Method and device for evaluating working state of nuclear-grade pump based on neural network activation function, storage medium and electronic equipment

Abstract

The invention relates to a method and device for evaluating the working state of a nuclear-grade pump based on a neural network activation function, a storage medium and electronic equipment. The method comprises the following steps that factory data of a to-be-evaluated nuclear-grade pump are obtained; calculating based on factory data to obtain a relation between factory flow and lift; determining an assembly deviation variable of the nuclear-grade pump to be evaluated; constructing a neural network based on the assembly deviation variable, the input flow of the to-be-evaluated nuclear-grade pump and the correspondingly generated lift; acquiring operation data of the to-be-evaluated nuclear-grade pump; calculating based on the operation data and a neural network to obtain a weight parameter and an offset parameter; and evaluating the working state of the nuclear-grade pump to be evaluated by combining the activation function of the neurons according to the weight parameter, the bias parameter and the relationship between the delivery flow and the lift. The operation state of the nuclear-grade pump can be monitored in real time, potential faults can be found in advance, it is avoided that shutdown time is prolonged due to sudden faults of the nuclear-grade pump, and the influence of shutdown on the productivity of the nuclear power plant is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of nuclear power plant pump status monitoring, and more specifically, to a method, device, storage medium and electronic equipment for evaluating the working status of a nuclear-grade pump based on a neural network activation function. Background Art

[0002] As a core component of a nuclear power plant, nuclear-grade pumps maintain effective circulation in the cooling system, prevent reactor overheating, and effectively mitigate the risk of radioactive material leakage, safeguarding the health and safety of workers and the public. Furthermore, the efficient and stable operation of nuclear-grade pumps reduces unnecessary maintenance costs and power generation losses due to downtime, playing a crucial role in improving the economic efficiency and sustainable development of nuclear power plants. Therefore, ensuring the long-term stable operation of nuclear-grade pumps is an essential component of the nuclear power industry.

[0003] After long-term operation and multiple repairs and installations, the structural parameters of nuclear-grade pumps may change, causing performance to deviate from expectations. Currently, nuclear power plants generally use the factory acceptance standards for centrifugal pumps as the performance evaluation criteria. However, under actual installation and operating conditions, the pump's operating point offset is uncertain, making it difficult to accurately predict before installation whether the pump will meet the performance guarantee points specified in preliminary documents such as the technical specifications. In addition, the above standards also include complex parameters such as tolerance ranges and system uncertainties, which are often difficult to directly measure and estimate. When the performance of a nuclear-grade pump exceeds the control standard, the only judgment is based on the experience of the operator, but this often involves a high degree of subjectivity and uncertainty. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method, device, storage medium and electronic device for evaluating the working status of a nuclear-grade pump based on a neural network activation function in response to the problems existing in the prior art.

[0005] The technical solution adopted by the present invention to solve the technical problem is to construct a method for evaluating the working status of a nuclear-grade pump based on a neural network activation function, comprising the following steps:

[0006] Obtain factory data of the nuclear-grade pump to be evaluated;

[0007] Calculate based on the factory data to obtain the relationship between the factory flow rate and lift;

[0008] determining an assembly deviation variable of the nuclear-grade pump to be evaluated;

[0009] Constructing a neural network based on the assembly deviation variable, the input flow rate of the nuclear-grade pump to be evaluated, and the corresponding generated head;

[0010] Obtaining operating data of the nuclear-grade pump to be evaluated;

[0011] Performing calculations based on the operating data and the neural network to obtain weight parameters and bias parameters;

[0012] The working state of the nuclear-grade pump to be evaluated is evaluated based on the weight parameter, the bias parameter, and the relationship between the flow rate and the head at the factory, combined with the activation function of the neuron.

[0013] In the method for evaluating the working status of a nuclear-grade pump based on a neural network activation function of the present invention, the factory data includes: factory flow data and factory head data;

[0014] The calculation based on the factory data to obtain the factory flow rate and head relationship includes:

[0015] Based on the factory data, the relationship between the factory flow rate and the lift is obtained through a neural network or by using a quadratic function fitting.

[0016] In the method for evaluating the working status of a nuclear-grade pump based on a neural network activation function according to the present invention, constructing a neural network based on the assembly deviation variable, the input flow rate of the nuclear-grade pump to be evaluated, and the corresponding generated head includes:

[0017] defining the assembly deviation variable and the input flow as an input layer of the neural network;

[0018] Defining a hidden layer of the neural network according to the influence of the assembly deviation variable and the input flow rate on the lift;

[0019] Define the head as the output layer of the neural network.

[0020] In the method for evaluating the working status of a nuclear-grade pump based on a neural network activation function according to the present invention, the operating data includes: the actual flow rate of the nuclear-grade pump to be evaluated, the actual assembly deviation variable of the nuclear-grade pump to be evaluated, the factory head of the nuclear-grade pump to be evaluated, and the actual head of the nuclear-grade pump to be evaluated;

[0021] The step of performing calculation based on the operating data and the neural network to obtain weight parameters and bias parameters includes:

[0022] The weight parameters and the bias parameters are obtained by performing calculations based on the actual flow rate of the nuclear-grade pump to be evaluated, the actual assembly deviation variable of the nuclear-grade pump to be evaluated, the factory head of the nuclear-grade pump to be evaluated, and the actual head of the nuclear-grade pump to be evaluated, combined with the loss function of the neural network.

[0023] In the method for evaluating the working status of a nuclear-grade pump based on a neural network activation function according to the present invention, evaluating the working status of the nuclear-grade pump to be evaluated based on the weight parameter, the bias parameter, and the factory flow rate and head relationship in combination with the neuron activation function includes:

[0024] Substituting the weight parameter and the bias parameter into the loss function to obtain a value of the loss function;

[0025] The working state of the nuclear-grade pump to be evaluated is evaluated according to the value of the loss function and in combination with the activation function of the neuron.

[0026] In the method for evaluating the working state of a nuclear-grade pump based on a neural network activation function according to the present invention, substituting the weight parameter and the bias parameter into a loss function for solving the loss function to obtain the value of the loss function includes:

[0027] Using a first calculation method, substituting the weight parameter and the bias parameter into the loss function to perform partial derivative calculation to obtain a first value; the first value is a calculated value in a linear region;

[0028] Using the second calculation method, the weight parameter and the bias parameter are substituted into the loss function to perform partial derivative calculation to obtain a second value; the second value is a calculated value in the linear region.

[0029] In the method for evaluating the working state of a nuclear-grade pump based on a neural network activation function according to the present invention, evaluating the working state of the nuclear-grade pump to be evaluated according to the value of the loss function and in combination with the activation function of the neuron includes:

[0030] comparing the first value to the second value;

[0031] If the first value is less than the second value, it is determined that the activation function of the neuron is in a linear region, and it is determined that the working state of the nuclear-grade pump to be evaluated is normal;

[0032] If the first value is greater than the second value, it is determined that the activation function of the neuron is in a nonlinear region, and it is determined that the working state of the nuclear-grade pump to be evaluated is abnormal.

[0033] The present invention also provides a device for evaluating the working state of a nuclear-grade pump based on a neural network activation function, comprising:

[0034] A factory data acquisition unit, used to obtain the factory data of the nuclear-grade pump to be evaluated;

[0035] A relationship determination unit, configured to calculate based on the factory data to obtain a relationship between the factory flow rate and the lift;

[0036] a deviation variable determining unit, configured to determine an assembly deviation variable of the nuclear-grade pump to be evaluated;

[0037] A neural network construction unit, configured to construct a neural network based on the assembly deviation variable, the input flow rate of the nuclear-grade pump to be evaluated, and the corresponding generated head;

[0038] An operating data acquisition unit, configured to acquire operating data of the nuclear-grade pump to be evaluated;

[0039] a parameter calculation unit, configured to perform calculations based on the operating data and the neural network to obtain weight parameters and bias parameters;

[0040] A state evaluation unit is used to evaluate the working state of the nuclear-grade pump to be evaluated based on the weight parameter, the bias parameter and the factory flow rate and head relationship in combination with the activation function of the neuron.

[0041] The present invention also provides a storage medium storing a computer program, wherein the computer program is suitable for loading by a processor to execute the steps of the method for evaluating the working status of a nuclear-grade pump based on a neural network activation function as described above.

[0042] The present invention also provides an electronic device comprising a memory and a processor, wherein a computer program is stored in the memory, and the processor executes the steps of the method for evaluating the working status of a nuclear-grade pump based on a neural network activation function as described above by calling the computer program stored in the memory.

[0043] The method, device, storage medium, and electronic device for evaluating the working status of a nuclear-grade pump based on a neural network activation function of the present invention have the following beneficial effects: comprising the following steps: obtaining factory data of the nuclear-grade pump to be evaluated; performing calculations based on the factory data to obtain the factory flow rate and head relationship; determining the assembly deviation variable of the nuclear-grade pump to be evaluated; constructing a neural network based on the assembly deviation variable, the input flow rate of the nuclear-grade pump to be evaluated, and the corresponding head generated; obtaining operating data of the nuclear-grade pump to be evaluated; performing calculations based on the operating data and the neural network to obtain weight parameters and bias parameters; and evaluating the working status of the nuclear-grade pump to be evaluated based on the weight parameters, bias parameters, and the factory flow rate and head relationship in combination with the activation function of the neuron. The present invention can monitor the operating status of the nuclear-grade pump in real time and detect potential faults in advance, thereby avoiding extended downtime caused by sudden nuclear-grade pump failures and reducing the impact of downtime on the productivity of nuclear power plants. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0045] Figure 1This is a flow chart of an embodiment of a method for evaluating the working status of a nuclear-grade pump based on a neural network activation function provided by the present invention;

[0046] Figure 2 It is a schematic diagram of the nuclear-grade pump neural network and transmission relationship provided by the present invention;

[0047] Figure 3 This is a logic block diagram of an embodiment of a device for evaluating the working status of a nuclear-grade pump based on a neural network activation function provided by the present invention. DETAILED DESCRIPTION

[0048] 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 the embodiments. 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.

[0049] To address the current issues with nuclear power plant status monitoring of nuclear-grade pumps, the present invention utilizes a large amount of historical nuclear-grade pump data and a neural network algorithm to develop a method for evaluating the operating status of nuclear-grade pumps based on a neural network activation function. This method can more accurately fit the performance curve of nuclear-grade pumps through neural network activation functions and / or spatial distance calculation methods, thereby accurately evaluating the operation of nuclear-grade pumps under different operating conditions. This method provides accurate performance evaluation results to technicians, who can then optimize and adjust the operating parameters of nuclear-grade pumps based on the provided performance evaluation results, ensuring that the nuclear-grade pumps operate near their optimal efficiency points, reducing energy waste and reduced production efficiency caused by inefficient pump operation, and thereby improving the overall production efficiency of nuclear power plants.

[0050] refer to Figure 1 In a preferred embodiment, the method for evaluating the working status of a nuclear-grade pump based on a neural network activation function comprises the following steps:

[0051] Step S100: Obtain the factory data of the nuclear-grade pump to be evaluated.

[0052] Optionally, in an embodiment of the present invention, the factory data includes factory flow data and factory head data. The factory data can be obtained in various existing ways, which are not specifically limited in the present invention.

[0053] Step S200: Calculate based on factory data to obtain the relationship between factory flow rate and lift.

[0054] Optionally, in some embodiments, calculating based on factory data to obtain the factory flow rate and head relationship includes: obtaining the factory flow rate and head relationship based on the factory data using a neural network or a quadratic function fitting. That is, the factory flow rate and head relationship can be obtained by neural network calculation, or can also be obtained by direct calculation using a quadratic function fitting.

[0055] Step S300: Determine the assembly deviation variable of the nuclear-grade pump to be evaluated.

[0056] Specifically, in some embodiments, the assembly deviation variable of the nuclear-grade pump to be evaluated can be determined by actual installation conditions, which is not specifically limited in the present invention.

[0057] Step S400: Constructing a neural network based on the assembly deviation variable, the input flow rate of the nuclear-grade pump to be evaluated, and the corresponding generated head.

[0058] Optionally, in some embodiments, constructing a neural network based on the assembly deviation variable, the input flow of the nuclear-grade pump to be evaluated, and the corresponding generated head includes: defining the assembly deviation variable and the input flow as the input layer of the neural network; defining the hidden layer of the neural network according to the influence of the assembly deviation variable and the input flow on the head; and defining the head as the output layer of the neural network.

[0059] Specifically, such as Figure 2 As shown, the neural network consists of three layers, the first layer is the input layer, the second layer is the hidden layer, and the third layer is the output layer. The input layer includes: input flow and detection change parameter X, which are q, x1, ..., x n , where q is the input flow, x i is the assembly deviation variable. The hidden layer includes: y1,…,y n The output layer includes: lift h. The second hidden layer includes two parts, one is the lift generated by the input flow q, and the other is x1,…,x n The lift generated by the input flow q can be obtained through neural network training, or it can be regarded as a large neuron, directly forming an activation function from the fitting function, with a weight coefficient of 1 and a bias of 0.

[0060] like Figure 2 As shown, the relationship between layers is as follows:

[0061]

[0062] (1) In this formula, the value of n is determined by the number of assembly deviation variables, and w i,j is the weight parameter, b j is the bias parameter.

[0063] x0=q(2);

[0064] p i =f(y j )(i=1,…,n)(3);

[0065]

[0066] p0=h0(q)(5).

[0067] (5) where h0(q) is the value without x1,…,x n The relationship between flow and head when parameters are affected, h0(q) can be generated by neural network or obtained by direct fitting.

[0068] Furthermore, in the embodiment of the present invention, the activation function f(y j )for:

[0069]

[0070] (6) In the formula, c is the proportional coefficient, which is 5%. By using a quadratic function to act on the nonlinearity of the hidden layer [2, 3], the quadratic function can be made to fit the performance curve of the nuclear-grade pump to obtain better results.

[0071] Furthermore, define is the amplification factor of the jth neuron, when When , the neuron is considered to be in the linear working area, and the nuclear pump can be determined to be in normal working state; when or When the neuron is in the nonlinear working area, it can be determined that the nuclear-grade pump is in an abnormal working state, and the operation of the nuclear-grade pump needs to be monitored.

[0072] Step S500: Acquire the operating data of the nuclear-grade pump to be evaluated.

[0073] Optionally, in some embodiments, the operating data includes: the actual flow rate of the nuclear-grade pump to be evaluated, the actual assembly deviation variable of the nuclear-grade pump to be evaluated, the factory head of the nuclear-grade pump to be evaluated, and the actual head of the nuclear-grade pump to be evaluated.

[0074] Step S600: Calculate based on the operating data and the neural network to obtain weight parameters and bias parameters.

[0075] Optionally, in some embodiments, calculations are performed based on operating data and a neural network to obtain weight parameters and bias parameters, including: calculating according to the actual flow rate of the nuclear-grade pump to be evaluated, the actual assembly deviation variable of the nuclear-grade pump to be evaluated, the factory head of the nuclear-grade pump to be evaluated, and the actual head of the nuclear-grade pump to be evaluated, combined with the loss function of the neural network to obtain weight parameters and bias parameters.

[0076] Step S700: Evaluate the working status of the nuclear-grade pump to be evaluated based on the weight parameters, bias parameters, and the factory flow rate and head relationship in combination with the activation function of the neuron.

[0077] Optionally, in some embodiments, evaluating the working status of the nuclear-grade pump to be evaluated based on the weight parameters, bias parameters and the relationship between the factory flow rate and head in combination with the activation function of the neuron includes: substituting the weight parameters and bias parameters into the loss function for solving to obtain the value of the loss function; evaluating the working status of the nuclear-grade pump to be evaluated based on the value of the loss function and in combination with the activation function of the neuron.

[0078] Optionally, in an embodiment of the present invention, substituting the weight parameters and bias parameters into the loss function for solution to obtain the value of the loss function includes: using a first calculation method, substituting the weight parameters and bias parameters into the loss function for partial derivative calculation to obtain a first value; the first value is a calculated value in the linear region; using a second calculation method, substituting the weight parameters and bias parameters into the loss function for partial derivative calculation to obtain a second value; the second value is a calculated value in the linear region. The first calculation method and the second calculation method can be the same or different. For example, both can be calculated using a neural network.

[0079] In some embodiments, evaluating the working state of the nuclear-grade pump to be evaluated based on the value of the loss function and in combination with the activation function of the neuron includes: comparing a first numerical value with a second numerical value; if the first numerical value is less than the second numerical value, judging that the activation function of the neuron is in a linear region, and judging that the working state of the nuclear-grade pump to be evaluated is normal; if the first numerical value is greater than the second numerical value, judging that the activation function of the neuron is in a nonlinear region, and judging that the working state of the nuclear-grade pump to be evaluated is abnormal.

[0080] In the embodiment of the present invention, in order to solve the weight parameters and bias parameters, a loss function can be used to solve them. Specifically, the loss function is defined as:

[0081]

[0082] (7) In the formula, (X1, X2,…, X k ) is a vector composed of k samples, where any vector, X j =(q j ,x j,1, x j,2, …,x j,n ).h t is the actual head corresponding to the flow rate.

[0083] In order to make Loss((X1,X2,…,X k )) is extremely small and needs to satisfy the following formula:

[0084]

[0085] By solving equations (8) and (9), we can get the weight parameter w i,j and bias parameter b i When a single value cannot be obtained through equations (8) and (9), it is necessary to increase the value of n to provide more X j , and finally solve the weight parameters and bias parameters.

[0086] In some other embodiments, the solution of weight parameters and bias parameters can also be solved by the neural network gradient reflection propagation method, and can also be solved in linear and nonlinear cases, or, the two regions can be directly approximated as a differentiable function using a smoothing method for calculation.

[0087] After solving the weight parameters and bias parameters of each neuron in the neural network, the working state of the nuclear pump can be evaluated by judging whether the activation of the neuron is in the linear region. When any activation function is in the nonlinear section, that is, or When the nuclear grade pump is in a nonlinear state, it is necessary to monitor the operation. The nuclear-grade pumps are within the expected range and can be considered to be operating normally.

[0088] refer to Figure 3 , Figure 3 This is a logic block diagram of an embodiment of the device for evaluating the working status of a nuclear-grade pump based on a neural network activation function provided by the present invention.

[0089] like Figure 3 As shown, the device for evaluating the working status of a nuclear-grade pump based on a neural network activation function includes:

[0090] The factory data acquisition unit 301 is used to acquire the factory data of the nuclear-grade pump to be evaluated.

[0091] The relationship determination unit 302 is used to perform calculations based on factory data to obtain the factory flow rate and lift relationship.

[0092] The deviation variable determining unit 303 is used to determine the assembly deviation variable of the nuclear-grade pump to be evaluated.

[0093] The neural network construction unit 304 is used to construct a neural network based on the assembly deviation variable, the input flow rate of the nuclear-grade pump to be evaluated, and the corresponding generated head.

[0094] The operation data acquisition unit 305 is used to acquire the operation data of the nuclear-grade pump to be evaluated.

[0095] The parameter calculation unit 306 is used to perform calculations based on the operating data and the neural network to obtain weight parameters and bias parameters.

[0096] The state evaluation unit 307 is used to evaluate the working state of the nuclear-grade pump to be evaluated based on the weight parameters, bias parameters and the factory flow rate and head relationship combined with the activation function of the neuron.

[0097] The present invention will be described below by taking a nuclear-grade pump system of a nuclear power plant as an example.

[0098] Table 1 shows the factory performance curve points (i.e. factory flow rate and head) of a nuclear-grade pump.

[0099] Table 1 Pump factory performance curve points

[0100]

[0101] By analyzing the data in Table 1, we can find that all points can be fitted by using a quadratic function, and its R 2 =0.9996, the fitting function is:

[0102] h0(q)=―3.54×10 ―6 q 2 +9.423×10 ―2 q+61.552(10);

[0103] Table 2 Effect of front ring axial clearance and flow rate on lift

[0104]

[0105] It can be seen from Table 2 that the assembly deviation variable that affects the lift is the axial clearance of the front ring, that is, there is only one input parameter. Therefore, based on formula (7), it can be obtained:

[0106] Loss((X1,X2,…,X k ))=(c(3500)+f(w 0,1 3500+w 1,1 0.5+b1)―h t (3500)) 2 +(h0(4000)+f(w 0,1 4000+w 1,1 0.2+b1)―h t (4000)) 2 +(h0(4300)+f(w 0,1 4300+w 1,1 0.5+b1)―h t (4300)) 2 ;

[0107] Assume that f(y j ) is in the linear region, according to formula (8) and formula (9), we can get:

[0108]

[0109]

[0110] Solving the above three equations we can get: w 0,1 =-0.00125; w 1,1 =1.0833333333; b1 = 5.8333333333. The weight parameter (i.e. w 0,1 =-0.00125; w 1,1 =1.08333333333) and the bias parameter (i.e. b1=5.8333333333) are substituted into the loss function for calculation, and the value of the loss function is 0. This indicates that the current state is in the linear region, indicating that the best fitting effect has been achieved. There is no need to enter the nonlinear region for calculation. It can be calculated that the activation function satisfies:

[0111]

[0112] From (10), we can determine that the activation function is in the linear region, and therefore, we can determine that the nuclear-grade pump is in normal working condition.

[0113] The present invention utilizes a large amount of historical data on nuclear-grade pumps and adopts a neural network method to develop a new set of evaluation criteria. Before installation, it can accurately predict whether the nuclear-grade pump can meet the performance guarantee points specified in the technical specifications and other preliminary documents; it can also predict complex parameters such as tolerance range and system uncertainty, which are used to accurately evaluate the safety of the nuclear-grade pump during operation. Through methods such as neural network activation functions and / or spatial distance calculations, the performance curve of the nuclear-grade pump can be fitted more accurately, thereby accurately evaluating its operation under different operating conditions. Based on the accurate performance evaluation results, technicians can optimize and adjust the operating parameters of the nuclear-grade pump so that it operates near the optimal efficiency point, reducing energy waste and reduced production efficiency caused by low pump operation efficiency, thereby improving the overall production efficiency of the nuclear power plant.

[0114] This invention monitors the operating status of nuclear-grade pumps in real time and detects potential faults in advance, allowing maintenance personnel ample time to perform preventive maintenance and repairs, thus avoiding extended downtime caused by sudden nuclear-grade pump failures. Compared with traditional fault repair methods, this reduces the impact of downtime on nuclear power plant productivity.

[0115] Specifically, the specific coordination operation process between the various units in the device for evaluating the working status of a nuclear-grade pump based on a neural network activation function can refer to the above-mentioned method for evaluating the working status of a nuclear-grade pump based on a neural network activation function, and will not be repeated here.

[0116] In addition, an electronic device of the present invention includes a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program to implement a method for evaluating the working status of a nuclear-grade pump based on a neural network activation function as described above. Specifically, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed by an electronic device and, when executed, performs the above-mentioned functions defined in the method of the embodiment of the present invention. The electronic device in the present invention can be a terminal such as a notebook, desktop, tablet computer, smart phone, or a server.

[0117] In addition, the present invention provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above methods for evaluating the operating status of a nuclear-grade pump based on a neural network activation function. Specifically, it should be noted that the storage medium of the present invention may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. Computer-readable storage media may be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or components, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device, or component. In the present invention, a computer-readable signal medium may include a data signal transmitted in baseband or as part of a carrier wave, which carries computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. Program code embodied on a computer-readable medium may be transmitted using any suitable medium, including but not limited to wires, optical cables, RF (radio frequency), or any suitable combination thereof.

[0118] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0119] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Reference can be made to the common and similar parts between the various embodiments. For the devices disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method description.

[0120] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0121] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0122] The above embodiments are intended only to illustrate the technical concepts and features of the present invention. Their purpose is to enable those skilled in the art to understand the present invention and implement it accordingly. They are not intended to limit the scope of protection of the present invention. All equivalent variations and modifications within the scope of the claims of the present invention are intended to be covered by the claims of the present invention.

Claims

1. A method for evaluating the working status of a nuclear-grade pump based on a neural network activation function, characterized in that: The following steps are involved: Obtain factory data of the nuclear-grade pump to be evaluated; Calculate based on the factory data to obtain the relationship between the factory flow rate and lift; determining an assembly deviation variable of the nuclear-grade pump to be evaluated; Constructing a neural network based on the assembly deviation variable, the input flow rate of the nuclear-grade pump to be evaluated, and the corresponding generated head; Obtaining operating data of the nuclear-grade pump to be evaluated; Performing calculations based on the operating data and the neural network to obtain weight parameters and bias parameters; The working state of the nuclear-grade pump to be evaluated is evaluated based on the weight parameter, the bias parameter, and the relationship between the flow rate and the head at the factory, combined with the activation function of the neuron.

2. The method for evaluating the working status of a nuclear-grade pump based on a neural network activation function according to claim 1, characterized in that: The factory data includes: factory flow data and factory head data; The calculation based on the factory data to obtain the factory flow rate and head relationship includes: Based on the factory data, the relationship between the factory flow rate and the lift is obtained through a neural network or by using a quadratic function fitting.

3. The method for evaluating the working status of a nuclear-grade pump based on a neural network activation function according to claim 1, wherein: The neural network is constructed based on the assembly deviation variable, the input flow rate of the nuclear-grade pump to be evaluated, and the corresponding generated head, including: defining the assembly deviation variable and the input flow as an input layer of the neural network; Defining a hidden layer of the neural network according to the influence of the assembly deviation variable and the input flow rate on the lift; Define the head as the output layer of the neural network.

4. The method for evaluating the working status of a nuclear-grade pump based on a neural network activation function according to claim 1, wherein: The operation data includes: the actual flow rate of the nuclear-grade pump to be evaluated, the actual assembly deviation variable of the nuclear-grade pump to be evaluated, the factory head of the nuclear-grade pump to be evaluated, and the actual head of the nuclear-grade pump to be evaluated; The step of performing calculation based on the operating data and the neural network to obtain weight parameters and bias parameters includes: The weight parameters and the bias parameters are obtained by performing calculations based on the actual flow rate of the nuclear-grade pump to be evaluated, the actual assembly deviation variable of the nuclear-grade pump to be evaluated, the factory head of the nuclear-grade pump to be evaluated, and the actual head of the nuclear-grade pump to be evaluated, combined with the loss function of the neural network.

5. The method for evaluating the working status of a nuclear-grade pump based on a neural network activation function according to claim 1, characterized in that: The step of evaluating the working state of the nuclear-grade pump to be evaluated based on the weight parameter, the bias parameter, and the relationship between the flow rate and the head at the factory and in combination with the activation function of the neuron includes: Substituting the weight parameter and the bias parameter into the loss function to obtain a value of the loss function; The working state of the nuclear-grade pump to be evaluated is evaluated according to the value of the loss function and in combination with the activation function of the neuron.

6. The method for evaluating the working status of a nuclear-grade pump based on a neural network activation function according to claim 5, characterized in that: Substituting the weight parameter and the bias parameter into a loss function to obtain a value of the loss function includes: Using a first calculation method, substituting the weight parameter and the bias parameter into the loss function to perform partial derivative calculation to obtain a first value; the first value is a calculated value in a linear region; Using the second calculation method, the weight parameter and the bias parameter are substituted into the loss function to perform partial derivative calculation to obtain a second value; the second value is a calculated value in the linear region.

7. The method for evaluating the working status of a nuclear-grade pump based on a neural network activation function according to claim 6, characterized in that: The evaluating the working state of the nuclear-grade pump to be evaluated according to the value of the loss function and in combination with the activation function of the neuron includes: comparing the first value to the second value; If the first value is less than the second value, it is determined that the activation function of the neuron is in a linear region, and it is determined that the working state of the nuclear-grade pump to be evaluated is normal; If the first value is greater than the second value, it is determined that the activation function of the neuron is in a nonlinear region, and it is determined that the working state of the nuclear-grade pump to be evaluated is abnormal.

8. A device for evaluating the working status of a nuclear-grade pump based on a neural network activation function, characterized in that: include: A factory data acquisition unit, used to obtain the factory data of the nuclear-grade pump to be evaluated; A relationship determination unit, configured to calculate based on the factory data to obtain a relationship between the factory flow rate and the lift; a deviation variable determining unit, configured to determine an assembly deviation variable of the nuclear-grade pump to be evaluated; A neural network construction unit, configured to construct a neural network based on the assembly deviation variable, the input flow rate of the nuclear-grade pump to be evaluated, and the corresponding generated head; An operating data acquisition unit, configured to acquire operating data of the nuclear-grade pump to be evaluated; a parameter calculation unit, configured to perform calculations based on the operating data and the neural network to obtain weight parameters and bias parameters; A state evaluation unit is used to evaluate the working state of the nuclear-grade pump to be evaluated based on the weight parameter, the bias parameter and the factory flow rate and head relationship in combination with the activation function of the neuron.

9. A storage medium, characterized in that: The storage medium stores a computer program, which is suitable for being loaded by a processor to execute the steps of the method for evaluating the working status of a nuclear-grade pump based on a neural network activation function as described in any one of claims 1 to 7.

10. An electronic device, characterized in that: The method comprises a memory and a processor, wherein a computer program is stored in the memory, and the processor executes the steps of the method for evaluating the working status of a nuclear-grade pump based on a neural network activation function as described in any one of claims 1 to 7 by calling the computer program stored in the memory.

Images