] Among them, l i represents the weight coefficient of the i-th single model, e it represents the absolute error of the i-th single model at time t, x t represents the observation value at time t, x it It represents the predicted value of the i-th single model at time t, m represents the number of single models, and N represents the time length of the prediction sequence.

[0105] In this embodiment, the initial LSTM model and the initial ELM model are iteratively trained according to the modal components corresponding to the first multidimensional time series sequence to obtain the trained LSTM model and the trained ELM model, and then the initial weighting coefficient is obtained. According to the initial weighting coefficient, the trained LSTM model and the trained ELM model are weightedly fused to obtain the initial prediction model. In this way, data prediction can be performed based on the weighted fused combined prediction model, thereby improving the prediction accuracy of the data.

[0106] In one embodiment, a first prediction model for single-loop temperature prediction is constructed based on the modal components corresponding to the first multidimensional time series sequence, and the method also includes: obtaining a second multidimensional time series sequence corresponding to the target parameter based on the operating data of the unit to be controlled in the second historical period; using a sliding window mechanism, iteratively updating the initial prediction model based on the second multidimensional time series sequence and the fusion error threshold to obtain a target prediction model, and using the target prediction model as the first prediction model.

[0107] Optionally, FIG3 is a schematic diagram of the tuning process of the initial prediction model. As shown in FIG3 , the server first obtains a second multi-dimensional time series sequence corresponding to the target parameter based on the operating data of the unit to be controlled in the second historical period.

[0108] When the first initial prediction model predicts the i-th data in the second multidimensional time series, the server determines whether the i-th data is the last data. In the case that the i-th data is the last data, the model tuning is completed, and the first target prediction model is obtained, and the first target prediction model is used as the first prediction model; in the case that the i-th data is not the last data, the i-th data is updated to the sliding window as the new sample data, and then it is determined whether the combined prediction error corresponding to the i-th data is greater than the preset fusion error threshold. If the combined prediction error is not greater than the fusion error threshold, the model continues to predict the i+1-th data in the second multidimensional time series, and the model is not updated; if the combined prediction error is greater than the fusion error threshold, the model parameters and model weights are adjusted according to the new samples accumulated in the sliding window, and then the i+1-th data is continued to be predicted.

[0109] The training method of the target prediction model of the reactor power can refer to the training method of the target prediction model of the first circuit temperature. Specifically, in an optional embodiment, a second prediction model for reactor power prediction is constructed based on the modal components corresponding to the first multidimensional time series sequence, and further includes: obtaining a second multidimensional time series sequence corresponding to the target parameter based on the operating data of the unit to be controlled in the second historical period, adopting a sliding window mechanism, and iteratively updating the second initial prediction model according to the second multidimensional time series sequence and the fusion error threshold to obtain a second target prediction model, and using the second target prediction model as the second prediction model. The specific implementation method is the same as the specific implementation method in the above embodiment, and will not be repeated here.

[0110] In this embodiment, the second multidimensional time series sequence corresponding to the target parameter is obtained based on the operating data of the unit to be controlled in the second historical period, and a sliding window mechanism is adopted to iteratively update the initial prediction model according to the second multidimensional time series sequence and the fusion error threshold to obtain the target prediction model. In this way, the model can adapt to the dynamic characteristics of the variable operating condition process, thereby improving the prediction accuracy of the data.

[0111] In one embodiment, the start-up and shutdown control of the controlled unit is performed according to the predicted operating conditions, including: when the predicted operating conditions are the same as the current operating conditions of the startup separator, the target control parameters are determined according to the reactor outlet temperature deviation value, the direct steam generator steam pressure deviation value and the primary circuit outlet flow deviation value; when the predicted operating conditions are different from the current operating conditions of the startup separator, the preset control parameters corresponding to the predicted operating conditions are used as the target control parameters; according to the target control parameters, the load regulating valve, the startup separator outlet isolation valve, the main feed water pump speed, the main feed water low load regulating valve, the main feed water pump return water regulating valve and the startup separator steam pipeline regulating valve are adjusted.

[0112] The server is pre-configured with corresponding control parameters for each start-stop condition of the start separator.

[0113] Optionally, after determining the predicted operating condition for starting the separator, the server determines a corresponding control strategy according to the predicted operating condition for starting the separator and the current operating condition.

[0114] When the predicted operating conditions are identical to the current operating conditions, the server obtains the reactor outlet temperature deviation, the once-through steam generator steam pressure deviation, and the primary circuit outlet flow deviation based on the feedback and setpoint values ​​of the reactor outlet temperature, the feedback and setpoint values ​​of the once-through steam generator steam pressure, and the feedback and setpoint values ​​of the primary circuit outlet flow. Furthermore, based on the reactor outlet temperature deviation, the once-through steam generator steam pressure deviation, and the primary circuit outlet flow deviation, feedback control parameters are obtained. Based on the feedback control parameters, feedback adjustment is performed on the load regulating valve, the start-up separator outlet isolation valve, the main feedwater pump speed, the main feedwater low-load regulating valve, the main feedwater pump return water regulating valve, and the start-up separator steam pipeline regulating valve.

[0115] When the predicted operating conditions are different from the current operating conditions, the server will use the preset control parameters corresponding to the predicted operating conditions as the target control parameters, and adjust the load regulating valve, the start-up separator outlet isolation valve, the main feed water pump speed, the main feed water low load regulating valve, the main feed water pump return water regulating valve and the start-up separator steam pipeline regulating valve according to the target control parameters.

[0116] In this embodiment, the target control parameters are adjusted through the feedback mechanism when the starting separator does not switch the operating conditions, and the target control parameters are adjusted through preset parameters when the starting separator switches the operating conditions. Then, according to the target control parameters, the load regulating valve, the starting separator outlet isolation valve, the main feed water pump speed, the main feed water low load regulating valve, the main feed water pump return water regulating valve and the starting separator steam pipeline regulating valve are adjusted. In this way, the unit to be controlled can be automatically controlled during the start-up and shutdown stages, thereby ensuring the safe and stable operation of the unit to be controlled.

[0117] In one embodiment, as shown in FIG4 , a method for controlling the start and stop of a metal stack is provided, the method comprising the following steps:

[0118] Based on the operating data of the unit to be controlled in the first historical period, an initial time series corresponding to each target parameter is obtained. The target parameters include primary circuit temperature and reactor power, and may also include one or more of pressurizer pressure, pressurizer water level, primary circuit pressure, reactor coolant outlet temperature, reactor coolant inlet temperature, coolant flow time, primary circuit flow, primary circuit boron concentration, steam generator pressure, steam generator water level, high-pressure cylinder exhaust pressure, first-stage reheater extraction pressure, low-pressure cylinder inlet and outlet superheated steam pressure, low-pressure cylinder inlet and outlet superheated steam temperature, condenser operating pressure, feedwater temperature, and steam reheat degree.

[0119] Data preprocessing operations are performed on the initial time series corresponding to each target parameter to obtain the sub-time series corresponding to each target parameter. The data preprocessing operations include missing value filling and outlier replacement.

[0120] A first multidimensional time series sequence is obtained according to the sub-time series sequences corresponding to each target parameter, and a modal component corresponding to the first multidimensional time series sequence is obtained.

[0121] According to the modal components corresponding to the first multidimensional time series sequence, the initial LSTM model and the initial ELM model are iteratively trained to obtain the trained LSTM model and the trained ELM model, obtain the initial weighting coefficient, and according to the initial weighting coefficient, the trained LSTM model and the trained ELM model are weightedly fused to obtain the initial prediction model.

[0122] According to the operating data of the unit to be controlled in the second historical period, the second multidimensional time series sequence corresponding to the target parameter is obtained. The sliding window mechanism is adopted to iteratively update the initial prediction model according to the second multidimensional time series sequence and the fusion error threshold to obtain the target prediction model, and the target prediction model is used as the first prediction model or the second prediction model.

[0123] The real-time time series corresponding to the target parameters is obtained, and the real-time time series is input into the first prediction model and the second prediction model respectively to obtain the primary circuit temperature prediction sequence and the reactor power prediction sequence.

[0124] The predicted operating conditions of the start-up separator are obtained based on the primary circuit temperature prediction sequence and the reactor power prediction sequence.

[0125] When the predicted operating conditions are the same as the current operating conditions of the startup separator, the target control parameters are determined based on the reactor outlet temperature deviation value, the direct steam generator steam pressure deviation value and the primary circuit outlet flow deviation value; when the predicted operating conditions are different from the current operating conditions of the startup separator, the preset control parameters corresponding to the predicted operating conditions are used as the target control parameters.

[0126] According to the target control parameters, the load regulating valve, the start-up separator outlet isolation valve, the main feed water pump speed, the main feed water low load regulating valve, the main feed water pump return water regulating valve and the start-up separator steam pipeline regulating valve are adjusted.

[0127] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0128] Based on the same inventive concept, embodiments of the present application also provide a metal stack start-stop control device for implementing the aforementioned metal stack start-stop control method. The solution provided by this device is similar to the solution described in the aforementioned method. Therefore, the specific limitations of the embodiments of the metal stack start-stop control device provided below can be found in the aforementioned limitations of the metal stack start-stop control method and will not be further elaborated here.

[0129] In an exemplary embodiment, as shown in FIG5 , a start-stop control device for a metal stack is provided, comprising: an acquisition module 510 , a construction module 520 , a prediction module 530 , and a control module 540 , wherein:

[0130] An acquisition module 510 is configured to acquire a first multidimensional time series sequence corresponding to a target parameter and a modal component corresponding to the first multidimensional time series sequence based on the operating data of the unit to be controlled in a first historical period; the target parameter includes a primary circuit temperature and a reactor power;

[0131] A construction module 520 is configured to construct a first prediction model for primary circuit temperature prediction based on the modal components corresponding to the first multidimensional time series, and to construct a second prediction model for reactor power prediction based on the modal components corresponding to the first multidimensional time series; the first prediction model and the second prediction model are obtained by combining multiple single models;

[0132] The prediction module 530 is used to obtain a real-time time series sequence corresponding to the target parameter, input the real-time time series sequence into the first prediction model and the second prediction model respectively, and obtain a primary circuit temperature prediction sequence and a reactor power prediction sequence;

[0133] The control module 540 is used to obtain the predicted operating conditions for starting the separator according to the primary circuit temperature prediction sequence and the reactor power prediction sequence, and to perform start and stop control on the unit to be controlled according to the predicted operating conditions.

[0134] In one embodiment, the target parameters also include at least one of pressurizer pressure, pressurizer water level, primary circuit pressure, reactor coolant outlet temperature, reactor coolant inlet temperature, coolant flow time, primary circuit flow, primary circuit boron concentration, steam generator pressure, steam generator water level, high-pressure cylinder exhaust pressure, first-stage reheater extraction pressure, low-pressure cylinder inlet and outlet superheated steam pressure, low-pressure cylinder inlet and outlet superheated steam temperature, condenser operating pressure, feed water temperature and steam reheat degree.

[0135] In one embodiment, the acquisition module 510 is also used to obtain a first multidimensional time series sequence corresponding to the target parameters, including: obtaining an initial time series sequence corresponding to each target parameter; performing data preprocessing operations on the initial time series sequence corresponding to each target parameter to obtain a sub-time series sequence corresponding to each target parameter; the data preprocessing operations include missing value filling and outlier replacement; and obtaining a first multidimensional time series sequence based on the sub-time series sequence corresponding to each target parameter.

[0136] In one embodiment, the construction module 520 is also used to iteratively train the initial LSTM model and the initial ELM model according to the modal components corresponding to the first multidimensional time series sequence to obtain the trained LSTM model and the trained ELM model; obtain the initial weighting coefficient; and according to the initial weighting coefficient, perform weighted fusion on the trained LSTM model and the trained ELM model to obtain the initial prediction model, and use the initial prediction model as the first prediction model.

[0137] In one embodiment, the construction module 520 is also used to obtain a second multidimensional time series sequence corresponding to the target parameter based on the operating data of the unit to be controlled in the second historical period; using a sliding window mechanism, the initial prediction model is iteratively updated according to the second multidimensional time series sequence and the fusion error threshold to obtain a target prediction model, and the target prediction model is used as the first prediction model.

[0138] In one embodiment, the control module 540 is also used to determine the target control parameters based on the reactor outlet temperature deviation value, the direct current steam generator steam pressure deviation value and the primary circuit outlet flow deviation value when the predicted operating conditions are the same as the current operating conditions of the startup separator; when the predicted operating conditions are different from the current operating conditions of the startup separator, the preset control parameters corresponding to the predicted operating conditions are used as the target control parameters; and according to the target control parameters, the load regulating valve, the startup separator outlet isolation valve, the main feed water pump speed, the main feed water low load regulating valve, the main feed water pump return water regulating valve and the startup separator steam pipeline regulating valve are adjusted.

[0139] Each module in the aforementioned metal stack start-stop control device can be implemented in whole or in part through software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a computer device memory in the form of software, so that the processor can call and execute the corresponding operations of each module.

[0140] In an exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be shown in Figure 6. The computer device includes a processor, a memory, an input / output interface (I / O) and a communication interface. The processor, the memory and the input / output interface are connected via a system bus, and the communication interface is connected to the system bus via the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store preset control parameters and other data corresponding to the start-stop operating conditions. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a start-stop control method for a metal pile is implemented.

[0141] Those skilled in the art will understand that the structure shown in FIG6 is merely a block diagram of a portion of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0142] In an exemplary embodiment, a computer device is provided, comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the following steps when executing the computer program: obtaining a first multidimensional time series sequence corresponding to a target parameter based on operating data of a unit to be controlled in a first historical period, and obtaining a modal component corresponding to the first multidimensional time series sequence; the target parameters include a primary circuit temperature and a reactor power; constructing a first prediction model for primary circuit temperature prediction based on the modal component corresponding to the first multidimensional time series sequence, and constructing a second prediction model for reactor power prediction based on the modal component corresponding to the first multidimensional time series sequence; the first prediction model and the second prediction model are obtained based on a combination of multiple single models; obtaining a real-time time series sequence corresponding to the target parameter, inputting the real-time time series sequence into the first prediction model and the second prediction model respectively, and obtaining a primary circuit temperature prediction sequence and a reactor power prediction sequence; obtaining a predicted operating condition for starting a separator based on the primary circuit temperature prediction sequence and the reactor power prediction sequence, and performing start and stop control on the unit to be controlled according to the predicted operating condition.

[0143] In one embodiment, the target parameters involved when the processor executes the computer program also include at least one of pressurizer pressure, pressurizer water level, primary circuit pressure, reactor coolant outlet temperature, reactor coolant inlet temperature, coolant flow time, primary circuit flow, primary circuit boron concentration, steam generator pressure, steam generator water level, high-pressure cylinder exhaust pressure, first-stage reheater extraction pressure, low-pressure cylinder inlet and outlet superheated steam pressure, low-pressure cylinder inlet and outlet superheated steam temperature, condenser operating pressure, feed water temperature and steam reheat degree.

[0144] In one embodiment, when the processor executes the computer program, it further implements the following steps: obtaining an initial time series sequence corresponding to each target parameter; performing data preprocessing operations on the initial time series sequence corresponding to each target parameter to obtain a sub-time series sequence corresponding to each target parameter; the data preprocessing operations include missing value filling and outlier replacement; and obtaining a first multidimensional time series sequence based on the sub-time series sequence corresponding to each target parameter.

[0145] In one embodiment, when the processor executes the computer program, the following steps are also implemented: according to the modal components corresponding to the first multidimensional time series sequence, the initial LSTM model and the initial ELM model are iteratively trained to obtain the trained LSTM model and the trained ELM model; the initial weighting coefficient is obtained; according to the initial weighting coefficient, the trained LSTM model and the trained ELM model are weightedly fused to obtain the initial prediction model, and the initial prediction model is used as the first prediction model.

[0146] In one embodiment, when the processor executes the computer program, it also implements the following steps: obtaining a second multidimensional time series sequence corresponding to the target parameter based on the operating data of the unit to be controlled in the second historical period; using a sliding window mechanism, iteratively updating the initial prediction model based on the second multidimensional time series sequence and the fusion error threshold to obtain a target prediction model, and using the target prediction model as the first prediction model.

[0147] In one embodiment, when the processor executes the computer program, the following steps are also implemented: when the predicted operating condition is the same as the current operating condition of the startup separator, the target control parameters are determined based on the reactor outlet temperature deviation value, the direct steam generator steam pressure deviation value and the primary circuit outlet flow deviation value; when the predicted operating condition is different from the current operating condition of the startup separator, the preset control parameters corresponding to the predicted operating condition are used as the target control parameters; according to the target control parameters, the load regulating valve, the startup separator outlet isolation valve, the main feed water pump speed, the main feed water low load regulating valve, the main feed water pump return water regulating valve and the startup separator steam pipeline regulating valve are adjusted.

[0148] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented: based on the operating data of the unit to be controlled in the first historical period, a first multidimensional time series sequence corresponding to the target parameter is obtained, and the modal components corresponding to the first multidimensional time series sequence are obtained; the target parameters include the primary circuit temperature and the reactor power; based on the modal components corresponding to the first multidimensional time series sequence, a first prediction model for primary circuit temperature prediction is constructed, and based on the modal components corresponding to the first multidimensional time series sequence, a second prediction model for reactor power prediction is constructed; the first prediction model and the second prediction model are obtained based on a combination of multiple single models; a real-time time series sequence corresponding to the target parameter is obtained, and the real-time time series sequence is input into the first prediction model and the second prediction model respectively to obtain the primary circuit temperature prediction sequence and the reactor power prediction sequence; based on the primary circuit temperature prediction sequence and the reactor power prediction sequence, the predicted operating conditions of the start-up separator are obtained, and the unit to be controlled is started and stopped according to the predicted operating conditions.

[0149] In one embodiment, the target parameters involved when the computer program is executed by the processor also include at least one of pressurizer pressure, pressurizer water level, primary circuit pressure, reactor coolant outlet temperature, reactor coolant inlet temperature, coolant flow time, primary circuit flow, primary circuit boron concentration, steam generator pressure, steam generator water level, high-pressure cylinder exhaust pressure, first-stage reheater extraction pressure, low-pressure cylinder inlet and outlet superheated steam pressure, low-pressure cylinder inlet and outlet superheated steam temperature, condenser operating pressure, feed water temperature and steam reheat degree.

[0150] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: obtaining an initial time series sequence corresponding to each target parameter; performing data preprocessing operations on the initial time series sequence corresponding to each target parameter to obtain a sub-time series sequence corresponding to each target parameter; the data preprocessing operations include missing value filling and outlier replacement; and obtaining a first multidimensional time series sequence based on the sub-time series sequence corresponding to each target parameter.

[0151] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: iteratively training the initial LSTM model and the initial ELM model according to the modal components corresponding to the first multidimensional time series sequence to obtain the trained LSTM model and the trained ELM model; obtaining the initial weighting coefficient; and weightedly fusing the trained LSTM model and the trained ELM model according to the initial weighting coefficient to obtain the initial prediction model, and using the initial prediction model as the first prediction model.

[0152] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: based on the operating data of the unit to be controlled in the second historical period, a second multidimensional time series sequence corresponding to the target parameter is obtained; using a sliding window mechanism, the initial prediction model is iteratively updated according to the second multidimensional time series sequence and the fusion error threshold to obtain a target prediction model, and the target prediction model is used as the first prediction model.

[0153] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: when the predicted operating condition is the same as the current operating condition of the startup separator, the target control parameters are determined based on the reactor outlet temperature deviation value, the direct steam generator steam pressure deviation value and the primary circuit outlet flow deviation value; when the predicted operating condition is different from the current operating condition of the startup separator, the preset control parameters corresponding to the predicted operating condition are used as the target control parameters; according to the target control parameters, the load regulating valve, the startup separator outlet isolation valve, the main feed water pump speed, the main feed water low load regulating valve, the main feed water pump return water regulating valve and the startup separator steam pipeline regulating valve are adjusted.

[0154] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps: obtaining a first multidimensional time series sequence corresponding to a target parameter based on operating data of a unit to be controlled within a first historical period, and obtaining a modal component corresponding to the first multidimensional time series sequence; the target parameters include a primary circuit temperature and a reactor power; constructing a first prediction model for primary circuit temperature prediction based on the modal component corresponding to the first multidimensional time series sequence, and constructing a second prediction model for reactor power prediction based on the modal component corresponding to the first multidimensional time series sequence; the first prediction model and the second prediction model are obtained based on a combination of multiple single models; obtaining a real-time time series sequence corresponding to the target parameter, inputting the real-time time series sequence into the first prediction model and the second prediction model respectively, and obtaining a primary circuit temperature prediction sequence and a reactor power prediction sequence; obtaining a predicted operating condition for starting a separator based on the primary circuit temperature prediction sequence and the reactor power prediction sequence, and performing start and stop control on the unit to be controlled according to the predicted operating condition.

[0155] In one embodiment, the target parameters involved when the computer program is executed by the processor also include at least one of pressurizer pressure, pressurizer water level, primary circuit pressure, reactor coolant outlet temperature, reactor coolant inlet temperature, coolant flow time, primary circuit flow, primary circuit boron concentration, steam generator pressure, steam generator water level, high-pressure cylinder exhaust pressure, first-stage reheater extraction pressure, low-pressure cylinder inlet and outlet superheated steam pressure, low-pressure cylinder inlet and outlet superheated steam temperature, condenser operating pressure, feed water temperature and steam reheat degree.

[0156] In one embodiment, when the computer program is executed by the processor, the following steps are further implemented: obtaining an initial time series sequence corresponding to each target parameter; performing data preprocessing operations on the initial time series sequence corresponding to each target parameter to obtain a sub-time series sequence corresponding to each target parameter; the data preprocessing operations include missing value filling and outlier replacement; and obtaining a first multidimensional time series sequence based on the sub-time series sequence corresponding to each target parameter.

[0157] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: iteratively training the initial LSTM model and the initial ELM model according to the modal components corresponding to the first multidimensional time series sequence to obtain the trained LSTM model and the trained ELM model; obtaining the initial weighting coefficient; and weightedly fusing the trained LSTM model and the trained ELM model according to the initial weighting coefficient to obtain the initial prediction model, and using the initial prediction model as the first prediction model.

[0158] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: based on the operating data of the unit to be controlled in the second historical period, a second multidimensional time series sequence corresponding to the target parameter is obtained; using a sliding window mechanism, the initial prediction model is iteratively updated according to the second multidimensional time series sequence and the fusion error threshold to obtain a target prediction model, and the target prediction model is used as the first prediction model.

[0159] In one embodiment, when the computer program is executed by the processor, the following steps are also implemented: when the predicted operating condition is the same as the current operating condition of the startup separator, the target control parameters are determined based on the reactor outlet temperature deviation value, the direct steam generator steam pressure deviation value and the primary circuit outlet flow deviation value; when the predicted operating condition is different from the current operating condition of the startup separator, the preset control parameters corresponding to the predicted operating condition are used as the target control parameters; according to the target control parameters, the load regulating valve, the startup separator outlet isolation valve, the main feed water pump speed, the main feed water low load regulating valve, the main feed water pump return water regulating valve and the startup separator steam pipeline regulating valve are adjusted.

[0160] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processor involved in the various embodiments provided herein may be, but are not limited to, a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, and the like.

[0161] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0162] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for starting and stopping a metal pile, wherein: The method comprises: According to the operation data of the unit to be controlled in the first historical period, a first multidimensional time series sequence corresponding to the target parameter is obtained, and a modal component corresponding to the first multidimensional time series sequence is obtained; the target parameter includes a primary circuit temperature and a reactor power; According to the modal components corresponding to the first multidimensional time series, a first prediction model for primary circuit temperature prediction is constructed, and according to the modal components corresponding to the first multidimensional time series, a second prediction model for reactor power prediction is constructed; the first prediction model and the second prediction model are obtained based on a combination of multiple single models; Acquire a real-time time series sequence corresponding to the target parameter, input the real-time time series sequence into the first prediction model and the second prediction model respectively, and acquire a primary circuit temperature prediction sequence and a reactor power prediction sequence; According to the primary circuit temperature prediction sequence and the reactor power prediction sequence, the predicted operating condition of the start-up separator is obtained, and the start and stop control of the unit to be controlled is performed according to the predicted operating condition.

2. The method according to claim 1, wherein: The target parameters also include at least one of pressurizer pressure, pressurizer water level, primary circuit pressure, reactor coolant outlet temperature, reactor coolant inlet temperature, coolant flow time, primary circuit flow, primary circuit boron concentration, steam generator pressure, steam generator water level, high-pressure cylinder exhaust pressure, first-stage reheater extraction pressure, low-pressure cylinder inlet and outlet superheated steam pressure, low-pressure cylinder inlet and outlet superheated steam temperature, condenser operating pressure, feed water temperature and steam reheat.

3. The method according to claim 1, wherein: The obtaining of the first multi-dimensional time series corresponding to the target parameter includes: Obtaining the initial time series corresponding to each target parameter; Performing data preprocessing operations on the initial time series corresponding to each target parameter to obtain a sub-time series corresponding to each target parameter; the data preprocessing operations include missing value filling and outlier value replacement; The first multidimensional time series sequence is acquired according to the sub-time series sequences corresponding to each target parameter.

4. The method according to claim 1, wherein: The single model includes a long short-term memory (LSTM) model and an extreme learning machine (ELM) model; the first prediction model for primary circuit temperature prediction is constructed according to the modal components corresponding to the first multidimensional time series, including: Iteratively training the initial LSTM model and the initial ELM model according to the modal components corresponding to the first multidimensional time series sequence to obtain a trained LSTM model and a trained ELM model; Obtaining initial weighting coefficients; According to the initial weighting coefficient, the trained LSTM model and the trained ELM model are weightedly fused to obtain an initial prediction model, and the initial prediction model is used as the first prediction model.

5. The method according to claim 4, wherein: The constructing a first prediction model for primary circuit temperature prediction according to the modal components corresponding to the first multidimensional time series sequence further includes: Acquire a second multi-dimensional time series sequence corresponding to the target parameter according to the operation data of the unit to be controlled in the second historical period; The sliding window mechanism is adopted to iteratively update the initial prediction model according to the second multidimensional time series and the fusion error threshold to obtain a target prediction model, and the target prediction model is used as the first prediction model.

6. The method according to claim 1, wherein: The starting and stopping control of the unit to be controlled according to the predicted operating condition includes: When the predicted operating condition is the same as the current operating condition of the start-up separator, determining the target control parameter according to the reactor outlet temperature deviation value, the once-through steam generator steam pressure deviation value and the primary circuit outlet flow deviation value; In the case where the predicted operating condition is different from the current operating condition of the start-up separator, using the preset control parameter corresponding to the predicted operating condition as the target control parameter; According to the target control parameters, the load regulating valve, the start separator outlet isolation valve, the main feed water pump speed, the main The feed water low load regulating valve, the main feed water pump return regulating valve and the start-up separator steam pipeline regulating valve are regulated.

7. A start-stop control device for a metal pile, wherein: The device comprises: an acquisition module, configured to acquire a first multidimensional time series sequence corresponding to a target parameter and a modal component corresponding to the first multidimensional time series sequence according to the operation data of the unit to be controlled in a first historical period; the target parameter includes a primary circuit temperature and a reactor power; A construction module, configured to construct a first prediction model for primary circuit temperature prediction according to the modal components corresponding to the first multidimensional time series, and to construct a second prediction model for reactor power prediction according to the modal components corresponding to the first multidimensional time series; the first prediction model and the second prediction model are obtained based on a combination of multiple single models; A prediction module, used for obtaining a real-time time series sequence corresponding to the target parameter, inputting the real-time time series sequence into the first prediction model and the second prediction model respectively, and obtaining a primary circuit temperature prediction sequence and a reactor power prediction sequence; The control module is used to obtain the predicted working condition of the start-up separator according to the primary circuit temperature prediction sequence and the reactor power prediction sequence, and to start and stop the unit to be controlled according to the predicted working condition.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, wherein: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product comprising a computer program, wherein: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images