Thorium-based molten salt reactor primary loop molten salt flow prediction and control method and system

Abstract

The invention discloses a thorium-based molten salt reactor primary loop molten salt flow prediction and control method and system, and the method comprises the steps: obtaining an original molten salt flow signal of each node in a thorium-based molten salt reactor primary loop, and obtaining a multi-scale molten salt flow component; inputting the multi-scale fused salt flow component into an LSTM network adaptive gating mechanism network model to obtain a fused salt flow initial prediction value of each node; obtaining an error coefficient of the fused salt flow based on D-SMAPE according to the initial predicted value of the fused salt flow of each node; the error coefficient of the fused salt flow is input into the trained ResNet prediction model, and an optimal fused salt flow prediction value is obtained; according to the optimal fused salt flow predicted value, the fused salt flow of the thorium-based fused salt reactor primary loop is controlled based on the DEM-DAC algorithm, and the method and system can solve the technical problems that during reactor variable-load operation, the fused salt flow fluctuation is large, and the unit load adjusting performance is poor.

Description

Technical Field

[0001] The present invention belongs to the technical field of nuclear power engineering, and relates to a method and system for predicting and controlling the flow rate of molten salt in a primary circuit of a thorium-based molten salt reactor. Background Art

[0002] Thorium-based molten salt reactors use liquid fluoride salts as a coolant and nuclear fuel carrier. This molten salt has excellent thermophysical properties and chemical stability, enabling operation at high temperatures and efficient heat transfer. The technical advantages of thorium-based molten salt reactors include: 1) Abundant fuel resources: Thorium is more abundant and widely distributed on Earth than the uranium used in traditional nuclear reactors. 2) High inherent safety: With a high negative temperature coefficient, reactivity automatically decreases as the core temperature rises, thus preventing serious accidents such as core overheating and meltdown. Furthermore, molten salt reactors operate at atmospheric pressure, eliminating the risk of radioactive material leakage from a rupture of the high-pressure system. 3) Low nuclear waste generation: The nuclear waste produced by thorium-based molten salt reactors contains relatively low levels of long-lived radioactive nuclides, resulting in low radioactive toxicity and half-lives, significantly reducing the difficulty and risk of nuclear waste treatment and disposal. Thorium-based molten salt reactors have become a priority for future advanced nuclear fission energy development.

[0003] Existing thorium-based molten salt reactors face technical challenges such as poor operational flexibility and transient response to large and rapid load changes. This is primarily due to the poor prediction accuracy and unstable control of molten salt flow prediction and control technologies. Traditional signal decomposition methods, such as wavelet transform (WT) and empirical mode decomposition (EMD), have limitations when processing non-stationary signals and are unable to adaptively capture local signal features. Existing methods typically require preset parameters or basis functions, making them difficult to adapt to dynamic signal changes. Existing deep learning methods, such as LSTM, perform well when processing time series data, but their performance is limited when processing data with complex spatial dependencies. While GNNs can capture spatial dependencies, they lack the ability to model temporal dependencies. Traditional error assessment methods, such as mean squared error (MSE) and mean absolute error (MAE), have limitations when processing data with varying dimensions or zero values. While symmetric mean absolute percentage error (SMAPE) can address these issues to some extent, it remains highly sensitive to extreme values and zero values. Furthermore, in terms of actuator control, traditional DEM and DAC technologies suffer from nonlinear distortion, component mismatch, and high power consumption during signal conversion. While existing DEM algorithms can mitigate component mismatch, their implementation is complex and may introduce additional noise. Based on this analysis, the development of more accurate molten salt flow prediction and control methods is crucial for the safe and stable operation of thorium-based molten salt reactor nuclear power plants. Summary of the Invention

[0004] The purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art and provide a method and system for predicting and controlling the molten salt flow rate in the primary circuit of a thorium-based molten salt reactor. This method and system can solve the technical problems of large fluctuations in the molten salt flow rate and poor unit load regulation performance during the reactor's variable load operation.

[0005] To achieve the above-mentioned object, the present invention discloses a method for predicting and controlling the flow rate of molten salt in the primary circuit of a thorium-based molten salt reactor, comprising:

[0006] Obtaining the original molten salt flow signal of each node in the primary circuit of the thorium-based molten salt reactor, performing ASDA decomposition on the original molten salt flow signal to obtain decomposed molten salt flow sub-signals, and performing real-time signal processing and joint feature extraction on the decomposed molten salt flow sub-signals to obtain multi-scale molten salt flow components;

[0007] Inputting the multi-scale molten salt flow components into the LSTM network adaptive gating mechanism network model to obtain the initial predicted value of the molten salt flow at each node;

[0008] According to the initial predicted value of the molten salt flow rate of each node, an error coefficient of the molten salt flow rate is obtained based on D-SMAPE;

[0009] The error coefficient of the molten salt flow rate is input into the trained ResNet prediction model to obtain the optimized molten salt flow rate prediction value;

[0010] According to the optimized molten salt flow prediction value, the molten salt flow in the primary circuit of the thorium-based molten salt reactor is controlled based on the DEM-DAC algorithm.

[0011] The further improvement of the method for predicting and controlling the primary circuit molten salt flow of a thorium-based molten salt reactor described in the present invention is:

[0012] Furthermore, the process of controlling the molten salt flow rate of the primary circuit of the thorium-based molten salt reactor based on the DEM-DAC algorithm according to the optimized molten salt flow rate prediction value is as follows:

[0013] The optimized molten salt flow prediction value is converted into an electrical signal based on the DEM-DAC algorithm, and the electrical signal is then input into the frequency converter of the molten salt pump to control the variable frequency operation of the molten salt pump, thereby controlling the molten salt flow in the primary circuit of the thorium-based molten salt reactor.

[0014] Furthermore, after obtaining the original molten salt flow signal of each node in the primary circuit of the thorium-based molten salt reactor, the method further includes:

[0015] The original molten salt flow signal is normalized to eliminate dimension and amplitude differences.

[0016] Furthermore, the training process of the LSTM network adaptive gating mechanism network model is as follows:

[0017]

[0018] Among them, η t is the learning rate of the tth iteration; η0 is the initial learning rate.

[0019] Furthermore, the loss function of the ResNet prediction model during training is:

[0020]

[0021] Where N is the number of samples, Correct the predicted value for the molten salt flow rate.

[0022] The present invention discloses a thorium-based molten salt reactor primary circuit molten salt flow prediction and control system, comprising:

[0023] An acquisition module is used to obtain the original molten salt flow signal of each node in the primary circuit of the thorium-based molten salt reactor, perform ASDA decomposition on the original molten salt flow signal to obtain decomposed molten salt flow sub-signals, and perform real-time signal processing and joint feature extraction on the decomposed molten salt flow sub-signals to obtain multi-scale molten salt flow components;

[0024] A first prediction module is used to input the multi-scale molten salt flow components into the LSTM network adaptive gating mechanism network model to obtain an initial prediction value of the molten salt flow of each node;

[0025] A calculation module, configured to obtain an error coefficient of the molten salt flow rate based on the D-SMAPE according to the initial predicted value of the molten salt flow rate of each node;

[0026] The second prediction module is used to input the error coefficient of the molten salt flow into the trained ResNet prediction model to obtain the optimized molten salt flow prediction value;

[0027] The control module is used to control the molten salt flow rate of the primary circuit of the thorium-based molten salt reactor based on the DEM-DAC algorithm according to the optimized molten salt flow rate prediction value.

[0028] The further improvement of the primary circuit molten salt flow prediction and control system of the thorium-based molten salt reactor described in the present invention is:

[0029] Furthermore, the process of controlling the molten salt flow rate of the primary circuit of the thorium-based molten salt reactor based on the DEM-DAC algorithm according to the optimized molten salt flow rate prediction value is as follows:

[0030] The optimized molten salt flow prediction value is converted into an electrical signal based on the DEM-DAC algorithm, and the electrical signal is then input into the frequency converter of the molten salt pump to control the variable frequency operation of the molten salt pump, thereby controlling the molten salt flow in the primary circuit of the thorium-based molten salt reactor.

[0031] Furthermore, after obtaining the original molten salt flow signal of each node in the primary circuit of the thorium-based molten salt reactor, the method further includes:

[0032] The original molten salt flow signal is normalized to eliminate dimension and amplitude differences.

[0033] The present invention discloses a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the method for predicting and controlling the primary-loop molten salt flow of a thorium-based molten salt reactor are implemented.

[0034] The present invention discloses a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for predicting and controlling the primary-loop molten salt flow of a thorium-based molten salt reactor are implemented.

[0035] The present invention has the following beneficial effects:

[0036] The method and system for predicting and controlling the molten salt flow rate of a primary circuit of a thorium-based molten salt reactor described in the present invention adopts an adaptive signal decomposition algorithm (ASDA) to decompose the original molten salt flow rate during specific operation, and significantly improves the processing capability of non-stationary signals by dynamically adjusting the decomposition parameters and optimizing the decomposition process. Based on the adaptive graph attention long short-term memory network (AG-LSTM), by combining the graph attention mechanism and the LSTM network, the modeling capability of complex spatiotemporal dependency data is significantly improved. The molten salt flow error coefficient is obtained based on the dynamic symmetric mean absolute percentage error (D-SMAPE), and the accuracy and robustness of the thorium-based molten salt reactor molten salt flow prediction model are significantly improved by dynamically adjusting the error evaluation weight. The residual network (ResNet) of deep learning is also introduced to obtain the optimized prediction value of the molten salt flow rate, which solves the problems of gradient vanishing and gradient explosion that are prone to occur in the training process of existing deep learning methods, and greatly improves the final prediction accuracy of the molten salt flow rate of the thorium-based molten salt reactor. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] The accompanying drawings, which constitute part of the present invention, are intended to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are intended to explain the present invention and do not constitute an undue limitation of the present invention. In the accompanying drawings:

[0038] Figure 1 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0039] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0040] In the description of the present invention, it is to be understood that the terms “include” and “comprise” indicate the presence of the described features, wholes, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0041] It should also be understood that the terms used in the present specification are only for the purpose of describing particular embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0042] It should be further understood that the term "and / or" as used in the present specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in the present invention generally indicates that the associated objects are in an "or" relationship.

[0043] It should be understood that although the terms "first," "second," and "third" may be used to describe preset ranges in embodiments of the present invention, these preset ranges should not be limited to these terms. These terms are merely used to distinguish one preset range from another. For example, without departing from the scope of embodiments of the present invention, the first preset range may also be referred to as the second preset range, and similarly, the second preset range may also be referred to as the first preset range.

[0044] The word "if," as used herein, may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to the determination" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)," depending on the context.

[0045] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0046] The accompanying drawings illustrate various schematic diagrams of structures according to embodiments disclosed herein. These figures are not drawn to scale; for clarity, some details are exaggerated and some details may be omitted. The shapes of the various regions and layers shown in the figures, as well as their relative sizes and positional relationships, are merely exemplary and may deviate in practice due to manufacturing tolerances or technical limitations. Those skilled in the art may design regions / layers with different shapes, sizes, and relative positions as needed.

[0047] Example 1

[0048] refer to Figure 1 The method for predicting and controlling the primary circuit molten salt flow rate of a thorium-based molten salt reactor according to the present invention comprises the following steps:

[0049] 1) Obtaining the original molten salt flow signal of each node in the primary circuit of a thorium-based molten salt reactor, normalizing the original molten salt flow signal to eliminate dimension and amplitude differences; then performing ASDA decomposition on the original molten salt flow signal to obtain decomposed molten salt flow sub-signals; performing real-time signal processing and joint feature extraction on the decomposed molten salt flow sub-signals to obtain multi-scale molten salt flow components;

[0050] 2) Inputting each multi-scale molten salt flow component into the AG-LSTM to construct an LSTM network adaptive gating mechanism network model, perform multi-scale feature extraction and adaptive learning, and obtain the initial prediction value of the molten salt flow at each node;

[0051] 3) according to the initial predicted value of the molten salt flow rate of each node, obtaining the error coefficient of the molten salt flow rate based on D-SMAPE;

[0052] 4) Input the error coefficient of the molten salt flow rate into the trained ResNet prediction model to obtain the optimized molten salt flow rate prediction value;

[0053] 5) Based on the DEM-DAC algorithm, the optimized molten salt flow prediction value is converted into an electrical signal, which is then input into the frequency converter of the molten salt pump to control the variable frequency operation of the molten salt pump, thereby controlling the molten salt flow in the primary circuit of the thorium-based molten salt reactor.

[0054] The process of performing ASDA decomposition on the original molten salt flow signal in step 1) is as follows:

[0055] 11) Signal preprocessing

[0056] The input original molten salt flow signal is standardized to eliminate the dimension effect.

[0057] Detect the local characteristics of the molten salt flow signal (such as frequency, amplitude changes, etc.).

[0058] 12) Adaptive decomposition parameter adjustment;

[0059] Dynamically adjust the decomposition parameters according to the local characteristics of the molten salt flow signal:

[0060]

[0061] Where θ(t) is the decomposition parameter at time t, x(t) is the input molten salt flow signal, c j (t; θ) is the j-th molten salt flow sub-signal, R(θ) is the regularization term used to control the complexity of the decomposition, and λ is the regularization parameter.

[0062] 13) Multi-scale decomposition framework;

[0063] Design a multi-scale decomposition framework to capture the different time scale characteristics of the molten salt flow signal:

[0064]

[0065] Among them, c j，k (t;θ k ) is the jth molten salt flow quantum signal at the kth scale, θ k is the decomposition parameter of the kth scale.

[0066] 14) Optimize the objective function;

[0067] The optimization objective function is introduced to ensure that the decomposed molten salt flow quantum signal has a clear physical meaning:

[0068]

[0069] Among them, λ1 and λ2 are weight parameters; the second and third terms are used for the smoothness and change rate of the molten salt flow quantum signal, respectively.

[0070] 15) Real-time signal processing;

[0071]

[0072] Where η is the learning rate, is the gradient of the objective function with respect to the molten salt flow decomposition parameter.

[0073] 16) Joint feature extraction;

[0074] The decomposed molten salt flow quantum signal is combined with the multi-scale feature extraction module to improve the expression ability of molten salt flow signal analysis:

[0075] f k =Conv1D (c j,k (t;θ k ), k=1, 2, ..., K)

[0076] Among them, f k is the feature of the kth scale, and Conv1D is a one-dimensional convolution operation.

[0077] In step 2), each multi-scale molten salt flow component is input into the AG-LSTM to construct the LSTM network adaptive gating mechanism network model, and multi-scale feature extraction and adaptive learning are performed to obtain the initial prediction value of the molten salt flow of each node. The process is as follows:

[0078] 21) Adaptive graph attention mechanism;

[0079] Introducing an adaptive graph attention mechanism to dynamically learn the relationship weights between nodes:

[0080] e ij =LeakyReLU(α T [Wh i ||Wh j ])

[0081] Among them, e ij is the attention score between molten salt flow node i and node j, α is the learnable attention vector, W is the learnable weight matrix, h i and h j is the characteristic representation of molten salt flow node i and node j, α ij is the normalized attention weight.

[0082] 22) Space-time fusion module;

[0083] 221) Construct an LSTM network with adaptive gating mechanism;

[0084] Forget Gate:

[0085]

[0086] Among them, f t is the output of the forget gate at time t, σ is the Sigmoid function, whose output value is between 0 and 1, W f is the weight matrix of the forget gate, Indicates that the hidden state h of the previous moment t-1 and the current input x t Stitched together, b f is the bias term of the forget gate, g(t) is a time-dependent adaptive function used to dynamically adjust the weight of the forget gate, α f is the adaptive coefficient.

[0087] Input Gate:

[0088] The activation value of the input gate is:

[0089]

[0090] Among them, i t is the output of the input gate at time t, W i is the weight matrix of the input gate, b i is the corresponding bias term, α i is the adaptive coefficient.

[0091] Memory unit status:

[0092] Calculate the state value of the candidate memory unit for:

[0093]

[0094] Calculate the candidate memory unit state value C t :

[0095]

[0096] Among them, the tanh function maps the output value to between -1 and 1, W C is the weight matrix of the candidate memory cell state, b C is the corresponding bias term.

[0097] Output Gate:

[0098] The activation value of the output gate is calculated as:

[0099]

[0100] Among them, t is the output of the output gate at time t, W o is the weight matrix of the output gate, b o is the bias term, αo is the adaptive coefficient.

[0101] Update hidden state:

[0102] h t =o t tanh(C t )

[0103] Among them, the tanh function maps the output value to between -1 and 1.

[0104] 222) Design a spatiotemporal fusion module that combines the graph attention mechanism with the LSTM network;

[0105]

[0106] in, is the hidden state of molten salt flow node i at time step t; N(i) is the set of neighbor nodes of molten salt flow node i.

[0107] 23) Multi-scale feature extraction;

[0108] Through the multi-scale feature extraction module, the spatiotemporal features of different time scales are captured:

[0109] H multi-scale =Concat(AvgPool(H,k1),AvgPool(H,k2),…)

[0110] Among them, H is the molten salt flow characteristic matrix output by the spatiotemporal fusion module; k1, k2, ... are different time scales.

[0111] 24) Adaptive learning rate adjustment;

[0112] Introducing an adaptive learning rate adjustment strategy to optimize the model training process:

[0113]

[0114] Among them, η t is the learning rate of the tth iteration; η0 is the initial learning rate.

[0115] 25) Real-time prediction and update;

[0116] Supports real-time data input and model updates, suitable for dynamically changing scenarios:

[0117] y=Linear(H multi-scale )

[0118] Among them, y is the predicted value of molten salt flow, and Linear is the linear transformation layer.

[0119] The process of step 3) is:

[0120] 31) Calculate the predicted value y of the molten salt flow rate at the i-th node i The dynamic error e i :

[0121]

[0122] 32) Error aggregation;

[0123] Average all errors over the molten salt flow series:

[0124]

[0125] Segment aggregation:

[0126]

[0127] 33) Output error coefficient;

[0128] The average value obtained in step 32) is multiplied by 100% to obtain the final SMAPE value, which is expressed as a percentage of the error coefficients ΔY1, ΔY2, ..., ΔY of the molten salt flow rate. n .

[0129] This algorithm, based on symmetry processing, solves the problem of MAPE being unable to be calculated when the actual value or predicted value is zero. The core idea is to use symmetry processing and introduce a dynamic adjustment factor to dynamically adjust the sensitivity of the error calculation based on the difference between the actual value and the predicted value, making the error calculation more reasonable.

[0130] In step 4), the error coefficient of the molten salt flow rate is input into the trained ResNet prediction model to obtain the optimized molten salt flow rate prediction value of each node as follows:

[0131] 41) Construct residual block

[0132] Introduce dynamic residual connection, according to the input molten salt flow error coefficient Δy i The dynamic characteristics of the residual connection weight coefficient α(Δy i )for:

[0133] y i =F(Δy i ,{w i})+α(Δy i )·Δy i

[0134] Among them, y i is the predicted value of molten salt flow, Δy i is the input feature, is the error coefficient of molten salt flow rate; F(Δy i ,{wi}) is the residual function; σ is the Sigmoid activation function, and α(Δy i ) is limited to the range [0, 1].

[0135] α(Δy i ) is calculated by a small neural network:

[0136] α(Δy i )=σ(W α ·Δy i +b α )

[0137] 42) Construct a residual network;

[0138] The molten salt flow residual network is constructed by stacking multiple residual blocks.

[0139] Assume that the network has l molten salt flow residual blocks, and the input of the lth molten salt flow residual block is X l-1 , the output is X l , then:

[0140] x1=ResBlock(x 1.1 )

[0141] Among them, ResBlock represents the operation of the residual block.

[0142] Let the output of the last residual block be X l , through a fully connected layer to obtain the predicted value y of molten salt flow i for:

[0143] y i =w out x1+b out

[0144] Among them, W out is the weight matrix of the output layer, b out is the bias vector.

[0145] 43) Model training;

[0146] 431) Define the loss function;

[0147] The loss function is calculated as:

[0148]

[0149] Where N is the number of samples, Correct the predicted value for the molten salt flow rate.

[0150] 432) Select optimizer;

[0151] Calculate the gradient: Let the parameter be θ, and the gradient of the loss function with respect to the parameter is:

[0152]

[0153] Compute first- and second-order moment estimates:

[0154]

[0155] Where β1 and β2 are decay rates, usually set to 0.9 and 0.999 respectively.

[0156] Update parameters:

[0157]

[0158] Among them, α is the learning rate and ∈ is the factor constant.

[0159] 433) Circuit training;

[0160] Forward propagation: Input xi into the model and calculate the predicted value of molten salt flow through each layer

[0161] Calculate loss: Calculate the loss L of the current batch according to the loss function.

[0162] Backpropagation: Calculate the gradient of loss with respect to model parameters

[0163] Parameter update: Use the optimizer to update the model parameters θ.

[0164] 434) output prediction results;

[0165] The final feature is mapped to the output space through the fully connected layer to obtain the final predicted value of the molten salt flow for:

[0166]

[0167] in, and are the weight matrix and bias term of the fully connected layer respectively.

[0168] In step 5), the process of converting the predicted value of the optimized molten salt flow of each node into an electrical signal based on the DEM-DAC algorithm is as follows:

[0169] 51) Adaptive weight allocation

[0170] Dynamically assign weights to DAC components through an optimization formula to reduce the impact of component mismatch:

[0171]

[0172] Among them, ω j is the weight of the jth element, is the digital input signal of the i-th optimized molten salt flow rate, s ij is the response of the jth element to the i-th signal.

[0173] 52) Establish a mathematical optimization model;

[0174] A mathematical model based on least squares method and stochastic optimization is established to dynamically adjust the parameters of DEM and DAC:

[0175]

[0176] Wherein, Zi is the output electrical signal of the i-th optimized molten salt flow rate; ω is the weight vector of the DAC element; is the actual output molten salt flow electrical signal; λ is the regularization parameter used to control the complexity of the model.

[0177] 53) Noise prediction and compensation;

[0178] The mathematical model is used to predict the quantization noise, and the compensation formula is used to reduce the impact of the noise on the molten salt flow electrical signal:

[0179]

[0180] in, is the predicted quantization noise, and α is the compensation coefficient.

[0181] 54) Power consumption optimization;

[0182] Reducing power consumption of DEM and DAC through mathematical optimization:

[0183]

[0184] Among them, p j (ω J ) is the power consumption of the jth component, and M is the total number of components.

[0185] 55) Real-time correction;

[0186] Real-time correction algorithm based on mathematical model ensures the accuracy of output molten salt flow electrical signal:

[0187]

[0188] Where η is the learning rate, is the gradient of the loss function.

[0189] Example verification

[0190] To verify the reliability of the present invention, experimental data (i.e., the original lead-bismuth alloy flow rate) was obtained from a thorium-based molten salt test reactor with a load adjustment range of 60% to 100%, and the test period was one month. Data from the first 15 days were selected, and a total of 4,252 data samples at 5-minute intervals were provided to train the prediction model. After the model's self-learning function was completed, data from the last 15 days was selected to test the performance of the proposed model. To compare the advanced performance of the present invention, the dynamic symmetric mean absolute percentage error (D-SMAPE) was used for evaluation, and traditional WT, EMD, LSTM, and GNN were selected for comparison with the present invention.

[0191] The error evaluation index SMAPE is used as the evaluation standard for each model, and the experimental results are shown in Table 1.

[0192] Table 1

[0193] Evaluation indicators WT EMD LSTM GNN Model of this application D-SMAPE (%) 47.11 32.10 28.09 16.70 10.01

[0194] As shown in Table 1, the present invention has the smallest error coefficient when compared with all other benchmark models. Compared with the WT and EMD models, the error coefficients of the LSTM, GNN, and present invention are significantly reduced, demonstrating that the present invention can significantly improve prediction accuracy compared with the WT and EMD models. Furthermore, compared with the LSTM and GNN models, the error coefficients of the present invention are smaller than those of the other two models, demonstrating that the present invention can improve the prediction accuracy of molten salt flow in thorium-based molten salt reactors.

[0195] Example 2

[0196] The thorium-based molten salt reactor primary circuit molten salt flow prediction and control system of the present invention comprises:

[0197] An acquisition module is used to obtain the original molten salt flow signal of each node in the primary circuit of the thorium-based molten salt reactor, perform ASDA decomposition on the original molten salt flow signal to obtain decomposed molten salt flow sub-signals, and perform real-time signal processing and joint feature extraction on the decomposed molten salt flow sub-signals to obtain multi-scale molten salt flow components;

[0198] A first prediction module is used to input the multi-scale molten salt flow components into the LSTM network adaptive gating mechanism network model to obtain an initial prediction value of the molten salt flow of each node;

[0199] A calculation module, configured to obtain an error coefficient of the molten salt flow rate based on the D-SMAPE according to the initial predicted value of the molten salt flow rate of each node;

[0200] The second prediction module is used to input the error coefficient of the molten salt flow into the trained ResNet prediction model to obtain the optimized molten salt flow prediction value;

[0201] The control module is used to control the molten salt flow rate of the primary circuit of the thorium-based molten salt reactor based on the DEM-DAC algorithm according to the optimized molten salt flow rate prediction value.

[0202] In this embodiment, the process of controlling the molten salt flow rate of the primary circuit of the thorium-based molten salt reactor based on the DEM-DAC algorithm according to the optimized molten salt flow rate prediction value is as follows:

[0203] The optimized molten salt flow prediction value is converted into an electrical signal based on the DEM-DAC algorithm, and the electrical signal is then input into the frequency converter of the molten salt pump to control the variable frequency operation of the molten salt pump, thereby controlling the molten salt flow in the primary circuit of the thorium-based molten salt reactor.

[0204] In this embodiment, after obtaining the original molten salt flow rate signal of each node in the primary circuit of the thorium-based molten salt reactor, the method further includes:

[0205] The original molten salt flow rate signal is normalized to eliminate dimension and amplitude differences.

[0206] The division of modules in the embodiments of the present application is illustrative and is merely a logical functional division. In actual implementation, other division methods may be used. Furthermore, the functional modules in the various embodiments of the present application may be integrated into a single processor, or may exist physically separately, or two or more modules may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or software functional modules.

[0207] Example 3

[0208] A computer device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for predicting and controlling the molten salt flow rate in the primary circuit of a thorium-based molten salt reactor is implemented. For example, the method comprises: obtaining an original molten salt flow rate signal of each node in the primary circuit of the thorium-based molten salt reactor, performing ASDA decomposition on the original molten salt flow rate signal to obtain a decomposed molten salt flow sub-signal, performing real-time signal processing and joint feature extraction on the decomposed molten salt flow sub-signal to obtain a multi-scale molten salt flow component; inputting the multi-scale molten salt flow rate component into an LSTM network adaptive gating mechanism network model to obtain an initial predicted value of the molten salt flow rate of each node; obtaining an error coefficient of the molten salt flow rate based on the D-SMAPE according to the initial predicted value of the molten salt flow rate of each node; inputting the error coefficient of the molten salt flow rate into a trained ResNet prediction model to obtain an optimized predicted value of the molten salt flow rate; and controlling the molten salt flow rate of the primary circuit of the thorium-based molten salt reactor based on a DEM-DAC algorithm according to the optimized predicted value of the molten salt flow rate. The memory may include internal memory, such as high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device. The processor, network interface, and memory are interconnected via an internal bus. This internal bus may be an Industry Standard Architecture bus, a Peripheral Component Interconnect Standard bus, an Extended Industry Standard Architecture bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. The memory is used to store programs. Specifically, the program may include program code, and the program code includes computer operating instructions. The memory may include internal memory and non-volatile memory, and provides instructions and data to the processor.

[0209] Example 4

[0210] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the method for predicting and controlling the molten salt flow rate of a primary circuit of a thorium-based molten salt reactor. For example, the method includes: obtaining an original molten salt flow rate signal of each node in the primary circuit of a thorium-based molten salt reactor, performing ASDA decomposition on the original molten salt flow rate signal to obtain a decomposed molten salt flow sub-signal, performing real-time signal processing and joint feature extraction on the decomposed molten salt flow sub-signal to obtain a multi-scale molten salt flow component; inputting the multi-scale molten salt flow rate component into an LSTM network adaptive gating mechanism network model to obtain an initial predicted value of the molten salt flow rate of each node; obtaining an error coefficient of the molten salt flow rate based on the D-SMAPE according to the initial predicted value of the molten salt flow rate of each node; inputting the error coefficient of the molten salt flow rate into a trained ResNet prediction model to obtain an optimized predicted value of the molten salt flow rate; and controlling the molten salt flow rate of the primary circuit of the thorium-based molten salt reactor based on a DEM-DAC algorithm according to the optimized predicted value of the molten salt flow rate. Specifically, the computer-readable storage medium includes, but is not limited to, volatile memory and / or non-volatile memory. The volatile memory may include random access memory and / or cache memory, etc. The non-volatile memory may include read-only memory, hard disk, flash memory, optical disk, magnetic disk, etc.

[0211] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code.

[0212] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the steps in the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0213] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0214] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0215] Those skilled in the art will readily identify other embodiments of the present invention after considering the specification and disclosure of the invention. This application is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the invention being indicated by the following claims.

[0216] It should be understood that the present invention is not limited to the exact construction described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

[0217] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any way. Any simple modification, change and equivalent structural change made to the above embodiment based on the technical essence of the present invention shall still fall within the scope of protection of the technical solution of the present invention.

Claims

1. A method for predicting and controlling the flow rate of molten salt in the primary circuit of a thorium-based molten salt reactor, characterized in that: include: Obtaining the original molten salt flow signal of each node in the primary circuit of the thorium-based molten salt reactor, performing ASDA decomposition on the original molten salt flow signal to obtain decomposed molten salt flow sub-signals, and performing real-time signal processing and joint feature extraction on the decomposed molten salt flow sub-signals to obtain multi-scale molten salt flow components; Inputting the multi-scale molten salt flow components into the LSTM network adaptive gating mechanism network model to obtain the initial predicted value of the molten salt flow at each node; According to the initial predicted value of the molten salt flow rate of each node, an error coefficient of the molten salt flow rate is obtained based on D-SMAPE; The error coefficient of the molten salt flow rate is input into the trained ResNet prediction model to obtain the optimized molten salt flow rate prediction value; According to the optimized molten salt flow prediction value, the molten salt flow in the primary circuit of the thorium-based molten salt reactor is controlled based on the DEM-DAC algorithm.

2. The method for predicting and controlling the primary circuit molten salt flow of a thorium-based molten salt reactor according to claim 1, characterized in that: The process of controlling the molten salt flow rate of the primary circuit of the thorium-based molten salt reactor based on the DEM-DAC algorithm according to the optimized molten salt flow rate prediction value is as follows: The optimized molten salt flow prediction value is converted into an electrical signal based on the DEM-DAC algorithm, and the electrical signal is then input into the frequency converter of the molten salt pump to control the variable frequency operation of the molten salt pump, thereby controlling the molten salt flow in the primary circuit of the thorium-based molten salt reactor.

3. The method for predicting and controlling the primary circuit molten salt flow of a thorium-based molten salt reactor according to claim 1, characterized in that: After obtaining the original molten salt flow signal of each node in the primary circuit of the thorium-based molten salt reactor, the method further includes: performing standardization processing on the original molten salt flow signal to eliminate dimension and amplitude differences.

4. The method for predicting and controlling the primary circuit molten salt flow of a thorium-based molten salt reactor according to claim 1, characterized in that: The training process of the LSTM network adaptive gating mechanism network model is: Among them, η t is the learning rate of the tth iteration; η0 is the initial learning rate.

5. The method for predicting and controlling the primary circuit molten salt flow of a thorium-based molten salt reactor according to claim 1, characterized in that: The loss function of the ResNet prediction model during training is: Where N is the number of samples, Correct the predicted value for the molten salt flow rate.

6. A thorium-based molten salt reactor primary circuit molten salt flow prediction and control system, characterized in that: include: An acquisition module is used to obtain the original molten salt flow signal of each node in the primary circuit of the thorium-based molten salt reactor, perform ASDA decomposition on the original molten salt flow signal to obtain decomposed molten salt flow sub-signals, and perform real-time signal processing and joint feature extraction on the decomposed molten salt flow sub-signals to obtain multi-scale molten salt flow components; A first prediction module is used to input the multi-scale molten salt flow components into the LSTM network adaptive gating mechanism network model to obtain an initial prediction value of the molten salt flow of each node; A calculation module, configured to obtain an error coefficient of the molten salt flow rate based on the D-SMAPE according to the initial predicted value of the molten salt flow rate of each node; The second prediction module is used to input the error coefficient of the molten salt flow into the trained ResNet prediction model to obtain the optimized molten salt flow prediction value; The control module is used to control the molten salt flow rate of the primary circuit of the thorium-based molten salt reactor based on the DEM-DAC algorithm according to the optimized molten salt flow rate prediction value.

7. The thorium-based molten salt reactor primary circuit molten salt flow prediction and control system according to claim 6, characterized in that: The process of controlling the molten salt flow rate of the primary circuit of the thorium-based molten salt reactor based on the DEM-DAC algorithm according to the optimized molten salt flow rate prediction value is as follows: The optimized molten salt flow prediction value is converted into an electrical signal based on the DEM-DAC algorithm, and the electrical signal is then input into the frequency converter of the molten salt pump to control the variable frequency operation of the molten salt pump, thereby controlling the molten salt flow in the primary circuit of the thorium-based molten salt reactor.

8. The thorium-based molten salt reactor primary circuit molten salt flow prediction and control system according to claim 6, characterized in that: After obtaining the original molten salt flow signal of each node in the primary circuit of the thorium-based molten salt reactor, the method further includes: The original molten salt flow rate signal is normalized to eliminate dimension and amplitude differences.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the method for predicting and controlling the primary circuit molten salt flow of a thorium-based molten salt reactor as described in any one of claims 1 to 5 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for predicting and controlling the primary circuit molten salt flow of a thorium-based molten salt reactor as described in any one of claims 1 to 5 are implemented.

Images