Switching power intelligent management and control method and system
By acquiring equipment operation data and using a long short-term memory network model to train an equipment fault prediction model, the problem of poor equipment fault prediction performance in existing technologies is solved, and higher accuracy equipment fault prediction is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- URBAN RAIL TRANSIT ENGINEERING CO LTD OF CHINA RAILWAY FIRST GROUP CO LTD
- Filing Date
- 2025-03-17
- Publication Date
- 2026-05-15
AI Technical Summary
The accuracy of equipment failure prediction in existing technologies depends on the data quality of a large number of training samples, resulting in poor equipment failure prediction performance and making it difficult to apply on a large scale.
By acquiring the equipment's operating data, the first, second, and third average operating status voiceprints are determined, and a long short-term memory network model is used to train the equipment fault prediction model. These voiceprints are then input in real time to improve prediction accuracy.
By using multiple acoustic signatures to characterize the operating status of equipment, the fault prediction process is simplified, and the accuracy and efficiency of equipment fault prediction are improved.
Smart Images
Figure CN120337710B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent control technology for switching power supplies, and more specifically to an intelligent control method and system for switching power supplies. Background Technology
[0002] Early equipment failure prediction models were mainly data-driven. Some data-driven machine learning models, such as distributed gradient boosting models, random forest models, support vector machine models, and artificial neural networks, have been widely used in equipment failure prediction applications.
[0003] However, the accuracy of empirical statistical models in predicting equipment failures depends heavily on the quality of a large number of training samples and the representativeness of the model. Their accuracy in predicting equipment failures remains limited, and their estimation results are unsatisfactory. In practical applications, equipment failure prediction relies on a large number of observation samples, which is resource-intensive and difficult to implement on a large scale. Furthermore, with a deeper understanding of equipment operating conditions, analyzing these conditions is more conducive to equipment failure prediction, and its prediction accuracy is superior to that of empirical statistical models.
[0004] Therefore, how to propose an intelligent control method and system for switching power supplies to solve the problem of poor equipment fault prediction in the existing technology and improve the intelligent control effect of switching power supplies is an urgent problem to be solved by those skilled in the art. Summary of the Invention
[0005] In view of this, the present invention provides an intelligent management and control method and system for switching power supplies, which solves the problem of poor equipment fault prediction in the prior art and improves the intelligent management and control effect of switching power supplies. To achieve the above objectives, the present invention adopts the following technical solution:
[0006] A method for intelligent control of switching power supplies, comprising:
[0007] Acquire the device's operating data, and determine the device's first average operating state voiceprint, second average operating state voiceprint, and third average operating state voiceprint based on the device's operating data;
[0008] Based on the first average operating state voiceprint, the second average operating state voiceprint, the third average operating state voiceprint of multiple equipment samples and the equipment fault prediction results, a neural network is trained to obtain an equipment fault prediction model.
[0009] The first average operating state voiceprint, the second average operating state voiceprint, and the third average operating state voiceprint of the device are input into the device fault prediction model in real time to obtain the fault prediction result of the device output by the device fault prediction model.
[0010] Optionally, the first average operating state voiceprint is the average operating state voiceprint from the start-up to the stable operation phase, the second average operating state voiceprint is the average operating state voiceprint extracted from the time sequence segment of the stable operation phase, and the third average operating state voiceprint is the average operating state voiceprint from the stable operation phase to the shutdown phase.
[0011] Optionally, the first average operating state soundprint, the second average operating state soundprint, the third average operating state soundprint, and the equipment fault prediction results of the multiple equipment samples are obtained by simulating multiple different scenarios through an equipment fault prediction model. Each scenario corresponds to a set of scenario parameter combinations, which include combinations of current parameters, voltage parameters, temperature parameters, environmental parameters, and management measure parameters.
[0012] Optionally, the inputs to the equipment fault prediction model are current parameters, voltage parameters, temperature parameters, environmental parameters, and management measure parameters, and the output is the fault prediction result.
[0013] Optionally, acquiring the device's operating data includes:
[0014] The startup type of the device is determined based on the device's startup method;
[0015] Based on the device's startup type, determine a first time range from startup to stable operation, a second time range from stable operation to shutdown, and a third time range; obtain corresponding operating data based on the first time range, the second time range, and the third time range.
[0016] Optionally, the neural network is a long short-term memory network.
[0017] Optionally, the method for determining the average operating state voiceprint includes:
[0018] The average operating status signal formed by current, voltage, temperature, and environmental parameters indicates the equipment status over a certain period of time. The intensity of the operating status signal at different time points within a defined time range is collected. The intensity of the operating status signal at the measured time point, measured by parameter type, is expressed as: {φ} ij (t), t=1,…,k(k≥1)};where, φ ij (t) represents the signal strength of the operating status measured in the tth time, k represents the number of measurements, i represents the i-th measured time point, and j represents the j-th parameter type. The signal strength of the operating status at the i-th measured time point is obtained through the j-th parameter type, and the average operating status voiceprint library ψ is constructed.
[0019] Optionally, the equipment failure prediction model includes:
[0020] The forget gate reads the output information h from the previous time step. t-1 and the input information x at the current moment t Output a value f between 0 and 1. t For each number in the cell state, 1 means to keep it completely and 0 means to discard it completely.
[0021] The value i to be updated is determined by the sigmoid layer of the input gate. t A new candidate vector of cell states is created through the tanh layer. The old cell state C t-1 f obtained from the forgetting gate t Multiply, discard the information that needs to be discarded, and add the updated value i obtained from the input gate. t With the new cell candidate value vector The product of these two factors yields the new cell state C. t ;
[0022] Current input information x t It will go through a sigmoid layer to determine the result to be output. t Next, output the result o. t Multiply by C the cell state after tanh layer treatment t The output h determined at the current time is obtained. t .
[0023] Optionally, it also includes: verifying the estimation accuracy of the equipment failure prediction model, including verification using measured sample points. The measured sample point verification data comes from sample data from different plant areas, and the accuracy verification index includes the coefficient of determination R. 2 Root mean square error (RMSE) and mean relative error (MRE), R 2 The higher the value, the smaller the RMSE and MRE, indicating a better performance of the equipment failure prediction model.
[0024] Optionally, a switching power supply intelligent management and control system includes:
[0025] Data acquisition module: Used to acquire the device's operating data;
[0026] Preprocessing module: used to determine the first average operating state voiceprint, the second average operating state voiceprint, and the third average operating state voiceprint of the device based on the device's operating data;
[0027] Model building module: used to train a recurrent neural network to obtain an equipment fault prediction model based on the first average operating state voiceprint, the second average operating state voiceprint, the third average operating state voiceprint and the equipment fault prediction results of multiple equipment samples.
[0028] Prediction module: used to input the first average operating state voiceprint, the second average operating state voiceprint, and the third average operating state voiceprint of the device into the device fault prediction model in real time, and obtain the fault prediction result of the device output by the device fault prediction model.
[0029] As can be seen from the above technical solutions, compared with the prior art, the present invention discloses an intelligent control method and system for switching power supplies, which has the following beneficial effects:
[0030] This invention proposes an intelligent management and control method for switching power supplies, comprising: acquiring equipment operating data, and determining a first average operating state soundprint, a second average operating state soundprint, and a third average operating state soundprint based on the equipment operating data; training a neural network to obtain an equipment fault prediction model based on the first average operating state soundprint, the second average operating state soundprint, the third average operating state soundprint, and the equipment fault prediction results from multiple equipment samples; and inputting the first average operating state soundprint, the second average operating state soundprint, and the third average operating state soundprint into the equipment fault prediction model in real time to obtain the equipment fault prediction result output by the equipment fault prediction model. This invention solves the problem of poor equipment fault prediction performance in existing technologies. It better characterizes the equipment operating state through the first average operating state soundprint, the second average operating state soundprint, and the third average operating state soundprint. By using current parameters, voltage parameters, temperature parameters, environmental parameters, and management measure parameters as input indicators, and leveraging a long short-term memory network model, it simplifies the equipment fault prediction process and improves prediction accuracy. Attached Figure Description
[0031] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0032] Figure 1 This is a schematic diagram of a method for intelligent control of switching power supplies provided by the present invention.
[0033] Figure 2 The present invention provides a structural framework diagram of an intelligent control system for switching power supplies. Detailed Implementation
[0034] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0035] This invention discloses an intelligent control method for switching power supplies, such as... Figure 1 As shown, it includes:
[0036] Acquire the device's operating data, and determine the device's first average operating state voiceprint, second average operating state voiceprint, and third average operating state voiceprint based on the device's operating data;
[0037] Based on the first average operating state voiceprint, the second average operating state voiceprint, the third average operating state voiceprint of multiple equipment samples and the equipment fault prediction results, a neural network is trained to obtain an equipment fault prediction model.
[0038] The first average operating state voiceprint, the second average operating state voiceprint, and the third average operating state voiceprint of the device are input into the device fault prediction model in real time to obtain the fault prediction result of the device output by the device fault prediction model.
[0039] Furthermore, the first average operating state voiceprint is the average operating state voiceprint from the start-up to the stable operation phase, the second average operating state voiceprint is the average operating state voiceprint extracted from the time sequence segment of the stable operation phase, and the third average operating state voiceprint is the average operating state voiceprint from the stable operation phase to the shutdown phase.
[0040] Furthermore, the first average operating state soundprint, the second average operating state soundprint, the third average operating state soundprint, and the equipment fault prediction results of the multiple equipment samples are obtained by simulating multiple different scenarios through an equipment fault prediction model. Each scenario corresponds to a set of scenario parameter combinations, which include combinations of current parameters, voltage parameters, temperature parameters, environmental parameters, and management measure parameters.
[0041] Furthermore, the inputs to the equipment fault prediction model are current parameters, voltage parameters, temperature parameters, environmental parameters, and management measure parameters, and the output is the fault prediction result. The management measure parameters include, for example, the detection date.
[0042] Furthermore, acquiring the device's operating data includes:
[0043] The startup type of the device is determined based on the device's startup method;
[0044] Based on the device's startup type, determine a first time range from startup to stable operation, a second time range from stable operation to shutdown, and a third time range; obtain corresponding operating data based on the first time range, the second time range, and the third time range.
[0045] Furthermore, determining the startup type of the equipment based on its startup method includes: determining a set of different equipment categories based on the startup method (e.g., star-delta startup, autotransformer step-down startup, or soft startup) and corresponding load parameters. Then, based on these different equipment categories, determining a first time range from startup to stable operation, a second time range of stable operation, and a third time range from stable operation to shutdown.
[0046] Specifically, the second average operating state voiceprint is the average operating state voiceprint extracted from the time sequence segment of the stable operating phase. It includes a time segment of length (first time range + third time range) / 2 extracted from the stable operating phase as the second time range, and the second average operating state voiceprint is determined through the second time range.
[0047] Furthermore, the neural network is a long short-term memory network.
[0048] Furthermore, the method for determining the average operating status voiceprint includes: using an average operating status signal formed by current parameters, voltage parameters, temperature parameters, and environmental parameters to indicate the equipment status over a certain time period; collecting the intensity of the operating status signal at different time points within a defined time range; and representing the operating status signal intensity at the measured time point through parameter type as: {φ ij (t), t=1,…,k(k≥1)};where, φ ij (t) represents the signal strength of the operating status measured in the tth time, k represents the number of measurements, i represents the i-th measured time point, and j represents the j-th parameter type. The signal strength of the operating status at the i-th measured time point is obtained through the j-th parameter type, and the average operating status voiceprint library ψ is constructed.
[0049] In a specific implementation, the method for determining the average operating state voiceprint specifically includes:
[0050] The average operating status signal formed by current, voltage, temperature, and environmental parameters indicates the equipment status over a certain period of time. The intensity of the operating status signal at different time points within a defined time range is collected, and the intensity of the operating status signal at the measured time point, measured by parameter type, is expressed as:
[0051] {φ ij(t), t=1,…,k(k≥1);
[0052] Where, φ ij (t) represents the operating status signal strength of the t-th measurement, k represents the number of measurements, i represents the i-th measured time point, and j represents the j-th parameter type. The operating status signal strength of the i-th measured time point is obtained through the j-th parameter type, and the average operating status fingerprint database ψ is constructed.
[0053]
[0054] Where m is the number of parameter types, n is the number of time point groups to be measured, and ψ is the i-th row vector of ψ. i =[φ i1 ,φ i2 ,...,φ im This refers to the fingerprint feature values of m parameter types measured at the time point under test. Different feature quantities are measured by different types of parameters to form a multi-dimensional feature quantity dimension. When a device failure occurs at the time point under test, the information measured by each parameter type is the operating status signal strength.
[0055] In a specific implementation, the above-mentioned method for determining the average operating state voiceprint is adopted. Based on the startup type of the device, a first time range from the device to the stable operation stage, a second time range of the stable operation stage, and a third time range from the stable operation stage to the shutdown stage are determined. Based on the first time range, the second time range, and the third time range, corresponding operating data are obtained, and the first average operating state voiceprint, the second average operating state voiceprint, and the third average operating state voiceprint of the device are determined based on the operating data of the device.
[0056] Furthermore, the equipment failure prediction model includes:
[0057] The forget gate reads the output information h from the previous time step. t-1 and the input information x at the current moment t Output a value f between 0 and 1. t For each number in the cell state, 1 means to keep it completely and 0 means to discard it completely.
[0058] The value i to be updated is determined by the sigmoid layer of the input gate. t A new candidate vector of cell states is created through the tanh layer. The old cell state C t-1 f obtained from the forgetting gate t Multiply, discard the information that needs to be discarded, and add the updated value i obtained from the input gate. t With the new cell candidate value vector The product of these two factors yields the new cell state C. t ;
[0059] Current input information x t It will go through a sigmoid layer to determine the result to be output. t Next, output the result o. t Multiply by C the cell state after tanh layer treatment t The output h determined at the current time is obtained. t .
[0060] In specific implementations, Long Short-Term Memory (LSTM) networks are a widely used machine learning algorithm for time-series signal prediction and classification. LSTM is a special type of RNN that can learn long-term dependency information. Its basic structure is the same as that of an RNN, a chain of repetitive neural network modules. LSTM has four special layers for the interaction between input and output.
[0061] LSTM networks are like constantly updating cells, with the key concept being the cell state. LSTMs have the ability to remove or add information to the cell state through carefully designed structures called "gates." Gates allow information to pass selectively; 0 represents no information can pass, and 1 represents any information can pass. LSTMs have three gates for updating the cell state: the forget gate, the input gate, and the output gate. The forget gate and input gate are the main components in Long Short-Term Memory networks that maintain network activity. As the network iterates and updates, they determine the information that needs to be discarded and retained. Specifically:
[0062] The first step is to determine the information to discard from the cell state. This is accomplished through a sigmoid layer with a "forget gate," which reads the output information h from the previous time step. t-1 and the input information x at the current moment t Output a value f between 0 and 1. t For each number in the cell state, 1 means to keep it completely and 0 means to discard it completely.
[0063] f t =σ(W f ·[h t-1 x t ]+b f );
[0064] The second step is to determine the new information stored in the cell state. First, the sigmoid layer of the "input gate" determines the value i to be updated. tSecond, a new candidate vector of cell states is created through the tanh layer.
[0065] i t =σ(W f ·[h t-1 x t ]+b i );
[0066]
[0067] The old cell state C t-1 f obtained from the "Forgotten Gate" t Multiply, discard the information that needs to be discarded, and add the updated value i obtained from the "input gate". t The product of the new cell candidate value vector and the new cell state C is obtained. t ,
[0068]
[0069] The third step is to determine the output (prediction result). The output result will be based on the current cell state C. t The input information x at the current moment. t It will go through a sigmoid layer to determine the result to be output. t Next, the result to be output is o. t Multiply by C the cell state after tanh layer treatment t The output h determined at the current time is obtained. t .
[0070] o t =σ(W o ·[h t-1 x t ]+b o );
[0071] h t =o t *tanh(C t );
[0072] The input to an LSTM network consists of current, voltage, temperature, and environmental parameters, defined as X. t =[x t RH x t RK x t RA x t LH The output is the predicted equipment failure result Y for the next moment. t+1 =[yt+1 RH y t+1 RK y t+1 RA y t+1 LH ].
[0073] To evaluate the performance of LSTM networks in device failure prediction, the root mean square error (RMSE) is used. RMS This is used to quantify the difference between predicted and actual observed values.
[0074]
[0075] In this specific implementation, the LSTM model is built in the MATLAB environment. The LSTM network architecture is defined as follows: the number of input features and the number of output responses are both 4, and the number of hidden units in the network is set to 200. The solver for training the model is set to 'Adam' and 250 training rounds are performed. To prevent gradient explosion, the gradient threshold is set to 1, and the initial learning rate is 0.005. After every 50 training rounds, the learning rate is reduced by multiplying by a reduction factor of 0.5. The dataset is derived from device operation data, including current parameters (RH), voltage parameters (RK), temperature parameters (RA), and environmental parameter data (LH). To prevent training divergence, the training data is first standardized to a dataset with zero mean and unit variance. The dataset is divided into training and testing datasets in a 6:4 ratio. The goal is to predict the device failure outcome at the next time step using historical device data, with the prediction time step defined as 1 sampling point. The historical device data vector serves as the input vector of the LSTM network, and the predicted device failure outcome is the output vector of the LSTM.
[0076] In a specific implementation, during the model training process, the model optimizer is Adam, the loss function is mean squared error, and the network is trained multiple times to find the optimal values of parameters such as the number of model layers, number of neurons (units), number of epochs, batch size, and dropout.
[0077] Furthermore, this also includes: verifying the estimation accuracy of the equipment failure prediction model, including verification using measured sample points. The measured sample point verification data comes from sample data from different plant areas, and the accuracy verification index includes the coefficient of determination R. 2 Root mean square error (RMSE) and mean relative error (MRE), R 2 The higher the value, the smaller the RMSE and MRE, indicating a better performance of the equipment failure prediction model.
[0078]
[0079] Where, x i and y i These represent the actual forecast value and the estimated forecast value, respectively. It is the average of the actual predicted values.
[0080] In a specific implementation, a switching power supply intelligent management and control system, such as Figure 2 As shown, it includes:
[0081] Data acquisition module: Used to acquire the device's operating data;
[0082] Preprocessing module: used to determine the first average operating state voiceprint, the second average operating state voiceprint, and the third average operating state voiceprint of the device based on the device's operating data;
[0083] Model building module: used to train a recurrent neural network to obtain an equipment fault prediction model based on the first average operating state voiceprint, the second average operating state voiceprint, the third average operating state voiceprint and the equipment fault prediction results of multiple equipment samples.
[0084] Prediction module: used to input the first average operating state voiceprint, the second average operating state voiceprint, and the third average operating state voiceprint of the device into the device fault prediction model in real time, and obtain the fault prediction result of the device output by the device fault prediction model.
[0085] In a specific embodiment, an electronic device is also included, comprising a processor, a communication interface, a memory, and a communication bus. The processor, communication interface, and memory communicate with each other via the communication bus. The processor can invoke logical instructions in the memory to execute a smart control method for a switching power supply. Furthermore, the aforementioned logical instructions in the memory can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0086] In a specific embodiment, a non-transitory computer-readable storage medium is also included, on which a computer program is stored, which, when executed by a processor, is implemented to perform the intelligent control method for switching power supplies provided by the above methods.
[0087] In a specific implementation, a computer program product is also included. The computer program product includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute a switching power supply intelligent control method provided by the above methods.
[0088] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.
[0089] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for intelligent control of a switching power supply, characterized in that, include: Acquire the device's operating data, and determine the device's first average operating state voiceprint, second average operating state voiceprint, and third average operating state voiceprint based on the device's operating data; The first average operating state voiceprint is the average operating state voiceprint from the start-up to the stable operation phase, the second average operating state voiceprint is the average operating state voiceprint extracted from the time sequence segment of the stable operation phase, and the third average operating state voiceprint is the average operating state voiceprint from the stable operation phase to the shutdown phase. The method for determining the average operating state voiceprint includes: The average operating status signal formed by current, voltage, temperature, and environmental parameters indicates the equipment status over a certain period of time. The intensity of the operating status signal at different time points within a defined time range is collected, and the intensity of the operating status signal at the measured time point, measured by parameter type, is expressed as: ;in, Let k represent the number of measurements, k represent the number of measurements, i represent the i-th measured time point, and j represent the j-th parameter type. The operating status signal strength at the i-th measured time point is obtained through the j-th parameter type, and an average operating status voiceprint database is constructed. ; Based on the first average operating state voiceprint, the second average operating state voiceprint, the third average operating state voiceprint of multiple equipment samples and the equipment fault prediction results, a neural network is trained to obtain an equipment fault prediction model. The first average operating state voiceprint, the second average operating state voiceprint, and the third average operating state voiceprint of the device are input into the device fault prediction model in real time to obtain the fault prediction result of the device output by the device fault prediction model.
2. The intelligent control method for a switching power supply according to claim 1, characterized in that, The first average operating state soundprint, the second average operating state soundprint, the third average operating state soundprint, and the equipment fault prediction results of the multiple equipment samples are obtained by simulating various different scenarios through an equipment fault prediction model. Each scenario corresponds to a set of scenario parameter combinations, which include combinations of current parameters, voltage parameters, temperature parameters, environmental parameters, and management measure parameters.
3. The intelligent control method for a switching power supply according to claim 1, characterized in that, The inputs to the equipment fault prediction model are current parameters, voltage parameters, temperature parameters, environmental parameters, and management measure parameters, and the output is the fault prediction result.
4. The intelligent control method for a switching power supply according to claim 1, characterized in that, The acquisition of device operating data includes: The startup type of the device is determined based on the device's startup method; Based on the device's startup type, determine a first time range from startup to stable operation, a second time range from stable operation to shutdown, and a third time range; obtain corresponding operating data based on the first time range, the second time range, and the third time range.
5. The intelligent control method for a switching power supply according to claim 1, characterized in that, The neural network is a long short-term memory network.
6. The intelligent control method for a switching power supply according to claim 1, characterized in that, The equipment failure prediction model includes: The forget gate reads the output information h from the previous time step. t-1 and the input information x at the current moment t Output a value f between 0 and 1. t For each number in the cell state, 1 means to keep it completely and 0 means to discard it completely. The value i to be updated is determined by the sigmoid layer of the input gate. t A new candidate vector of cell states is created through the tanh layer. , put the old cell state C t-1 f obtained from the forgetting gate t Multiply, discard the information that needs to be discarded, and add the updated value i obtained from the input gate. t With the new cell candidate value vector The product of these two states yields the new cell state C. t ; Current input information x t It will go through a sigmoid layer to determine the result to be output. t Next, output the result o. t Multiply by C the cell state after tanh layer treatment t The output h determined at the current time is obtained. t .
7. The intelligent control method for a switching power supply according to claim 1, characterized in that, Also includes: The accuracy of the equipment failure prediction model was verified, including verification using measured sample points. The measured sample point verification data came from sample data from different plant areas. The accuracy verification index included the coefficient of determination R. 2 Root mean square error (RMSE) and mean relative error (MRE), R 2 The higher the value, the smaller the RMSE and MRE, indicating a better performance of the equipment failure prediction model.
8. A smart control system for switching power supplies, characterized in that, include: Data acquisition module: Used to acquire the device's operating data; Preprocessing module: used to determine the first average operating state voiceprint, the second average operating state voiceprint, and the third average operating state voiceprint of the device based on the device's operating data; The first average operating state voiceprint is the average operating state voiceprint from the start-up to the stable operation phase, the second average operating state voiceprint is the average operating state voiceprint extracted from the time sequence segment of the stable operation phase, and the third average operating state voiceprint is the average operating state voiceprint from the stable operation phase to the shutdown phase. The method for determining the average operating state voiceprint includes: The average operating status signal formed by current, voltage, temperature, and environmental parameters indicates the equipment status over a certain period of time. The intensity of the operating status signal at different time points within a defined time range is collected, and the intensity of the operating status signal at the measured time point, measured by parameter type, is expressed as: ;in, Let k represent the number of measurements, k represent the number of measurements, i represent the i-th measured time point, and j represent the j-th parameter type. The operating status signal strength at the i-th measured time point is obtained through the j-th parameter type, and an average operating status voiceprint database is constructed. ; Model building module: used to train a recurrent neural network to obtain an equipment fault prediction model based on the first average operating state voiceprint, the second average operating state voiceprint, the third average operating state voiceprint and the equipment fault prediction results of multiple equipment samples. Prediction module: used to input the first average operating state voiceprint, the second average operating state voiceprint, and the third average operating state voiceprint of the device into the device fault prediction model in real time, and obtain the fault prediction result of the device output by the device fault prediction model.