CN114636882A - A transformer bias detection system and method based on digital twin - Google Patents

Abstract

The invention discloses a transformer bias detection system and method based on digital twin. The method adopts the digital twin technology to establish a simulation transformer model in a workstation, collects transformer data information, and remotely imports it into the simulation transformer to form a digital twin transformer. The twin transformer and the transformer information are consistent, and the data information is fused together through the multi-sensor data fusion technology to jointly judge the bias information of the transformer, and establish the characteristic curve of the neutral point current and the magnetic field strength of the transformer core in the workstation. The data and bias information are superimposed on the digital twin transformer, so that users do not need to go to the site to view the bias information of the transformer and other characteristic changes caused by the bias through the workstation, and view the neutral point current through the information superimposed on the digital twin transformer. The influence of the transformer bias, and the influence of the transformer bias on the transformer.

Description

A transformer bias detection system and method based on digital twin

technical field

The invention belongs to the field of transformer detection, and in particular relates to a transformer bias detection system and method based on digital twins.

Background technique

The DC bias of the transformer is mainly caused by the DC current intruding into the AC system by the operation of the high-voltage DC transmission unipolar circuit, the geomagnetically induced current caused by the solar magnetic storm, and the inflow into the power grid caused by the inverter, controller and other power electronic devices. A DC component, etc., causes a DC current to flow into the transformer windings. Especially when the high-voltage power grid operates in a single-pole ground loop mode, due to a certain potential difference between each grounding point, this potential difference will cause the neutral line on one side of the transformer to inject a certain DC current into the transformer. Moreover, with the development of urban rail transit, the stray current leaking into the ground when the locomotive is running intrudes into the neutral point of the transformer along the line, which will offset the working magnetization curve of the transformer core and cause the DC bias effect of the transformer. Compared with the geomagnetic field changes caused by HVDC unipolar operation and solar magnetic storms, the geomagnetic field changes caused by rail transit stray currents have obvious periodicity and are more common.

After the DC flows into the transformer, a DC magnetic flux will be generated in the iron core, resulting in a series of problems such as half-cycle saturation of the transformer core, serious distortion of the excitation current, consumption of a large amount of reactive power, and increased vibration and noise. The problem of vibration and noise caused by DC bias has gradually attracted people's attention. The vibration and noise of transformers seriously affect people's normal life and physical and mental health. Mechanical vibration performance is also one of the technical indicators for evaluating the working state of a transformer. A transformer under abnormal vibration for a long time will cause its structure to loosen and its mechanical strength to decrease. In severe cases, it will also cause structural wear and insulation strength. Secondly, the DC bias of the transformer will cause the temperature of the transformer to rise. If the transformer runs at a high temperature for a long time, it will shorten the life of the internal insulating cardboard, make the insulating cardboard brittle, prone to rupture, and lose its proper insulating effect. , causing accidents such as breakdown, serious aging of winding insulation, and accelerated deterioration of insulating oil, affecting service life.

In the prior art, in order to ensure the stable operation of the transformer, the transformer is generally shut down for maintenance on a regular basis, and the temperature change of the transformer and the vibration of the transformer are checked on-site mainly through a manual hand-held infrared thermal imager. The second is to monitor the bias of the transformer through various sensors, and transmit it to the monitoring center for monitoring through wired or wireless transmission. However, manual on-site inspection of the transformer operation will increase the labor intensity of the staff, and the cost will be high and the maintenance period will be long. And there is a certain danger. The existing sensor detection method is relatively simple, which is prone to misjudgment. Once the sensor has a problem, it cannot be detected, and the method of sensor detection is to digitally display the collected data information, and users cannot more intuitively. Check the transformer bias.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a transformer bias detection system and method based on digital twin, which can use digital twin technology to establish a simulated transformer model in 3D MAX software according to transformer parameters, install magnetic field sensor, current sensor, temperature sensor on the transformer Sensors, vibration sensors, and noise detectors are used to monitor the transformer's core magnetic field change information, neutral point current information, primary current information, secondary current information, temperature change information, vibration information, and generated noise information. Through multi-sensor data The fusion technology fuses these data information together to judge the bias information of the transformer, transmits the detected data information to the monitoring center, and automatically imports the data information into the simulation transformer model in the monitoring center to form a digital twin transformer. The core magnetic field change information, neutral point current information, primary side current information, secondary side current information, temperature change information, vibration information, and generated noise information are consistent with the transformer information, and can be superimposed on the digital twin transformer for display without the need for users to If you go to the site frequently, you can check the bias information of the transformer and other characteristic changes caused by the bias through the computer, and check the influence of the neutral point current on the bias of the transformer and the bias of the transformer through the information superimposed on the digital twin transformer. impact on the transformer.

In order to achieve the above purpose, the technical scheme adopted in the present invention is: a method for detecting bias magnetism of transformers based on digital twin, the method steps are as follows:

In step 1, the change information of the iron core magnetic field, the neutral point current information, the primary voltage information, the primary current information, the secondary current information, the temperature change information, the vibration information and the generated noise information of the transformer are detected by the sensor;

Step 2, transmit the detected data information to the industrial computer, perform signal amplification, filtering, AD conversion and data preprocessing on the collected data information, and then remotely transmit the preprocessed data information to the workstation;

Step 3, the workstation establishes a simulation transformer model according to the transformer parameters through the digital twin technology, and then imports the preprocessed data information into the simulation transformer model to form a digital twin transformer;

Step 4: Through multi-sensor data fusion technology, the core magnetic field change information, primary current information, secondary current information, temperature change information, vibration information and generated noise information are fused together to judge the bias information of the transformer, and superimposed on the transformer bias information. The digital twin transformer is displayed on the workstation;

Step 5: Establish the characteristic curve between the neutral point current and the magnetic field strength of the transformer core in the workstation, and superimpose the characteristic curve between the neutral point current and the core magnetic field strength on the digital twin transformer. information to view the effect of neutral point current on transformer bias.

Further, in step 4, the multi-sensor data fusion technology is used to determine the bias information scheme of the transformer as follows: the bias of the transformer will cause the magnetic field strength of the transformer core to change, the vibration and noise will increase, and its temperature will also increase. The magnetism will cause the change of the excitation current, and the change of the excitation current will cause the change of the primary current and the secondary current of the transformer. Therefore, the change information of the iron core magnetic field, the primary current information, the secondary current information, and the temperature change generated by the bias magnetization of the transformer can be integrated. Information, vibration information, noise information and other parameter information to judge the bias of the transformer, by establishing a gradient data fusion model to detect the bias of the transformer, first set up a data fusion center, where the sensors include voltage sensors, current sensors, magnetic field sensors , temperature sensor, vibration sensor and noise detector, the current sensor includes a first current sensor and a second current sensor, wherein the first current sensor is used to detect the primary current information and secondary current information of the transformer, and the second current sensor is used to detect Transformer neutral point current information, set up an intermediate station between the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, noise detector and data fusion center, in the intermediate station, the first current sensor, magnetic field sensor , temperature sensor, vibration sensor and the data information collected by the noise detector are correlated, and then data-level fusion is performed to achieve local fusion, and feature extraction is performed on the locally fused data information in the data fusion center. Attributes are judged, and then the judged data information is associated for decision-level fusion, thereby realizing global fusion, and outputting the fusion result to judge the bias status of the transformer.

Further, in step 5, after the second current sensor detects the current at the neutral point, it is compared with the data information collected by the magnetic field sensor to determine the influence of the stray current on the bias magnetization of the transformer.

Further, in the process of judging the influence of the neutral point current on the bias magnetization of the transformer, the magnetic field change information of the iron core is detected by the magnetic field sensor, and the neutral point current information detected by the second current sensor and the magnetic field sensor and the magnetic field change of the iron core are detected. The information is transmitted to the workstation together, and the characteristic curve between the neutral point current and the magnetic field strength of the transformer core is established in the workstation, and the characteristic curve between the neutral point current and the core magnetic field strength is superimposed on the digital twin transformer, and the neutral point moves upward. When entering the stray current, the information superimposed on the digital twin transformer can be used to check the influence of the neutral point stray current on the transformer bias.

Further, when building a digital twin transformer in step 3, first import the specific parameters of the transformer into the BIM software Revit to generate a .RVT file, and import the .RVT file into the 3D max software for editing. In the 3D max software, the BIM is converted by element type. The same parts of the model are merged into one object, the redundant points, lines and surfaces are deleted, the merged model is given material texture and characteristic parameters, the edited model is exported in 3D max software, loaded into the engine, and the secondary material is carried out according to the transformer material properties Adjust the effect, and then render a virtual transformer with the same parameters as the transformer, establish a virtual industrial computer, and then pass the data information collected by the voltage sensor, current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector through the physical industrial computer. After aggregation, it is transmitted to the virtual industrial computer, and the data interface between the virtual industrial computer and the virtual transformer is established. According to the collected data information, data analysis and scene rendering are performed in the virtual transformer, and finally a digital twin transformer is generated, so that the digital twin transformer can receive The obtained data information is consistent with the real characteristic data of the transformer.

Further, in the process of data-level fusion of the data information collected by the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector, the collected data information is locally fused by using the fuzzy logic control method, and X ₁ , X ₂ , X ₃ , X ₄ , and X ₅ are respectively used as fuzzy sets of the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector, wherein X ₁ includes the data collected by the first current sensor at each moment. The primary side current information and the secondary side current information, wherein the primary side current information and the secondary side current information collected at the kth moment are expressed as x ₁ (k); X ₂ includes the core magnetic field change information collected by the magnetic field sensor at each moment, Among them, the magnetic field change information of the iron core collected at the kth time is expressed as x ₂ (k); X ₃ includes the temperature change information collected by the temperature sensor at each time, and the temperature change information collected at the kth time is expressed as x ₃ (k) X ₄ includes the vibration information collected at each moment of the vibration sensor, wherein the vibration information collected at the kth moment is represented as x ₄ (k); X ₅ includes the noise information collected by the noise detector at every moment, and wherein the kth moment The collected noise information is expressed as x ₅ (k); the similarity between X ₁ , X ₂ , X ₃ , X ₄ , and X ₅ is evaluated by the degree of closeness to identify the bias situation. At the kth moment, each sensor The data matrix between is:

α_mn(k)表示第一电流传感器、磁场感应器、温度传感器、振动传感器、噪声检测仪中任意两个传感器之间的贴近度表达式，x_m(k)、x_n(k)分别表示第一电流传感器、磁场感应器、温度传感器、振动传感器、噪声检测仪在k时刻采集到的数据信息，当m＝1、n＝2时，x₁(k)表示第一电流传感器在k时刻采集到的数据信息，x₂(k)表示磁场感应器在k时刻采集到的数据信息，当α_mn(k)小于预设值M时，则认为x_m(k)和x_n(k)不相似，那么就可以得出α_mn(k)＝0，即有：

The elements in matrix A are the expressions of proximity between each sensor, namely:

α _mn (k) represents the closeness expression between any two sensors in the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector, and x _m (k) and x _n (k) represent respectively Data information collected by the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector at time k. When m=1, n=2, x ₁ (k) represents the first current sensor at time k. The collected data information, x ₂ (k) represents the data information collected by the magnetic field sensor at time k, when α _mn (k) is less than the preset value M, it is considered that x _m (k) and x _n (k) are not similar, then it can be concluded that α _mn (k) = 0, that is:

式中α_ij(k)为k时刻传感器在数据矩阵中的第i行，第j列数据信息，α_ij(k)为x₁(k)、x₂(k)、x₃(k)、x₄(k)或x₅(k)。The calculation formula for the consistency measurement of the sampling values of the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector is:

where α _ij (k) is the data information of the ith row and jth column of the sensor in the data matrix at time k, and α _ij (k) is x ₁ (k), x ₂ (k), x ₃ (k), x ₄ (k) or x ₅ (k).

式中x₁(k)表示为k时刻第一电流传感器采集到的数据信息，x₂(k)表示为k时刻磁场感应器采集到的数据信息，x₃(k)表示为k时刻温度传感器采集到的数据信息，x₄(k)表示为k时刻振动传感器采集到的数据信息，x₅(k)表示为k时刻噪声检测仪采集到的数据信息，ω_i(k)表示数据矩阵中的第i行权重，根据信息分享原理，最优融合估计的信息量之和可分解为若干个测量数据的信息量之和，即一个信息可被若干个子系统所分享，有

将第一电流传感器、磁场感应器、温度传感器、振动传感器、噪声检测仪的模糊贴近度做归一化处理，得到各自的权重

式中，c₁(k)，c₂(k)，c₃(k)，c₄(k)，c₅(k)表示一组非负数。The calculation formula of data fusion is expressed as:

where x ₁ (k) is the data information collected by the first current sensor at time k, x ₂ (k) is the data information collected by the magnetic field sensor at time k, and x ₃ (k) is the temperature sensor at time k The collected data information, x ₄ (k) represents the data information collected by the vibration sensor at time k, x ₅ (k) represents the data information collected by the noise detector at time k, ω _i (k) represents the data in the matrix The weight of the i-th row of , according to the principle of information sharing, the sum of the information volume estimated by the optimal fusion can be decomposed into the sum of the information volume of several measurement data, that is, one information can be shared by several subsystems, there are

Normalize the fuzzy closeness of the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector to obtain their respective weights

In the formula, c ₁ (k), c ₂ (k), c ₃ (k), c ₄ (k), and c ₅ (k) represent a set of non-negative numbers.

Further, in the decision-level fusion process of the data fusion center, the transformer bias conditions are divided into four levels: normal F ₁ , mild F ₂ , severe F ₃ , and severe F ₄ . Let Θ be an identification frame, and Ω( Θ) represents the set of all subsets in Θ, then the transformer bias level is expressed as:

Ω(Θ) ₌ {φ,{F1 _} ,{F2 _} ,{F3},{ _{F4 }} ,{F1,F2 _} , _{ F2, _F3 },...},φ is an empty set, assuming Q={F ₁ , F ₂ }∈Θ, it means that Q is either normal F ₁ or mild F ₂ , that is, Ω(Θ) is any one of these states;

式中G(Q)称为基本概率赋值函数，表示状态识别模型对一个特定状态的置信度，用信度函数Bel来表示识别框架Θ上的映射Bel:Q→[0，1]，满足：

再用似真函数Pl表示下一个映射Pl:Ω(Θ)→[0，1]，且满足：

式中Q^c表示Q的补集，Pl(Q)表示命题Q的似真度，其中Pl(Q)＝0表示证据拒绝Q，Pl(Q)＝1表示证据支持Q；After determining the recognition frame Θ, define a set function G that maps Ω(Θ) to a number on the interval [0,1], expressed as:

In the formula, G(Q) is called the basic probability assignment function, which represents the confidence of the state recognition model for a specific state, and the confidence function Bel is used to represent the mapping Bel on the recognition framework Θ: Q→[0, 1], which satisfies:

Then use the plausibility function Pl to represent the next mapping Pl:Ω(Θ)→[0,1], and satisfy:

where Q ^c represents the complement of Q, Pl(Q) represents the plausibility of proposition Q, where Pl(Q)=0 means that the evidence rejects Q, and Pl(Q)=1 means that the evidence supports Q;

K表示归一化后的系数，表示为：

G₁(Q_i)为第1组证据Q_i的基本概率赋值函数，G₂((Ω(Θ))_j)为第2组证据(Ω(Θ))_j的基本概率赋值函数，G(Q)反映了两个证据之间的冲突程度，当K≠0时，用正交和的形式表示两组证据间的组合；当K＝0，表示不存在正交和，两个证据之间属于矛盾关系；Let G _i represent the basic probability assignment function of the ith group of evidences defined on the recognition framework Θ, where i=1, 2...n, the basic probability assignment function after combining any two groups of evidences is expressed as :

K represents the normalized coefficient, which is expressed as:

G ₁ (Q _i ) is the basic probability assignment function of the first group of evidence Qi _, G ₂ ((Ω(Θ)) _j ) is the basic probability assignment function of the second group of evidence (Ω(Θ)) _j , G( Q) reflects the degree of conflict between the two evidences. When K≠0, the combination between the two sets of evidences is expressed in the form of an orthogonal sum; when K=0, it means that there is no orthogonal sum, and the a contradictory relationship;

即可计算出E₁、E₂、E₃、E₄、E₅融合后的基本概率赋值函数，根据融合后的基本概率赋值函数来判断变压器偏磁情况处于哪一种状态，例如在识别框架Θ上构建E₁的基本概率赋值函数，如表1所示，其中对样本1进行分析，样本1的基本概率赋值函数最大值在F₁，表明变压器偏磁情况处于正常状态。After that, the basic probability assignment functions of the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector are constructed, and the identification results of the first current sensor, magnetic field sensor, temperature sensor, vibration sensor and noise detector are used as five Evidence E ₁ , E ₂ , E ₃ , E ₄ , E ₅ , and four levels of normal F ₁ , mild F ₂ , severe F ₃ , and severe F ₄ are used as the identification domain, and Θ={F ₁ ,F ₂ , F ₃ , F ₄ } constitute the common identification frame of E ₁ , E ₂ , E ₃ , E ₄ , and E ₅ , and construct the basic probability assignment function of each evidence on the identification frame Θ. According to the formula

The basic probability assignment function after the fusion of E ₁ , E ₂ , E ₃ , E ₄ , and E ₅ can be calculated. According to the basic probability assignment function after fusion, it can be judged which state the transformer bias is in. For example, in the identification framework The basic probability assignment function of E ₁ is constructed on Θ, as shown in Table 1. Sample 1 is analyzed, and the maximum value of the basic probability assignment function of sample 1 is F ₁ , indicating that the transformer bias is in a normal state.

Table ₁ Basic probability assignment function of E1

sample F₁ F₂ F₃ F₄ Sample category 1 0.69 0.15 0.06 0.10 F₁ 2 0.45 0.43 0.06 0.06 F₂ 3 0.09 0.02 0.77 0.12 F₃ 4 0.11 0.05 0.03 0.81 F₄

A digital twin-based transformer bias detection system includes a transformer, a voltage sensor, a current sensor, a magnetic field sensor, a temperature sensor, a vibration sensor, a noise detector, an industrial computer, a 5G module and a workstation. The voltage sensor and current sensor, the voltage sensor and current sensor are respectively connected to the bus of the transformer, the industrial computer is installed on the outer surface of the transformer, the 5G module is installed on the industrial computer, the magnetic field sensor, temperature sensor, vibration sensor, noise detection are installed next to the transformer core The instrument is installed inside the transformer, and the magnetic field sensor, temperature sensor, vibration sensor, noise detector, current sensor and voltage sensor are respectively connected to the industrial computer, and the industrial computer is connected to the workstation through the 5G module.

Further, the current sensor includes a first current sensor and a second current sensor, there are six first current sensors, which are respectively used to detect the primary current of the transformer and the secondary current of the transformer, and one second current sensor is used to detect the current in the transformer. There are three voltage sensors, which are used to detect the primary voltage of the transformer bus.

Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention uses the digital twin technology to reflect the operation of the transformer under the magnetic bias condition, and uses the 5G technology to convert the information on the magnetic field change of the iron core and the neutral point current information in the transformer. , primary side current information, secondary side current information, temperature change information, vibration information, and generated noise information are transmitted to the digital twin transformer, so that the information of the digital twin transformer is consistent with that of the transformer, and can be superimposed on the digital twin transformer for display. , users can check the bias information of the transformer and other characteristic changes caused by the bias through the computer without frequent trips to the site, and check the influence of the neutral point current on the bias of the transformer through the information superimposed on the digital twin transformer, and The effect of transformer bias on the transformer. Through the analysis of transformer simulation data, the efficiency of fault monitoring is improved. 2. Through multi-sensor data fusion technology, these data information are fused together to judge the bias information of the transformer. The system no longer relies on the data information collected by a certain sensor to judge the bias of the transformer, even if one of the sensors fails. It does not affect the detection and effectively improves the recognition accuracy. 3. By establishing the characteristic curve of the neutral point current and the magnetic field strength of the transformer core in the workstation, and superimposing the characteristic curve between the neutral point current and the core magnetic field strength on the digital twin transformer, and by superimposing the characteristic curve on the digital twin transformer. information to view the effect of neutral point current on transformer bias. The relationship between the stray current and the transformer bias can be matched, especially for the transformer next to the rail transit, which can provide an effective maintenance reference for the staff.

Description of drawings

Fig. 1 is the connection block diagram of each component of the present invention;

Fig. 2 is the structural representation of the present invention;

Fig. 3 is the flow chart of digital twin transformer modeling of the present invention;

FIG. 4 is a flow chart of the multi-sensor data fusion of the present invention.

In the picture: 1. Voltage sensor, 2. Current sensor, 3. Magnetic field sensor, 4. Temperature sensor, 5. Vibration sensor, 6. Noise detector, 7. Industrial computer, 8.5G module.

Detailed ways

1 to 4 , a method for detecting bias magnetism of a transformer based on a digital twin, the method steps are as follows:

In step 1, the change information of the iron core magnetic field, the neutral point current information, the primary voltage information, the primary current information, the secondary current information, the temperature change information, the vibration information and the generated noise information of the transformer are detected by the sensor;

Step 2, then transmit the detected data information to the industrial computer 7, perform signal amplification, filtering, AD conversion and data preprocessing on the collected data information, and then remotely transmit the preprocessed data information to the workstation;

Step 3, the workstation establishes a simulation transformer model according to the transformer parameters through the digital twin technology, and then imports the preprocessed data information into the simulation transformer model to form a digital twin transformer,

Step 4: Through multi-sensor data fusion technology, the core magnetic field change information, primary current information, secondary current information, temperature change information, vibration information and generated noise information are fused together to judge the bias information of the transformer, and superimposed on the transformer bias information. The digital twin transformer is displayed on the workstation; therefore, when judging the bias of the transformer, it no longer relies on the data information collected by a certain sensor to judge the bias of the transformer, which effectively improves the recognition accuracy;

Step 5: Establish the characteristic curve between the neutral point current and the magnetic field strength of the transformer core in the workstation, and superimpose the characteristic curve between the neutral point current and the core magnetic field strength on the digital twin transformer. information to view the effect of neutral point current on transformer bias.

As shown in Figure 4, among them, in step 4, the bias information scheme of the transformer is judged by the multi-sensor data fusion technology: the bias magnetization of the transformer will cause the magnetic field strength of the transformer core to change, the vibration and noise will increase, and its temperature will also rise. High, because the transformer bias will cause the excitation current to change, and the excitation current change will cause the transformer primary current and secondary current to change. The current information, temperature change information, vibration information, noise information and other parameter information can be used to judge the bias of the transformer, and the bias of the transformer is detected by establishing a gradient data fusion model. First set up a data fusion center, where the sensors include voltage sensors 1, 1 and 2. A current sensor 2, a magnetic field sensor 3, a temperature sensor 4, a vibration sensor 5 and a noise detector 6, the current sensor 2 includes a first current sensor and a second current sensor, wherein the first current sensor is used to detect the transformer primary current information and Secondary current information, the second current sensor is used to detect the neutral point current information of the transformer, and a In the intermediate station, the data information collected by the first current sensor, the magnetic field sensor 3, the temperature sensor 4, the vibration sensor 5 and the noise detector 6 is associated in the intermediate station, and then data-level fusion is performed to achieve local fusion, and in the middle station. The data fusion center performs feature extraction on the locally fused data information, judges the data attributes according to the feature extraction results, and then associates the judged data information for decision-level fusion, thereby realizing global fusion, and outputs the fusion results. To judge the bias of the transformer.

Wherein, in step 5, after the second current sensor detects the current at the neutral point, it is compared with the data information collected by the magnetic field sensor 3 to determine the influence of the stray current on the bias of the transformer.

Among them, in the process of judging the influence of the neutral point current on the bias magnetization of the transformer, the magnetic field sensor 3 is used to detect the change information of the iron core magnetic field, and the neutral point current information detected by the second current sensor and the magnetic field sensor and the change of the iron core magnetic field are detected. The information is transmitted to the workstation together, and the characteristic curve between the neutral point current and the magnetic field strength of the transformer core is established in the workstation, and the characteristic curve between the neutral point current and the core magnetic field strength is superimposed on the digital twin transformer, and the neutral point moves upward. When entering stray current, the information superimposed on the digital twin transformer can be used to check the influence of the neutral point stray current on the bias magnetism of the transformer. It is mainly to check the change of the core magnetic field strength of the transformer along the rail transit line when the locomotive passes, and then obtain The influence of the stray current generated when the locomotive is running on the transformer bias.

As shown in Figure 3, when building a digital twin transformer in step 3, first import the specific parameters of the transformer into the BIM software Revit to generate a .RVT file, and import the .RVT file into the 3D max software for editing. In the 3D max software Merge the same parts of the BIM model into one object by element type, delete redundant points, lines and surfaces, assign the merged model to material texture and feature parameters, export the edited model in 3D max software, and load it into the engine. The properties of the secondary material effect are adjusted, and then a virtual transformer with the same parameters as the transformer is rendered, a virtual industrial computer is established, and then the voltage sensor 1, current sensor 2, magnetic field sensor 3, temperature sensor 4, vibration sensor 5, noise detection The data information collected by the instrument 6 is aggregated by the physical industrial computer 7 and then transmitted to the virtual industrial computer, establishing a data interface between the virtual industrial computer and the virtual transformer, and performing data analysis and scene rendering in the virtual transformer according to the collected data information. , and finally generate a digital twin transformer, so that the data information received by the digital twin transformer is consistent with the real feature data of the transformer. The data information collected by the voltage sensor 1 and the current sensor 2 used to detect the bus will also be transmitted to the digital twin transformer. At the same time, the operating voltage and current input of the digital twin transformer and the transformer are consistent, so as to ensure all aspects of the digital twin transformer and the transformer. Data consistency.

Among them, the data information collected by the first current sensor, the magnetic field sensor 3, the temperature sensor 4, the vibration sensor 5, and the noise detector 6 is locally fused by using the fuzzy logic control method during the data-level fusion process. , take X ₁ , X ₂ , X ₃ , X ₄ , and X ₅ as the fuzzy sets of the first current sensor, magnetic field sensor 3, temperature sensor 4, vibration sensor 5, and noise detector 6, respectively, where X ₁ includes the first The primary current information and secondary current information collected by the current sensor at each moment, wherein the primary current information and secondary current information collected at the kth moment are represented as x ₁ (k); X ₂ includes the magnetic field sensor 3 at each moment The collected iron core magnetic field change information, wherein the iron core magnetic field change information collected at the kth time is represented as x ₂ (k); X ₃ includes the temperature change information collected by the temperature sensor 4 at each time, wherein the temperature collected at the kth time Change information is expressed as x ₃ (k); X ₄ includes the vibration information collected at each moment of the vibration sensor 5, wherein the vibration information collected at the kth moment is expressed as x ₄ (k); X ₅ includes the noise detector 6 at every moment The generated noise information is collected, and the generated noise information collected at the kth moment is represented as x ₅ (k); the bias magnets are identified between X ₁ , X ₂ , X ₃ , X ₄ , and X ₅ by evaluating the closeness The similarity of the situation, at the kth moment, the data matrix between each sensor is:

矩阵A中的元素即为各个传感器之间的贴近度表达式，即：

α_mn(k)表示第一电流传感器、磁场感应器、温度传感器、振动传感器、噪声检测仪中任意两个传感器之间的贴近度表达式，x_m(k)、x_n(k)分别表示第一电流传感器、磁场感应器、温度传感器、振动传感器、噪声检测仪在k时刻采集到的数据信息，当m＝1、n＝2时，x₁(k)表示第一电流传感器在k时刻采集到的数据信息，x₂(k)表示磁场感应器在k时刻采集到的数据信息，当α_mn(k)小于预设值M时，则认为x_m(k)和x_n(k)不相似，那么就可以得出α_mn(k)＝0，即有：

The elements in matrix A are the expressions of proximity between each sensor, namely:

α _mn (k) represents the closeness expression between any two sensors in the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector, and x _m (k) and x _n (k) represent respectively Data information collected by the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector at time k. When m=1, n=2, x ₁ (k) represents the first current sensor at time k. The collected data information, x ₂ (k) represents the data information collected by the magnetic field sensor at time k, when α _mn (k) is less than the preset value M, it is considered that x _m (k) and x _n (k) are not similar, then it can be concluded that α _mn (k) = 0, that is:

式中α_ij(k)为k时刻传感器在数据矩阵中的第i行，第j列数据信息，α_ij(k)为x₁(k)、x₂(k)、x₃(k)、x₄(k)或x₅(k)。The calculation formula for the consistency measurement of the sampling values of the first current sensor, the magnetic field sensor 3, the temperature sensor 4, the vibration sensor 5, and the noise detector 6 is:

where α _ij (k) is the data information of the ith row and jth column of the sensor in the data matrix at time k, and α _ij (k) is x ₁ (k), x ₂ (k), x ₃ (k), x ₄ (k) or x ₅ (k).

式中x₁(k)表示为k时刻第一电流传感器采集到的数据信息，x₂(k)表示为k时刻磁场感应器采集到的数据信息，x₃(k)表示为k时刻温度传感器采集到的数据信息，x₄(k)表示为k时刻振动传感器采集到的数据信息，x₅(k)表示为k时刻噪声检测仪采集到的数据信息，ω_i(k)表示数据矩阵中的第i行权重，根据信息分享原理，最优融合估计的信息量之和可分解为若干个测量数据的信息量之和，即一个信息可被若干个子系统所分享，有

将第一电流传感器、磁场感应器、温度传感器、振动传感器、噪声检测仪的模糊贴近度做归一化处理，得到各自的权重

式中，c₁(k)，c₂(k)，c₃(k)，c₄(k)，c₅(k)表示一组非负数。The calculation formula of data fusion is expressed as:

where x ₁ (k) is the data information collected by the first current sensor at time k, x ₂ (k) is the data information collected by the magnetic field sensor at time k, and x ₃ (k) is the temperature sensor at time k The collected data information, x ₄ (k) represents the data information collected by the vibration sensor at time k, x ₅ (k) represents the data information collected by the noise detector at time k, ω _i (k) represents the data in the matrix The weight of the i-th row of , according to the principle of information sharing, the sum of the information volume estimated by the optimal fusion can be decomposed into the sum of the information volume of several measurement data, that is, one information can be shared by several subsystems, there are

Normalize the fuzzy closeness of the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector to obtain their respective weights

In the formula, c ₁ (k), c ₂ (k), c ₃ (k), c ₄ (k), and c ₅ (k) represent a set of non-negative numbers.

Among them, in the decision-level fusion process of the data fusion center, the transformer bias conditions are divided into four levels: normal F ₁ , mild F ₂ , severe F ₃ , and severe F ₄ , and set Θ as an identification framework, Ω(Θ ) represents the set of all subsets in Θ, then the transformer bias level is expressed as:

Ω(Θ) ₌ {φ,{F1 _} ,{F2 _} ,{F3},{ _{F4 }} ,{F1,F2 _} , _{ F2, _F3 },...},φ is an empty set, assuming Q={F ₁ , F ₂ }∈Θ, it means that Q is either normal F ₁ or mild F ₂ , that is, Ω(Θ) is any one of these states;

式中G(Q)称为基本概率赋值函数，表示状态识别模型对一个特定状态的置信度，用信度函数Bel来表示识别框架Θ上的映射Bel:Q→[0，1]，满足：

再用似真函数Pl表示下一个映射Pl:Ω(Θ)→[0，1]，且满足：

式中Q^c表示Q的补集，Pl(Q)表示命题Q的似真度，其中Pl(Q)＝0表示证据拒绝Q，Pl(Q)＝1表示证据支持Q；After determining the recognition frame Θ, define a set function G that maps Ω(Θ) to a number on the interval [0,1], expressed as:

In the formula, G(Q) is called the basic probability assignment function, which represents the confidence of the state recognition model for a specific state, and the confidence function Bel is used to represent the mapping Bel on the recognition framework Θ: Q→[0, 1], which satisfies:

Then use the plausibility function Pl to represent the next mapping Pl:Ω(Θ)→[0,1], and satisfy:

where Q ^c represents the complement of Q, Pl(Q) represents the plausibility of proposition Q, where Pl(Q)=0 means that the evidence rejects Q, and Pl(Q)=1 means that the evidence supports Q;

K表示归一化后的系数，表示为：

G₁(Q_i)为第1组证据Q_i的基本概率赋值函数，G₂((Ω(Θ))_j)为第2组证据(Ω(Θ))_j的基本概率赋值函数，G(Q)反映了两个证据之间的冲突程度，当K≠0时，用正交和的形式表示两组证据间的组合；当K＝0，表示不存在正交和，两个证据之间属于矛盾关系；Let G _i represent the basic probability assignment function of the ith group of evidences defined on the recognition framework Θ, where i=1, 2...n, the basic probability assignment function after combining any two groups of evidences is expressed as :

K represents the normalized coefficient, which is expressed as:

G ₁ (Q _i ) is the basic probability assignment function of the first group of evidence Qi _, G ₂ ((Ω(Θ)) _j ) is the basic probability assignment function of the second group of evidence (Ω(Θ)) _j , G( Q) reflects the degree of conflict between the two evidences. When K≠0, the combination between the two sets of evidences is expressed in the form of an orthogonal sum; when K=0, it means that there is no orthogonal sum, and the a contradictory relationship;

即可计算出E₁、E₂、E₃、E₄、E₅融合后的基本概率赋值函数，根据融合后的基本概率赋值函数来判断变压器偏磁情况处于哪一种状态，例如在识别框架Θ上构建E₁的基本概率赋值函数，如表1所示，其中对样本1进行分析，样本1的基本概率赋值函数最大值在F₁，表明变压器偏磁情况处于正常状态。Then construct the basic probability assignment function of the first current sensor, magnetic field sensor 3, temperature sensor 4, vibration sensor 5, noise detector 6, first current sensor, magnetic field sensor 3, temperature sensor 4, vibration sensor 5 and noise The recognition results of the detector 6 are used as five evidences E ₁ , E ₂ , E ₃ , E ₄ , and E ₅ , and four levels of normal F ₁ , mild F ₂ , severe F ₃ , and severe F ₄ are used as the recognition domains. The common recognition frame of E ₁ , E ₂ , E ₃ , E ₄ , and E ₅ is formed by Θ={F ₁ , F ₂ , F ₃ , F ₄ }, and the basic probability assignment function of each evidence is constructed on the recognition frame Θ, According to the formula

The basic probability assignment function after the fusion of E ₁ , E ₂ , E ₃ , E ₄ , and E ₅ can be calculated. According to the basic probability assignment function after fusion, it can be judged which state the transformer bias is in. For example, in the identification framework The basic probability assignment function of E ₁ is constructed on Θ, as shown in Table 1. Sample 1 is analyzed, and the maximum value of the basic probability assignment function of sample 1 is F ₁ , indicating that the transformer bias is in a normal state.

Table ₁ Basic probability assignment function of E1

sample F₁ F₂ F₃ F₄ Sample category 1 0.69 0.15 0.06 0.10 F₁ 2 0.45 0.43 0.06 0.06 F₂ 3 0.09 0.02 0.77 0.12 F₃ 4 0.11 0.05 0.03 0.81 F₄

1 and 2, a digital twin-based transformer bias detection system includes a transformer, a voltage sensor 1, a current sensor 2, a magnetic field sensor 3, a temperature sensor 4, a vibration sensor 5, a noise detector 6, and an industrial computer. 7. 5G module 8 and workstation, install voltage sensor 1 on the incoming line of the transformer bus, and install voltage sensor 1 and current sensor 2 on the outer surface of the transformer. Voltage sensor 1 and current sensor 2 are respectively connected to the bus of the transformer. Industrial control The machine 7 is installed on the outer surface of the transformer, the 5G module 8 is installed on the industrial computer 7, the magnetic field sensor 3 is installed next to the transformer core, the temperature sensor 4, the vibration sensor 5, and the noise detector 6 are installed inside the transformer, and the magnetic field sensor 3, The temperature sensor 4 , the vibration sensor 5 , the noise detector 6 , the current sensor 2 and the voltage sensor 1 are respectively connected to the industrial computer 7 , and the industrial computer 7 is connected to the workstation through the 5G module 8 .

Among them, the current sensor 2 includes a first current sensor and a second current sensor. There are six first current sensors, which are respectively used to detect the primary current of the transformer and the secondary current of the transformer, and one second current sensor is used to detect the current in the transformer. There are three voltage sensors 1, which are used to detect the primary voltage of the transformer bus.

The working principle and working process of the present invention are as follows:

As shown in Figures 1 to 4, voltage sensor 1 and current sensor 2 are used to collect the voltage and current on the bus connected to the transformer, respectively. Voltage sensor 1, current sensor 2, magnetic field sensor 3, temperature sensor 4, and vibration sensor 5 , The noise detector 6 transmits the detected information to the industrial computer 7, performs signal amplification, filtering, AD conversion, data preprocessing and other operations on the collected data information in the industrial computer 7, and controls 5G through 5G technology. Module 8 transmits the data to the workstation, where the digital twin technology is used to establish a simulation transformer model in the 3D MAX software according to the transformer parameters, and the preprocessed data information is automatically imported into the simulation transformer model to form a digital twin transformer, and the digital twin transformer receives The obtained data information is consistent with the real characteristic data of the transformer, and then the core magnetic field change information, primary current information, secondary current information, temperature change information, vibration information and generated noise information are fused together through multi-sensor data fusion technology. The bias information of the transformer is judged and displayed on the digital twin transformer. By establishing the characteristic curve between the neutral point current and the magnetic field strength of the transformer core in the workstation, and superimposing the characteristic curve between the neutral point current and the core magnetic field strength on the digital twin transformer, the information superimposed on the digital twin transformer can See the effect of neutral point current on transformer bias.

Finally, it should be noted that the above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, the The technical solutions described in the foregoing embodiments may be modified, or some technical features thereof may be equivalently replaced. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims (9)

1. a transformer bias detection method based on digital twin is characterized in that: In step 1, the change information of the iron core magnetic field, the neutral point current information, the primary voltage information, the primary current information, the secondary current information, the temperature change information, the vibration information and the generated noise information of the transformer are detected by the sensor; Step 2, transmit the detected data information to the industrial computer, perform signal amplification, filtering, AD conversion and data preprocessing on the collected data information, and then remotely transmit the preprocessed data information to the workstation; Step 3, the workstation establishes a simulation transformer model according to the transformer parameters through the digital twin technology, and then imports the preprocessed data information into the simulation transformer model to form a digital twin transformer; Step 4: Through multi-sensor data fusion technology, the core magnetic field change information, primary side current information, secondary side current information, temperature change information, vibration information and generated noise information are fused together to judge the bias information of the transformer, and superimposed on the transformer bias information. The digital twin transformer is displayed on the workstation; Step 5: Establish the characteristic curve between the neutral point current and the magnetic field strength of the transformer core in the workstation, and superimpose the characteristic curve between the neutral point current and the core magnetic field strength on the digital twin transformer. information to view the effect of neutral point current on transformer bias. 2. a kind of transformer bias detection method based on digital twin according to claim 1, is characterized in that: described sensor comprises voltage sensor, current sensor, magnetic field sensor, temperature sensor, vibration sensor and noise detector, current The sensor includes a first current sensor and a second current sensor, wherein the first current sensor is used to detect the primary current information and secondary current information of the transformer, and the second current sensor is used to detect the neutral point current information of the transformer; In the process of data fusion, first set up a data fusion center, and set up an intermediate station between the first current sensor, magnetic field sensor, temperature sensor, vibration sensor and noise detector and the data fusion center. The data information collected by the sensor, magnetic field sensor, temperature sensor, vibration sensor and noise detector are correlated, and then data-level fusion is performed to realize local fusion, and the data information after local fusion is extracted in the data fusion center. The feature extraction result judges the data attributes, and then associates the judged data information for decision-level fusion, thereby realizing global fusion, and outputting the fusion result to judge the magnetic bias of the transformer. 3. a kind of transformer bias detection method based on digital twin according to claim 2, it is characterized in that: in step 5, after the second current sensor detects the current of the neutral point, by the data collected with the magnetic field sensor The information is compared to judge the influence of stray current on transformer bias. 4. a kind of transformer bias magnetism detection method based on digital twin according to claim 3, is characterized in that: in judging the influence process of neutral point current on transformer bias magnetism, detect iron core magnetic field change information by magnetic field sensor , transmit the neutral point current information and the core magnetic field change information detected by the second current sensor and the magnetic field sensor to the workstation together, establish the characteristic curve of the neutral point current and the magnetic field strength of the transformer core in the workstation, and transfer the neutral point current The characteristic curve between the current and the magnetic field strength of the core is superimposed on the digital twin transformer. When the stray current enters the neutral point, the information superimposed on the digital twin transformer can be used to check the effect of the neutral point stray current on the transformer bias. Magnetic influence. 5. a kind of transformer bias magnetism detection method based on digital twin according to claim 2, is characterized in that: when step 3 constructs digital twin transformer, first builds the model consistent with transformer, and then renders and has the same parameter as transformer. Virtual transformer, and then establish a virtual industrial computer, the data information collected by the voltage sensor, current sensor, magnetic field sensor, temperature sensor, vibration sensor, noise detector is aggregated through the physical industrial computer and transmitted to the virtual industrial computer to establish a virtual industrial control. Based on the data interface between the computer and the virtual transformer, data analysis and scene rendering are performed in the virtual transformer according to the collected data information, and finally a digital twin transformer is generated, so that the data information received by the digital twin transformer is consistent with the real characteristic data of the transformer. 6. a kind of transformer bias detection method based on digital twin according to claim 2 is characterized in that: the data information collected by the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, noise detector is at the data level In the fusion process, the collected data information is locally fused by using the fuzzy logic control method, and X ₁ , X ₂ , X ₃ , X ₄ , and X ₅ are used as the first current sensor, magnetic field sensor, temperature sensor, and vibration sensor respectively. Fuzzy set of sensors and noise detectors, where X ₁ includes the primary current information and secondary current information collected by the first current sensor at each moment, and the primary current information and secondary current information collected at the kth moment are expressed as x ₁ (k); X ₂ includes the core magnetic field change information collected by the magnetic field sensor at each moment, wherein the iron core magnetic field change information collected at the kth moment is represented as x ₂ (k); X ₃ includes the temperature sensor collected at each moment. The temperature change information of , wherein the temperature change information collected at the kth moment is represented as x ₃ (k); X ₄ includes the vibration information collected at each moment of the vibration sensor, and the vibration information collected at the kth moment is represented as x ₄ ( k); X ₅ includes the generated noise information collected by the noise detector at each moment, wherein the generated noise information collected at the kth moment is represented as x ₅ (k); X ₁ , X ₂ , X ₃ , The similarity between X ₄ and X ₅ of the identified bias situation, at the kth moment, the data matrix between each sensor is:

The elements in matrix A are the expressions of proximity between each sensor, namely:

α _mn (k) represents the closeness expression between any two sensors in the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector, and x _m (k) and x _n (k) represent respectively Data information collected by the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector at time k. When m=1, n=2, x ₁ (k) represents the first current sensor at time k. The collected data information, x ₂ (k) represents the data information collected by the magnetic field sensor at time k, when α _mn (k) is less than the preset value M, it is considered that x _m (k) and x _n (k) are not similar, then it can be concluded that α _mn (k) = 0, that is:

The calculation formula for the consistency measurement of the sampling values of the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector is:

where α _ij (k) is the data information of the ith row and jth column of the sensor in the data matrix at time k, and α _ij (k) is x ₁ (k), x ₂ (k), x ₃ (k), x ₄ (k) or x ₅ (k); The calculation formula of data fusion is expressed as:

where x ₁ (k) is the data information collected by the first current sensor at time k, x ₂ (k) is the data information collected by the magnetic field sensor at time k, and x ₃ (k) is the temperature sensor at time k The collected data information, x ₄ (k) represents the data information collected by the vibration sensor at time k, x ₅ (k) represents the data information collected by the noise detector at time k, ω _i (k) represents the data in the matrix The weight of the i-th row of , according to the principle of information sharing, the sum of the information volume estimated by the optimal fusion can be decomposed into the sum of the information volume of several measurement data, that is, one information can be shared by several subsystems, there are

Normalize the fuzzy closeness of the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector to obtain their respective weights

In the formula, c ₁ (k), c ₂ (k), c ₃ (k), c ₄ (k), and c ₅ (k) represent a set of non-negative numbers. 7. A kind of transformer bias detection method based on digital twin according to claim 6, it is characterized in that: in the data fusion center to carry out the decision-level fusion process, the transformer bias condition is divided into normal F ₁ , mild F _2. There are four levels of severe F ₃ and severe F _4. Let Θ be an identification frame, and Ω(Θ) represents the set of all subsets in Θ, then the transformer bias level is expressed as: Ω(Θ)={φ,{ F ₁ }, {F ₂ }, {F ₃ }, {F ₄ }, {F ₁ ,F ₂ },{F ₂ ,F ₃ },...}, φ is the empty set, suppose Q={F ₁ , F ₂ }∈Θ, it means that Q is either normal F ₁ or mild F ₂ , that is, Ω(Θ) is either state; After determining the recognition frame Θ, define a set function G that maps Ω(Θ) to a number on the interval [0,1], expressed as:

In the formula, G(Q) is called the basic probability assignment function, which represents the confidence of the state recognition model for a specific state, and the confidence function Bel is used to represent the mapping Bel on the recognition framework Θ: Q→[0, 1], which satisfies:

Then use the plausibility function Pl to represent the next mapping Pl:Ω(Θ)→[0,1], and satisfy:

where Q ^c represents the complement of Q, Pl(Q) represents the plausibility of proposition Q, where Pl(Q)=0 means that the evidence rejects Q, and Pl(Q)=1 means that the evidence supports Q; Let G _i represent the basic probability assignment function of the ith group of evidences defined on the recognition framework Θ, where i=1, 2...n, the basic probability assignment function after combining any two groups of evidences is expressed as :

K represents the normalized coefficient, which is expressed as:

G ₁ (Q _i ) is the basic probability assignment function of the first group of evidence Qi _, G ₂ ((Ω(Θ)) _j ) is the basic probability assignment function of the second group of evidence (Ω(Θ)) _j , G( Q) reflects the degree of conflict between the two evidences. When K≠0, the combination between the two sets of evidences is expressed in the form of an orthogonal sum; when K=0, it means that there is no orthogonal sum, and the a contradictory relationship; After that, the basic probability assignment functions of the first current sensor, magnetic field sensor, temperature sensor, vibration sensor, and noise detector are constructed, and the identification results of the first current sensor, magnetic field sensor, temperature sensor, vibration sensor and noise detector are used as five Evidence E ₁ , E ₂ , E ₃ , E ₄ , E ₅ , and four levels of normal F ₁ , mild F ₂ , severe F ₃ , and severe F ₄ are used as the identification domain, and Θ={F ₁ ,F ₂ , F ₃ , F ₄ } constitute the common identification frame of E ₁ , E ₂ , E ₃ , E ₄ , and E ₅ , and construct the basic probability assignment function of each evidence on the identification frame Θ. According to the formula

Then the basic probability assignment function of E ₁ , E ₂ , E ₃ , E ₄ , and E ₅ can be calculated. According to the basic probability assignment function after fusion, it can be judged which state the transformer bias is in. 8. A digital twin-based transformer bias detection system, characterized in that it includes a transformer, a voltage sensor, a current sensor, a magnetic field sensor, a temperature sensor, a vibration sensor, a noise detector, an industrial computer, a 5G module and a workstation, and A voltage sensor and a current sensor are installed on the outer surface of the transformer. The voltage sensor and current sensor are respectively connected to the busbar of the transformer. The industrial computer is installed on the outer surface of the transformer. The 5G module is installed on the industrial computer. Magnetic field sensors and temperature sensors are installed beside the transformer core. , Vibration sensor, noise detector are installed inside the transformer, magnetic field sensor, temperature sensor, vibration sensor, noise detector, current sensor and voltage sensor are respectively connected to the industrial computer, and the industrial computer is connected to the workstation through the 5G module. 9. A digital twin-based transformer bias detection system according to claim 8, wherein the current sensor comprises a first current sensor and a second current sensor, and the first current sensors are six, which are respectively used to detect Transformer primary side current and transformer secondary side current, one second current sensor is used to detect the current at the neutral point of the transformer, and three voltage sensors are used to detect the primary side voltage of the transformer bus.

Images