Power transformer-oriented sensor output data intelligent calibration method and device
By employing an intelligent calibration method for power transformer sensors, which utilizes a multilayer perceptron neural network to evaluate performance and perform online calibration, the problem of sensor performance degradation in the field environment is solved, thereby improving data reliability and system security.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- STATE GRID HEBEI ELECTRIC POWER RES INST
- Filing Date
- 2025-06-13
- Publication Date
- 2026-07-31
AI Technical Summary
Sensors deployed on-site at power transformers are affected by factors such as electromagnetic fields, vibration, and temperature, leading to performance degradation or failure. Existing technologies make it difficult to effectively perform intelligent calibration of sensor output data.
The system collects current and historical output data from the sensors, performs performance evaluation using a multilayer perceptron neural network, calculates performance degradation scores, and performs online calibration using data fitting or a calibration model if an anomaly is detected. This includes sensor attribute clustering and training the calibration model using the Levenberg-Marquardt backpropagation algorithm.
This has improved the reliability and data accuracy of sensors in the field environment, reduced maintenance frequency and labor costs, and enhanced the intelligent operation and maintenance level and system security of power transformers.
Smart Images
Figure CN120927048B_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of power transformer technology, and in particular to a method and apparatus for intelligent calibration of sensor output data for power transformers. Background Technology
[0002] Power transformers are critical equipment in power systems, and their operational status directly affects the stability of the power grid and the reliability of power supply. During the operation of power transformers, various sensors are used to monitor parameters such as temperature, humidity, voltage, current, and vibration, helping maintenance personnel to understand the real-time operating status of the power transformers. However, during long-term operation in the field, sensors are affected by many factors, such as electromagnetic fields, vibration, temperature, humidity, and internal and external impact signals. These factors can all cause sensor performance degradation or even failure. Therefore, how to conduct intelligent calibration of sensor output data after deployment in the field is a key issue that needs to be considered. Summary of the Invention
[0003] In a first aspect, embodiments of this disclosure provide an intelligent calibration method for sensor output data of power transformers, the method comprising:
[0004] Collect output data from sensors deployed on the power transformer in the field;
[0005] The sensor's performance is evaluated based on the current output data and historical output data collected from the sensor.
[0006] If the sensor's performance evaluation result is abnormal, then the sensor's current output data and subsequent output data are calibrated.
[0007] In some possible implementations of the first aspect, the sensor performance is evaluated based on the current output data and historical output data of the sensor, including:
[0008] Based on the current operating conditions and environment of the sensor, query the historical output data of the sensor to find the first target output data corresponding to the current operating conditions and environment;
[0009] Calculate the sensor's performance evaluation index based on the sensor's current output data and the first target output data;
[0010] The sensor's performance degradation score is calculated based on the sensor's performance evaluation indicators, and this score is used to determine whether the sensor's performance is abnormal.
[0011] In some possible implementations of the first aspect, the sensor performance is evaluated based on the current output data and historical output data of the sensor, including:
[0012] Based on the current operating condition and environment of the sensor, query the historical output data of the sensor to find the first target output data corresponding to the current operating condition and environment; based on the current operating condition and environment of each reference sensor, query the historical output data of each reference sensor to find the second target output data corresponding to the current operating condition and environment.
[0013] Calculate the sensor's performance evaluation index based on the sensor's current output data and the first target output data; calculate the reference sensor's performance evaluation index based on the reference sensor's current output data and the second target output data;
[0014] Calculate the sensor's performance degradation score based on the sensor's performance evaluation metrics; calculate the reference sensor's performance degradation score based on the reference sensor's performance evaluation metrics;
[0015] Calculate the difference between the performance degradation score of the sensor and the performance degradation scores of each reference sensor, and determine whether the sensor's performance is abnormal based on the calculated difference.
[0016] In some possible implementations of the first aspect, the respective reference sensors corresponding to the sensor are determined in the following manner:
[0017] Acquire attribute information of the sensor and multiple candidate reference sensors deployed on the power transformer in the field;
[0018] The sensor and multiple candidate reference sensors are clustered based on the acquired attribute information;
[0019] Select a candidate reference sensor from the cluster to which the sensor belongs as the reference sensor for the corresponding sensor.
[0020] In some possible implementations of the first aspect, if the sensor's performance evaluation result is abnormal, the sensor's current output data and subsequent output data are calibrated, including:
[0021] If the sensor's performance evaluation result is abnormal, then the sensor's current output data is fitted with the first target output data, and the calibration parameters are determined based on the fitting result.
[0022] The sensor's current output data and subsequent output data are calibrated using calibration parameters.
[0023] In some possible implementations of the first aspect, if the sensor's performance evaluation result is abnormal, the sensor's current output data and subsequent output data are calibrated, including:
[0024] If the sensor's performance evaluation result is abnormal, the sensor's calibration parameters are calculated based on the current output data of a reference sensor whose performance degradation score is lower than a preset threshold and the sensor's current output data.
[0025] The sensor's current output data and subsequent output data are calibrated using calibration parameters.
[0026] In some possible implementations of the first aspect, if the sensor's performance evaluation result is abnormal, the sensor's current output data and subsequent output data are calibrated, including:
[0027] If the sensor's performance evaluation result is abnormal, then the calibration model is used to calibrate the sensor's current output data and subsequent output data.
[0028] The calibration model is trained through the following steps:
[0029] Acquire the sensor output data to be calibrated and the expected output data of the power transformer;
[0030] Construct a multilayer perceptron (MLP) neural network, which includes an input layer, a hidden layer, and an output layer. The activation function of the hidden layer is the sigmoid function, and the logistic regression from the hidden layer to the output layer is softmax regression.
[0031] The MLP neural network is trained using the output data to be calibrated as the input and the desired output data as the desired output. The Levenberg-Marquardt backpropagation algorithm is used to obtain the calibration model.
[0032] In some possible implementations of the first aspect, the sensor output data to be calibrated and the desired output data of the power transformer are acquired, including:
[0033] The raw output data to be calibrated and the raw expected output data of the sensor of the power transformer are obtained. The raw output data to be calibrated and the raw expected output data are normalized to obtain the output data to be calibrated and the expected output data.
[0034] In some possible implementations of the first aspect, the MLP neural network is trained using the output data to be calibrated as the input and the desired output data as the desired output, employing the Levenberg-Marquardt backpropagation algorithm to obtain a calibration model, including:
[0035] The output data to be calibrated is input into the MLP neural network, which outputs calibration output data. The total error of the MLP neural network is calculated based on the calibration output data and the expected output data. If the total error is greater than a preset threshold, the Jacobian matrix is calculated, and the parameter change of the MLP neural network is calculated based on the Jacobian matrix. The parameters of the MLP neural network are updated based on the parameter change. The latest MLP neural network outputs calibration output data, and the total error of the MLP neural network is calculated again. The learning factor is adjusted based on the trend of the total error change. This process is iterated until the total error is less than or equal to the preset threshold. The final MLP neural network is then used as the calibration model.
[0036] Secondly, embodiments of this disclosure provide an intelligent calibration device for sensor output data of power transformers, the device comprising:
[0037] The data acquisition module is used to collect the output data of sensors deployed on the power transformer in the field;
[0038] The evaluation module is used to evaluate the performance of the sensor based on the current output data and historical output data of the sensor.
[0039] The calibration module is used to calibrate the sensor's current output data and subsequent output data if the sensor's performance evaluation result is abnormal.
[0040] Thirdly, embodiments of this disclosure provide an electronic device comprising: at least one processor; and a memory communicatively connected to the at least one processor; the memory storing instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the method described above.
[0041] Fourthly, embodiments of this disclosure provide a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the methods described above.
[0042] In the embodiments of this disclosure, the output data of sensors deployed on power transformers can be analyzed in real time, the sensor performance can be evaluated using intelligent algorithms, and online calibration can be performed when the sensor performance is abnormal, thereby improving the reliability of the sensors and the accuracy of the data in the field environment.
[0043] It should be understood that the description in the Summary of the Invention is not intended to limit the key or essential features of the embodiments of this disclosure, nor is it intended to restrict the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description
[0044] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. The drawings are provided for a better understanding of the invention and are not intended to limit the scope of this disclosure. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein:
[0045] Figure 1 A flowchart is shown below illustrating an intelligent calibration method for sensor output data of a power transformer, provided by an embodiment of this disclosure.
[0046] Figure 2 A schematic diagram of the structure of an MLP neural network provided by an embodiment of the present disclosure is shown;
[0047] Figure 3 A structural diagram of an intelligent calibration device for sensor output data of a power transformer, provided by an embodiment of this disclosure, is shown.
[0048] Figure 4 A structural diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure is shown. Detailed Implementation
[0049] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.
[0050] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.
[0051] To address the problems in the background art, embodiments of this disclosure provide a method, apparatus, device, and storage medium for intelligent calibration of sensor output data for power transformers. Specifically, the method involves collecting output data from sensors deployed on the power transformer; evaluating the sensor's performance based on the collected current output data and historical output data; and calibrating the sensor's current output data and subsequent output data if the performance evaluation result is abnormal.
[0052] In this way, the output data of sensors deployed on power transformers can be analyzed in real time, the performance of sensors can be evaluated using intelligent algorithms, and online calibration can be performed when the sensor performance is abnormal, thereby improving the reliability of sensors and the accuracy of data in the field environment.
[0053] The following detailed description, with reference to the accompanying drawings and specific embodiments, illustrates a method, apparatus, device, and storage medium for intelligent calibration of sensor output data for power transformers provided by the present disclosure.
[0054] Figure 1 A flowchart illustrating an intelligent calibration method for sensor output data of a power transformer, as provided in an embodiment of this disclosure, is shown. Figure 1 As shown, the intelligent calibration method 100 for sensor output data may include the following steps:
[0055] S110 collects output data from sensors deployed on the power transformer.
[0056] In some embodiments, the output data of the sensors (such as temperature, humidity, voltage, current, and vibration sensors) can be collected and recorded periodically or continuously using a real-time data acquisition module.
[0057] S120 evaluates the sensor's performance based on the current output data and historical output data collected from the sensor.
[0058] In some embodiments, the first target output data corresponding to the current operating condition and operating environment can be queried from the historical output data of the sensor according to the current operating condition and operating environment of the sensor. The sensor's performance evaluation index (such as accuracy, stability, and drift degree) is calculated based on the sensor's current output data and the first target output data. The sensor's performance degradation score is calculated based on the sensor's performance evaluation index. If the performance degradation score is greater than a preset threshold, the sensor's performance is determined to be abnormal; otherwise, the sensor's performance is determined to be normal.
[0059] For example, the performance evaluation metrics of a sensor can be accuracy, stability, and drift, and their calculation formulas can be as follows:
[0060]
[0061] Where Accuracy represents precision; Stability represents stability; Drift represents the degree of drift; n represents the number of data points; y i This represents the i-th current output data point; y target,i This represents the i-th output data point of the first target; t represents the average value of the first target output data points;i This represents the i-th time point.
[0062] Based on this, the formula for calculating the performance degradation score can be as follows:
[0063] P=ω1Accuracy -1 +ω2Stability -1 +ω3Drift (4)
[0064] Where P represents the performance degradation score; ω1, ω2, and ω3 represent the weighting coefficients corresponding to accuracy, stability, and drift, respectively.
[0065] In other embodiments, a first target output data corresponding to the current operating condition and environment of the sensor can be queried from the historical output data of the sensor based on the current operating condition and environment of the sensor. Second target output data corresponding to the current operating condition and environment of each reference sensor can be queried from the historical output data of each reference sensor based on the current operating condition and environment of the sensor. A performance evaluation index of the sensor is calculated based on the current output data of the sensor and the first target output data. A performance evaluation index of the reference sensor is calculated based on the current output data of the reference sensor and the second target output data. A performance degradation score of the sensor is calculated based on the performance evaluation index of the sensor. A performance degradation score of the reference sensor is calculated based on the performance evaluation index of the reference sensor. The difference between the performance degradation score of the sensor and the performance degradation scores of each reference sensor is calculated. If the average difference is greater than a preset threshold, the sensor performance is determined to be abnormal; otherwise, the sensor performance is determined to be normal.
[0066] It is worth noting that the calculation formulas for the performance evaluation indicators and performance degradation scores of the reference sensor are similar to those described above, and will not be repeated here.
[0067] Alternatively, the reference sensors corresponding to the sensor can be determined in the following way:
[0068] The system acquires attribute information of the sensor and multiple candidate reference sensors deployed on the power transformer. Then, it clusters the sensor and multiple candidate reference sensors based on the acquired attribute information, and selects the candidate reference sensor in the cluster to which the sensor belongs as the corresponding reference sensor.
[0069] S130, if the sensor's performance evaluation result is abnormal, then the sensor's current output data and subsequent output data are calibrated.
[0070] In some embodiments, if the sensor's performance evaluation result is abnormal, the sensor's current output data can be fitted with the first target output data, and calibration parameters can be determined based on the fitting result. Then, the calibration parameters are used to calibrate the sensor's current output data and subsequent output data.
[0071] In other embodiments, if the performance evaluation result of the sensor is abnormal, the calibration parameters of the sensor can be calculated based on the current output data of a reference sensor whose performance degradation score is lower than a preset threshold and the current output data of the sensor. Then, the calibration parameters are used to calibrate the current output data and subsequent output data of the sensor.
[0072] The calculation process for the calibration parameters can be illustrated by the following formula:
[0073] I = {i|P} i <T} (5)
[0074]
[0075] Where I represents the set of reference sensors whose performance degradation score is below a preset threshold; T represents the preset threshold; C represents the calibration parameter; R i Let P represent the current output data of the i-th reference sensor; S represents the current output data of the sensor; P represents the current output data of the i-th reference sensor. i This represents the performance degradation score of the i-th reference sensor.
[0076] In some other embodiments, if the sensor's performance evaluation results are abnormal, a calibration model can be used to calibrate the sensor's current output data and subsequent output data.
[0077] The calibration model is trained through the following steps:
[0078] (1) Obtain the sensor output data to be calibrated and the expected output data of the power transformer.
[0079] Specifically, the raw output data to be calibrated and the raw expected output data of the sensor of the power transformer are obtained, and the raw output data to be calibrated and the raw expected output data are normalized to obtain the output data to be calibrated and the expected output data.
[0080] For example, for any sensor, its original output data to be calibrated is x', and its original expected output data is t'. Normalization is performed within the range [0,1] to facilitate subsequent operations:
[0081]
[0082] Where x is the output data to be calibrated, x′ maxx′ represents the maximum value of the original output data to be calibrated. min Let t' be the minimum value of the original output data to be calibrated, and t' be the desired output data. max t′ represents the maximum value of the original expected output data. min This represents the minimum value of the original expected output data.
[0083] The goal of calibration is to produce calibration output data that is a straight line with a unit slope. The relative error is defined as follows:
[0084] e = yt (9)
[0085] Where y represents the calibration output data corresponding to the sensor.
[0086] (2) Construct a multilayer perceptron (MLP) neural network.
[0087] The structure of an MLP neural network can be as follows: Figure 2 As shown, it includes an input layer, a hidden layer, and an output layer.
[0088] The relationship between output and input can be expressed as:
[0089]
[0090] Where x represents the input of the MLP neural network, which is the output data to be calibrated here; y represents the output of the MLP neural network, which is the output data to be calibrated here; a is the weight vector; b is the bias vector; the activation function of the hidden layer is the sigmoid function; and the logistic regression from the hidden layer to the output layer is softmax regression.
[0091] Therefore, all the parameters of an MLP neural network are simply the weights and biases between each layer. These parameters need to be determined through training.
[0092] (3) The MLP neural network is trained using the output data to be calibrated as the input and the expected output data as the expected output of the MLP neural network. The Levenberg-Marquardt backpropagation algorithm is used to obtain the calibration model.
[0093] Specifically, the output data to be calibrated is input into the MLP neural network, which outputs calibration output data. The total error of the MLP neural network is calculated based on the calibration output data and the expected output data. If the total error is greater than a preset threshold, the Jacobian matrix is calculated, and the parameter change of the MLP neural network is calculated based on the Jacobian matrix. The parameters of the MLP neural network are updated based on the parameter change. The latest MLP neural network is used to output calibration output data, and the total error of the MLP neural network is calculated again. The learning factor is adjusted based on the trend of the total error change. This process is iterated until the total error is less than or equal to the preset threshold. The final MLP neural network is then used as the calibration model.
[0094] For example, assuming the sensor has N calibration points, then to obtain its output data to be calibrated x, expected output data t, and calibration output data y, we need to:
[0095] {x1,t1,y1},{x2,t2,y2},...,{x N ,t N ,y N} (11)
[0096] (3.1) Define the error function (MSE)
[0097] For N calibration points, the total error is calculated as follows:
[0098]
[0099] (3.2) Constructing the Jacobian matrix J
[0100] The Jacobian matrix represents the partial derivative of the relative error at each calibration point with respect to each parameter. The values of each element in the matrix are:
[0101]
[0102] Among them, e i =y i -t i The relative error at each calibration point, These are the parameters of the MLP neural network.
[0103] (3.3) Calculate the parameter changes of the MLP neural network
[0104] The changes in the parameters (weights a and bias b) of the MLP neural network are:
[0105] Δw k =[J T (w k )J(w k )+μk I] -1 J T (w k )e(w k (14)
[0106] Where J is the parameter w k The calculated Jacobian matrix, μ k The learning factor controls the convergence speed and stability, where I is the identity matrix, and e(w) k )=[e1,e2,…e N ] represents the error vector.
[0107] (3.4) Update MLP neural network parameters
[0108] Update MLP neural network parameters (weights a and bias b):
[0109] w k+1 =w k +Δw k (15)
[0110] (3.5) Recalculate the total error
[0111] Re-output y using the updated MLP neural network k+1 And recalculate the total error e mse,k+1 If the error decreases, i.e., e mse,k+1 <e mse,k Then we have:
[0112] μ k+1 =μ k / β (16)
[0113] β is a constant greater than 1. Then return to (3.2) to construct the Jacobian matrix.
[0114] If the error decreases, i.e., e mse,k+1 >e mse,k Then we have:
[0115] μ k+1 =βμ k (17)
[0116] Then return to (3.2) to reconstruct the Jacobian matrix until the total error is less than or equal to the preset threshold.
[0117] Furthermore, after S130, the average value of the difference between the sensor's performance degradation score and the performance degradation scores of each reference sensor can be calculated. If the average value is greater than a preset threshold and calibration fails multiple times in a row, an alarm message is output and a maintenance prompt message is generated to guide the management personnel to perform on-site maintenance or replacement of the sensor.
[0118] In summary, this disclosure achieves at least the following technical effects:
[0119] Improve the reliability of output data: Through real-time evaluation and intelligent calibration, a self-calibration mechanism that requires no manual intervention can be achieved, effectively solving the sensor deviation problem and improving the reliability of sensor output data.
[0120] Reduced maintenance costs: No need to disassemble the sensor for offline calibration, reducing the frequency of on-site maintenance and labor costs, making it suitable for large-scale applications.
[0121] Intelligentization and automation: Utilizing machine learning algorithms to achieve adaptive calibration of sensors reduces human intervention and improves the level of intelligent operation and maintenance of power transformers.
[0122] Enhancing system safety: Monitoring the transformer's operating status through precise sensor data helps to detect potential faults in a timely manner and prevent safety accidents caused by equipment failures.
[0123] At the same time, this disclosure has at least the following potential applications:
[0124] This disclosure can be widely applied to various transformer equipment in power systems, providing effective sensor data support for operation and maintenance personnel, and improving the operational safety and maintenance efficiency of transformers. By realizing intelligent and automated sensor calibration, this disclosure is expected to be widely used in smart grids, substations, and high-voltage equipment, providing technical support for the digital transformation of the power industry.
[0125] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this disclosure is not limited to the described order of actions, because according to this disclosure, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this disclosure.
[0126] The above is an introduction to the method embodiments. The following describes the solution described in this disclosure further through device embodiments.
[0127] Figure 3 A structural diagram of an intelligent calibration device for sensor output data of a power transformer, as provided in an embodiment of this disclosure, is shown. Figure 3 As shown, the intelligent calibration device 300 for sensor output data may include:
[0128] The acquisition module 310 is used to acquire the output data of sensors deployed on the power transformer in the field.
[0129] The evaluation module 320 is used to evaluate the performance of the sensor based on the current output data and historical output data of the sensor.
[0130] The calibration module 330 is used to calibrate the current output data and subsequent output data of the sensor if the performance evaluation result of the sensor is abnormal.
[0131] Understandable, Figure 3 Each module / unit in the intelligent calibration device 300 for sensor output data shown has the ability to implement... Figure 1 The functions of each step in the intelligent calibration method 100 for sensor output data shown are explained, and their corresponding technical effects are achieved. For the sake of brevity, these will not be elaborated here.
[0132] Figure 4 A structural diagram of an exemplary electronic device capable of implementing embodiments of the present disclosure is shown. Electronic device 400 is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic device 400 may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.
[0133] like Figure 4 As shown, the electronic device 400 may include a computing unit 401, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 402 or a computer program loaded from a storage unit 408 into a random access memory (RAM) 403. The RAM 403 may also store various programs and data required for the operation of the electronic device 400. The computing unit 401, ROM 402, and RAM 403 are interconnected via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.
[0134] Multiple components in electronic device 400 are connected to I / O interface 405, including: input unit 406, such as keyboard, mouse, etc.; output unit 407, such as various types of displays, speakers, etc.; storage unit 408, such as disk, optical disk, etc.; and communication unit 409, such as network card, modem, wireless transceiver, etc. Communication unit 409 allows electronic device 400 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.
[0135] The computing unit 401 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 401 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 401 performs the various methods and processes described above, such as method 100. For example, in some embodiments, method 100 may be implemented as a computer program product, including a computer program tangibly contained in a computer-readable medium, such as storage unit 408. In some embodiments, part or all of the computer program may be loaded and / or installed on device 400 via ROM 402 and / or communication unit 409. When the computer program is loaded into RAM 403 and executed by the computing unit 401, one or more steps of method 100 described above may be performed. Alternatively, in other embodiments, the computing unit 401 may be configured to perform method 100 by any other suitable means (e.g., by means of firmware).
[0136] The various embodiments described above can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), payload programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.
[0137] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.
[0138] In the context of this disclosure, a computer-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of computer-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.
[0139] It should be noted that this disclosure also provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute method 100 and achieve the corresponding technical effects achieved by the embodiments of this disclosure in executing the method. For the sake of brevity, these will not be elaborated here.
[0140] In addition, this disclosure also provides a computer program product including a computer program that implements method 100 when executed by a processor.
[0141] To provide interaction with a user, the embodiments described above can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).
[0142] The embodiments described above can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with the implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication (e.g., a communication network) of any form or medium. Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.
[0143] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.
[0144] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.
[0145] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.
Claims
1. A method for intelligent calibration of sensor output data for power transformers, characterized in that, The method includes: Collect output data from sensors deployed on the power transformer in the field; The sensor's performance is evaluated based on the current output data and historical output data collected from the sensor. If the performance evaluation result of the sensor is abnormal, then the current output data and subsequent output data of the sensor shall be calibrated. The performance evaluation of the sensor based on the current output data and historical output data of the sensor includes: Based on the current operating condition and operating environment of the sensor, query the historical output data of the sensor to obtain the first target output data corresponding to the current operating condition and operating environment; based on the current operating condition and operating environment of each reference sensor corresponding to the sensor, query the historical output data of each reference sensor to obtain the second target output data corresponding to the current operating condition and operating environment. The performance evaluation index of the sensor is calculated based on the current output data of the sensor and the first target output data; the performance evaluation index of the reference sensor is calculated based on the current output data of the reference sensor and the second target output data. Calculate the performance degradation score of the sensor based on the performance evaluation index of the sensor; calculate the performance degradation score of the reference sensor based on the performance evaluation index of the reference sensor; Calculate the difference between the performance degradation score of the sensor and the performance degradation scores of each reference sensor, and determine whether the performance of the sensor is abnormal based on the calculated difference; Each reference sensor corresponding to the aforementioned sensor is determined in the following manner: Acquire the attribute information of the sensor and multiple candidate reference sensors deployed on the power transformer in the field; The sensor and multiple candidate reference sensors are clustered based on the acquired attribute information; Select a candidate reference sensor from the cluster to which the sensor belongs as the reference sensor corresponding to the sensor; If the performance evaluation result of the sensor is abnormal, then the current output data and subsequent output data of the sensor are calibrated, including: If the performance evaluation result of the sensor is abnormal, the current output data and subsequent output data of the sensor are calibrated using a calibration model. The calibration model is trained through the following steps: Acquire the sensor output data to be calibrated and the expected output data of the power transformer; Construct a multilayer perceptron (MLP) neural network, which includes an input layer, a hidden layer, and an output layer. The activation function of the hidden layer is the sigmoid function, and the logistic regression from the hidden layer to the output layer is softmax regression. The MLP neural network is trained using the output data to be calibrated as the input and the desired output data as the desired output. The Levenberg-Marquardt backpropagation algorithm is used to obtain the calibration model. The process of acquiring the sensor output data to be calibrated and the expected output data of the power transformer includes: The raw output data to be calibrated and the raw expected output data of the sensor of the power transformer are obtained. The raw output data to be calibrated and the raw expected output data are normalized to obtain the output data to be calibrated and the expected output data. The calibration model is obtained by using the output data to be calibrated as the input of the MLP neural network, the expected output data as the expected output of the MLP neural network, and training the MLP neural network using the Levenberg-Marquardt backpropagation algorithm, including: The output data to be calibrated is input into the MLP neural network, which outputs calibration output data. The total error of the MLP neural network is calculated based on the calibration output data and the expected output data. If the total error is greater than a preset threshold, the Jacobian matrix is calculated, and the parameter change of the MLP neural network is calculated based on the Jacobian matrix. The parameters of the MLP neural network are updated based on the parameter change. The latest MLP neural network outputs calibration output data, and the total error of the MLP neural network is calculated again. The learning factor is adjusted based on the trend of the total error change. This process is iterated until the total error is less than or equal to the preset threshold. The final MLP neural network is then used as the calibration model.
2. A smart calibration device for sensor output data of power transformers, characterized in that, The apparatus is used to perform the method of claim 1, comprising: The data acquisition module is used to collect the output data of sensors deployed on the power transformer in the field; An evaluation module is used to evaluate the performance of the sensor based on the current output data and historical output data of the sensor. A calibration module is used to calibrate the current output data and subsequent output data of the sensor if the performance evaluation result of the sensor is abnormal.